The challenge ahead for Social Security is primarily a matter of rising costs.  With an outlay total of more than $500 billion in 2005, Social Security now equals 4.3 percent of the nation’s economy (GDP) and 11 percent of workers’ pay (taxable payroll).  Over the next 30 years, as today’s children move into the workforce and today’s workers begin to retire, Social Security will place a mounting burden on the budget and the economy.  By 2035, the program’s projected annual cost of $1.3 trillion will equal 6.3 percent of GDP and 17.4 percent of taxable payroll — an increase of roughly 50 percent by either measure.

The options lawmakers have for resolving Social Security’s financing problem break down into three fundamental choices: increase revenue, constrain outgo, or borrow. 

Borrowing is not a viable option

Borrowing our way out of the problem is not a viable option because the fiscal challenge facing the system is not a temporary phenomenon.  The retirement of the post World War II baby boomers, along with increasing life spans, will sharply expand Social Security’s costs and those of other age-related federal entitlements in the coming decades.  And because there is no time in the future when those costs are projected to decline, resorting to borrowing — even if politically expedient — would be economically unsustainable. Incurring ever-rising levels of debt to plug the gap would result in huge interest costs and consume the savings needed to spur economic growth.  Moreover, with mounting concern about chronic budget deficits, a “solution” that simply amounts to running up the national debt rather than making hard choices would signal to increasingly wary financial markets that Washington has no intention of doing what is necessary to get its fiscal house in order.

The real choices for resolving the problem thus require lawmakers to change current law by either raising future taxes or constraining future benefits. 

The case for constraining future benefits

Raising future taxes is certainly a substantive option, and one that is more fiscally responsible than unlimited borrowing, but it ignores or dismisses the magnitude of the looming demands that Social Security and other entitlements will place on the income of future workers.  Levying higher taxes to meet those costs could hinder an economy that will also have to cope with near stagnant workforce growth.  Yet even assuming that future workers will be able to afford higher taxes, there is another more fundamental reason why this should not be the first option for reform — it is generationally inequitable.  Ultimately, choosing to raise future taxes to meet current law costs is similar to borrowing in that it places a claim on the expected earnings of today’s children — in effect confiscating their economic progress.  If future generations want to sustain these higher costs it should be their choice, not the consequence of the current generation’s refusal to plan responsibly for a known problem.

It is worth noting in this regard that since 1960 the federal tax burden has remained about the same — averaging 18 percent of GDP over that time — despite the fact that as a nation we are more than 2.5 times wealthier now than we were in 1960.  This suggests a certain resistance among the American public to taxes much above that level for any extended period of time.  And yet financing current law benefit promises, for Social Security, Medicare and Medicaid would add about 8 percent of GDP to the federal tax burden by 2040 even under conservative assumptions.  Will the American public in the future accept a permanent level of taxation that is 40 to 50 percent higher than it has been over the past 40 years?

Maybe it will, but there is no guarantee.  Thus, aside from the dubious generational ethics of deciding today how our children should spend their money, relying on tax increases to fund current law benefit promises risks an intractable political dilemma for future lawmakers — choosing between unacceptable tax levels or abrupt benefit cuts. 

This leaves the third option: phasing in benefit constraints now that will gradually reduce costs.  Unlike tax increases, benefit constraints reduce the claim on the production of future workers. Moreover, if today’s assumptions about growing entitlement spending prove to be exaggerated, there would be a windfall for future politicians and taxpayers.  It would be far easier to unexpectedly raise benefits than to take them away, or to cut taxes rather than raise them.  In any event, it would allow future policy makers to have more of a say in setting their own fiscal priorities.

More importantly, benefits can be reduced from projected levels without producing a “cut” relative to the value of today’s benefits.  Social Security benefits are not, and never have been, fixed from one group of new recipients to the next.  As now scheduled by law, the benefits for successive groups of future retirees are projected to grow in value when gauged either by their purchasing power or the lengthening lifetimes over which they will be received.

In other words, Social Security’s future benefit levels are automatically scheduled now to rise not only nominally, but in real value.  They will be worth more because they will rise faster than inflation.  From a future recipient’s perspective, they will buy more goods and services than the benefits of similarly situated recipients today.  And this rise in the value of benefits contributes to the imbalance between the program’s future income and outgo almost as much as the increase in the number of its future recipients.

A recent analysis by the Congressional Budget Office (CBO) showed that because future benefits are projected to grow faster than inflation, their purchasing power for an age 65 retiree in 2035 could be 25 percent greater than for someone retiring at age 65 in 2003.  Moreover, because that future retiree is projected to live two years longer, his or her lifetime benefits would be 35 percent greater.  The analysis further illustrated that even if those future benefits were reduced by 10 percent, their purchasing power would still be 12 percent greater than for someone retiring at age 65 in 2003, and their lifetime value, 20 percent greater.

As CBO summed up the issue —

“…a proposed reduction in the Social Security program’s overall future costs does not necessarily mean a reduction in the value of benefits for future recipients relative to the value for people today.”

If the public understood this, it might be more willing to reduce future Social Security benefits beneath what is promised by current law.  Unfortunately many people assume that future recipients’ benefits will have the same value as they do for today’s recipients.  Thus, the downside to proposing benefit constraints is that it is easy to demagogue and requires everyone to accept the notion that there is no free lunch.  And whether spurred by demagoguery or not, the reluctance of politicians to suggest restraints on public benefits is inevitable.  One gets fewer votes by doing so.  However, if current efforts to restructure Social Security are to be effective, cost saving measures must be an important part of the solution.

Major sources of cost growth

Over the next 75 years, Social Security’s revenues are projected to hover in a narrow range around 13 percent of the nation’s taxable payrolls.  The program’s costs, on the other hand, are projected to grow rapidly from 11 percent of the nation’s payrolls today to 15.6 percent in 2025 and more gradually thereafter to more than 19 percent by 2080.

While the conventional view is that those rising costs will be due to the aging of the population, that view is incomplete.  A deliberate policy of paying ever-higher real benefits is also a significant factor.  Under the Social Security trustees’ assumptions, the purchasing power of the average earner’s benefits at retirement is expected to roughly double between now and 2075. When the CBO analyzed this issue in 2003, it estimated that approximately 55 percent of the rise in Social Security’s cost over the next 75 years would be due to an increase in the number of beneficiaries and improvements in life expectancy.  The remaining 45 percent would be due to a projected increase in the real value of the benefits (see Figure 1).

Significantly, CBO estimated that the cost of Social Security as a share of the economy would rise only temporarily if the value of future benefits were to remain the same through time, i.e., if the benefits simply kept up with inflation (see Figure 2).

Thus, from a policy perspective, if the aim of reform is to address Social Security’s financing problem at its source — rising costs — either adjusting the program for increasing longevity or constraining the growing value of its scheduled monthly benefits are the two most logical solutions.

Wage indexing and the cost of benefits

Under rules that have been in effect since 1979, Social Security benefits for each new group of recipients are adjusted to keep up with the growth of wages in the economy.  This is referred to as wage indexing.  Over the postwar years, wages have risen faster than prices by about 1 percent per year on average.  The difference is largely due to productivity gains, and in a sense, represents the worker’s share in the nation’s real economic growth.  Wage indexing is thus aimed at ensuring that the living standard for successive groups of retirees keeps pace with society’s overall living standard.  If an average wage earner retiring at age 65 today gets 42 percent of his or her pre-retirement earnings replaced by Social Security benefits, a wage adjusted system will give approximately the same “replacement rate” to all future retirees.

This is economic speak for saying that the role of the program as a source of retirement income will not contract over time as the nation gets wealthier.  It will keep rising at the same pace.

On the surface this may sound like a reasonable policy, but on closer examination it raises several important questions regarding the future.  For example, as the nation gets wealthier, must future retirees rely so heavily on Social Security as a source of retirement income?  Does the government have to maintain as significant a role 30 years from now as it does today… or as it did when it began paying Social Security benefits in 1940?  Shouldn’t other means of saving for retirement take a larger role?

It is also relevant to ask why, if the program has a long-range financing problem, should the real value of Social Security benefits be allowed to go up automatically?  Why should the purchasing power of future retirees’ benefits be increased when the tax burden on future workers has to rise sharply to achieve such generosity?  Moreover, if the trend toward increasing longevity continues, why should future generations receive not only higher valued benefits, but receive them for a longer period of retirement than today’s and past generations of recipients?

It is sometimes asserted that a core, and perhaps essential, element of Social Security is to keep the ratio of pre-retirement earnings to initial benefits constant through time.  However, constant replacement ratios have hardly been the experience of the program.  Over the program’s first five decades, those ratios varied greatly ranging from below 20 percent in the 1940s to a peak of 52 percent in 1981.  While keeping them fixed was the objective of amendments enacted in 1977, those ratios will decline over the next two decades with the phasing in of the rise in the age at which full-benefits are payable enacted in 1983.  They presumably would be fixed thereafter, but with the anticipated exhaustion of the Social Security trust funds in 2041, they would decline again as the program’s revenues would be insufficient to pay for the benefits now scheduled under the law (see Figure 3).

Thus, asserting that the program’s future value can somehow be gauged by retrospective comparisons of replacement rates is problematic.  It presupposes that some conventional level of replacement rates has existed through time.  In practice, none has existed.  Assessing the future worth of benefits based on replacement rate comparisons is more of a subjective judgment about what replacement rates ought to be than on any past notions of them set in the Social Security law.

Aging and the cost of benefits

There have been major improvements in longevity since Social Security’s inception 65 years ago.  Males born in 1940 could expect to live to age 70; females, to age 76.  People born today can expect to live 10 years longer.  The Social Security trustees project that by 2075 males will live 16 years longer, and females, 13 years longer.  All told, if the projections hold true, life spans over that 135-year period from 1940 to 2075 will have risen by roughly 20 percent.

With the baby boomers now approaching their retirement years and longevity continuing to rise, the trustees estimate that the population age 65 and older will increase from 37 million today (12 percent of the population) to 62 million in 2025 and to 94 million in 2075 (23 percent of the population).

The growing share of the population comprised of people age 65 and older and the significant improvements in longevity have led many actuaries, economists, and demographers to suggest that as people live longer they should be expected to work longer. And with the large economic pressures looming on future workers, whose numbers will not grow nearly as fast as the number of retirees, the proportion of one’s lifetime spent in retirement should not necessarily increase as life spans grow.

Source: 2005 Social Security trustees’ report

When Social Security began paying benefits in 1940, on average people reaching age 65 could expect to get benefits for close to 14 years.  In 2000, they could expect to get benefits for 18 years.  And for those reaching age 65 in 2050, the figure could be 21 years.  In other words, a man retiring in 2050 could expect to receive 18 percent more over his lifetime than someone retiring in 2000, and 54 percent more than someone retiring in 1940–simply because he had a longer life span.  For a woman, the increases would be 15 percent and 52 percent, respectively. And because of liberalizations enacted over the years, monthly benefit levels have risen in relative terms.  Thus, even these numbers are low.

Reform proposals

There are numerous ways to constrain the growth of future benefits, but the two basic categories that are directed at the main sources of automatic cost growth are:

(1) Adjusting the system for rising longevity and

(2) Changing the way benefits are calculated

These two approaches encompass a broad range of possible changes.  While the labels suggest different types of change, the manner in which the two approaches limit the growth of future benefits could be very similar.

1.  Adjusting the system for rising longevity 

Two key ages are pertinent to the receipt of Social Security retirement benefits:  the age when benefits can begin (or the age of eligibility), which is 62, and the age at which full benefits can be paid, sometimes referred as the full-benefit or “normal” retirement age.  For people born before 1938, the full-benefit age is 65.  As a result of legislation enacted in 1983, for those born in 1938, that age will be higher as it phases up to 67 for persons born in 1960 and later.  The higher full-benefit age phases up in two steps — to 66, by two months a year for people born between 1938 and 1943 — and to 67, by two months a year for people born between 1955 and 1960.  People who elect to get benefits before the full-benefit age must take reduced benefits to reflect the fact they will collect benefits for a longer period of time.

If the full-benefit age were raised again as it was in 1983, the reductions for retiring “early” would be larger.  For instance, if the full-benefit age were raised to 70, the reduction for taking benefits at age 62 would be around 45 percent (assuming it was based on the same actuarial principle of assuring there is no lifetime advantage to retiring early).  A number of proposals would actually increase the reductions (or tilt them) to create a financial incentive for people to remain in the workforce. 

For the same purpose, some proposals would raise the first age of eligibility to a later age than 62.  If the actuarial reductions remained the same, raising the initial age might not produce benefit savings for the program (as over a person’s lifetime the benefits would be approximately the same), but it might induce people to work longer and more taxes would potentially flow in.

Some proposals would raise the full-benefit age in the future (i.e., beyond age 67) to age 68, 70, or 72.  Others would attempt to set up an automatic provision, referred to as “longevity indexing,” that would have the full-benefit or initial age rise periodically as longevity increases (perhaps based on estimates of future improvement or through a “look back” measure assessing actual improvement).  Raising the eligibility age to a higher fixed target may balance the system for a while.  But without longevity indexing, the system will drift out of balance again.  There are two ways to index Social Security to longevity.  The minimum eligibility age for benefits could itself be indexed–that is, the early retirement age could be raised in tandem with average life expectancy.  Or else annual benefits could be reduced so as to offset the greater number of years that will be spent collecting those benefits.  This is the equivalent of indexing the so-called normal retirement age, the age at which full or unreduced benefits are payable. 

In total these changes could be designed to address some, and perhaps all, of the gap between Social Security’s future receipts and expenditures. The dimension of the potential savings achieved by different measures is reflected in recent estimates by CBO and the Social Security actuaries.  CBO estimated the impact of raising the full-benefit age to 70 by 2029.  The rise would be by two months a year starting next year (the rise to age 67 scheduled under current law would begin in 2006 rather than in 2017).  By their estimates, a large portion, but not all, of the gap between the program’s future income and outgo would be closed.  For instance, the deficit in the 75^th year of their valuation period would be reduced by 60 percent.

The Social Security actuaries made estimates of more gradual increases, pegging the rise in the full-benefit age to projected improvements in longevity.  One measure would raise that age to 68 over a 30-year period and then keep it there.  The other would raise it to 70 over a 78-year period.

Given that life expectancy is expected to grow by about four years between 2005 and 2075, the more rapid rise in the full-benefit age reflected in the proposal scored by CBO would take into account not only future longevity improvements, but a portion of the improvement that occurred in the past.  The two proposals scored by the Social Security actuaries would only follow a path of future improvements.

2.  Change the calculation of benefits

The other broad category of possible constraints involves changes in the way benefits are calculated.  Shifting to a process called “price indexing” from today’s wage indexing is one such option.  As CBO describes the concept–

“…the most straightforward method of reducing the growth in Social Security spending is to slow the rates at which initial benefits rise from one cohort to the next…  The benefits awarded to them would still rise in nominal terms but only enough to keep up with inflation. That approach would not alter the benefits of those already on the rolls prior to its implementation.”

Procedures under current law base the benefits on workers’ past earnings, which are expressed as an average level of earnings over their working lifetime (35 years’ worth is used for retirement purposes).  The earnings used are those on which workers and their employers paid Social Security taxes (up to the taxable maximum, now $90,000).  Much of those earnings are indexed to reflect average growth of wages in the economy, and this produces what is technically referred to as average indexed monthly earnings, or AIME.  Initial benefits for new recipients are then computed by multiplying a “progressive” three-part benefit formula against that average.  The progressive nature of the formula causes a higher proportion of pre-retirement earnings to be replaced for people with low average earnings than it does for those with higher earnings.

The following formula is used for workers who reach age 62 in 2005:

As with earnings used in calculating the AIME, the points between the three brackets of the formula–referred to as “bend” points (the $627 and $3,779 in the formula above)–are indexed annually to grow with average earnings in the economy.  This annual adjustment, coupled with the indexing of the recipient’s earnings history, keeps future workers from having greater shares of their AIMEs computed into benefits in the lower-yielding second and third brackets of the formula.  Because wages tend to rise faster than prices over time, the wage-indexing adjustments made in the benefit calculation cause the real value of initial benefits for successive groups of recipients to rise.

Price indexing would change the way benefits are calculated so that the real value of initial benefits would no longer rise.  It would link the growth in initial benefits to a price index, rather than to a wage index, and by so doing, it would ensure that the purchasing power of future benefits would stay the same from one group of new recipients to the next.

CBO estimated the impact of this option assuming implementation in 2009.  For illustrative purposes, CBO also assumed that currently promised benefits would somehow be paid in full beyond the date of trust fund insolvency even though Social Security’s tax revenues would only pay for about three-quarters of those benefits. 

In this illustration, each group of newly retired and disabled workers thereafter would receive benefits that were lower than what the current system promises.  The difference would increase over time–for each year’s new beneficiaries–with its size determined by how much real wages grew.  If the growth of real wages (i.e., the amount that average wages exceed inflation) remains about 1 percent per year, for example, the projected impact on future benefits would be quite large.

For example, workers becoming eligible for benefits in 2030 would receive nearly 20 percent less than they are promised by current rules, and workers becoming eligible in 2075 would receive about 50 percent less.  The value of the benefits would similarly drop as a share of pre-retirement earnings.  CBO estimates that by 2075 this option would reduce Social Security spending by about 40 percent from what it would be if benefits were paid as currently prescribed.  In so doing, it would result in substantial reductions in the deficits in the earlier years, and by 2075 it would bring down the costs such that a small surplus would exist then.

Another approach, illustrated by the Social Security actuaries, would take a more progressive approach by greatly mitigating the effects of a large formula change on low and moderately low-income workers.  Under this proposal the benefit rates in the second and third brackets of the benefit formula would be gradually reduced by one third over the next 30 years (instead of a 32 percent rate in the second bracket, it would phase down to 21 percent; and the rate in the third bracket would drop from 15 percent to 10 percent).  Wage indexing would continue but for recipients whose AIMEs fall above the first bracket, a gradual reduction in the bracket rates would offset its effects, producing lower long-term benefit levels for moderate and high-income workers.  This approach is sometimes called “progressive price indexing.” 

Much of the design of a “progressive” approach to constraining benefits has to do with how the benefits of the lower income segments of the population are to be protected.  “Progressive indexing” simply conveys one approach.  The protection for low-income workers can be afforded through means other than tilting the benefit formula in their favor.  The benefit formula could be the same for everyone as it is today but other features of the program could be changed.  In other words, price indexing could be the general rule, but higher benefits could be paid for long working low-wage earners through a long service bonus, benefit limits that now apply to families of young survivor and disabled workers could be raised, low-earnings years could be dropped from the benefit calculation, etc.  Some have also suggested that a portion of the cost saving from changes in how benefits are calculated could be redirected toward strengthening the safety net features of the program. There is no single or best approach to designing a progressive constraint on future benefits.  It is more the idea that whatever constraints are devised, low-income workers should be spared.

What’s a benefit cut?

Reform options that constrain the automatic growth of future benefits are often described as “benefit cuts.”  This description can be misleading — and politically caustic — because it implies that benefit checks for current recipients will be cut.  None of the options described above, however, would affect current beneficiaries. Moreover, even for future recipients, these options would not reduce benefits below the inflation-adjusted value of today’s benefits.

Finally, it is important to ask whether the supposed “cut” is from the level of benefits the current system promises or the level that it can afford to pay.  According to the 2005 Trustees Report, Social Security’s dedicated tax revenues will be enough to pay just 74 percent of promised benefits by the 2040s.

As CBO Director Douglas Holtz-Eakin has explained, “With benefits reduced annually to match available revenue (as they will be under current law when the trust funds run out), the program can be continued or sustained forever. Of course, many people may not consider a sudden cut in benefits of 20 percent to 30 percent to be desirable policy, but it is sustainable from a financing perspective. What is not sustainable is continuing to provide the present level of scheduled benefits (those based on the benefit formulas that exist today) given the present financing. Under current formulas, outlays for scheduled benefits are projected to exceed available revenues forever after about 2020. That gap cannot be sustained without continual–and substantial–injections of funds from the rest of the budget.”

It is certainly fair to criticize reform options on policy grounds. But it is fundamentally unfair to judge them against a standard that assumes the current system can deliver everything it promises. The relevant comparison is with what current law can deliver, not what it promises.

Opening a productive dialogue requires public recognition that benefits are scheduled to automatically grow in value

By keeping the full-benefit age at a specific level, now age 66, and after 2022, 67, while longevity improves, the Social Security program automatically builds increases into lifetime benefits for future recipients. It also builds in automatic increases in the purchasing power of benefits for each new group of recipients.

Automatically increasing the real value of benefits came relatively late in the program’s development.  During Social Security’s first 40 years, benefits were raised only as policymakers saw a need and perceived that the resources would be there.  Retaining or modifying the current practice involves judging the merits of making sizable automatic transfers to the aged and disabled relative to other budgetary priorities and, ultimately, other claims on the economy.  By absorbing a larger share of the federal budget, Social Security spending could crowd out other functions of government, require higher taxes on tomorrow’s workers, or require greater borrowing.  The alternative is to require future workers to be responsible for more of their retirement income. That is an unavoidable trade-off that policymakers have yet to confront.

In supplemental views, a number of members of the 1979 Social Security Advisory Council put it this way–

“At the levels of real income prevailing in the 1930s (or perhaps even the 1950s), it can well be argued that it was appropriate, indeed, highly desirable–perhaps even necessary for the preservation of our society — that government should, by law, have guaranteed to the aged and disabled and their dependents replacement income sufficient to avoid severe hardship, and to have required workers (and their employers) to finance this system with a kind of ‘forced saving‘ through payroll tax contributions.  But as real incomes continue to rise, it is not easy to justify the requirement that workers and their employers ‘save‘ through payroll contributions to finance ever higher replacement rates far above those needed to avoid severe hardship…”

Those members stated further that–

“…future Congresses will be better equipped than today’s Congress to determine the appropriate level and composition of benefits for future generations… Congress might elect to give more money to certain groups of beneficiaries than to others, or to provide protection against new risks that now are uncovered.  But precisely because we cannot forecast what form those desirable adjustments might take, we feel the commitment to large increases in benefits and taxes implied under current law will deprive subsequent Congresses, who will be better informed about future needs and preferences, of needed flexibility to tailor social security to the needs and tastes of generations to come.”

Explaining to the public that Social Security benefits are pre-programmed to grow in real terms over longer periods of time, and that this automatic growth is the cause of its future deficits, would help put the trade-offs in a more meaningful context.  Amending the “third rail of American politics” may be politically dangerous, but to maintain that there is a free lunch to be had is a generationally irresponsible prescription for political entrenchment and eventual economic disruption.  It solves nothing.  The initial acknowledgement that difficult changes have to be made to fix Social Security is an essential first step in opening a dialogue.

APPENDIX

A Note about Disability Benefits

A key issue with constraints on future Social Security benefits is what, if anything, should change in the disability part of the program (officially referred to as Disability Insurance).  Disability benefits, which can be awarded up to the full-benefit age, are not reduced, and the recipients receive those higher benefits until their disability ends or for the rest of their lives.  In one way, this helps to mitigate any hardship for people who need to retire early due to injury or disease.  The disability criteria for older workers are somewhat more lenient than for younger workers. Thus, if an older worker qualifies for disability benefits, the greater reductions for retiring early from raising the full-benefit age wouldn’t apply.  But there are people who become impaired earlier in life whose earnings have been constrained in their later working years due to limitations from their conditions.  They don’t achieve their full earnings potential but may not be impaired enough to qualify for disability.  For them, collecting early retirement benefits may not be a choice but a necessity.  Measures that would increase the special minimum benefit (a provision under Social Security that offers larger benefits than the regular formula to those with long period of low-wage work) might offset the reductions for this group.  Other measures might do so by further liberalizing the disability rules for older workers.

The difficulty posed by the disability issue is that by keeping full-benefits for those who do qualify for disability, there may be a huge incentive to file for disability rather than retirement.  Even today, a person who wins a disability award at age 62 gets a 33 percent higher benefit for life than a person of comparable earnings taking early retirement benefits.  When the age for full retirement benefits reaches age 67, the advantage of a disability award at age 62 will be 43 percent.  For this reason, a number of proposals to raise the full-benefit age would set the basic disability benefit at lower levels (e.g., reducing it by the same amount applicable to someone retiring at age 65), and offset that reduction for the younger disabled by allowing greater number of low years of earnings to be dropped in determining the average earnings on which their benefits are computed, or by raising the level of family benefits.  Similar accommodations could be made to accompany a benefit formula constraint such as moving to price indexing.