Social Security benefits are becoming less adequate for today's and tomorrow's retirees, due to changes already enacted into law, changes in family work patterns and the likelihood of future financial problems. Some of the forces affecting Social Security have already happened; others are more speculative. But the trend is clear.
Measures of Adequacy
A common measure of retirement–income adequacy is the replacement rate – that is, retirement income relative to pre–retirement earnings (sometimes adjusted in some way to reflect inflation). The philosophy underlying the use of replacement rates is that retired people should be able to maintain their pre–retirement standard of living (although the relationship to adequacy per se may be legitimately questioned). Retirees need to receive a determinable percentage of pre–retirement earnings in order to accomplish this goal. The lifestyle–maintaining replacement rates, which vary by income level for a variety of well–known reasons such as varying marginal tax rates and declining work–related expenses, are not too difficult to compute, and many analysts have done this. The generally accepted figure is around 70 percent for middle–income people. Social Security provides only about half of this target figure.
While we can determine fairly easily the replacement rates of yesterday's and today's retirees, determining the corresponding figures for tomorrow's retirees depends to some extent on the projection model used. Of course, computer models can be only as good as the assumptions on which they are based (if they are even that good!). In this situation, the assumptions should be regarded with considerable skepticism, simply because they necessarily extend several decades into the future.
Because demographic experience is not very volatile, demographic assumptions may be forecast with considerable confidence. Replacement rates, however, do not depend very much on demographics, except for the fairly predictable life expectancy – and even that can be ignored when examining defined–benefit plans like Social Security (meaning Old–Age, Survivors and Disability Insurance, or OASDI). Economic assumptions are an entirely different matter.
Analyses of short–range economic forecasts by even the most highly respected economists demonstrate the great uncertainty that exists in this field. Economists (and everybody else, for that matter) have demonstrated limited ability to forecast many years ahead. Even so, some well–known economic models attempt to forecast the situation 30 or 40 years into the future. Actuaries at the Social Security Administration (SSA) top them all, attempting the perhaps impossible task of projecting economic and demographic variables for the next 75 years, as required by law! And in 2003, Social Security's trustees decided to add infinite–time–horizon projections to their annual report on the program's financial condition. No comment on the reliability of such assumptions seems necessary! (If you don't have anything good to say...)
Very long–range projections can be put into perspective in the following way: Consider an economist (or actuary) sitting in his office in 1975 and trying to predict what the United States economy would be like 35 years hence, in 2010. How close to reality would he (or she) have been? Then make the enormous leap to an economist in the year 1935, when the Social Security program was enacted into law during the Great Depression, trying to forecast 75 years down the road to the present time. Obviously, none of these people – no matter how brilliant – would have had any chance of coming even close to projecting what actually occurred. Because today's economists are probably not much better at predicting the far–distant future than their predecessors were, since there are a lot of elements that are unforeseen we need to take all of these projections simply as that, projections. The use of computers really should not give us unwarranted confidence in the results. No systems can predict the future only the likelihood of what can happen.
Social Security's Contribution to Retirement Income
The magnitudes of the most likely future sources of retirement income are quite difficult to project. We have no way of accurately estimating future investment income, for example, because we cannot project with any real confidence how much people will save during their working years or what future earnings rates will be on invested assets. One item that seems to be an exception to this general rule, however, is Social Security.
Social Security has been regarded for 75 years now as the "floor of protection" for working Americans. It is the foundation on which other retirement programs are built. Thus, Social Security's contribution to the typical retiree's replacement rate is a matter of considerable interest. That contribution increased for the first several decades of the program's existence but has been declining slightly since about 1979. It is expected to decline much more in the future, for three reasons:
- Changes already enacted into law will reduce the adequacy of Social Security benefits, especially for workers born after 1959.
- Changes in family work patterns will reduce or eliminate certain "free" spousal benefits that have historically been provided to married couples.
- Future financial problems are likely to require additional reductions in Social Security benefits.
The present–law OASDI benefit formula produces absolutely stable and predictable replacement rates under any economic conditions. Very briefly, a worker with average earnings in every year (about $41,000 in 2008) who retires at his or her normal retirement age (NRA) will receive benefits at retirement that are about 41 percent of earnings in the last year before retirement. Similarly, an otherwise identical worker with maximum OASDI–covered earnings ($106,800 in 2010) will, in the long run, receive benefits at retirement that are about 25 percent of the last year's earnings.
These replacement rates, and the continuum of replacement rates for workers at other earnings levels, were essentially locked into place by the Social Security Amendments of 1977 (Public Law 95–216). Absent changes in the law, Social Security's contribution to future retirement income can be predicted with greater confidence than any other source. Still, Social Security has some surprises in store for the unwary retiree.
Changes Already Enacted Into Law
The stable replacement rates cited above are for workers retiring at NRA. The NRA was age 65 when Social Security began, and it remains 65 for workers born before 1938 (who first became eligible in 2000 for Social Security's early–retirement benefits). For workers born in 1938 and later, however, the NRA rises, under provisions enacted into law as part of the Social Security Amendments of 1983 (Public Law 98–21). Eventually, for workers born after 1959, the NRA reaches 67. Thus, to get the same replacement rate from Social Security that a worker got through 2003 at age 65, retirees in 2027 and later will need to wait 2 years, until age 67.
The fact that the increase in NRA is a benefit reduction, which may be obvious enough, becomes much more obvious when considered in light of the overwhelming percentage of beneficiaries who claim early–retirement benefits from Social Security. Currently, about three–fourths of the non–disabled eligible population claim Social Security retirement benefits at age 62 or shortly thereafter. At exact age 62, the early–retirement reduction required by law is 20 percent for those with NRA of 65; therefore, these early retirees receive 80 percent (or slightly more) of the benefit that they could receive if they waited until age 65 (ignoring the usually small effects of additional earnings). Their replacement rates are therefore lower. For example, the hypothetical average earner described above could receive 41 percent of his or her last year's earnings from Social Security at NRA; at age 62, the replacement rate would be only 33 percent.
Under current law, early–retirement benefits continue to be available at age 62, but because the number of years of early retirement will increase (from 3 to 5), the early–retirement reduction factor also increases. Starting in 2022, workers who retire at exact age 62 will receive just 70 percent of the benefit that would be payable at NRA, instead of 80 percent today. This represents a 12½ percent relative reduction in benefit amount. The replacement rate for our hypothetical average earner who retires at age 62 would be just 29 percent in 2022, instead of 33 percent today. Clearly, unless retirement behavior changes, Social Security will contribute less toward the adequacy of retirement income in the future than it does today.
The question of whether retirement behavior will change is worth investigating. Everyone knows that Americans have been retiring earlier and earlier for many decades, although the decline stopped in the mid–'90s. Of course, Americans have been living longer, too, and we might anticipate that some of this extra longevity will be reflected in longer working lives, rather than going entirely to more leisure time. On the other hand, we could speculate that greater affluence will allow the trend toward earlier retirement to continue. Perhaps most Americans do not like working so long and will retire as soon as they can afford to do so. This issue raises questions that we simply cannot answer, and it serves to emphasize the uncertainty of all long–range projections.
The fairness of Social Security's early–retirement reductions is worth considering, too. Most people understand that early retirees choose to get benefits for a longer time – their age at death presumably doesn't change – and therefore should get less each month. The actuarially fair reduction depends on the assumed mortality rates. Higher mortality should translate into larger actuarial reductions for early retirement; lower mortality means smaller reductions. Because the Social Security law is totally gender–neutral today, one set of reduction factors is applied to both men and women, even though men have significantly higher mortality experience (and shorter life expectancies). Thus, the reductions factors tend to be too low for men and too high for women, but the "errors" are small. As mortality continues to decline, the reduction factors will eventually become too large for both men and women, but that won't happen for many decades.
So the early–retirement reduction factors are roughly fair, but only for people who have no "special" knowledge of their own mortality. What does that mean? Well, consider a 62–year–old worker who is thinking about retiring and claiming Social Security immediately but might work until NRA instead. If he (or she) has average mortality, then the decision is actuarially neutral in that regard – although maybe not with respect to other factors, of course. However, if this hypothetical worker goes to the doctor and finds that he is terminally ill, that changes everything. With this new information, waiting until NRA to claim Social Security seems foolish, because the worker expects to be dead by then. If he wants any Social Security benefits, he must claim them immediately. The actuarial reduction is the least of his worries.
But he needs to consider something else. If he is married, his spouse's benefit as a widow will be limited to the benefit that he received as a retired worker. If he never claimed benefits before his death, then she can get an amount equal to his unreduced benefit (although possibly reduced based on her age). If he claimed early–retirement benefits, her widow's benefit would be limited to what he had been receiving (plus cost–of–living adjustments, of course). In other words, his reduction could affect her for many years, costing her much more than the few months of benefits he received. The decision to claim early–retirement benefits from Social Security is not so easy!
Even that's not the end of the analysis. If a beneficiary is working, the amount of his or her benefit can be reduced on account of earnings from employment or self–employment, even in employment that is not covered by Social Security. This "retirement earnings test" is designed to ensure that early–retirement benefits are not paid to workers who have not substantially retired. Benefits are not reduced on account of pensions, interest, dividends, capital gains, etc., because these are not attributable to current work. For beneficiaries who have not yet reached NRA in 2010, the benefit reduction is $1 for every $2 by which annual earnings exceed $14,160. A special monthly test applies only in the first calendar year of benefit receipt. For beneficiaries who will reach NRA in 2010 but have not yet reached it, the reduction is $1 for every $3 by which earnings in the months preceding attainment of NRA exceed $37,680. Beneficiaries can earn any amount without affecting their benefits starting with the month in which they reach NRA.
Workers who claim Social Security retirement benefits before they have fully retired have to consider whether those benefits will actually be paid, taking into consideration their anticipated level of earnings. At the same time, however, Social Security benefits are automatically recomputed at NRA to "give back" actuarial reductions for months in which no benefits were received on account of the retirement earnings test. In other words, a worker who claimed benefits 36 months before his or her NRA would have a 20–percent benefit reduction imposed initially, but if he or she had benefits withheld for 18 of those months due to excess earnings, the benefit would be recomputed at NRA to reflect just a 10–percent reduction, which would continue for life. The benefit amount would also reflect any increase due to the additional earnings, of course.