1994-96 Advisory Council Report

Supplemental Statement

Edith U. Fierst


Primarily because I believe Social Security should be a defined benefit plan which promises assured benefits instead of the uncertain benefits payable by a defined contribution plan, I support the Maintenance of Benefits (MB) plan. However, I do not support all of its particulars. Each of the first three sections of this paper discusses an issue on which I disagree with the MB plan. These are as follows:

1. Extending the Computation Period. I do not favor increasing the computation period from 35 to 38 years because it would have a disparate negative impact on women who spend a few years out of the labor force bringing up their children.

2. Raising Normal Retirement Age. I favor raising normal retirement age as life expectancy grows because it is the intuitive way of curbing the costs of longer life and, therefore, seems the least objectionable way of realizing the money necessary to close the Social Security deficit.

3. Increasing the Payroll Tax Rate is Unnecessary. I do not believe it will be necessary to increase the rate of the payroll tax, even 50 years from now, if for example normal retirement age is raised and a small tax is imposed on some fringe benefits.

The next two sections raise issues which I deem important but which were omitted from the MB plan. These sections urge:

1. Increasing Survivor Benefits. I favor increasing survivor benefits to 75 percent of the combined benefits of husband and wife if that is greater than 100 percent of the benefit of either. This would be helpful to survivors of two-earner couples who are currently disadvantaged in widowhood by the dual entitlement rule. It would also have the beneficial effect of mitigating poverty for many elderly widows who are the poorest part of the population. To provide partial payment for this increase, I would accept a decrease in spouse benefits from 50 percent to 33-1/3 percent of the higher earner's benefit. The rest of the cost must come from other sources within Social Security.

2. Reducing the Exemption from Tax of Fringe Benefits. I believe current law is unfair in allowing some employees to receive substantial untaxed benefits while others must pay both the payroll and income tax on their earned income before they can use it to purchase benefits, such as health insurance and pensions. As a consequence, many employees, especially low-paid employees, lack these basic benefits. Narrowing the present discrepancy in assessment of the payroll tax (I do not propose eliminating it) would make the system fairer and also help solve much of the expected Social Security deficit.

The remaining three sections supplement the views of the Maintain Benefits group. I believe they are generally consistent with the philosophy behind the MB plan.

Specifically these papers argue:

1. Financial Security in Old Age Should Not Be Dependent on The Participant Being a Knowledgeable Investor. Investors need sophisticated knowledge to invest successfully. But most investors lack the sophistication to make intelligent selection of equities or bonds; they will need professional help, not all of which is disinterested or available at reasonable prices. Some complex issues relevant to the decision where to invest, and the difficulty any layman faces in trying to understand how these issues affect the value of a particular company are outlined below. Examples are the competence of management, the company's competitive situation, the impact of new technologies on its prospects, any unfunded promises it may have made to pay deferred compensation to highly-paid employees, and the complexities of its financial statement, such as whether it uses FIFO (first-in, first out) or LIFO (last in, first out) in its accounting.

2. Investment Earnings under Social Security Should Not Be Exempt from Income Tax. The proposal made by the PSA plan (and as an alternative offered by the IA plan) to exempt earnings from Social Security investments from the income tax would be costly to the Government and unfair to workers whose earnings are low.

3. Money's Worth and Replacement Rates Should Not Be Treated as Major Factors. Far less important than money's worth and replacement rates is a good balance between adequacy and equity. Adequacy requires not only that workers with full careers have decent old age incomes, but also that workers who become disabled be adequately cared for, that sufficient benefits be available for the dependent young survivors of those workers who die, and that reasonable benefits be available to women who spend some years (or a full career) at home caring for their families instead of in the labor force. None of these groups is adequately cared for under the PSA and the IA plans, as explained more fully below.

There are also technical reasons, outlined in the paper, why the money's worth and replacement rate analyses are misleading.



Under one alternative of the MB plan, the computation period would be extended from the current 35 years to 38 years. I oppose this enlargement because it would have a disproportionately negative effect on women, including women of the future, who are projected to work longer years than those of the past or present.

Much of the continuing differential between men and women in years of covered employment is due to years spent at home with young children.

The figures provided by Social Security's actuaries show that of the women expected to retire in 1999, only 15.4 percent -- compared to 56.7 percent of the men -- will have had 38 years of covered employment. This is a differential of more than 40 percentage points. Lengthening the computation period to 38 years will cause women who retire in 1999 to lose an average of about 3 percent of their earned benefits, compared to about 2 percent for men. 

An additional 14 percent of women and one percent of men retiring in 2020 are expected to have worked 38 years, thus narrowing the differential between them. At that time 28.9 percent of women and 57.7 percent of men are expected to have completed 38 years of covered employment, shrinking the relative difference to about 30 percentage points. However, despite the much greater numbers of women with careers of 38 years and the small increase in such careers for men, both men and women are expected to incur further decreases in benefits, with women suffering greater losses. Extending the computation period is projected to reduce women's benefits by an average of 3.9 percent and men's benefits by three percent. 2/ Some women will lose as much as 6 percent of their benefits.

The figures cited above for the working lives of men and women are based on expectations under current law. If the computation period were increased, both men and women might work longer periods. But it would be unfortunate if mothers, pressured by fear of poverty in old age, gave up opportunities to stay at home with young children. Like many other Americans, I believe the option of child care provided by the mother, especially in early childhood years, is a choice families should be able to make.

In an ideal world, the father of her children would take care of the mother's retirement needs when the couple jointly decide that she should provide home care for their children, but in reality her decision to stay home is not necessarily a joint decision. The father may be absent, and even when he is there, he may have disagreed with her decision or given only grudging consent. And there are many other possibilities reflecting our highly diverse population. Some women will have no entitlement to their mate's Social Security because they never married and/or divorced after less than ten years of marriage. Even those who were married for ten years may face inadequate Social Security spouse or survivor benefits -- perhaps because their husbands had low earnings or years of involuntary unemployment. This would become more serious if the computation period is extended. Many married women also have no right to share in their husband's other assets, if any, including pension benefits. Their rights to these benefits may disappear because of abandonment, divorce, separation, or death. Considering how often women have to rely on their own earnings, we must be careful not to recommend changes in Social Security that will reduce the benefits women earned themselves.

Finally, several Council members (of whom I am one) are recommending that spouse benefits be reduced as part of a plan to pay larger survivor benefits to two-earner couples (see below). Part of the rationale for that change is that more married women in the future will be earning their own benefits. But if their earned benefits are also reduced, there would no longer be a fail-safe against cuts in spouse benefits. Women would be hurt both as spouses and as workers.


At present half of all Americans start to draw Social Security benefits by age 62, an earlier age than previously, and most recipients of Social Security are living to much older ages than in the past. (In 1995, 65-year old men could expect to live to 80.3 years, and 65-year old women to 84.1 years.) This stretching of the years in retirement is putting Social Security in a financial bind. Raising the normal retirement age is the intuitive way of curbing the costs of longer life and, therefore, seems the least objectionable way of realizing the money necessary to close the Social Security deficit.

The loss to Social Security from workers retiring at age 62, rather than at normal retirement age (now 65, but scheduled to increase to 67 early in the next century), is primarily due to not receiving the payroll tax from retirees. It is not from the cost of paying early retirement benefits, since that is largely taken care of by actuarial reductions. Nevertheless the difference in dollars is considerable.

At the other end of life, no actuarial decrease offsets the cost of longer life. Benefits must be paid for longer periods and no additional taxes are received, thus creating one of the major reasons for the anticipated shortfall. To cure this deficit, it appears necessary to raise the normal retirement age to 68, and, thereafter, to index normal retirement age proportionate to further increases in life expectancy. If this were done, as life expectancy increases, normal retirement age would also rise sufficiently to maintain the relationship between working and retirement years. This increase should be added incrementally after 67 becomes the normal retirement age. If we do this, retirees will still enjoy longer periods in retirement than they did in the early years of Social Security, 3/ but at least the retirement phase of life will not continue to grow relative to working years.

Social Security's actuaries estimate that delaying retirement age to 68 would save 0.35 to 0.40 percent of payroll. This is more money than the MB plan is expecting to accrue from enlarging the computation period. The extra money should be used to pay for increased survivor annuities, as proposed below, and to avoid raising taxes in the middle of the next century. If there is any left, it should be used toward increased benefits for those who work longer than the minimum career.4/ This is different from reducing benefits for those who do not work longer years.

A question has been raised about the effect on widow(er)s of raising the retirement age. At present widow(er)s can start receiving benefits at age 60, or if disabled, at age 50, with reductions in benefits up to 28.5 percent. Were normal retirement age to be delayed, a decision would be needed whether the above ages should also be delayed for widow(er)s or kept the same. If they are kept the same, should there or should there not be further actuarial reductions for early retirement? My view is that the age for retirement of healthy widow(er)s should be raised so that it has the same relationship to early retirement with the same actuarial reductions as under current law. Very early retirement was originally added to Social Security as a bridge for homemakers on the assumption that they would be lifetime dependents. In the future, however, most women at age 60 will be employed or have had substantial employment. They can continue or resume working. They will not be solely reliant, as were prior generations of women, on the earnings of their husbands.

Of course, one can't help being concerned about those persons, widow(er)s, married, or single, who need to find jobs in middle age or later, whether because they have been widowed or lost their jobs. Nevertheless, Social Security cannot be expected to support all those who are without work before early retirement age.

The situation for disabled widows is different: they cannot work. Moreover, under the recency-of-work rule, those who have been employed but taken time out to care for their families will have lost their entitlement to a worker's disability benefit after 5 years away from employment. Accordingly, if disabled homemakers become widows, they may be in real trouble. The recency-of-work rule should be relaxed, or such women (and men) should continue to be able to collect Social Security at age 50, if not earlier, with no further reduction in benefits.


I do not favor increasing the rate of payroll tax even on a deferred basis (the MB plan would increase the payroll tax rate in 2045). I believe the necessary money to avoid making such increases can be found through a combination of delaying the normal retirement age, allowing the percentage of surplus funds invested in equities to grow to 50 percent, and taxing fringe benefits.

Investments in the private markets by the trust fund will generate income to the Social Security Trust Fund that can be expected to grow. If the proceeds were indefinitely reinvested, the percentage can be expected to grow over time beyond the 40 percent initial investment favored by the MB plan. Keeping the maximum invested in private markets at 40 percent will, therefore, eventually require the trust fund to stop reinvesting dividends or sell some of its private equities and bonds. Instead, I would favor retaining investments up to the level of 50 percent, and selling only when the portion of the trust fund in private equities and bonds exceeds that percent.

With respect to fringe benefits as a potential source of income to Social Security, please see the next section of my Supplementary Statement. As is explained therein, it would take very minor taxes to raise large sums of money. If necessary, they alone could produce the extra income necessary to avoid the payroll tax increase proposed by the MB plan.


The above papers describe issues on which I differ with the majority of those favoring the MB plan. The following sections describe issues on which the MB plan is silent. I think these two proposals are important and should be included:


The current law exempting fringe benefits from payroll taxes needs to be rethought and revised. I believe the exemption should be narrowed to reduce the Social Security deficit and to make the system fairer.


Some employee compensation paid by private employers 5/ is not subject to the payroll tax or counted in determining Social Security benefits because it is paid in the form of non-cash fringe benefits. The exempt compensation includes $263 billion for private group health insurance, $87.7 billion paid for private pension and profit sharing plans, and $7.2 billion for private group life insurance; 6/ together these make up 8.9 percent of employee compensation.

Some non-cash compensation would be exempt from Social Security even if paid in cash because it exceeds the taxable maximum. However, in 1994 (the last year for which figures are available), up to about 76 percent of fringes was earned by workers who earned less than the taxable maximum. 7/

Thus, up to about 6.8 percent of employee compensation paid by private employers (76 percent of 8.9 percent) is in the form of fringe benefits exempt from Social Security tax. This portion of employee compensation is not counted in computing Social Security benefits.

The exempt percentage of earned income gets larger each year as employers increase the proportion of compensation paid in fringe benefits. This trend is expected to continue. The Trustees estimated in their 1995 report that under the intermediate assumptions (Alternative II), future reductions in wages as a portion of employee compensation will approximate 0.2 percent per annum, 8/ with commensurate decreases in taxes received by the Trust Fund.

The consequences of the exemption from payroll tax for fringe benefits below the taxable maximum of 6.8 percent of compensation are so large that the actuaries estimate that were it to be eliminated, Social Security's deficit would be reduced from the now projected long range deficit (75 years) of 2.17 percent of payroll to 1.15 percent of payroll. This estimate of the potential for deficit reduction takes into consideration the additional benefits that would accrue to participants if fringes were part of the Social Security tax and benefit base. Obviously even a partial enlargement of the tax base would be a substantial help in resolving the Social Security deficit for years ahead. On the other hand, doing nothing means that the deficit will get worse in the future, as predicted by the Trustees and taken into account by the actuaries.

One reason the cost would not go up commensurate with the new tax receipts is the progressive benefit formula of Social Security which returns smaller increments to higher paid employees, including most of those receiving fringes. While the benefits of all those with earnings below the taxable maximum would be increased, their gains would not be large enough to erase most of the increased income to the trust fund from the additional tax base.


The current law exempting fringe benefits distorts the fairness of the system by giving recipients of fringes major advantages over those who do not receive them. Employees without fringes must pay taxes, both FICA and income, on their total compensation and then, if they want health insurance, pensions etc., purchase them with after-tax money. Unfortunately, many uncovered workers cannot afford to buy additional benefits with their remaining income, and those who can may be unable to get into health benefit plans comparable to those offered through employment. Some with poor health or prior disabilities cannot buy health insurance coverage at all; many millions of Americans are without any coverage.

This unfairness is compounded by the fact that fringe benefits tend to go to higher paid employees. By contrast, employees without fringe benefits work disproportionately at low wages. 9/

The distortion in coverage can be illustrated by the following facts: currently only about 40 percent of workers in private employment are enrolled in employer-sponsored pension plans, leaving approximately 60 percent of the employees of private companies without coverage. 10/ Approximately 70 percent of employees of private employers (including the self-employed) were covered directly or indirectly in 1994 by employer-provided health insurance plans, leaving approximately 30 percent without such employer-provided insurance. 11/ The proportion of employees with employer-funded life insurance is not known, but plainly such coverage is not universal. Employees of large companies are far more likely to be covered by fringes of any kind than are employees of small companies.

Another aspect of the present unfairness stems from the difference in the tax ultimately paid. There is no payroll tax on pensions at any time. (Some of the exempt income tax is assessed on pensions when received.) Recipients of employer-funded health insurance never pay a tax, payroll or income, on their health benefit coverage.

Why the Council didn't recommend changes

The Council was aware in its deliberations of these huge discrepancies in tax and benefits, but was unable to agree on what, if anything, to do about them. Some members opposed doing anything. These included members who were appointed from organized labor or business 12/, groups that now enjoy a tax break they do not want to lose. I, on the other hand, was appointed from the self-employed; they and unorganized workers 13/ are the

most disadvantaged by the unfairness of the current law. Self-employed workers may deduct contributions to pension plans from both the income and payroll tax, but may not make deductions from the payroll tax for other benefits. As to the income tax, until recently health benefits premiums paid by the self-employed were fully taxable; in recent years deductions have begun to be phased in gradually. In 1997, 40 percent of the cost will be deductible from income tax. By 2006, the self-employed will be able to deduct 80 percent of premium payments for health benefits from income tax.

There is nothing on the books, present or prospective, giving the self-employed deductions from income tax for other benefits, such as life insurance, education, adoption, day care, or other forms of benefits which employers often provide tax-free to their employees.

As for employees whose employers do not provide fringe benefits (most likely unorganized workers), they are permitted to deduct from the payroll and income tax their own contributions to pension plans, but may not deduct payments toward other types of benefits from either tax.

Issues To Be Reviewed

Some objections to placing a partial tax on fringe benefits are substantive, primarily concern that employers would reduce fringe benefits if they were subject to the payroll tax; this fear exists although economists are persuaded that the cost of fringes is shifted to employees. Others are technical problems. The following discussion of options recognizes that if a decision is made to make fringe benefits subject to the payroll tax, it will be necessary to consider the size of the tax, the choice of benefits (e.g., health insurance, pensions, cafeteria plans, or other) on which to impose it, and whether to tax when compensation is paid or when benefits are received.

1. Health insurance.

The potential for raising money by placing a FICA tax on health insurance is very great. Social Security's actuaries estimate that if the cost of private employer-provided group health and life insurance were taxed as though it were cash compensation, about 0.8 percent of payroll would be raised, after subtracting the cost of additional credit toward Social Security benefits. This sum would be enough by itself to solve over 35 percent of the shortfall predicted for Social Security. This may be more than is appropriate to obtain from one source. If it is decided to tax health benefits, therefore, it might be desirable to assess a lower tax rate, spreading the pain of reducing the deficit cost among recipients of many fringe benefits and other workers and retirees.

Another, perhaps simpler, approach would be to set a cap on deductible contributions to health benefit plans either at the maximum premium payable today or at a level below that. Now there is no limit on the amount that is exempt if used for contributions to health insurance. Such a cap would restrain the growth of tax-free benefits. Especially if the cost of health insurance keeps rising, it could generate significant revenue.

Another approach might be a 3 or 4 percent payroll tax on the employer's overall costs in providing health and group life insurance, whether paid through premiums or self-insurance. Only the employer would pay the tax, and no determination would be made of the cost per employee.14/ Nor would credit be given toward Social Security benefits of employees as a result of this tax.


2. Pensions.

An initial question is whether employers should be taxed differently on defined contribution and defined benefit plans. As a practical matter, employer contributions to defined contribution plans are made for specific individuals, and, therefore, are easily identified for tax purposes. This is not true of contributions to defined benefit plans. They are hard to attribute to an individual employee because they are based on many factors, including previous gains and losses from investments by the plan, the number of employees who are predicted to leave without vesting, and mortality among the employee population. But taxing the two types of plans differently might have an unwanted impact on employer choice of the type of plan to adopt.

One solution would be to impose the tax on benefits when received, as the income tax is now assessed, rather than on benefits when accrued.

Another issue is whether to tax contributions at the full rate or at a rate lower than the payroll tax. Social Security's actuaries estimate that a gain of 0.15 percent of payroll could be achieved by assessing a 3 percent FICA tax on private pension and profit-sharing benefits when paid. This FICA tax is assumed to be assessed only on benefits derived from contributions not previously taxed, and only to the extent such benefits do not exceed the taxable maximum. The taxed payments would be credited toward the earnings record of the recipient and serve to increase his or her subsequent Social Security benefits.

3. Flexible benefit plans.

Contributions to flexible benefit plans under section 125 of the Internal Revenue Code, also known as cafeteria plans and flexible spending accounts, are currently exempt from the payroll tax. Unlike some health insurance plans, they are not self-insured, a fact that would make it fairly simple to subject them to the payroll tax.

In 1994 the Inspector General of HHS urgently recommended that 2.9 percent, 15/ the portion of FICA allocated to Hospital Insurance under Medicare, be assessed on contributions to such plans. She cited Treasury Department estimates that losses to HI from the lack of such a tax would total $2.1 billion over the next 5 years. Comparable losses to Social Security would be about 3 to 4 times that figure since the portion of the payroll tax allocated to OASDI is 12.4 percent. 16/ The actuaries estimate this change would save Social Security about 0.04 percent of payroll.

4. Monies withdrawn before retirement age.

Another possibility would be to impose the payroll tax on monies withdrawn before retirement age. Some persons withdraw the contributions they have made to their pension plans before retirement. The law tries to discourage such withdrawals, and, thereby, to preserve an individual's assets for retirement, by assessing a penalty of 10 percent in addition to regular income tax on such withdrawals. This withdrawal penalty could be changed to include a FICA tax. For example, it would be feasible to:

a. Increase the penalty for early withdrawal of assets in pension plans from 10 percent to 15 percent, with the additional 5 percent going to FICA. This would discourage withdrawals from all types of pensions, whether or not totally paid for by employers.

b. Assess the full 15.3 percent payroll tax (including the portion for Medicare) on contributions placed in pension plans and withdrawn in less than a year. This would virtually eliminate the "scam" now being practiced by some employers to avoid paying the payroll tax. These employers put the equivalent of the FICA tax in private pensions to avoid the tax; employees think they have nothing to lose since they can withdraw the pension contributions and pay only a 10 percent tax rather than the full 15.3 (including the portion for Medicare), but, in fact, they lose credit toward their Social Security benefits if contributions are not made. In addition their current law tax is 10 percent, not 7.65 percent (half of 15.3 percent).

Other ideas may be developed and should be explored. The objective would be to come up with a plan that is fairer, increases revenue to a reasonable extent, and does not excessively deter employers from providing fringe benefits.


Survivor benefits must be increased to alleviate poverty among elderly widows, today's and those of the future, since neither group is likely to benefit from the increasing employment of married women. If higher survivor annuities were provided for women with earnings of their own, equity would also improve. Some money for this can be found by reducing spouse benefits. The rest will have to come from other sources, such as raising the retirement age, or taxing fringe benefits.

Poverty Among Elderly Widows and Widowers

While few elderly married couples are poor, as many as 40 percent of widows and widowers aged 65 and over have incomes below the poverty line (1991 figures). Both men and women in that age bracket are poor, but since women outlive their husbands by average of about six years 17/, most of the elderly poor are women. 18/Depend

Background Explanation

Despite the sharp increase in working women (nearly 60 percent of those 16 and over were in the labor force in 1994), the figures on poverty among widows and widowers have not changed much and are unlikely to do so in the future. (This fact is overlooked by the proponents of the PSA plan in their discussion of women's increasing labor force participation.) The reason is that women have lower earnings than their husbands, and the dual entitlement rule limits the possibility of their benefiting from both their own and their husbands' benefits. The result of the portion of the dual entitlement rule applicable to survivors is that:

A widow(er) gets the higher of the benefit to which she or he is entitled as a worker or as a survivor. The survivor benefit is equal to that earned by the deceased spouse.

Most women are entitled to survivor benefits greater than the benefits they earned themselves. Among women who were born in the 1930s or 1940s and are now or soon will be eligible to retire, only 10 to 15 percent are expected to have earned benefits greater than those earned by their husbands. Moreover, little change in this regard is expected for future generations. Only about 20 percent of women retiring in 2015 are expected to have earned benefits greater than those of their husbands.

There are two reasons: (1) women's earnings on average remain lower than those of men and (2) many women do not work a full career. Motherhood and other family responsibilities often keep women home for several years or more, and for those who can possibly afford it, this pattern of periodic absence from the work force for family reasons is likely to continue in the years ahead. Thus women are not likely to get increased benefits in widowhood as a result of their own increasing work patterns unless a change is made in Social Security.

Improvement in Couple's Benefits from Wife's Work

Retired wives, in contrast to survivors, will do better if they work more, as most are now doing. The dual entitlement rule provides that retired wives get the larger of the worker benefits they earned themselves or a "spouse benefit" equal to half the dollar amount of the benefits earned by their husbands. If a woman earns a worker benefit that is less than her spouse benefit, her worker benefit will be paid -- plus a spouse benefit large enough to bring the total to half her husband's benefit. If she earns more than her spouse benefit, as is happening more and more, she will get the benefit she earned.

Note that because Social Security benefits are computed on a progressive formula 19/ , it is not necessary for a wife to earn half as much as her husband in order for her worker benefits to exceed half of his. The consequence is that large numbers of women earn worker benefits larger than their spouse benefits.

The Social Security Administration projects that roughly 60 percent of wives who retire in 2015 will receive worker benefits and no spouse benefits. The remainder will either be dually entitled (roughly 35 percent) or receive spouse benefits only (roughly 5 percent).

Lack of Equity for Two Earner Couples.

From an equity standpoint, women who work and pay FICA taxes should reap greater Social Security benefits, or at least not fare worse, than those who do not. But as explained above, wives get no increased benefit from their own work while their husbands are alive unless they have earned more than half his benefit, and they rarely get an increase in widowhood. This is true even though 12.4 percent of their earnings up to the taxable maximum, now $62,700, like comparable earnings of everyone else, must be paid in FICA tax by the workers and their employers. Thus, couples with two workers are likely to have paid a higher tax for the same retirement and survivor benefits (excluding disability and child survivors) than couples with only one worker. In widowhood the survivor of a couple that had two workers may get little additional benefit for the taxes paid by the second earner.20/ In fact, unless a woman has earned more than her husband, the greater the share of the combined earnings of the couple earned by the wife, the lower her rate of return in widowhood and the less the widow(er) receives in actual dollars.

The following chart illustrates the results of different distributions of earnings on benefits. Here, each of the couples earned a total of $1,000 a month, one through the efforts of one earner, the others through different allocations of earnings between husband and wife. 21/




HUSBAND $1,000 $564  
WIFE 0 282 $ 564
COMBINED $1,000 $845  
HUSBAND $ 844 $ 510  
WIFE 166 255 $ 510
COMBINED $1,000 $ 680  
HUSBAND $ 667 $ 458  
WIFE 333 300 $ 458
COMBINED $1,000 $ 765  
HUSBAND $ 500 $404  
WIFE 500 404 $ 404
COMBINED $1,000 $808  

Dependence of Elderly Widows on Social Security

The poverty of most elderly widows is directly related to their Social Security benefit levels. Other assets a couple had at retirement are likely to have been spent by the time of the husband's death, leaving the widow primarily dependent upon her Social Security survivor benefit, supplemented by SSI if necessary to bring her income up to $470 a month (in 1996).

Private pensions are payable to only about 40 percent of retired Americans, and they are usually persons with higher incomes. The median pension earned by women is only half of the median earned by men. Most low-income persons have no private pension and are unlikely to be able to accumulate them in the future.22/ Workers with low earnings find their incomes stretched so far that saving for old age is very difficult. Similarly, despite the non-discrimination rules of ERISA (the Employee Retirement Income Security Act) which deny tax advantages to an employer who provides pensions disproportionately in favor of highly paid staff, many more employer-funded pensions go to workers with higher earnings.23/

Options for Change

In trying to determine how to help surviving spouses, especially those who are totally dependent on Social Security, the Advisory Council considered several potentially helpful options. One route that reflects women's increasing employment patterns is to reduce the spouse benefit, which worker benefits are replacing for many women, and increase the survivor benefit, which is usually unaffected by women's work. 24/

Reducing spouse benefits would have no effect on the sixty percent of married women who are expected in 2015 to have worker benefits larger than their spouse benefits. The thirty-five percent who are expected to have worker benefits plus residual spouse benefits would incur some reduction, depending upon the new level of the spouse benefit and the amount of the worker benefit of each. Only five percent are not expected to have earned worker benefits; their benefits would be decreased to the new level of spouse benefits.

Among those who receive spouse benefits and no worker benefits, other financial resources generally exist to prevent a decrease from having a devastating impact. As mentioned at the beginning of this paper, it is unusual for couples in retirement to be poor. Husband and wife share expenses and, in addition to the wife's spouse or worker benefits, they have the husband's benefits and whatever assets or pensions the couple has accumulated before retirement. The spending of savings is less likely to wipe out the couple's assets while both are alive than at widowhood which occurs later in life.

It is possible to reduce benefits payable to couples enough to pay for an increase in survivor benefits. Taking into consideration the age differences between the typical husband and wife and their respective life expectancies, Social Security estimates a decrease in benefits to couples of $1.00 can finance an increase in benefits to survivors of $1.45. Spouse benefits would have to be reduced to about 20 percent to finance survivor benefits at the level of 75 percent of the couple's combined benefits.

However, I believe this would be too steep a drop, especially because of its impact on divorced women. Those who were married ten years or more receive the same benefits as married women, but during the period of time, often extended, that their former husbands are alive, they do not share expenses with them and rarely have access to their assets. The result is poverty for nearly 30 percent of divorced or separated women whose husbands (or in the case of divorced women, former husbands) are alive today. 25/ More divorced, and probably more separated, women, are earning worker benefits, and will be able to weather some reduction in spouse benefits, but the extent of their improved worker earnings may not be enough. And their survivor annuities are unaffected.

Some policy analysts suggest that we pay more to divorced women living apart from their former husbands than to married women. Although it seems unlikely many women would seek a divorce merely to get better benefits, it is possible that paying higher benefits to the divorced than to the separated-but-still-married would provide an effective divorce incentive to some. Whether it would or not, paying more to the divorced than to the still-married is offensive to many people. 26/ Furthermore, a policy that preferred the divorced would raise such questions as discrimination against those who are separated but for religious or other reasons are not divorced.

In my opinion, a better solution is to treat all women entitled to spouse or survivor benefits the same, even if it raises costs somewhat. The proposal that makes the most sense in this context would be to reduce spouse benefits from half the husband's benefit to one-third, and raise survivor benefits to 75 percent of the combined benefits of husband and wife (including in such benefits both the worker's own benefit and any spouse benefit) if that is greater than 100 percent of the benefit payable to either. This would apply to both one and two-earner couples, but would have its beneficial impact on two-earner couples.27/

Under this proposal, survivor benefits would rise proportionate to the earnings record of the lower earner. As equity suggests, the higher her worker benefits, the higher the survivor annuity. If she earned as much as he did, the survivor annuity payable to the surviving spouse would be 150 percent of their combined worker benefits, assuring two-earner couples higher benefits than one-earner couples. Benefits payable to the latter would remain the same.

Moreover, this formula would bring the benefit for the survivor of a two-earner couple closer to what is needed relative to the official poverty line. (This assumes that a one-person household needs 80 percent of the income of a two-person household to achieve an equivalent standard of living.)

The following chart uses the same couples as Chart A and shows the impact of this proposal more specifically:



HUSBAND $1,000 $564      
WIFE 0 188 - $ 94    
COMBINED $1,000 752 658 $564 $ 0
HUSBAND $ 844 $ 510      
WIFE 166 170 - $ 85    
COMBINED $1,000 $ 680 $595 $510 $ 0
HUSBAND $ 667 $ 458      
WIFE 333 300 0    
COMBINED $1,000 $ 758 $758 $569 + $111
HUSBAND $ 500 $404      
WIFE 500 404 0    
COMBINED $1,000 $808 $808 $606 + $202

As Chart B clearly shows, the proposed change would increase both adequacy and equity for working wives and their husbands during their joint lives and in survivorship. While it is true that couples with wives who worked little or not at all would lose, there would not be many of them. The change would be at the relatively modest cost of 0.15 percent 28/of payroll. 29/I therefore urge its adoption.


The following papers supplement the views of those favoring the MB plan; I believe they are consistent with the philosophy behind the MB plan.


Investors need sophisticated knowledge to invest successfully, a sophistication millions of participants in Social Security lack. In order to do reasonably well over time -- nothing is guaranteed -- investors must be able to assess the value of the companies whose stocks and bonds are offered to them. But many, perhaps 100 million participants in Social Security, lack requisite knowledge, such as the market served by individual companies, ways of judging the competence of management, relevant changes in technology and whether an individual company is able to keep current, the competitive situation, both local and sometimes international, the company's unfunded promises to pay deferred compensation to highly paid employees, and many other factors.

Some relevant issues will be included in the accounting statement in the company's annual report, but most retirees do not know how to interpret this information even if provided to them by purveyors of securities. Without adequate knowledge, privatization will leave many who are required to substitute investments for assured Social Security benefits without the ability to protect themselves against potential disaster.

For example, most investors probably assume that the earnings of a stock are an important index to its value, but few of them understand how easily and legally apparent earnings can be manipulated to make them seem greater or smaller. If a manufacturing or retail business follows the accounting principle of first in, first out (FIFO) in valuing its inventory, its earnings will often seem relatively high because the cost of the items sold will have been set some time ago and will not reflect the price increases which have occurred in the interim.

On the other hand, if the company values its inventory on the accounting principle of last in, first out (LIFO), its costs will have been set close to the time of sale. Costs will, therefore, seem higher, giving the impression that the company's earnings are relatively low

No other single accounting practice has the impact on reported earnings of these inventory valuation choices. The information is in the footnotes of the accounting statement included in the corporation's report to the public, but few laymen have the skill to make the necessary analysis of even the basics, much less the exotic techniques sometimes used to bolster earnings in a down year.

Another example: If one company acquires another and pays for it in shares of the acquiring company's common stock, it will likely be able to record the assets of the acquired company at the depreciated costs on the books of the acquired company rather than at its cost to the acquiring company. Assuming the various conditions for a so-called pooling of interests treatment are satisfied, the company will not need to show the actual price paid, although it may have been far higher. The result is that the costs charged against earnings in the accounting statements of future years will be far more modest than if the company had paid the same price in cash or other securities, and its earnings will seem proportionately higher. Sophisticated investors will understand, at least for a while, that their managements have been able to ignore part of the price paid in their accounting statements, but for the rest of us the mystery of what happened to the difference between the price paid and the assets shown on the books will be impossible to comprehend.

Another example: only in recent years has the Financial Accounting Standards Board (FASB) begun to require employers to show their contractual liability for health insurance for retirees. Previously it was a rare company that carried this obligation on its books. The amounts of money for some companies (e.g., General Motors) when required to be reported turned out to be in the billions of dollars. I understand FASB is now looking at the way in which derivatives and affiliated companies are handled in financial reports, with potentially significant affect on reported income and values.

Another factor is the schedule on which a company depreciates its plant and equipment. Alternative schedules are available, and doubtless carefully studied by each company's management and accountants before a course of action is determined. The ordinary investor will not be able to understand these decisions in their full ramifications, but they may greatly affect the value of investments.

Since analysis of the effect of all these alternatives in the company's assets will be extremely difficult for most lay persons, investors who are aware of and worry about these and other complexities are likely to seek the help of professionals to guide them. The advice of investment advisers does not come free; charges are assessed in a number of ways. Some advisers charge a flat or sliding fee for giving advice. Stockbrokers usually charge a percentage of the cost of purchases as commissions. Other professionals may bundle a group of investments together into a mutual fund, and sometimes charge the purchaser a "load" for this service. Professionals who manage mutual funds and other investment vehicles charge management fees. And there are many other ways that fees can be computed.

The Advisory Council's consultant concluded that the annual charge would average about one percent of the value of the securities concerned. But this is an average; the maximum for any individual or transaction may be much higher. Moreover, some of the charges may be abusive, such as those for churning or where the investment adviser has a conflict of interest.

If instead of workers investing individually in the market, the trust fund makes the investments in indexed funds, the average combined gross earnings derived from the two approaches will be the same. There will be no difference attributable to an inherent likelihood that large numbers of individual investors will make wiser investments than the average of the market. For them to do so would be an oxymoron.

The trust fund will have higher net earnings, attributable to: (1) earnings of the trust fund being unreduced by the extra cost of paying commissions and for advice etc., whereas individual investors will incur these costs, and (2) the trust fund rather than individuals benefiting from the gains from good investments and bearing the risk of loss from poor ones. Any advantage for the PSA plan shown on the money's worth charts comes from both the greater sums available for investment as a result of the tax increase that is part of their plan and the presumption that individuals will invest half their assets in the stock market over a lifetime, while the MB plan at present calls for such investment of only 40 percent. It has been charted accordingly.


One major difference among the three plans is their proposed income tax consequences. Under the MB plan, beneficiaries would be exempt from tax on benefits up to the amount of their FICA contributions. This is the current practice with respect to other forms of defined benefit pension plans, both private and public, to which employees contribute after-tax money, and is proposed by proponents of all three plans with respect to benefits paid by the trust fund.

But the advocates of the PSA plan urge a different tax policy for contributions to, and income from, the personal savings plans they would establish. Contributions to these accounts would be taxed, as contributions to the trust fund are currently taxed, but earnings on investments in the beneficiary's personal savings would be exempt from income tax.

The IA plan treats the PSA approach to income tax as one of the acceptable options.

The technical argument for exempting earnings on investment is that it "would have about the same net effect in present value terms as the deferred tax treatment that is now received by most other defined contribution pension savings." (Quote is from page 35 of the Council's Report.)

The rationale for this statement starts from the fact that today contributions to defined contribution plans are generally tax exempt, while benefits paid from such plans are taxable (the reverse of the treatment now proposed). The argument is that it makes no difference in present value or budget accounting (not cash flow) terms, whether taxes are assessed on contributions or on benefits because when the employee pays tax on his contribution, he gives up the opportunity to earn money on the amount taxed. That opportunity goes to the Government.

This proposition is illustrated as follows: Assume that the income tax rate at the time of payment of the FICA tax is 25 percent and the employee contributes $100; after tax, the employee has $75 left to invest. If he doubles his money, he gets $150. On the other hand, if he pays no tax on the contribution when made, but is subject to a 25 percent tax on benefits when paid, he would earn $200, but have to pay $50 in tax. Either way he would end up with $150. The supposition is that either way the government would get the other $50, either as a result of its investment of the $25 it collected up front, or later in cash.

But this supposition is true only under two conditions; it is improbable that either would be met. The first is that the tax rate is the same both when contributions are made and when benefits are paid. This is unlikely. Not only does Congress change tax rates from time to time, but individuals move from one tax bracket to another. After retirement, the applicable tax rate typically goes down, except for those who are highly successful, for whom the tax rate may go up. Some retirees will have so little income that they will be exempt from tax altogether. 30/

Thus, this proposal exempts high income recipients from tax on investment earnings. Low income recipients would be exempt from tax anyway. It also deprives the Government of the tax it would otherwise collect from high income recipients.

The second condition is that the Government must invest and reinvest the amount taxed at the same rate of return as the individual would have done if he had kept the money. But the Government does not invest the money at the same rate of return. Indeed, it does not invest the money at all. Instead, it uses its tax collections to pay bills, and if they are inadequate, it borrows money by selling bonds. Therefore, any savings to the Government from collecting tax are the result of not having to borrow money at 2.3 percent per annum. This interest is what the Treasury saves if it does not have to borrow.

The Council has estimated average real earning from investment in equities to be 7 percent, 4.7 percent more than the 2.3 percent paid by government bonds. Thus, the average loss to the Government from exempting investment income from the income tax (assuming the same tax rate) is the tax rate times 4.7 percent of the accumulated balance in equities or bonds, less the underlying contributions. This sum multiplied by all workers covered by Social Security could reach a very large amount indeed.

It is sometimes argued that exempting investment income gives taxpayers an incentive to save and invest their savings-- a good thing for the economy. But that argument is inapplicable to the proposed PSAs (or IAs) under Social Security since investment of the required 5 (or 1.6) percent of earnings up to the taxable maximum would be mandatory for all covered workers, and no additional investment would be allowed through Social Security.

Moreover, if this method of taxing were applied across the board, as in the flat tax proposals that have been made by Cong. Richard Armey, Steve Forbes, and others -- and there is no inherent reason why investment earnings under Social Security should be treated more favorably than investment income from other sources -- the amount of tax collected would fall substantially. In 1993, investment income was 16 percent of adjusted gross income. 31/ If this income were exempt from the income tax, those who earn their money by working would have to pay at a higher rate in order to keep the Government whole.

For all these reasons, I believe it would be unfair and unwise from a budgetary standpoint to exempt investment earnings from the income tax. Fairness to those who earn their income by working requires taxing individuals on their benefits, not only on their contributions. If workers are taxed on their contributions, the contribution on which they were taxed should be subtracted from investment proceeds, but the remainder should be subject to the income tax. This would be fairer and also save the national budget from major new deficits.


The Report states that money's worth, defined as a comparison between lifetime contributions to Social Security and lifetime benefits, is a critical issue for the future of Social Security. The Comparisons of the three different plans also stress replacement rates, a comparison of earnings just before retirement and benefits just after.

Neither of these measures, money's worth or replacement rates, should be of major interest. By putting them at center stage, I believe the Report provides the wrong tools for evaluating the three proposals for the future of Social Security.

Instead of giving participants a good return for their money, Social Security was designed to provide a balance of adequacy and equity for its participants. This is as true for the young, who are paying more in FICA tax than did previous generations of recipients, as it is for the old who paid for fewer years and at lower rates.

All of Social Security's funds come in the first instance from the payroll tax. While they can be supplemented with earnings on investment of surplus funds, whether from the small rate of interest paid by government bonds or in the greater amounts available from private stocks or bonds, it remains doubtful that adequacy can be provided to low-paid and needy beneficiaries without to some extent diminishing equity for the high-paid. Inevitably when the neediest get more than they contributed or than was contributed on their behalf, the high-paid will get lesser returns.

This is not a deficiency. Social Security should be compared to an insurance policy for disability and death, as well as to a vehicle for income replacement after retirement. Even though purchasers of private insurance, like beneficiaries of Social Security, get their actuarially determined money's worth whether or not they suffer catastrophes, people get major returns only if they incur disaster. Everyone understands that purchasers of private fire insurance will get much higher returns from their premiums if their property burns than if it remains intact. No one expects otherwise. Similarly under Social Security, those who are disabled, widowed or orphaned will get much greater returns than those who do not experience such disasters. For example, children who are newly orphaned this year will receive an average of $470.40 a month, an amount of money crucial to them and their families. 32/ This protection for the most needy among us is as intended. Its great value as a safety net for Americans cannot be assessed through money's worth or replacement rate analysis.

As a vehicle for income replacement, Social Security has had remarkable success, keeping all but 12 percent of older Americans out of poverty. (Without Social Security, 52 percent of elderly Americans would be poor.33/)

Any proposal which reduces the insurance and adequacy provided by Social Security, as both the IA and PSA plans would do, puts this achievement at risk. Both these alternative plans would cut the benefit payable by the trust fund significantly below current levels, substituting chancy returns from investments for some current law benefits.

Specifically, the PSA and IA plans would threaten the disabled, spouses and survivors, and the low-paid, as follows:

The Disabled. The PSA plan would give disabled workers a reduced benefit both before and after retirement age compared to current law. Before retirement age, disability benefits would be reduced actuarially by up to 30 percent as the normal retirement age is increased to 67 and perhaps later. After retirement the amount of the benefit would depend in major part on the extent to which each covered individual disabled person had been able to accumulate personal savings accounts before the onset of disability.

The IA plan would reduce the Primary Insurance Amount (PIA) of retirees and disabled persons from the current 90 percent of the first $437 of AIME (average indexed monthly earnings computed based on the worker's earnings in the 35 years of highest earnings), 32 percent of the increment from $437 to $2,635 AIME, and 15 percent of AIME above $2,635. (AIME replacement levels are indexed annually.) By 2030, the new percentages would be 90 percent, 24.5 percent and 10.5 percent respectively. These changes, plus the delay in the retirement age and the increase in the computation period, would result in benefits for the highly-paid being reduced by about 32 percent, for average workers by about 30 percent, and for low-paid workers by about 22 percent. (Low-paid workers are those paid the minimum wage; some of their benefits are in the middle bracket now 32 percent, to be 24.5 percent.)

Under both plans, those who were disabled young would not have accumulated personal savings or individual accounts sufficient to make up for the cut in their Social Security benefits paid by the trust fund.

Spouses and survivors: The PSA plan would not assure benefits from private investments to the low-paid or non-working spouse or widow(er) in each family. The personal savings account would belong to the partner who earned it, and the entitlement of the lower-earning spouse or survivor to a share would depend upon the generosity of the higher-earning spouse or the divorce courts. If the worker were to die early, the amount accumulated in the personal account before the worker died might be insufficient regardless of its distribution.

The IA plan would give spouses and former spouses no claim on the worker's individual account during his or her lifetime. After the worker's death, it would provide a joint and survivor (J & S) form of benefit to protect survivors. A J & S plan normally gives the survivor half the benefit earned during the employee's lifetime and is paid for by a reduction in the employee's benefit. For low earners, it would be significantly less than available under current law which provides a survivor a benefit equal to that of the deceased worker if it is greater than his or her own. If the survivor had earned more than the worker, the J & S would provide a greater benefit since higher earners do not qualify for a survivor benefit under current law. Beneficiaries of this change in treatment of survivors would primarily be men, who generally earn more over a lifetime than their wives.

In cases of divorce, neither the PSA nor the IA plan would provide an adequate substitute for current law which computes benefits for a former spouse or survivor based on the entire working career of the higher earning earner even if the high earner remarries. By contrast, the divorce courts are unlikely to award a spouse or survivor part of an individual or personal savings account which was not earned during marriage.

Loss of progressivity. Social Security is able to assure adequacy to low-paid workers by computing benefits on a progressive formula. In 1996, everyone pays 6.2 percent of earned income up to $62,700 to Social Security (indexed annually) and another similar amount is contributed by their employers, but benefits vary by AIME level as described above. Those with low income receive a much larger percentage of their contributions in benefit than do high-paid participants. Under the PSA plan, the low-paid would do relatively well because the flat benefit payable to all beneficiaries is a larger part of their total benefit than for higher paid, but the careful and gradual redistributional formula which assures incremental progressivity would be lost. Under IA, benefits payable to everyone would be decreased, although less for the low-paid than for the highly paid. Changes to both plans would undercut adequacy protection for middle, and sometimes low, income workers. Neither plan would guarantee low-paid workers the level of benefits they get today.

In addition to the above policy reasons for objecting to money's worth and replacement rate as indices for evaluating the three proposals, there are the following technical reasons:

(1) The money's worth and replacement rate analysis compare the benefits payable to a couple under current law and the PSA and IA plans. They do not compare payments separately to husband and wife and thus may be misleading. If the individual accounts are divided equally between husband and wife, the money's worth and replacement rates for each individual would be as the charts show. But, as discussed above, equal division is not necessarily what would happen. If the higher earning spouse did not share the individual or personal savings account, both the replacement rate and money's worth payable to the worker would be much higher than under current law, but at a great cost to the lower earning spouse. The money's worth and replacement rate analyses are inappropriate for assessing these changes.

(2) The analyses treat the contributions of employers as if they were individually assignable to the employee. This stems from the hypothesis that employers reduce payments to employees by the amount they pay for benefits on their behalf. However, this is not necessarily mathematically true in each instance. For example, the cost of benefits may be treated by the employer as an unallocated lump sum, and each employee charged an average cost rather than an actual cost. This method of allocation is often used to even out the cost of health insurance when premiums vary by the age and health of the employee, but the employer does not wish to change cash wages.

Or the employer may follow a bookkeeping or compensation method that leads to different results. For example, some employers may reduce payments to employees only by the after-tax cost of such payments; obviously this would be less than the employer's actual payments.

Or the employee's wages may be depressed for other reasons. If the employee lacks the ability to bargain on equal terms, employers may pay them less than their true economic value. This means employers would not necessarily increase payments to employees to their full economic value even if no payroll tax were assessed.

For these technical reasons, as well as for the policy reasons outlined above, I believe the money's worth and replacement analyses should be looked at skeptically, and not treated as significant bases for evaluating the three plans.

Endnotes--Statement by Edith U. Fierst

1/ The benefit decrease for women in 1999 compared to that for men is smaller than one would expect considering the great discrepancy in the numbers who will have worked the full 38 years. The primary explanation is the protection from sharper decreases attributable to Social Security's progressive formula. Zero years have less impact on the final benefit for those with many years because the replacement rate for final years is much lower than that for the initial years of work.

2/ The decreases for both men and women are greater in 2020 than in 1999 because of the higher taxable maximum applicable throughout their working lives. With greater earnings credited toward their Social Security, it will be harder for retirees of 2020 to make up for the effects of having no earnings in a non-working year. In 1996, the taxable maximum is about $62,700, compared to under $10,000 as recently as the early 1970's (although, after indexing $10,000 then, is equal to about $30,000 now.)

3/ The increase at age 65 is from 11.9 years for men and 13.4 years for women in 1940 when Social Security benefits were first payable, to 15.3 and 19.1 respectively in 1995, a difference of over 3 years for men and nearly 6 for women. In 2015, life expectancy at age 65 is projected to be 16.2 years for men and 19.8 for women, a further increase of nearly a year for men and over 6 months for women. See the 1995 Trustees' Report at page 62.

4/ If Social Security had the money, I would greatly favor rewarding additional years of work beyond 35, or at least beyond 35 plus the customary 5 drop out years. Fairness requires that workers be assured of receiving extra benefits if they choose and are able to work more years than the computation period plus 5 drop out years. Under present law, they must pay FICA tax each year that they work, but they do not always get additional benefits in return for these taxes. Instead, they get a benefit based upon the 35 years of highest earnings, and no improvement in their benefits for years of lower earnings. Whenever additional earnings of a worker who has worked the full computation period are below those in the worker's highest earning 35 (often the case, for example, for those who work part-time as students or after retirement), the worker gets no credit toward an increased benefit. This is true whether the extra years are at the end of their careers, in the middle or at the start.

5/ Some employees of governmental units are covered by pension and health insurance, at least in part at employer expense. However, they are omitted from this discussion primarily because so many government employees are not covered by Social Security.

6/ The Commerce Department, Survey of Current Business, for January/February 1996, table 8.15, Supplements to Wages and Salaries by Type.

7/ According to the Social Security actuaries, total compensation paid by private employers to their employees in 1994 was 4.02 trillion dollars, wages in jobs covered by Social Security were 2.9 trillion and wages for workers below the taxable maximum were 2.6 trillion or about 76 percent of total covered wages. If fringes were distributed among employees roughly in proportion to the level of their wages, then about 76 percent of fringes could be attributed to workers with earnings below the taxable maximum.

8/ At page 151 of the 1995 Trustees' Report.

9/ Coverage by size of company and employee income is from The Committee for Economic Development (CED), Who will pay for your Retirement Income? The Looming Crisis, at pages 30 and 31.

10/ CED, op. cit. at 30 and 31.

11/ Issue Brief 170 published by Employee Benefit Research Institute, 1995.

12/ It is the conventional wisdom shared by most Council members that employers do not themselves bear the expense of the FICA tax or of fringe benefits, but shift these costs to their employees. Nevertheless, when there is a threat to tax exemptions on fringe benefits, representatives of business react as though they didn't believe that bit of the conventional wisdom.

13/ No members of the Council are from unorganized labor.

14/ This would also avoid the objection that a tax on health care costs paid by employees might be hard for older employees to pay since their health care costs more. Note, however, that taxing them even on the full cost of their individual benefits would not leave older workers worse off than their contemporaries whose employers do not provide health insurance and who must pay for health care themselves with after-tax income.

15/ June Gibbs Brown, Inspector General of HHS, Flexible Benefit Plans Negatively Impact Medicare's Hospital Insurance Program, report A-05-93-00066, transmitted August 1994.

16/ The losses are not a precise multiple of 2.9 percent because (1) there is no maximum on earned cash income to which the Medicare tax is applicable, and (2) some savings would be offset by increases in Social Security benefits of those who pay the tax. This latter is not true for Medicare whose beneficiaries receive full benefits regardless of the amount of their FICA contributions.

17/ On average, black persons 65 and over, both men and women, die two years earlier than whites, keeping the period of widowhood for both about six years. This figure, and all others not cited separately, come from Sandell and Iams, "Reducing Women's Poverty by Shifting Social Security Benefits from Retired Couples to Widows," an article in the spring 1997 issue of the Journal of Policy Analysis and Management.

18/ Because the problem of low benefits among the elderly affects widows primarily, and also for ease of exposition, this paper focuses on women. However, Social Security is gender neutral; therefore, both men and women in the same circumstances receive the same benefits and would be similarly affected by any changes in the program. When the husband is the lower earner and lives longer, everything said about the computation of benefits for a widow and their adequacy and equity for her is applicable to him.

19/ The formula gives a 90 percent return for earnings below approximately $5,000 a year, a 32 percent return for earnings between $5,000 and $30,000 a year, and a 15 percent return for earnings above approximately $30,000 a year. Both bendpoints are indexed annually to average wages.

20/ There is one benefit she does get. Wives who have worker benefits as well as spouse benefits can retire on their worker benefits at age 62 whether or not their husbands are retired. Spouse benefits are not available (except to the divorced) before the worker who earned them starts to draw benefits.

21/ The salary range shown in this chart is well below average, but analogous differences would arise even if the earned incomes of husband and wife were much higher.

22/ For example, the Pension and Welfare Benefits Administration of the U.S. Department of Labor reported in July 1995 that only 12 percent of those with annual pre-retirement incomes of less than $10,000 have private pensions, compared to 68 percent of those who earned $40,000 or more during their last year of work. See Retirement Benefits of American Workers, pp.6,8-10, 13.

23/ There are many reasons for this. One is the increasing prevalence of 401(k) plans under which contributions to pension plans must be paid primarily by employees out of their earnings. Most low-paid employees are unable to afford such contributions. Another is that small employers, for whom many low-wage persons work, are less likely than larger employers to offer pension plans. A third reason is that it is not a violation of ERISA's non-discrimination rules to exclude part-time workers, and women with minor children are more likely to work part-time. And finally there is Social Security integration under which tax exemptions are not denied for non-discrimination reasons to employers who reduce their contributions to pension plans by a portion of the FICA tax they pay on behalf of low wage persons.

24/ See the policy brief by Richard V. Burkhauser and Timothy Smeeding entitled Social Security Reform: A Budget Neutral Approach to Reducing Older Women's Disproportionate Risk of Poverty, published by the Center for Policy Research of Syracuse University, 1994.

25/ Unfortunately the proposal discussed in this paper does nothing to help never-married women. Of the 6 percent of women 65 and over in 1992 who never married, 21 percent were poor. In the future this problem is likely to become more serious as the number of elderly women who have never married is increasing, especially among black women whose marriage rates have fallen precipitously.

26/ Since enactment of the 1983 amendments, there is one difference in treatment of the married and the divorced: spouse benefits are payable beginning at age 62 to a lower earning spouse whose marriage ended in divorce two or more years earlier even if the former husband is still working. By contrast, if the couple is still married, the wife cannot start collecting spouse benefits until her husband retires. She can, of course, start collecting worker benefits at age 62.

27/ I do not favor plans that allow couples to choose whether their benefits during their joint lives should be reduced to pay for an enlarged survivor benefit. Even when both parties must agree to a decision against adopting such a plan, all sorts of considerations--inadequate information, impending divorce, disproportionate allocation of power between husband and wife, too much trust or lack of attention by the lower earner--can result in an unwise decision to elect against a joint and survivor (J & S) plan. Under ERISA, where the J & S is mandatory unless both parties waive it in writing, nearly a quarter of all couples sign waivers. I have been unable to find any statistical analysis of the consequence to widows of these waivers, but in some cases, at least, it must be seriously adverse. Currently J & S waivers affect private pensions only, leaving the widows who get no private pension survivor annuity with at least Social Security to underpin their retirement security. If the J & S under Social Security depended on the voluntary election of the worker, or even if it were mandatory unless waived by the spouse, many of those widows in greatest need would likely find themselves without survivor benefits.

28/ The full cost of increasing survivor benefits to 75 percent of the combined benefits of husband and wife would be 0.32 percent. Cutting spouse benefits from the current 50 percent to 33 1/3 percent would save 0.17 percent of payroll, leaving the net cost at 0.15 percent.

29/ A small saving might be made by eliminating the current waiver of actuarial reductions for early retirement from survivor annuities. Under current law, married women who retire early, whether as workers or wives, receive actuarially-reduced benefits until widowhood. Thereafter, their early retirement has no effect on the amount of their benefits if they shift to a larger widow's benefit. Thus, married women pay no price for their early retirement after they are widowed. There is no good reason why they should be singled out for this favoritism. By contrast, women who never married receive actuarially reduced benefits all their lives if they chose to retire early. They are often as poor as widows.

30/ In 1995, for example, no tax was payable by a couple if both husband and wife were 65 or over unless their combined income exceeded $13,050. For married couples where neither was 65 or over, the cutoff was $11,550. Moreover, CBO projects that twice as many families with an aged head as without will have a zero marginal tax bracket in 1997; similarly, it projects that two and a half times the number of families with an aged head will be in the 28 percent marginal tax rate compared to families without an aged head. Source: Congressional Budget Office table titled, "Distribution of Families and Persons by Marginal Federal Income Tax (rate projected for 1997).

31/ Individual Income Tax Returns for 1993, published by the Internal Revenue Service.

32/ Forthcoming Annual Statistical Supplement, 1996, published by the Social Security Administration.

33/Op. cit.