Social Security Reform: Improving Benefit Adequacy and Economic Security for Women

This policy brief is designed to raise awareness of the current and future economic circumstances of older women, and the ways in which Social Security reform can help alleviate their unmet needs. It considers the gaps in benefit adequacy and economic security that are not addressed by current Social Security reform proposals and then suggests a series of modest, low-cost reforms to help close these gaps. If our proposals are adopted, Social Security reform will not only close the long-run financial deficit, but it will also greatly reduce the future poverty status of older women, particularly those who live alone. This is an opportunity for progressive reform as well as for budgetary balance. The Social Security program was designed over 60 years ago for a world in which mothers worked at home, raised children, and were widowed young, but not divorced; where fathers worked in industrial settings; and where both men and women had much shorter life expectancies at older ages than those of succeeding generations. Back in 1935 the founders of Social Security did not anticipate that women would become the major beneficiaries of the program. Increasingly, women rely on Social Security as the major source of their economic security at older ages, much more so than do men. Therefore, women are the group with the most to gain or lose from reform of the Social Security system and modification of its benefit formulae. Future women beneficiaries will be different. Women's lives are changing rapidly in many ways. More women work outside the home today, and about half of all marriages end in divorce.


Introduction
This Policy Brief is designed to raise awareness of the current and future economic circumstances of older women, and the ways in which Social Security reform can help alleviate their unmet needs. It considers the gaps in benefit adequacy and economic security that are not addressed by current Social Security reform proposals and then suggests a series of modest, low-cost reforms to help close these gaps. If our proposals are adopted, Social Security reform will not only close the long-run financial deficit, but it will also greatly reduce the future poverty status of older women, particularly those who live alone. This is an opportunity for progressive reform as well as for budgetary balance.
The Social Security program was designed over 60 years ago for a world in which mothers worked at home, raised children, and were widowed young, but not divorced; where fathers worked in industrial settings; and where both men and women had much shorter life expectancies at older ages than those of succeeding generations. Back in 1935 the founders of Social Security did not anticipate that women would become the major beneficiaries of the program. Increasingly, women rely on Social Security as the major source of their economic security at older ages, much more so than do men. Therefore, women are the group with the most to gain or lose from reform of the Social Security system and modification of its benefit formulae.
Future women beneficiaries will be different. Women's lives are changing rapidly in many ways. More women work outside the home today, and about half of all marriages end in divorce.
Increasing numbers of children grow up in a single-parent family, typically that of the mother. The higher future benefits expected for women with their own careers in the labor market need to be balanced against the potentially bleak economic situation in old age for a large and growing number of divorced and never married women.

Current Economic Status of Older Women
Benefit adequacy and economic security for women in old age are our primary concerns. Social Security is the only guaranteed, inflation-protected, lifetime benefit for older people. We start by looking at the prevalence of women among older people, then move to the poverty status of older women and the proportion of elderly Social Security recipients who are women.
• More than two in three persons aged 75 and over are women; and almost three in four persons aged 85 and older are women (Social Security Administration 1998).
The population aged 85 and over is the fastest growing group among the old, and their economic and health care needs are of particular importance. The greater one's age, the greater the degree of insecurity among the elderly.
At least one member of the average couple who retires today and begins receiving Social Security benefits will live an average of 25 years (Social Security Administration 1998). Women at age 65 are expected to live an average 3.3 years longer than men, and because most women marry older men, women are three times more likely to be widowed in old age than are men (Anderson 1998). This produces an expectation of over 15 years that the average older female survivor spends as a widow (Schoen and Weinick 1993).
• Three of every four poor elderly people are women.
Poverty rates for the elderly are highest among divorced, never married, and widowed women-all about 20 percent-compared to a poverty rate of below 5 percent for married women ( Figure  1). And older United States women have poverty rates 1.5 to 20 times higher than those found in other western industrialized Organization for Economic Cooperation and Development (OECD) nations (Smeeding 1998;Burkhauser and Smeeding 1994).
Moreover, a recent report argues that the probability of ever being in poverty between the ages of 60 and 85 is 38.3 percent among both men and women. The probability is 37 percent among white, married older women with less than 12 years of education, and 63 percent if they are unmarried. For black, 4 married older women with less than 12 years of education, the probability rises to 66 percent, and to 88 percent if they are unmarried (Rank and Hirschl 1999).
• The typical elderly Medicare beneficiary spent almost 19 percent of her income on health care in 1998 (Moon 1999).
Out-of-pocket health care expenses act like a tax on income, forcing low-income elders to choose between health care or food and housing. If we follow the National Academy of Science recommendations and adjust incomes for taxes, in-kind benefits and out-of-pocket expenses for health care, the poverty rate for all older women living alone rises from 21 percent to 28-30 percent depending on whether other types of benefits are included in the income definition (Short, Garner, Johnson, and Doyle 1999). Both Medicare reform and Social Security reform need to address this issue.
• Without Social Security and barring any behavioral change, more than half of all older women would be poor today (Social Security Administration 1998).
Women are far less likely than men to qualify for private pensions (30 percent vs. 48 percent in 1994). Even when women do receive their own pensions, they qualify for benefits that are only about one-half the median benefit received by men. And about one-third of husbands still do not elect joint and survivor options for their private pensions upon retirement, despite federal legislation to increase such determinations.
Women make up over 60 percent of all Social Security beneficiaries. On average, unmarried women receive 72 percent of their incomes from Social Security. The percentage of income that comes from Social Security rises with age, rises among older women living alone, and rises as overall income declines. For instance, widows aged 80 to 84 with below median incomes rely on Social Security for more than 80 percent of those incomes (Macunovich 1999).
While poverty rates for older women are high, Social Security keeps far more older women out of poverty. Last year, Social Security reduced the poverty rate among older women from 53 to 14 percent (among men the rate was reduced from 41 percent to 8 percent). For elderly widows, as well as for women aged 85 and over, Social Security plays an even larger role, reducing the poverty rate from more than 60 percent to about 20 percent (Porter, Larin, and Primus 1999).

Older Women in the Future
Today's older women are at risk even in the current defined benefit (i.e., guaranteed benefit level) Social Security system. In particular, unmarried older women-never married, divorced, separated, or widowed-are economically vulnerable and face multiple economic and social risks. Some of these risks include poverty, chronic illness, physical and mental disability, and social isolation (O'Rand 1994).
Women have fewer years of earnings and also lower earnings due to their caretaking roles: women interrupt their careers to perform many years of unpaid work, caring for their children, elderly parents, or disabled spouses. Therefore, today's elderly women are less likely to have adequate private pensions and more likely to depend upon Social Security during their retirement years than are men. These risk factors are interrelated and contribute to the 38 percent chance of experiencing poverty before age 85 that older women face today.
What about older women in the future?
• Since more women are working in the paid labor market, can we expect more private pension participation, more Social Security benefits based on their own earnings, and increased savings to enhance the economic security of older women in the future?

6
The percentage of all women beneficiaries who receive benefits based on only their own earnings will rise from 37 percent in 2000 to 56 percent in 2030 according to the Social Security Administration (National Economic Council 1998). As times change and women's work histories improve, more women will collect higher private pensions and Social Security benefits based on their own earnings.
But women will still interrupt their work careers far more often than men to parent their children and to care for elderly parents.
Women are more likely than men to: • sacrifice market work to provide early life caregiving for children; • be forced to change careers or location to suit their husband's career job choices; • have their retirement timing determined by their husband's retirement date; • interrupt work to provide care to their elder parents, grandparents or siblings.
This will affect retirement income, because Social Security benefits depend on the 35 highest years of earnings and women will have far more years with no countable earnings ("zero years") than do men, even in the mid-21 st century (Smeeding, Estes, and Glasse 1999). In short, women's economic and social experiences are different from men's and will continue to be so as far into the 21 st century as we can foresee. Most of the life events listed above will lower market earnings and put women at greater risk of inadequate retirement income security compared to men. Combinations of these events compound their negative effects on retirement incomes as they accumulate. Hence, both the changing and unchanging experiences of women's economic lives should be taken into account by Social Security reformers.
Nearly one-half of all elderly women will continue to rely at least in part on their husbands' Social Security benefits. And married women will continue to rely on their husbands' pensions under joint and survivor options.
• Will there be changes in marital histories from today's elderly women? If so, what is the predicted impact on elderly women in the future? Table 1 shows that among women aged 45 to 49, the percentage married has dropped sharply from 81.4 percent in 1970 to 67.9 percent in 1997. The percentage divorced has increased sharply (from 5.3 to 17.7 percent), and the percentage never married has risen sharply (from 3.9 to 8.0 percent) between these cohorts.
While more women will have greater lifetime earnings in the future, they will be increasingly likely to be unmarried and will be at greater risk, due to higher rates of divorce and greater periods of single parenthood, than were previous generations of elders. For black women, the numbers will be much higher due to higher divorce rates. More women, both black and white will choose to have families outside of marriage, and the number of women who are predicted never to marry will also increase .
8 For a divorced spouse, age 62, to be eligible for retirement benefits based on a past marriage, that marriage must have lasted at least ten years. Yet the projected fraction of divorced women aged 62 or older in 2020 who are expected to have been married for less than ten years is 37.2 percent (Panis 1999). This, combined with the decrease in remarriage rates and the increasing prevalence of unmarried individuals, threatens to erode the future Social Security protection provided to many lower-income women.
• How will divorce and nonmarriage affect future women Social Security beneficiaries under current program rules?
The net outcome of these changes in earnings, pensions, and marital status is very difficult to predict with any degree of certainty. However, the Social Security Administration's Office of Policy has developed the Modeling Income in the Near-Term (MINT) Model to predict the distribution of retirement income for future Social Security beneficiaries. One way to use the model is to compare economic and demographic differences among the actual retired Social Security population (aged 62 and above) in 1991 and the same projected population for 2020, when persons born between 1926 and 1965, including the baby boom elderly, will be receiving Social Security benefits.
Before this model was developed, projections of the economic status of the baby boomers were limited to the "average" elderly person. This model allows us to move beyond averages and to consider future distributional changes in earnings, pensions, marriage, divorce, labor force participation, and other important economic and demographic factors. While there are limits to all projections, they are absolutely essential if the designers of Social Security reform schemes are to understand how different schemes affect women. Figure 2 presents some results based on this model Smith and Toder 1999;Butrica, Cohen, and Iams 1999).
One projection from the model is that in 2020 fewer women will receive benefits as married and widowed beneficiaries than at present. Declines in marriage rates will reduce the former, while better health for older men will reduce the number of widowed beneficiaries. In contrast, the fraction of divorced women will rise three-fold, from 6.4 percent to almost 19 percent, while the number of never married will rise from 4.1 to 5.6 percent (see Table 2 below). The broad picture from Figure 2 is that: • The poverty rate for older women receiving Social Security benefits in 2020 will be exactly the same as it was in 1991, 12 percent.
• Poverty rates will be a bit lower for married, divorced, and widowed women, but will stay high among divorced women (22 percent) and widowed women (15 percent).
• Poverty among never-married women, many of whom will be never-married mothers in 2020, will increase from 23 to 35 percent.
• Poverty among married women, which is only 4 percent in 1991, is expected to decline to 3 percent in 2020. In addition, older women in the future will continue to be alone after the death of a spouse, to need formal care after the onset of disability at older age, and to spend longer time in retirement at older ages.
• These results suggest that many of tomorrow's female Social Security recipients will be no better off than today's, and that poverty and insecurity will be as much a problem of older women in 2020 as in 1991 or 1999.
Despite the "good news" of greater labor force participation of women, which will increase the number of women with pensions and long earnings careers at higher earnings levels, there is also the "bad news" that divorce, never marrying, and the poor earnings futures of low-skilled women will yield poverty outcomes that mirror those we find today. Therefore, Social Security reformers should be wary of reform elements that put benefit adequacy or insurance protection at risk. Lower income older women will need Social Security just as much or more in the future as they do today.
Economic and social change will have both positive and negative effects on the economic security of tomorrow's women in old age. One can hope that women's earnings, Social Security benefits, and private pensions will grow to mirror those of men, but we must realize that tomorrow's women will experience many of the same insecurities and risks due to their social roles and career work and family patterns as do those of today (Iams and Sandell 1998). If we as a society want to reduce poverty among older women, we must take precautions now to provide even better levels of benefit adequacy and economic security through the reformed Social Security system. In particular, we should better protect those individual women who will lose more than is gained overall from these ongoing and tumultuous social and economic changes.

Proposals to Strengthen Benefit Adequacy and Economic Security for Older Women
The facts reviewed above suggest that older women are clearly at a substantial economic disadvantage compared to men. Social Security reform proposals are designed to provide financial stability by closing the projected long-run gap between revenues and liabilities of the Social Security system, but by and large they are not designed to increase benefit adequacy or economic security, or to lower poverty rates among older women. As we have seen, poverty rates among older women will be the same in 12 2020 as they are today unless we also move to improve the benefit adequacy and insurance features of the system.
Other measures are needed to increase the economic well-being of older women and provide a true floor to their Social Security income. The goal of this brief is to outline a set of interrelated strategies rather than to suggest one single strategy for reducing poverty in very old age. There are three elements to this package: (1) better survivors' benefits, (2) more equitable benefits for divorcees, and (3) a better income safety net. All three work together to play an important role in alleviating economic insecurity and poverty in old age.

Policy Options for Survivors' Benefits
The current Social Security program structure is built upon individual-based payroll taxes but family-based benefits. As a result, it provides different amounts of benefits to different workers who may have paid exactly the same lifetime payroll taxes, depending on their marital histories and their spouses' earnings. Each spouse in a married couple can collect either his or her retired worker benefit, or a spousal benefit based on their partner's earnings history, whichever is higher. For the spouse of a living retired worker, the full benefit is one-half the worker benefit; for a survivor, it is 100 percent of the deceased spouse's benefit.
Unfortunately, under these rules, a two-earner couple may be paid less than a one-earner couple in return for the same or higher total Social Security payroll taxes paid by both earners. For the survivors of such marriages, the contrast in benefits can even be greater. Indeed, if the two couples pay exactly the same total payroll taxes, the surviving spouse of a couple with equal-earning spouses can receive a benefit equal to one-half the benefit received by the survivor of a one-earner couple with the same taxable earnings! (See Burkhauser and Smeeding 1994;Devine 1998.) 13 The total benefits for a two-earner couple and the benefits for its surviving spouse depend directly on the division of total covered earnings between husband and wife, given the total payroll taxes paid. Because most wives have lower lifetime earnings than their husbands, and because most wives outlive their husbands and collect survivor benefits, this disadvantage in benefits for twoearner couples is generally viewed as a disadvantage for working wives. It creates a major inequity, particularly as the earnings of wives continue to rise relative to men as a fraction of total household earnings (Devine 1998).
Social Security reform should be aimed at strengthening rather than weakening survivors' benefits. Social Security survivors' benefits are the key feature of older women's economic wellbeing for the 15 years in old age the average female survivor spends as a widow. Survivor's benefits are crucial to the economic well-being of spouses with lower lifetime earnings.
To overcome this inequity, we urge consideration of plans that would provide a better return on earnings and a better survivors' benefit for older women.
• For example, a lower initial benefit for couples of 133 percent of the higher earner's initial individual benefit upon retirement (to reduce costs) has been proposed in combination with a higher benefit for the surviving spouse of 75 percent of both spouses' combined initial benefits. This option reduces the spousal benefit from one-half to one-third of worker benefits upon initial receipt, and raises survivor's benefits to threefourths of the couple's combined benefit, rearranging benefit levels over the "retired lifetime" of the couple.
Such a solution would go a long way toward helping lower earning women who are married at the time of Social Security receipt, once they reach widowhood and survivor status, because it raises benefits for older women with earnings and it shifts benefits from early in retirement, when the couple is economically better off, to later in retirement, when the surviving spouse depends more on Social Security to avoid poverty. The Social Security Administration has estimated that this plan would add about 7 percent (0.15 percent of taxable wages) to the current Social Security deficit figure of 2.1 percentage points of longterm taxable wages.
This policy change was recommended by the Social Security Administration's 1994-1996Advisory Council (1997, by Aaron and Reischauer (1998), and by President Clinton in October 1998.
Similar policy alternatives could also be adopted for survivors' benefits depending on budgetary circumstances. For instance, for each 1 percent reduction in the initial spouse benefit, 2 percent higher benefits could be paid to all survivors at no extra cost to the system (Iams and Sandell 1998;Burkhauser and Smeeding 1994). More costly options are also available. Both couples' initial benefits and the surviving spouse's benefits could be based on the combined earnings of both spouses. Improving the return on past contributions for spouses who currently receive little or nothing, while at the same time providing higher benefits to older survivors (in return for lower initial benefits in some cases), improves both individual equity and income security for twoearner couples. At the same time it improves the anti-poverty effectiveness of the Social Security system as a whole and better protects older women who are the large fraction of survivors.

Policy Options for Divorcees
As the fraction of divorced and not remarried females aged 62 or over rises from 6 to 18 percent of all women between 1991 and 2020, the number and percentage of divorced women with a less than ten-year marriage will also increase. As mentioned above, the Social Security MINT Model projects that in 2020 more than 37 percent of these divorcees will have a less than ten-year marriage history and will therefore not qualify for benefits based on their former spouse's Social Security earnings history (Panis 1999). This means that a large proportion of future older women 15 will receive no Social Security benefit from former marriages. There are at least two solutions to this problem: • Lower the number of years required to be married prior to divorce in order to collect benefits based on the earnings history of the former spouse. This strategy was followed earlier by Social Security when the years required were lowered from 20 to 10 years.
• Institute earnings sharing for all years of marriage-for divorced men and women only-thus making lifetime Social Security contributions proportionate to covered earnings over the number of years married, by splitting credits for earned income between the couple over the years married. In effect, this treats a Social Security account as one might treat any other asset acquired or built up during a marriage.
Since the amount of benefit for which a divorced woman qualifies under the first option (or under current rules) is independent of the number of years married beyond ten (Steuerle 1999), the second proposal above is liable to be more equitable to divorced women and for the system as a whole, as it is proportional to the length of the marriage without setting a specific minimum number of years. Moreover, as divorced women continue to increase their countable earnings before and after marriage, earnings credits earned for years of marriage plus own earnings are liable to add up to a respectable "own" earnings history and a higher initial individual level of old age benefits under Social Security.

Policy Options for Economically Vulnerable Older Women
The 133/75 percent benefit option proposed earlier does nothing to help a woman whose benefits, based on her own earnings, are less than one-third of her husband's. For these survivors, 75 percent of a 133 percent initial benefit is the same outcome as the current system (i.e., 67 percent of a 150 percent initial benefit).
And for widows in extreme old age, benefit adequacy will remain a paramount consideration even if spousal benefits are reformed as outlined above.
Furthermore, proposals to improve spouses' benefits do little, if anything, for divorced or never married women, the groups at highest risk of poverty in old age (Figure 2). Even our solution of earnings sharing at divorce is still liable to leave many older divorcees with below poverty benefits if their earnings histories are poor (e.g., from a later divorce combined with several years of child rearing).
The spousal benefit and divorce equity reform efforts therefore leave two types of economically vulnerable unmarried older women: those who are not married at the time of initial benefit receipt, and very old women.
The numbers of divorced and never married women are most certainly going to increase among the baby boom generation, as shown in Table 2. Iams and Butrica (1999) estimate that women unmarried at time of first benefit receipt will rise to 30 percent of women aged 62 or over in 2020.
There will be a much larger fraction of very old (aged 85 and older) women who have simply outlived their partners, assets, and savings. Because of early retirement and length of time spent depending on Social Security, these older women may find themselves living for 25 to 30 years on benefits that are only indexed to increases in prices. Given a reasonable rate of economic growth among the rest of the population, these women will have standards of living that continue to decline relative to the rest of the population despite a full cost-of-living-adjustment (COLA) for their Social Security benefits.
Some members of both groups will slip through the cracks and find themselves primarily reliant on Social Security at very old ages, and both groups include the large majority of poor elderly women. Our policy goal is to provide economic support to all vulnerable older women at reasonable cost and without creating a set of disincentives for economic self-support.
Two types of options would benefit this needy group: • An enhanced Supplemental Security Income (SSI) program The SSI program once covered 2.3 million elderly (1974) but by 1999 had fallen to 1.4 million elderly beneficiaries and is forecast to fall further to 1.0 million by 2007 (U.S. Congress 1998, Table  3-W:310). The SSI benefit guarantee for a single elderly person is 75 percent of the poverty level (Social Security Administration 1999), and 80 percent of the poverty level for a beneficiary who also has Social Security (the law allows the recipient to set aside $20 per month of Social Security in determining benefits). With few exceptions all other sources of income are "taxed" at 100 percent by SSI in determining benefits. Further, a single eligible beneficiary must have liquid assets of less than $2,000 (U.S. Congress 1998).
The combined effect of the income and asset provisions are to greatly reduce the number of elderly who are eligible for the program. Recent evidence suggests that these asset limits also discourage savings by the elderly (Hubbard, Skinner, and Zeldes 1995). And low-income elders who qualify for SSI but have Social Security benefits as well end up with just $20 per month more than if they never qualified for Social Security at all. The SSI Reform Commission (Social Security Administration 1992) recommended easing these asset limits to a level of $10,000 or more.
A proposal that eases asset limits, allows beneficiaries to protect up to $175 (not $20) per month of Social Security benefits, and has a qualifying age of 62 (not 65) would provide added protection for low-income women (and men) at low overall system cost. Under an expanded SSI system with these parameters, single elderly persons who receive $175 to $500 of Social Security benefits per month would find their incomes brought up to the poverty line by the combination of SSI and Social Security. Over time, the set aside would need to be adjusted to reflect consumer price increases, but for joint Social Security and SSI recipients with monthly Social Security benefits of at least $175, income poverty would be erased.
The shortcomings of a means-tested SSI-type system are that it would require recipients to submit a separate application, disclose their assets, and face the stigma of welfare participation, even if Social Security and SSI were received in one combined payment. One alternative is a plan that maintains targeting to low-income elders but at the same time avoids the welfare stigma of SSI.
• A new targeted minimum benefit guarantee As an alternative to SSI, federal policymakers should consider institution of a new income tested (but not asset tested) minimum benefit guarantee set at the poverty line (roughly $650 per month) within the Social Security system and separate from SSI. This plan would give credit to all Social Security recipients for a larger share of their benefit than does SSI (more than the current $20 per month). The program would have a guarantee of $650 per month for beneficiaries with less than $400 per month of Social Security benefits; provide a slightly lower than $200 subsidy for those with higher benefits ($400 or more); and would then phase out to zero for those receiving $800 per month (roughly 120 percent of the poverty line) The benefit guarantee could be wage indexed to increase with the average wages of all covered workers, or indexed to changes in the poverty line itself. One hundred percent of all other sources of income would be taxed by this program. Because the system would be run by the Social Security Administration as part of its regular operations, no stigma or take-up problems would arise as long as the beneficiaries were identified and contacted by SSA (e.g., those filing income taxes the year before, or those with low benefits who are likely to qualify). And such a system would benefit only those who qualify for Social Security to begin with.
One could think of this as a targeted minimum benefit. Canada has had great success with a similar system, a Guaranteed Income Supplement (GIS) that provides minimum benefits in old age. The program has reduced the rate of poverty among older single woman from 21 to 8 percent over ten years (Smeeding 1998).

Targeting the Benefits to Groups in Need
To raise benefits for low-income elders at a reasonable cost, benefits must be targeted, which means income testing and possibly other types of restrictions such as asset testing to keep costs down. For instance, benefits could be targeted to those above a certain age (e.g., age 75 or 80) or to those receiving benefits for more than a specified number of years if the budgetary costs of these guaranteed benefits programs are too high.
Both programs would provide a poverty-line income guarantee for all elderly persons who participated in the system long enough to collect benefits, but who were left for whatever reason with below-poverty Social Security benefits at older ages. Income-producing assets (e.g., stocks, bonds, checking accounts, cash management funds, etc.) would be counted against the poverty line guarantee to improve targeting.

Creating a Mutually Reinforcing Benefit Package
All three of these policies are mutually reinforcing and need to be instituted together. Improving benefits for survivors, and shared years of earnings credits for divorcees (coupled with own earning years) will together shrink the pool of those who need to rely on the minimum benefit package. Only never married women would rely on the minimum benefit option alone to avoid poverty. Other groups of women (over 93 percent of those aged 62 and over in 2020) would probably benefit from improved spousal benefits or divorcee benefits as well.

What Will It Cost?
Because of program and beneficiary interactions, the budgetary costs of this program are difficult to calculate. The Social Security Administration has estimated some of the spousal benefit options. Because earnings sharing upon divorce might simultaneously increase benefits for those with less than ten-year marriages and perhaps reduce benefits for those with greater than ten-year marriages, the net cost of this option is liable to be very small. The targeted benefit options can also be tailored to fit budgetary realities by changing parameters in the SSI system or in the new targeted benefit. Our preliminary estimates are that all three benefit changes would total less than one-half of one percent of the long run payroll gap between revenues and outlays, about $20.0 billion per year in 1998 dollars. However, the Social Security Administration needs to more precisely estimate these costs and the program interactions.
The long-run deficit issue requires closing a long-run payroll gap of less than 2.10 percent between revenue and outlays (Smeeding, Estes, and Glasse 1999, Table 4). Proposals that reduce poverty and provide a true floor to older women's incomes at an estimated total cost of less than 25 percent of this deficit do not seem to place extravagant demands on systemic reform, and targeting these benefits toward economically vulnerable older women or those with few other sources of income would lower even these modest costs substantially. For example, increasing Social Security benefits by 10 percent across the board, for widows and widowers alike, is estimated to cost $8.0 billion in 1999 dollars, about 2 percent of total Social Security outlays (Congressional Budget Office 1999).

Conclusions and Future Steps
Social Security has made great advances in increasing the retirement income of older people. But millions of older women who live alone have not been able to increase their economic security. Inequalities in incomes and assets have not declined, and the continuing economic boom of the 1990s has not benefited all Americans equally. Divorce rates continue to climb among older and middle-aged women, and more women are choosing to raise children alone. These trends and the Social Security projections presented here suggest that low-income elderly women will be no better off in the future than they are today. If we are to increase benefit adequacy and economic security for these vulnerable elders, it makes sense to incorporate an effective income floor into the Social Security system and changes in divorcee benefits at the same time that we restore actuarial balance to the system. Older women deserve such a commitment and Social Security financial reform presents the opportunity to all but eradicate poverty in old age.
In this Policy Brief we have discussed three separate but interrelated proposals to improve the economic well-being of the most vulnerable elderly: survivors, divorcees with less than a tenyear marriage history, and very low-income beneficiaries. Each of these groups are disproportionately women who will be at or below the poverty line without the programmatic changes suggested above. While details of these alternatives need further refinement, a package such as this would improve benefit adequacy of Social Security for older women and greatly reduce their future risk of poverty at low or modest cost. Further and better cost estimates by the Social Security Administration could pinpoint these costs.
If the economy, private pensions, and assets all continue to grow, the entire population, men and women alike, can look forward to an economically secure old age. However, this outcome is unlikely. Modest steps must be taken to assist those who are most vulnerable-older women-while at the same time restoring the financial health of the Social Security system. After all, older women are, today and in the future, the most important beneficiaries of the Social Security system.