Widows and Social Security

Introduction

In a moving letter to President Roosevelt in 1933, Mrs. M. A. Zoller asked for assistance for her 82-year-old widowed mother, writing in part:

She is helpless, suffering from Sugar Diabetes, which has affected her mind. She has to be cared for in the same manner as an infant. She is out of funds completely. Her son whom she used to keep house for is in a hospital in Waco, Texas—no compensation for either himself or her. I am a widow; have spent all my savings in caring for her.¹

Letters such as this were typical during the 1930s as the public asked elected officials for relief from the material hardship brought on by both the Great Depression and life events outside their control (health problems, job loss, death of a spouse). Though a wide variety of economic security plans were debated during the 1930s, policymakers ultimately produced two landmark pieces of legislation—the 1935 Social Security Act and the 1939 Amendments to the Act—that provided additional and immediate relief to low-income Americans and, for the longer term, a social insurance structure in which the payroll tax contributions of workers would fund benefits in retirement or upon the death of the wage earner. Many of the programs created over seven decades ago by these two pieces of legislation are easily recognizable even today, including Social Security, federal and state means-tested programs, and unemployment insurance.

Interestingly, the Social Security Act of 1935 provided only limited protection for survivors under the new Social Security program. A lump sum equal to 3.5 percent of total wages could be paid to the estate of a worker in certain cases. However, even before the program became truly operational, this approach began to be viewed as inadequate. The 1938 Social Security Advisory Council, using somewhat stark language, wrote:

A haunting fear in the minds of many older men is the possibility, and frequently, the probability, that their widow will be in need after their death. The day of large families and of the farm economy, when aged parents were thereby assured comfort in their declining years, has passed for a large proportion of our population. This change has had particularly devastating effect on the sense of security of the aged women of our country.²

Concluding that lump-sum benefits were unlikely to be adequate and likely to be spent by the recipient before her retirement, the advisory council recommended that the program include monthly benefit amounts for two classes of widows: aged widows and widows caring for children. The Social Security Board (the forerunner to the current Social Security Administration (SSA)) agreed with these recommendations on social insurance grounds, noting that most national insurance programs at that time provided for widows and orphans.³ Congress enacted these changes with the amendments of 1939.

The advent of widow benefits, which continue to this day, was a fundamental development in the history of the program for several reasons. Most directly, they have provided economic security to millions of widows by replacing income lost upon the death of a spouse at a point when the widow herself, because of age or family responsibilities, could not participate in the labor market. Less directly, the discussion over widow and other family benefits was related to important policy discussions under Social Security. For example, to provide meaningful benefits to widowed mothers and surviving children, the benefit computation would need to use average rather than total wages; otherwise, short working careers would translate into low benefits for survivors. This change, enacted in 1939 and applied to benefit computations for other types of benefits, had the effect of making benefit payments more generous in the earlier years of the program. In addition, the expansion of the program by the 1939 amendments to include family benefits was driven in part by debates over the use of reserves or trust funds. Some policymakers were concerned about the buildup of large reserves under the program; providing family benefits would both achieve socially desirable objectives and limit the buildup of a large reserve (Berkowitz 2002). Thus, policy discussions regarding widow benefits have touched on the main social insurance themes in the program's history: economic security for vulnerable groups, the relatively generous treatment of early participants in the program, and reserve versus pay-as-you-go funding.

Over the years, Congress has added a new type of widow benefit for disabled widows and modified the two original types of benefits by, among other things, making them gender neutral and allowing surviving divorced spouses to be eligible under certain conditions.⁴ Nevertheless, policymakers from the 1930s would recognize much of the structure of benefits paid at the start of 2010 and much of the current policy debate surrounding widow benefits. As was the case then, most of the current policy discussions focus on the adequacy of benefits and the related topic of the economic well-being of widows. This article examines these twin themes and provides policymakers context for understanding the history of Social Security widow benefits and the policy discussions likely to occur in the future. To provide context, it uses data from Social Security administrative records and federal household surveys, projections from a microsimulation model, and the recent research literature on widows and Social Security. The next section of the article presents general information on current benefits and beneficiaries, followed by detailed sections on each type of benefit and group of beneficiaries, a policy discussion, and a conclusion.

Background

This section describes current program rules for each of the three types of widow benefits. The effects of the rules can then be illustrated by examining data on current widow beneficiaries. Together, the description of program rules and a data profile of widow beneficiaries provide the necessary background for the policy discussions that occur in later sections of the article.

Program Rules

A comparison of current eligibility requirements and rules that determine monthly benefit amounts for the three types of Social Security widow benefits is presented in Table 1. At the most basic level, the basis for benefits is a condition that could make the widow's employment and earnings problematic, such as advanced age, caring for a young or disabled child, or having a disability. Further, benefits can only be paid if the deceased spouse worked enough in Social Security–covered employment to achieve the required insured status. All types of widow benefits have relationship requirements, and some of these requirements are similar across categories: The individual must have been married to the deceased worker and, with some exceptions, be currently unmarried. Other relationship requirements vary across category. For example, aged widows and disabled widows can remarry after meeting the basic eligibility requirements (age, or age and disability) without losing benefits, but child-in-care widows generally cannot. Divorced persons can be eligible for each type of widow benefit, but the aged and disabled categories have a substantial duration of marriage requirement (10 years or more), whereas the child-in-care widow category does not. Finally, some eligibility requirements are unique for a particular benefit type. For example, for disabled widows the disability must generally have occurred within 7 years of the worker's death (if the widow received child-in-care benefits in the past, the 7-year time limit would be based on when the widow last received those in-care benefits instead of when the worker died).

Monthly benefit amounts are based on a primary insurance amount (PIA), which is determined by an average of the deceased worker's earnings in Social Security–covered employment. The benefit rate applied to the PIA varies by benefit type. So, for example, an aged widow may receive a monthly benefit amount equal to the full PIA, whereas a child-in-care widow or disabled widow may receive 75 percent or 71.5 percent of the PIA, respectively. The monthly benefit amounts can be lower than that indicated by the benefit rate, and the reasons vary by type of benefit. For aged-widow benefits, the most common reasons are the claiming of benefits before the full retirement age (FRA, currently age 66) by the widow or the worker. An aged widow who takes benefits at age 60 receives 71.5 percent of the PIA, with higher benefits paid at later claiming ages (up to 100 percent of the PIA at the FRA or later). Aged-widow benefits are limited if the deceased worker received reduced retirement benefits and are increased if the worker received delayed retirement credits (DRCs). Child-in-care widows, on the other hand, often have benefits reduced because of Social Security's family maximum provisions. Generally, for these child-in-care families, individual benefits are reduced if there is more than one child on the deceased person's work record, with total family benefits being limited to a maximum amount that ranges from 150 percent to 187.5 percent of the PIA.

A widow who is aged 62 or older or disabled may be eligible for a retired-worker or disabled-worker benefit from Social Security that is based on his or her own work in Social Security–covered employment. This will often lead to dual entitlement (that is, the widow is entitled to both a worker benefit and a widow benefit). In dual entitlement cases, the widow receives the worker benefit plus a partial widow benefit. The total benefit, however, is often equal to or approximately equal to the full widow benefit. For purposes of this article, dually entitled widows are included in the statistics for widow beneficiaries and, except as noted otherwise, monthly benefit amounts reflect the total monthly benefit amount. This approach identifies all individuals receiving widow benefits, and the population estimates published here will be higher than those in the Annual Statistical Supplement to the Social Security Bulletin (SSA 2010, Table 5.A1), where dually entitled individuals are classified as retired workers.

In some cases, a widow will have a worker benefit that is high enough to prevent even a partial widow benefit from being paid. In these cases, the demographic definition of a widow will not match the programmatic definition (that is, the person's marital status is widowed, but they receive only a worker benefit from Social Security). Demographic and programmatic definitions will not align in other cases as well (for example, as noted earlier, a remarried person under some circumstances can be a "widow" beneficiary). There is, however, substantial overlap between the demographic and programmatic populations and reasons to study both groups. This article will present tabulations from both SSA's benefit records (generally using a programmatic definition of widow) and survey-based data (using a demographic definition).

The earnings test reduces benefits for persons below the FRA (but not after) when earnings exceed exempt amounts specified in the law. Benefits based on age (for example, those for aged widows) are increased at a later age to account for any months in which they were withheld. Child-in-care widows, on the other hand, do not have benefits increased at a later date if they are withheld because of the earnings test. Except where specifically noted, tables in this article showing current beneficiaries do not include individuals who are entitled to benefits, but who have their benefits withheld because of the earnings test.

A widow with low income and limited assets may receive payments from the Supplemental Security Income (SSI) program in addition to her Social Security benefits, provided the widow is aged 65 or older, or disabled. Because SSA administers the SSI program, its record systems contain information on whether a beneficiary receives SSI. Payments under SSI, however, are separate and are not part of the monthly Social Security benefit amount.

Program Profile

Data on current beneficiaries, by benefit type, can help illustrate the effects of the program rules as well as provide background for detailed discussions in later sections of this article on each benefit type. Tables 2 and 3 present statistics from a 1-in-100 sample of Social Security's benefit records on the characteristics and benefit levels of the current beneficiary populations under study.⁵ Some data in SSA's records, such as information on race and sex, do not have an administrative purpose, but rather are gathered for statistical purposes. Race is derived from voluntary reports on the Form SS-5 (Application for a Social Security Card). In addition, because of data limitations, race of the widow is assumed to be the race of the deceased worker.⁶

Although current program rules are gender neutral, monthly survivor benefits are of crucial importance to women who currently make up 98 percent, 93 percent, and 96 percent of aged, child-in-care, and disabled widows, respectively. Rising female labor force participation has led to large numbers of widows being dually entitled. Nearly half of aged-widow and disabled-widow beneficiaries are dually entitled to a retired-worker or disabled-worker benefit on their own work records.⁷ Child-in-care widows are rarely dually entitled because they generally would not meet the disability or age requirements for worker benefits.

Relatively few aged and child-in-care widow beneficiaries receive SSI in addition to their Social Security benefits. Disabled widows, however, have a high rate of SSI receipt, with about 1 in 7 drawing payments from this means-tested program. Relative to the other types of widow beneficiaries, disabled widows are more likely to be black and more likely to qualify for Social Security survivor benefits on the basis of a marriage that ended in divorce.

Average benefit amounts range, as of December 2009, from a low of $842 for child-in-care widows to a high of $1,204 for aged widows. The higher aged-widow benefit reflects the higher benefit rate for aged widows, and the lower child-in-care benefit reflects the effects of the family maximum in limiting individual benefits for some widows with children. Median amounts are similar to the average amounts, but there is substantial variation in monthly benefit amounts as seen by values for the 25^th and 75^th percentiles. This latter finding is particularly true for child-in-care widows and disabled widows. For disabled widows, 25 percent have somewhat low benefit amounts (less than $759), and 25 percent have somewhat high amounts (greater than $1,336). The 25^th and 75^th percentile values for child-in-care widows are $563 and $1,108, respectively.

An additional perspective on monthly benefit amounts is given in Table 3 (last column) by comparing the amounts to a standard, namely, the official poverty threshold. For aged and disabled widows, the comparison is of the individual widow's monthly benefit amount to the one-person threshold. For child-in-care widows, the comparison is of the total benefits received by all individuals on the deceased person's work record (including children) to the poverty threshold that corresponds to the number of persons on the work record. Although this comparison does not measure official poverty, which takes into account all sources of income and does not define families based on joint receipt of survivor benefits on a deceased person's work record, it is a useful exercise in assessing the adequacy of benefits and is consistent with general findings in later sections of this article. Social Security benefits are below the poverty threshold for about 22–23 percent of aged and child-in-care widows, and they are below the poverty threshold for about 37 percent of disabled widows.

Aged Widows

The policy discussions regarding aged widows in the years following the 1935 Social Security Act centered on two issues: whether to provide monthly benefits to widows and, if so, the appropriate benefit rate. The first of those issues was settled quickly as monthly widow benefits were added to the program with the amendments of 1939. However, the latter issue was the focus of policy discussions that lasted several decades. Those policy discussions produced the current-law framework for aged-widow benefits, resulting in the relatively high benefit rates and monthly benefit amounts reported in the previous section of this article. In addition, the historical policy debate has framed both policy and program discussions about benefit rates in the current period. Finally, the discussions over the benefit rate reflect a general concern of policymakers regarding the economic well-being of aged widows. This underlying concern has also produced several specialized provisions in the law that are nonetheless important to the economic security of a large number of widows.

Benefit Rate: Historical Policy Discussions

The original benefit rate for aged widows was set, by the amendments of 1939, at 75 percent of the basic benefit of the deceased worker, but discussion before the amendments reflected uncertainty about what the appropriate rate should be. Some policymakers believed a widow needed a benefit that equaled that of the deceased worker (100 percent), but others argued that the homemaking skills of women would allow them to get by with a smaller amount. The issue was crystallized by a question posed by Douglas Brown at a 1938 Social Security Advisory Council meeting:

Can a single woman adjust herself to a lower budget on account of the fact that she is used to doing her own housework whereas the man has to go to a restaurant?

Brown was attempting to clarify a preceding discussion on the topic and to call into question the assumptions underlying the rationale for a lower benefit rate for widows, but the council ultimately thought the answer to his question was "yes" (Berkowitz 2002, 24). The policy debates continued for decades and, in 1961, Congress took a step in the direction of equalizing benefit rates of workers and widows when it raised widow benefits to 82.5 percent of the basic benefit of the worker (a 10 percent increase). The debate that began in the 1930s concluded in 1972 when Congress set the benefit rate for widows at 100 percent of the deceased worker's basic benefit.

The increases in benefit rates over time were not only influenced by a concern that widows faced expenses that were as high as those of retired workers, but also a concern about whether widow benefits were high enough to prevent poverty or low levels of overall income late in life. Both of these concerns can be seen in the report on the amendments of 1972 by the Senate's Committee on Finance (1972):

It is the committee's view that the expenses of a widow living alone are no less than those of a single retired worker, and that there is therefore no reason for paying aged widows less than the amount which would be paid to their husbands as retirement benefits. … In addition, surveys of social security beneficiaries have shown that, on the average, women receiving widow's benefits have less other income than most other beneficiaries. (136)

Information on the economic status of widows in the years immediately leading up to the amendments of 1972 through the current period can be seen in Table 4. Poverty rates for both widows aged 65 or older who receive Social Security and for widows aged 65 or older who do not receive Social Security are shown. The latter group, which includes widows ineligible for benefits because they or their deceased spouses did not work in Social Security–covered employment, is a useful reference group for the purposes of analyzing changes in economic well-being as the result of changes in program provisions.⁸

The concern reflected in the Senate Finance Committee report about widows with too little overall income is borne out in the data. Nearly 30 years after aged-widow benefits were added to the Social Security program, economic hardship was still widespread among older widows: In 1967, 2 out of every 5 aged widows receiving Social Security had income below the official poverty line. By 1973, however, the poverty rate among aged widows receiving Social Security had fallen to just over 23 percent, whereas the poverty rate among aged widows without Social Security remained at its 1967 level of 40 percent.⁹

Poverty has gradually fallen among aged widows receiving Social Security since the 1970s and now stands at 13.4 percent, which exceeds the rate for all aged Social Security beneficiaries (7.8 percent), but is almost identical to the rate for the overall U.S. population (see Table 5).¹⁰ Modest income, however, is still somewhat common with about 38 percent of aged widows on Social Security having income below 150 percent of poverty (the corresponding estimates for all aged beneficiaries and the U.S. population as a whole are 22 percent and 23 percent, respectively). At least to a limited extent, then, the goals of Congress in 1972 have been achieved: Program rules now exist that establish a great deal of parity between the benefit amounts of widows and workers, and monthly benefits are high enough to typically provide at least a modest level of income in old age.

Benefit Rate: Current Issues

The current policy discussion over benefit rates is tied both generally and specifically to the historical debate on the topic. At a general level, as in the past, there is a focus on how much Social Security income an individual "needs" following the death of a spouse. At a specific level, there has been considerable discussion about certain features of the 1972 amendments (and subsequent legislation) that establish a link between the retirement decisions of workers and the benefit amounts received by their widows.

Equivalent income. Much of the current debate on how much income a widow needs centers on the implications of equivalence scales, which are used to equate income for families of different sizes. For example, based on equivalence scales used for the official U.S. poverty thresholds, an elderly person living alone would need 79 percent of the income of an elderly couple to have the same standard of living. Note that the poverty equivalence scale does not use a per capita adjustment, which would imply a need for 50 percent of a couple's income, because a two-person family can take advantage of economies of scale (housing, food, utility). That is, costs will be higher for a two-person family, but not double that of a single person. The basic Social Security benefit of a widow will be below 79 percent of the couple's basic benefit, which has led to several proposals to guarantee widows (or some subset of widows) a benefit equal to a specified percentage of the couple's benefit.

Box 1 shows widow benefits as a percent of the couple amount under four scenarios. For a retired-worker and wife beneficiary couple, the widow's benefit will equal 66 2/3 percent of the couple amount provided the worker and wife claimed benefits at their FRA. If benefits began at age 62, the widow benefit paid upon the husband's death will be 75 percent of the couple amount. A hypothetical two-earner couple is shown in the second panel (the husband and wife are assumed to have equal earnings in covered employment that led to equal PIAs). In this case, the widow benefit equals 50 percent or 55 percent of the couple amount, depending on whether retired-worker benefits were claimed at the FRA or age 62. Increased labor market participation among women will lead to an increasing percentage of two-earner couples, which will tend to lower the size of benefits received by the widow relative to the couple amount. In contrast, trends toward early retirement could increase the relative size of the widow benefit because early retirement, as shown in examples in Box 1, lowers the couple amount relatively more than it does the widow amount. Ultimately, whether benefits are adequate using the equivalence-scale criteria is an empirical question and will depend on the distribution of actual benefit amounts and the amount of retirement income other than Social Security.

Projections are shown in Table 6 of the ratio of income of the widow to the income of the couple (a few years before widowhood) for three groups of individuals: early baby boomers, late baby boomers, and generation Xers born around 1970.¹¹ Despite the wide range in birth cohorts—taken together, these birth-year groups will experience widowhood from the current period through roughly the first half of this century—there is little variation in the results by cohort. Counting all sources of income, the typical widow (as measured by the median) can expect to have income that equals about 71–73 percent of the couple income. This is below the equivalence ratio used in the United States for official statistics on income and poverty.¹² However, the median value, which rises slightly for later birth cohorts, is not far from the needed ratio.¹³ The table also presents values for the 25^th and 75^th percentiles. About three-quarters of widows have income below the equivalence-scale cutoff of 0.79, and about a quarter has income well below the cutoff (the 25^th percentile values range from 0.61–0.64).

It is useful to decompose the "All-income" results (first two columns in Table 6), as this helps identify which widows tend not to have equivalent income upon the death of a spouse. The second set of columns (All but asset income) in the table show results excluding income from assets.¹⁴ When asset income is excluded, the typical widow no longer has income close to the equivalent amount. These results are similar to those obtained when only Social Security income is examined (third set of columns), which should not be surprising given the importance of Social Security as a regular source of income. An important finding of this article is that, even for late baby boomers and generation Xers, Social Security typically replaces about two-thirds of the couple benefit.

In terms of policy, the results that exclude asset income suggest that changes to Social Security policy would be most effective at achieving equivalent income during widowhood if targeted toward groups with lower socioeconomic status. Using education as a fundamental proxy for socioeconomic status, Table 7 shows results for persons who did not finish high school. In this case, using all sources of income, lower-educated widows, compared with all widows, are less likely to have equivalent income (to that which was available when married). These results are, in some respects, consistent with research by Zick and Holden (2000) who find that the inclusion of income that could be derived from assets tended to make recent widows look more like their married counterparts, but only at the upper parts of the wealth distribution.

Legacy issues. The historical debate over benefit rates for widows has not only helped shape the current policy discussions regarding equivalent income, but it has also left a legacy of program provisions that have both policy and nonpolicy implications. To achieve parity of benefit amounts between workers and their widows, Congress not only increased the widow benefit rate to 100 percent of the PIA in 1972, but it also limited the widow's benefit amount if the deceased worker received reduced retirement benefits (a provision referred to as the widow's limit) and, in subsequent legislation, increased the widow amount if the deceased worker earned DRCs. As a result, the retirement decisions of workers are a major factor in the benefit amount ultimately received by their widows. This, in turn, has influenced policy debates such as the debate over whether the earnings test should be repealed at the early retirement age or, as was done in 2000, only at the FRA.¹⁵ It also has influenced discussions in academia, government, and the financial press over whether married workers fully understand the implications of early retirement.¹⁶ Because of the continuing importance of these program features, this section provides policymakers and others with information on the effects of the widow's limit and DRCs.

The widow's limit. Under the amendments of 1972, the widow's benefit is limited to the greater of the amount the deceased worker would be receiving if alive or 82.5 percent of the PIA. The basic intent was to create parity between worker and widow amounts in cases where the worker takes early retirement benefits (the 82.5 percent feature of this provision, however, reflects congressional desire not to have this provision lower benefits below the benefit rate (82.5 percent) that existed before these amendments). Of the approximately 8 million aged-widow beneficiaries currently on the rolls, about 3 million or 37 percent have their benefits reduced because their deceased spouses took early retirement benefits (Table 8). A substantial number of deceased workers, however, were disabled-worker beneficiaries or workers who died before becoming entitled to Social Security benefits, and benefits would not generally be reduced under these circumstances. Also, some individuals who claim widow benefits before the FRA will not reach the limit amount because their age-reduced benefit amounts are already below it. In general, however, the retirement decisions of workers often leave widows with reduced benefits. For cases where the deceased individual was a retired worker, widows have their benefits reduced 59.3 percent of the time because of the limit.

The size of the reduction in benefits is quantified by displaying the difference between the full PIA and the limit amount (see Table 9).¹⁷ Because the limit amount cannot be below 82.5 percent of the PIA, the maximum reduction is 17.5 percent. Average and median dollar reductions are $174.7 and $195.7, which on an annual basis translate to the lowering of potential benefits by about $2,096 and $2,348, respectively. The average and median percentage reductions in benefits are 13.2 percent and 17.5 percent.¹⁸ Table 9 also provides values for the 25^th and 75^th percentiles.

Delayed retirement credits. Workers who postpone receipt of benefits past their FRA, or who prior to the repeal of the earnings test at the FRA in 2000 had benefits withheld because of the test, receive DRCs that are inherited by aged-widow beneficiaries, which again helps achieve parity between worker and widow benefit amounts. The number and percentage of aged-widow beneficiaries with inherited DRCs are shown in Table 10, and the value of the DRCs in dollars and as a percentage of the PIA is shown in Table 11.

The effect of DRCs is somewhat modest. About 1 in 7 widows have higher monthly benefit amounts because of DRCs, and the average and median potential increase in basic benefits is about $898 and $431, respectively, in annualized terms. Even this modest effect is likely to decline in the future because repeal of the earnings test at the FRA has led to relatively few individuals receiving DRCs. Table 12 shows the percentage of men who have earned any DRCs, by the retired worker's year of birth. Note the clear effect of the repeal of the earnings test on the downward receipt of DRCs. The 1933–1934 birth cohorts were the last cohorts to reach FRA before the test was repealed in 2000, and 16.7 percent of men in these cohorts earned at least one DRC. The figure falls to 2.8 percent for the 1935–1936 cohorts. In addition, there was a scheduled increase in the FRA that affects workers born in 1938 or later. Such individuals will have to delay claiming past the higher age (not age 65) to earn DRCs.¹⁹ Note that the rate applied to DRCs has risen for successive cohorts; for persons born in 1943 or later, the rate will reach 8 percent per year.

Benefit Adequacy: Other Program Provisions

The discussion to this point has focused on benefit rates, but the underlying concern of policymakers regarding the economic well-being of widows has also produced several specialized provisions in the law. Though specialized, these provisions often affect large numbers of widows and, collectively, are part of the reason widows receiving Social Security are no longer generally characterized by economic hardship (Table 5). Examples of these provisions include the following:

• The early eligibility age for widow benefits is 60 as compared with age 62 for retired workers and spouse beneficiaries.
• The PIA used to compute retired-worker and spouse benefits, but not widow benefits, can be lowered by the Windfall Elimination Provision (WEP) if the worker established eligibility for a pension based on employment not covered by the Social Security program.
• A special PIA computation (the Widow's Indexing or WINDEX computation) is available for a person whose husband died at a relatively young age.²⁰
• Aged widows, but generally not spouses, can claim one type of benefit (for example, widow benefits) and then claim another type of benefit later (for example, retired-worker benefits).

The last provision, in particular, is worth discussion because it now affects a large number of widows. It is important to the growing number of women with strong attachments to the workforce and illustrates that programmatic analysis (that is, a focus only on widow benefits at a point in time) misses some of the benefit structure that Congress has provided to individuals who have been widowed.

Widows who are eligible for both a widow benefit and a retired-worker benefit can claim one benefit initially and then claim a higher one at a later date. For example, a widow can claim a widow benefit at age 60 and wait to claim a retired-worker benefit (with DRCs) at age 70. In this case, the widow would be a widow beneficiary initially and then only a retired-worker beneficiary. As another example, a widow might claim only a retired-worker benefit at age 62 and then claim an unreduced widow benefit at the FRA of 66. The widow, in this case, would initially be only a retired-worker beneficiary, but then would become a dually entitled widow beneficiary.

Almost 900,000 retired workers currently on the rolls use to be widow beneficiaries, but had those benefits ended because they claimed higher retirement benefits (Table 13). These individuals were often widowed in midlife, with an average age at widowhood of 54. Their average monthly retired-worker benefit is $1,201, which is very close to the amount ($1,204) paid to the 7,935,700 aged-widow beneficiaries currently on the rolls (see Table 3). The retirement-to-widow cases are much less common than widow-to-retirement cases (under 100,000 individuals), but benefits are somewhat higher. These individuals are also often widowed in midlife (average age at widowhood is 55). Thus, the ability to separate benefit receipt affects a large number of individuals widowed at earlier ages and brings their benefit amounts approximately in line with the benefits paid to the overall population of widow beneficiaries.

Child-in-Care Widows

In the early years of the Social Security program, child-in-care widow benefits were more common than aged-widow benefits. Over time, however, they have become a relatively small part of the annual awards made to the total widowed beneficiary population (aged, child-in-care, and disabled). In 1940, child-in-care widows accounted for over 83 percent all awards to widow beneficiaries, but by 2008 this figure had fallen to just over 5 percent. The reasons can be traced to policy choices, improving mortality among men, and striking demographic changes affecting whether mothers of surviving children meet the relationship requirements for child-in-care widow benefits.

Table 14 shows annual awards for child survivors, child-in-care widows, and the total for all types of widow beneficiaries (aged, child-in-care, and disabled) by selected year. Annual awards reflect new claims and thus differ from estimates of the total beneficiary population at a point in time (such as figures in Table 2). Awards increased fairly consistently through the 1970s as Social Security coverage gradually became close to universal and program expansions were legislated. In 1980, over 540,000 orphans and 107,000 child-in-care widows were awarded benefits. Child-in-care widows accounted for nearly 20 percent of benefits awarded to all widows in that year. Policy changes enacted in 1981 sharply reduced awards to child survivors and child-in-care widows: Child-in-care widows, effective for entitlements after August 1981, were required to be caring for a child under age 16 (previously under age 18) or disabled, and student benefits were phased out for those attending college. From 1980 through 1985, the number of awards to children of deceased workers and child-in-care widows fell by 38 percent and 33 percent, respectively.

Interestingly, awards to child survivors have stabilized, and the number of awards in 2008 was similar to the figure for 1985. Child-in-care widow awards have continued a marked decline, falling 55 percent from their 1985 levels. The differing trends among these younger survivor beneficiaries, in the presence of policy stability, suggest some demographic factors affecting child-in-care widow awards. Two possibilities are changes in marriage and divorce. To be eligible for child-in-care widow benefits, the survivor must have been married to the worker and must be currently unmarried. In a family structure where a worker and a spouse are married until the death of the worker, these requirements would naturally be met: The spouse was married to the worker, and upon the death of the worker the spouse would be unmarried. In a family structure where the mother of the children never married the worker or where the mother was married, but divorced the worker and subsequently remarried by the time of the worker's death, these requirements would not be met and only the children would be eligible for benefits. Table 15 provides some data that suggests changing family structure is a plausible, if not proven, partial explanation for the decline in child-in-care widow awards. Note in particular the very rapid and somewhat recent rise in children born to unmarried mothers.

Initial eligibility rules affect the number of awards, but the overall size of the population receiving child-in-care widow benefits is also affected by postentitlement rules on work and remarriage. Table 16 shows the number of child-in-care widows who were not receiving benefits because of the earnings test and the number who would otherwise be eligible but for the fact that remarriage terminated their benefits.²¹ Thus, without the earnings test or the termination provision for remarriage, the population receiving child-in-care widow benefits for December 2009 would be larger by 80,300 or 50 percent.

Characteristics of the CurrentChild-in-Care Widow Population

If demographic developments have led to the decline in awards for child-in-care widow benefits, the population on the rolls may be quite different than in the past. In an effort to provide policymakers with some information to broadly assess the characteristics of the child-in-care widow population, Table 17 presents tabulations from Social Security's benefit records.

The death of the wage earner typically occurs in middle age: The median age of the worker at death is 42, and at the 25^th and 75^th percentiles, the ages are 35 and 50, respectively. The widow is typically younger at the point of the worker's death: The median age at widowhood is age 38. At the time of death, the youngest child in the family is typically fairly young (median age at time of worker's death is age 6). Table 17 also shows the current age of the widow and the youngest child (median values are at ages 44 and 13, respectively). Zick, Fan, and Chang (2004) find that younger widows were at particular economic risk because the family was often not covered by Medicare, but had large medical expenses that were due to the spouse's death and because the family lost access to the spouse's labor market income. From an income perspective, at least, there is evidence that Social Security benefits are sufficient to prevent very low levels of income for these families. Family benefits tend to be relatively high and, by themselves, prevent poverty-level income for more than 75 percent of these families (see 25^th percentile in the last row of Table 17). If child-in-care widows have little in the way of income other than Social Security, it is likely that they are generally characterized by income levels that are modest, but not extremely low.²²

Disabled Widows

Disabled widow benefits were added to Social Security in 1968, following a recommendation of the 1965 Social Security Advisory Council and a request by President Johnson as part of a set of proposals outlined in a special message to Congress. The advisory council's recommendation and the president's proposal were somewhat general, but Congress legislated a tightly defined benefit structure. Kingson and others (2007) argue that cost considerations and uncertainty about the effects of a new type of benefit were important motivations behind the initial congressional focus on a narrow benefit structure. The initial requirements follow.

• The widow must be at least age 50 (still in place).
• The disability had to occur within 7 years of the husband's death (still in place).
• The benefits were actuarially reduced if claimed before age 60 (repealed in 1983).
• A stricter definition of disability for disabled widows than for disabled workers was required—a widow could not engage in any, as opposed to substantial, gainful activity (repealed in 1990).

Kingson and others (2007) and Veenis (2008) find that the two liberalizations led to higher benefit levels and an increase in the number of disabled widows. Table 18, using recent data, illustrates the effects of the two liberalizations. Real average widow benefits were nearly 22 percent higher in 1990 than in 1980, reflecting the effects of the 1983 legislation that eliminated, for those on the rolls and future beneficiaries, the additional actuarial reductions applied to benefits received before age 60. SSI receipt among these beneficiaries also fell 5 percentage points during this period. The number of disabled widows declined from 1980 through 1990, approximately doubled from 1990 through 2000, and then grew modestly over the next 8 years. The doubling of the disabled-widow population in the 1990s can be partly attributed to removing the stricter definition of disability (effective in 1991) that had applied to this beneficiary group.

Economic Status

Weaver (1997), using CPS data exactly matched to Social Security administrative records, finds that disabled-widow beneficiaries had the highest estimated poverty rate (37 percent) of any Social Security beneficiary group. Kingson and others (2007), using public-use CPS data to approximately identify individuals receiving disabled-widow benefits, find that 44 percent had income below poverty.²³ Veenis (2008), using Kingson and others' approximate method of identification and a large sample in the 2005 American Community Survey, finds that 36 percent of disabled-widow beneficiaries were poor. Kingson and others also examine the economic status of disabled widows aged 50–59 who were not receiving Social Security benefits, a group that contains individuals who are potentially eligible for benefits should disabled-widow benefits be further liberalized (for example, eliminating the requirement that disability occur within 7 years of the worker's death). This group was also characterized by very low income.

Table 19 updates Kingson and other's (2007) results with more recent data (2004–2009, as opposed to 1995–2000), using their methodology for identifying disabled-widow beneficiaries. These updated results tell the same basic story found in those authors' (2007) work: The measured poverty rate for disabled widows is high for both disabled widows receiving Social Security and disabled widows not eligible for Social Security (38.9 percent and 59.1 percent, respectively). Compared with nondisabled widows, both of these groups also have high measured levels of SSI receipt and low levels of education, which reinforces the finding that such widows are part of the low-income population. In addition, both groups of disabled widows have a higher percentage of individuals who report a race of black; notably, an estimated 1 in 3 disabled widows not receiving Social Security is black.

The findings from the survey data are also consistent with program experience. SSI is a means-tested program with strict asset limits and maximum federal payment amounts below the official poverty level. More than 2 in 5 disabled widows are receiving or have received benefits from the program, and an additional 1 in 20 applied for but were denied SSI (Table 20). Also, disabled widows are frequently widowed or disabled before age 50 (the earliest eligibility age for disabled-widow benefits). Thirty-five percent were widowed before age 50 and nearly half (46 percent) had a disability that started before their 50^th birthday (figures not shown in table). In short, early widowhood, disability, and frequent experience with a means-tested program suggest a population with a much lower economic status than the general U.S. or Social Security beneficiary populations.

Discussion

Aged-widow benefits were not included in the original Social Security Act, but over time few groups have received such sustained and often sympathetic consideration by policymakers during the history of the program. The group is noncontroversial for many reasons: The death of a spouse is beyond one's control and is naturally addressed through social insurance mechanisms, and advanced age will often not allow for these widows to financially adjust to the loss of a spouse. Even today, proposals routinely call for increased benefits for aged widows. The policy attention has achieved results: Social Security benefit increases have clearly played a role in the dramatic reductions in poverty among widows.

Today, the economic status of aged widows is not, generally, characterized by deep material hardship, but it is also not characterized by affluence. Poverty rates are above the overall population of Social Security beneficiaries, but about on par with the broader U.S. population. Widows, however, are more likely to have modest income compared with either group. This situation is likely to continue well into the future. Table 21 presents selected results for aged persons in 1992, 2020, and 2040.

For all aged persons, average projected income equals 5.45 times the poverty level in 2040, but for aged widows, the average value is about 30 percent lower (at 3.83). Absolute poverty rates are higher for widows, but decline for both the overall aged population and for aged widows—a result that is largely driven by the fact that poverty thresholds are adjusted by prices, and income is projected to increase with overall wage growth in the economy. Adjusting the poverty thresholds instead for wage growth yields a poverty measure that reflects well-being of widows relative to the nation's improved standard of living in future years (relative poverty); here again, widows have lower economic status than the overall older population.

The economic status of widows presents policymakers with some difficult choices: a population of interest whose well-being is neither clearly at a low-income level, nor clearly well off. As an illustration, Favreault, Sammartino, and Steuerle (2002) examine several proposals designed to help low-income beneficiaries and find that increases in survivor benefits (guaranteeing the widow 75 percent of the couple's benefits) did not target benefit increases to low-income women as well as other options, such as creating different types of new minimum benefits, changing the duration of marriage requirement for divorced benefits, or implementing child care credits. For the 75 percent of the couple's benefit option, only about 25 percent of the additional benefits reached those in the lowest lifetime earnings quintiles. One potential solution, which has appeared in several proposals, is to restrict benefit increases to low-benefit couples.²⁴ Such an approach is supported, to some extent, by the MINT analysis in this article (Table 7), in which widows from lower socioeconomic backgrounds were less likely to have income that was near the equivalent level of income in marriage. In addition, the couple's benefit option could be combined with other changes, such as lower spouse benefits, that would help control costs and make it more likely that the widow's income would be equivalent to that of the couple (Favreault 2009).

Another issue, particularly in light of changes in marital patterns, is whether program expansions should be directed at groups that meet particular marital-status requirements. Future retiree populations will have a greater share of never married individuals, a sizeable number of whom will have low income. For example, Tamborini (2007) finds that the percentage of retirees who have never been married is projected to more than double from 2020 through 2060 when they will represent about 10 percent of the retiree population. He also finds that never-married retirees have both high current poverty rates and projected rates for the future. Harrington Meyer, Wolf, and Himes (2006) present evidence that fewer women (especially black women) will likely qualify for survivor (and spouse) benefits because of the rise in the share that never married.

Policymakers concerned about low-income widows could alter aged-widow benefits or implement other options that would reach the widowed population. The percentage of SSI recipients aged 65 or older and the general U.S. population aged 65 or older, by marital status, is shown in Table 22. A large percentage of individuals on the SSI rolls are widows (39.0 percent) because they are a large part of the overall aged population (29.0 percent) and because they have more modest resources. Changes in the SSI program would not exclude other marital-status groups, but would affect a large number of widows. The basic result is not limited to SSI. Some proposed changes to Social Security, such as a new minimum benefit, though not targeted exclusively by marital status, could increase the income of many in the widowed population.²⁵

It should be noted that existing law offers married couples and widows the ability to substantially improve economic well-being in retirement. Delaying claiming of Social Security by workers past the early eligibility age has a sharp upward effect on the income available to their survivors in retirement. Similarly, individuals widowed early in life who have earned benefits in their own right have options under the law that allow them to significantly boost retirement income through delayed claiming of one of the benefits (the retirement or widow benefit). Efforts to educate the public about these options could improve retirement security for widows.

In addition to education, there are reasons to consider and evaluate policy changes to the DRC. At 8 percent per year, the rate is thought to be about actuarially fair for an individual worker (Myers 1993, 99). For a married worker, however, the rate would be more than fair because the increased benefits would be paid while either the worker or the spouse is alive. Also, Benitez-Silva and Yin (2009) find that, among very recent retirees, the small population that receives DRCs has a number of individuals with less robust earnings histories who may be using the DRC provisions as a way of "catching up" or securing an adequate retirement income. Orszag (2001) analyzes proposals to pay DRCs to workers as a lump-sum payment (widows could continue to receive the traditional DRC increment added to monthly benefits upon the death of the worker). He cites evidence indicating that individuals would be more likely to work longer and defer initial age of Social Security benefit receipt if actuarially equivalent lump sums were offered instead of smaller increments added to monthly benefit amounts. Additional work has a large effect on retiree well-being because retirement savings are increased rather than drawn down. Butrica, Smith, and Steuerle (2006) find that an additional 5 years of work would finance a 56 percent higher level of consumption in each year of retirement. One reason to consider DRC proposals is that, with the elimination of the earnings test at the FRA in 2000 and the increase in the FRA itself, relatively few workers and future widows benefit from the current structure of the credits.

Finally, the retirement security of aged widows depends strongly on access to resources in addition to Social Security. The microsimulation results presented in this article indicate that inherited assets (wealth and retirement accounts) are the difference between having approximately equivalent income in widowhood or having a lower standard of living upon the death of a spouse. Whether the wealth projections in the model unfold as projected, the underlying result for purposes of policy and planning are still informative. On the policy front, there have been proposals to expand workplace pensions using automatic enrollment, employer payroll systems, and existing individual retirement account structures (Iwry and John 2009). Such proposals target half the workforce (those not currently participating in an employer-sponsored retirement plan) and may ultimately offer widows greater security through their own retirement accounts or as wealth that is bequeathed them by their spouses.

The economic status of child-in-care widows, in broad terms, quite likely parallels that of aged widows: Social Security benefits prevent material hardship for a large percentage of the population, but the economic effects of widowhood leave overall income at modest levels. The driving policy issue in the future for child-in-care widows may be less about the adequacy of benefit levels (total family benefits are relatively high) and more about underlying program rules on marriage and work. Very large numbers of children in the United States (and many other developed countries) are born out of wedlock, and the mothers of surviving children may increasingly not meet the relationship requirements for child-in-care widow benefits. Policymakers may judge this appropriate (for example, if they believe marriage reflects dependence on the worker and therefore should be the basis for paying a benefit on his or her work record), but over time it will leave an increasing number of families with surviving children in which the head of the household does not receive Social Security. Further, marriage is a requirement for eligibility, but it is also a condition for termination of benefits. Aged widows and disabled widows can remarry and retain their benefits, but child-in-care widows cannot. Finally, with regard to work decisions, it is useful to note that the earnings test for aged beneficiaries does not reduce lifetime benefits under the program (Biggs 2008), but child-in-care widows face permanent losses in benefits because of the test as their benefits are not recomputed at a later date.

Perhaps more so than with the other two types of widow benefits, disabled-widow benefits were introduced into the system with the clear intent of potentially modifying them over time. As part of the large-scale solvency reforms of 1983, Congress enacted some program liberalizations that affected small but vulnerable groups including disabled widows. The 1983 and 1990 changes to these benefits are instructive as they reflect policymakers' view to create a benefit structure that follows principles applied to the much larger group of disabled beneficiaries (disabled workers). Before the change, disabled-widow benefits were actuarially reduced if claimed before age 60. After 1983, the reductions were removed making the benefit more similar to disabled-worker benefits where no actuarial reductions are applied. In 1990, the benefits were again made similar to the disabled-worker benefit structure by applying the same legal definition of disability for the two types of benefits. If policymakers want to further modify the benefit, changes to the early eligibility age and the current benefit rate are possibilities. Disabled-worker benefits are paid because of disability regardless of age, but disabled-widow benefits are not available before age 50. After the amendments of 1983, the benefit rate for disabled widows was set at 71.5 percent of the PIA, but disabled workers receive a benefit equal to the full PIA.

In addition to making disabled-widow benefits similar to disabled-worker benefits in structure, policy may be active in this area for two other reasons: Proposals are relatively inexpensive (Kingson and others 2007), and the population is characterized by low income. The relative low cost is due to natural limits on the size of the population that is both widowed at young and middle ages and has a severe impairment that will meet the legal definition of disability. In addition, such persons often qualify for other government programs such as SSI, meaning the net increase in costs from a total budget perspective is further limited. Unlike child-in-care and aged widows, the available evidence suggests that material hardship may be somewhat widespread in the disabled-widow population. Both poverty rates using the official thresholds and program data suggest that this group is of much lower economic status than the U.S. or general Social Security beneficiary populations.

Conclusion

Although just one personal story during the Depression, Mrs. M. A. Zoller's letter to President Roosevelt in 1933 was revealing. Her situation was desperate, being both a widow caring for her children and a daughter caring for her elderly widowed mother. Within 6 years, the nation had adopted a social insurance structure that would provide monthly benefit amounts for both young widows caring for children and for aged widows. The effects of the new social insurance program would prove to be remarkable. Since the program's inception, more than 28 million widows have been awarded monthly benefits (SSA 2010, Table 6.A1). Eligibility for benefits and subsequent program expansions have led to dramatic declines in poverty among aged widows, which— although not eliminated—has now reached a point where it is roughly on par with the overall U.S. population.

Congress has modified widow benefits several times in the program's history, including adding a new type of benefit for disabled widows in the 1960s. Legislative and policy proposals, even today, frequently include further expansions to Social Security aged-widow benefits. The program, today, however is a mature one, and large-scale liberalizations of aged-widow benefits, such as those that occurred in the 1970s, are less likely because of costs and the improved economic status of older widows. Smaller-scale proposals, such as those that target benefits increases to low-income aged widows, target benefit increases to individuals with limited income regardless of marital status, or that address some of the specific policy concerns of each group of widow beneficiaries (aged, child-in-care, and disabled), may occur on their own or as a part of broader Social Security legislation. Finally, numerous opportunities exist to improve the well-being of widows that go beyond Social Security policy. Efforts to educate the public about key Social Security provisions and how they impact income in both the near term and long term and efforts, through policy or education, to increase retirement savings hold promise in terms of providing widows with economic security late in life.