This is the accessible text file for GAO report number GAO-14-33 
entitled 'Retirement Security: Trends in Marriage and Work Patterns 
May Increase Economic Vulnerability for Some Retirees' which was 
released on February 26, 2014. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as 
part of a longer term project to improve GAO products' accessibility. 
Every attempt has been made to maintain the structural and data 
integrity of the original printed product. Accessibility features, 
such as text descriptions of tables, consecutively numbered footnotes 
placed at the end of the file, and the text of agency comment letters, 
are provided but may not exactly duplicate the presentation or format 
of the printed version. The portable document format (PDF) file is an 
exact electronic replica of the printed version. We welcome your 
feedback. Please E-mail your comments regarding the contents or 
accessibility features of this document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. Because this work 
may contain copyrighted images or other material, permission from the 
copyright holder may be necessary if you wish to reproduce this 
material separately. 

United States Government Accountability Office: 
GAO: 

Report to the Chairman, Special Committee on Aging, U.S. Senate: 

January 2014: 

Retirement Security: 

Trends in Marriage and Work Patterns May Increase Economic 
Vulnerability for Some Retirees: 

GAO-14-33: 

GAO Highlights: 

Highlights of GAO-14-33, a report to the Chairman, Special Committee 
on Aging, U.S. Senate. 

Why GAO Did This Study: 

Marriage has historically helped protect the financial health of 
couples and surviving spouses in old age. Based on their marriage, and 
independent of their own work history, spouses may receive retirement 
and survivor income through Social Security and some employer-
sponsored pension plans. Many of the federal requirements governing 
these benefits were developed at a time when family structures, work 
patterns, and pensions were very different from what they are today. 
In recent decades, marriage has become less common, more households 
have two earners rather than one, and many employers have shifted from 
DB plans to DC plans. In light of these trends, GAO was asked to 
examine the issue of marriage and retirement security. Specifically, 
GAO examined: (1) the trends in and status of marriage and labor force 
participation in American households, (2) how those trends have 
affected spousal benefits and retirement savings behavior within 
households today, and (3) the implications of these trends for future
retirement security. GAO analyzed nationally representative survey data
including the Survey of Consumer Finances, the Survey of Income and
Program Participation, and the Current Population Survey (CPS); 
conducted a broad literature review; and interviewed agency officials 
and a range of experts in the area of retirement security. 

What GAO Found: 

Over the last 50 years, the composition and work patterns of the 
American household have changed dramatically. During this period, the 
proportion of unmarried and never-married individuals in the 
population increased steadily as couples chose to marry at later ages 
and live together prior to marriage (see figure). At the same time, 
the proportion of single-parent households more than doubled. These 
trends were more pronounced for individuals with lower levels of
income and education and for certain racial and ethnic groups. Over 
the same period, labor force participation among married women nearly 
doubled. 

Figure: Marital Status over Time: 1960-2010[A]: 

[Refer to PDF for image: stacked horizontal bar graph] 

Year: 1960; 
Percentage of population 15 years old and over: 
Not married: 32%: 
Never married: 22%; 
Widowed: 8%; 
Divorced: 2%; 
Currently married: 68%. 

Year: 1980; 
Percentage of population 15 years old and over: 
Not married: 39%: 
Never married: 26%; 
Widowed: 7%; 
Divorced: 6%; 
Currently married: 61%. 

Year: 2000; 
Percentage of population 15 years old and over: 
Not married: 44%: 
Never married: 28%; 
Widowed: 6%; 
Divorced: 9%; 
Currently married: 56%. 

Year: 2010; 
Percentage of population 15 years old and over: 
Not married: 46%: 
Never married: 31%; 
Widowed: 6%; 
Divorced: 10%; 
Currently married: 54%. 

Source: GAO analysis of CPS data. 

[A] These statistics represent a snap shot of marital status at a 
particular point in time. Percentages within the not married sub-
categories may not sum to the not married total due to rounding. 

[End of figure] 

Taken together, these trends have resulted in a decline in the receipt 
of spousal and survivor benefits and married women contributing more 
to household retirement savings. From 1960 through 2011, the 
percentage of women aged 62 and older receiving Social Security 
benefits based purely on their spouse's (or deceased spouse's) work 
record declined from 56 to 25. At the same time, the percentage of 
women receiving benefits based purely on their own work records
rose from 39 to 48. Further, as of 2010, among married households 
receiving pensions, 40 percent had elected not to receive a survivor 
benefit. Rising labor force participation among married women enabled 
them to contribute more to household retirement savings. From 1992 to 
2010, married women's average contributions to household retirement 
savings increased from 20 to 38 percent. 

In the future, fewer retirees will receive spousal or survivor 
benefits from Social Security and private employer-sponsored pension 
plans, increasing vulnerabilities for some. Eligibility for Social 
Security spousal benefits among women is projected to decline, in 
part, because fewer women are expected to qualify based on marital 
history and more are expected to qualify for their own benefit based 
on their own work record. For many women, this shift will be
positive, reflecting their greater earnings and capacity to save for 
retirement. However, women with low levels of lifetime earnings and no 
spouse or spousal benefit may face greater risk of poverty in old age. 
For private plans, the shift from defined benefit (DB) to defined 
contribution (DC) plans increases the vulnerability of spouses because 
of different federal protections for spouses under these plans. DB 
plans are required to offer survivor benefits, which can only be 
waived with spousal consent. In contrast, DC plan participants generally
do not need spousal consent to withdraw funds from the account. 

What GAO Recommends: 

GAO is making no recommendations. GAO received technical comments on
a draft of this report from the Department of Labor and the Department 
of the Treasury, and incorporated them, as appropriate. 

View [hyperlink, http://www.gao.gov/products/GAO-14-33]. For more 
information, contact Charles Jeszeck at (202) 512-7215 or 
jeszeckc@gao.gov. 

[End of section] 

Contents: 

Letter: 

Background: 

The Composition and Work Patterns of the Family Have Changed 
Dramatically in the Past 50 Years: 

Trends in Marriage and Work Patterns Have Resulted in Fewer Women 
Receiving Social Security Spousal Benefits and Married Women 
Contributing More to Household Savings: 

The Decline in Receipt of Spousal Benefits Is Likely to Increase Some 
Retirees' Economic Vulnerability: 

Concluding Observations: 

Agency Comments: 

Appendix I: Objectives, Scope, and Methods: 

Appendix II: Selected Social Security Policy Options That May Offset 
Effects of Changes in Marital Patterns: 

Appendix III: GAO Contact and Staff Acknowledgments: 

Tables: 

Table 1: Selected Federal Requirements Related to Spousal Benefits or 
Protections in Social Security Retirement Benefits, Defined Benefit 
and Defined Contribution Plans, and Individual Retirement Accounts: 

Table 2: Selected Social Security Retirement Benefit Types and 
Characteristics: 

Table 3: Data Sources Used for Each Reporting Objective: 

Table 4: Characteristics of Female Social Security Beneficiaries: 2011: 

Table 5: Comparison of Wealth and Income Characteristics of Married 
Female Social Security Beneficiaries: 2011: 

Table 6: Characteristics of Households Currently Receiving Pension 
Income: 

Table 7: Characteristics of Working Age Households: 

Table 8: Selected Policy Options That Target Spouses and Survivors: 

Table 9: Selected Policy Options That Target All Elderly, Regardless 
of Marital Status: 

Figures: 

Figure 1: Poverty Rate Estimates of the Population Age 65 and Older, 
by Marital Status and Sex, 2012: 

Figure 2: Estimated Percentage of Population Age 15 and Older by 
Marital Status: 1960-2010: 

Figure 3: Estimated Number of Unmarried Couples (in Thousands): 1996-
2012: 

Figure 4: Among All Families, Estimated Percentage of Single-Parent 
Families by Sex of Head of Household: 1970-2012: 

Figure 5: Estimated Percentage of Married Individuals by Race and 
Gender over Time: 1960-2010: 

Figure 6: Estimated Percentage of Families Headed by a Single Parent 
by Race: 1970-2012: 

Figure 7: Among People Age 45-54, Estimated Percentage of Married Men 
and Women, and Unmarried Parents, by Highest and Lowest Income 
Quintiles: 1960-2010: 

Figure 8: Among People Age 18 and Older, Estimated Percentage of 
Married Men and Women, and Unmarried Parents, by Education Level: 1960-
2010: 

Figure 9: Estimated Labor Force Participation Rates by Sex and Marital 
Status: 1960-2010: 

Figure 10: Estimated Labor Force Participation Rates among Women with 
Children of Different Ages: 1975-2010: 

Figure 11: Among Married Households, Estimated Percentage of Married 
Households in Which Both Husband and Wife Have Earnings, Only the 
Husband Has Earnings, and Only the Wife Has Earnings: 1967-2010: 

Figure 12: Among Mothers Who Worked prior to the Birth of Their First 
Child, Length of Time off Taken for Caregiving, by Year First Child 
Was Born: 

Figure 13: Percentage of All Women Age 62 or Older by Social Security 
Benefit Status over Time: 

Figure 14: Estimated Percentage of Women Ineligible for Social 
Security Spouse or Widow Benefits Because of Marital History among 50-
59 Year Olds by Race: 1990 and 2009: 

Figure 15: Education Levels of Women Receiving Social Security 
Benefits Based on Their Own Work Record (Retired Worker), Their 
Spouse's Work Record (Spouse-only) or Both (Dually-entitled/Spouse): 
2011: 

Figure 16: Among Married Households Age 55-64 Years Old, Average 
Proportion of Household Retirement Savings (Defined Contribution Plans 
and Individual Retirement Accounts) Contributed by Each Spouse: 2010. 

Figure 17: Among All Households Age 50-64, Percentage of Households 
with Retirement Savings (in Defined Contribution Plans and Individual 
Retirement Accounts) and Among Households with Savings, the Median 
Levels of Savings 2010: 

Figure 18: Projected Distribution of Social Security Beneficiary Type 
by Sex, Age 62 and Older: 2012 to 2040: 

Figure 19: Projected Distribution of Social Security Beneficiary Type 
for Women by Race, Age 62 and Older: 2012 to 2040: 

Figure 20: Examples of Social Security Retirement and Survivor Income 
by Household Structure: 

Abbreviations: 

BLS: Bureau of Labor Statistics: 

Census: U.S. Census Bureau: 

CPS: Current Population Survey: 

DB: defined benefit: 

DC: defined contribution: 

DOMA: Defense of Marriage Act: 

ERISA: Employee Retirement Income Security Act of 1974: 

Federal Reserve: Board of Governors of the Federal Reserve System: 

IPUMS: Integrated Public Use Microdata Series: 

IRA: individual retirement account: 

IRC: Internal Revenue Code: 

QDRO: qualified domestic relations order: 

QJSA: qualified joint and survivor annuity: 

QPSA: qualified preretirement survivor annuity: 

MINT6: Modeling Income in the Near Term model, Version 6: 

SCF: Survey of Consumer Finances: 

SIPP: Survey of Income and Program Participation: 

SSA: Social Security Administration: 

[End of section] 

GAO:
United States Government Accountability Office: 
441 G St. N.W. 
Washington, DC 20548: 

January 15, 2014: 

The Honorable Bill Nelson: 
Chairman: 
Special Committee on Aging: 
United States Senate: 

Dear Chairman Nelson: 

Historically, the institution of marriage has served to help protect 
the financial health of couples, especially couples and surviving 
spouses in old age. Indeed, based on their marriage, spouses may 
receive retirement and survivor income from Social Security and some 
employer-sponsored pension plans.[Footnote 1] More specifically, 
Social Security retirement benefits--the most common type of 
retirement income--are available to individuals who are spouses, 
widows/widowers, and in some cases former spouses, of workers who 
qualify for Social Security retirement benefits based on their work 
history. In addition, private sector defined benefit (DB) plans are 
required by federal law to include a survivor's benefit for spouses. 
[Footnote 2] 

However, many of the policies governing spousal retirement benefits 
were developed at a time when family structures, labor force 
participation patterns, and retirement plans were very different than 
they are today. In recent decades, marriage has become less common, 
and for those who do marry, it is common for both spouses to work. 
Further, private sector employers have shifted away from DB plans to 
defined contribution (DC) plans, which generally confer fewer benefits 
for spouses.[Footnote 3] 

Given these trends, the Senate Special Committee on Aging requested 
that GAO examine the issue of marriage and retirement security. 
Specifically, we examined (1) the trends in and status of marriage and 
labor force participation in American households, (2) how those trends 
have affected spousal benefits and retirement savings behavior within 
households today, and (3) the implications of these trends for future 
retirement security. 

To address these questions, we analyzed several nationally-
representative datasets, conducted an extensive literature review, and 
consulted with a wide range of experts. Specifically, to understand 
trends in family structure and work patterns, we analyzed data from 
the 1996 and 2008 panels of the Survey of Income and Program 
Participation (SIPP); the Integrated Public Use Microdata Series 
(IPUMS); and recent U.S. Census Bureau (Census) and Bureau of Labor 
Statistics (BLS) publications, which report on analyses of nationally-
representative datasets such as the Current Population Survey (CPS). 
To determine the characteristics of the population receiving Social 
Security spousal benefits, we analyzed the 2008 restricted-use version 
of the SIPP data--known as the "Gold Standard" file--SIPP data that 
have been matched to administrative data on earnings and benefits from 
the Social Security Administration (SSA).[Footnote 4] To determine the 
characteristics of the population receiving spousal benefits through 
pensions, we used the 2010 version of the Survey of Consumer Finances 
(SCF), conducted by the Board of Governors of the Federal Reserve 
System (Federal Reserve). We also used the 2010 SCF to analyze 
retirement savings behavior among working-age households. To project 
the distribution of Social Security beneficiary types for women into 
the future, officials from the Office of Retirement Policy at SSA 
provided GAO with projections based on simulations they conducted with 
the Modeling Income in the Near Term model, Version 6 (MINT6). We 
conducted data reliability assessments of selected IPUMS, SIPP, SCF, 
and MINT6 data by conducting electronic data tests for completeness 
and accuracy, reviewing documentation on the dataset, or interviewing 
knowledgeable officials about how the data are collected and 
maintained and their appropriate uses. For the purposes of our 
analysis, we found the variables that we ultimately reported on to be 
sufficiently reliable.[Footnote 5] Lastly, to deepen our understanding 
of our data analyses and to understand the implications of demographic 
trends for retirement security, we conducted an extensive literature 
review and interviewed 49 experts from federal agencies, academia, 
advocacy groups, think tanks, and industry. We identified and 
interviewed experts who have specialized knowledge about spousal 
protections and retirement security in a specific area (e.g., law or 
social science), or who were recommended by one or more people or 
groups we had previously interviewed.[Footnote 6] 

We conducted this performance audit from September 2012 through 
January 2014 in accordance with generally accepted government auditing 
standards. Those standards require that we plan and perform the audit 
to obtain sufficient, appropriate evidence to provide a reasonable 
basis for our findings and conclusions based on our audit objectives. 
We believe that the evidence we obtained provides a reasonable basis 
for our findings and conclusions based on our audit objectives. 

Background: 

Financial Benefits of Marriage: 

A number of financial benefits associated with the institution of 
marriage relate to the ability of a couple to pool resources, 
specialize, and share risk. Specifically, married people can pool 
resources and share costs of household goods and services, which 
reduces the overall cost of items that do not need to be purchased 
individually and frees up resources for other types of investments. In 
addition, within married couples, even those in which both are 
working, each member can specialize in certain tasks, such as managing 
household finances. The financial benefits of marriage can carry over 
into retirement as couples continue to pool resources and assist each 
other as the risks of disability and deteriorating health increase. 

Conversely, several factors contribute to the economic vulnerability 
of single retirees. First, single retirees living alone do not benefit 
from sharing the cost of living expenses and caregiving. According to 
the Census Bureau's poverty measures, a single person age 65 or older 
needs 79 percent of the income of a two-person household.[Footnote 7] 
Moreover, unmarried individuals are more vulnerable to economic 
shocks, such as job loss, than their married counterparts. The 
unmarried, especially single parents, also tend to have fewer 
resources available to save for retirement during their working years. 
For example, a 2009 study shows that among all women age 35-54, never-
married women with children had the lowest level of pension plan 
participation among all family types.[Footnote 8] Taken together, 
these factors are likely to contribute to the higher poverty rates 
observed among unmarried individuals age 65 or older (see figure 1). 

Figure 1: Poverty Rate Estimates of the Population Age 65 and Older, 
by Marital Status and Sex, 2012: 

[Refer to PDF for image: vertical bar graph] 

Percentage below poverty line (estimated): 

Married: 
Men: 4.3%; 
Women: 5.0%; 
U.S. Average: 4.6%. 

Widowed: 
Men: 10.3%; 
Women: 14.3%; 
U.S. Average: 13.5%. 

Divorced: 
Men: 10.9%; 
Women: 18.1%; 
U.S. Average: 15.4%. 

Never married: 
Men: 17.3%; 
Women: 21.0%; 
U.S. Average: 19.4%. 

Source: GAO analysis of Current Population Survey data from the Census 
Bureau. 

Note: All estimates in this figure have 95 percent confidence 
intervals within +/-4.1 percentage points. Estimates of the poverty 
rates for married people have confidence intervals within +/-0.6 
percentage points. 

[End of figure] 

Federal law also confers a number of benefits and protections for 
married spouses or former spouses.[Footnote 9] In the case of certain 
types of retirement income, spouses or former spouses may receive 
retirement and survivor income based exclusively on marriage, 
independent of their own work history. This report focuses on four 
main sources of retirement income: (1) Social Security retirement 
benefits, (2) defined benefit (DB) plans, (3) defined contribution 
(DC) plans, and (4) individual retirement accounts (IRA).[Footnote 10] 
Each of these is subject to different federal requirements and 
provides varying types and levels of protections to spouses of workers 
who are eligible for Social Security, participate in an employer 
sponsored DB or DC plan, or own an IRA. Examples of these requirements 
are summarized in table 1 and described in greater detail below. 

Table 1: Selected Federal Requirements Related to Spousal Benefits or 
Protections in Social Security Retirement Benefits, Defined Benefit 
and Defined Contribution Plans, and Individual Retirement Accounts: 

Source of Retirement Income: Social Security Retirement Benefits; 
Marital Status at Retirement: 
Married: An eligible spouse is entitled to receive up to 50% of the 
retired worker's benefit, unless the spouse is entitled to a higher 
benefit based on his or her own work history[A]; 
Widowed: An eligible surviving spouse is entitled to receive up to 
100% of the deceased worker's benefit, unless the survivor is entitled 
to a higher benefit based on his or her own work history[A]; 
Divorced: If the marriage lasted at least 10 years, an eligible 
divorced spouse is entitled to receive up to 50% of the divorced 
retired worker's benefit (a surviving divorced spouse is entitled to 
receive up to 100% of the deceased worker's benefit), unless the spouse 
is entitled to a retired worker benefit that is higher.[A,B] 

Source of Retirement Income: Defined Benefit (DB) Plans; 
Marital Status at Retirement: 
Married or Widowed: DB plans and some DC plans must provide benefits 
in the form of a qualified joint and survivor annuity (QJSA) or 
qualified preretirement survivor annuity (QPSA), meaning the spouse is 
entitled to receive annuity payments for his or her life should the 
participant spouse die either before or after retirement. The QJSA or 
QPSA may be waived only with the written consent of the spouse, 
witnessed by a plan representative or notary. For the QJSA, the 
survivor benefit must be at least half of the benefit payment the 
participant received during their joint lives; 
Divorced: Benefits provided to a former spouse will depend 
on the plan and the divorce settlement. A DB or DC plan may assign 
benefits to a spouse or former spouse pursuant to a divorce if the 
assignment is governed by a qualified domestic relations order (QDRO). 

Source of Retirement Income: Defined Contribution (DC) Plans; 
Marital Status at Retirement: 
Married or Widowed: DC plans that meet the following criteria are not 
required to offer QPSAs or QJSAs[C]: (1) the plan provides that 100% 
of the participant's vested account balance is to be paid to the 
surviving spouse upon the death of the participant; (2) the 
participant does not elect a life annuity; and (3) the participant's 
account does not include transfers from a plan subject to the standard 
QJSA/QPSA benefit rules (e.g., transfers from a money purchase or a DB 
plan). Written spousal consent is required to designate a non-spousal 
beneficiary. DC plans that meet these criteria may allow the employee, 
during his or her lifetime, to make withdrawals from the account or 
roll over the balance into an IRA without spousal consent[D]; 
Divorced: Benefits provided to a former spouse will depend 
on the plan and the divorce settlement. A DB or DC plan may assign 
benefits to a spouse or former spouse pursuant to a divorce if the 
assignment is governed by a qualified domestic relations order (QDRO). 

Source of Retirement Income: Individual Retirement Accounts; 
Marital Status at Retirement: 
Married or Widowed: There is no federal requirement to provide 
Individual Retirement Account (IRA) benefits to a spouse unless he or 
she is a beneficiary. In general, an IRA owner may, during his or her 
lifetime, make withdrawals from the account without spousal consent.d 
Upon the death of the IRA owner, any remaining assets are to be 
distributed to the designated beneficiaries, if any. If there are 
none, assets will be distributed in accordance with the terms of the 
IRA documents and applicable state inheritance laws[E]; 
Divorced: Benefits provided to a former spouse will depend on the IRA 
documents and divorce settlement. An IRA owner may transfer IRA assets 
tax-free into an IRA for a spouse or former spouse under a divorce 
instrument[D]. 

Source: GAO analysis of applicable federal laws and regulations. 

Note: The specific benefits or protections a person may receive may 
vary depending on the individual circumstances, and additional 
requirements may apply that are not described here. 

This table is limited to the following types of retirement 
income: Old-Age and Survivors Insurance benefits under the Social 
Security Act, DC and DB plans subject to regulation under the Employee 
Retirement Income Security Act and the Internal Revenue Code, and IRAs 
subject to requirements of the Internal Revenue Code. 

[A] Other factors may affect the actual benefit amount. For example, 
actuarial reductions apply if workers/spouses claim benefits before 
full retirement age (which varies depending on the person's year of 
birth), with different size reductions for workers and spouses. 
Delayed retirement credits apply if they wait to collect benefits 
until after full retirement age. 

[B] Special rules apply if a worker has been married more than once. 

[C] Some DC plans, such as money purchase plans, are required to 
comply with the same requirements as DB plans (i.e., provide a QJSA or 
QPSA, and require the spouse's written consent to change this form of 
benefit). 

[D] Pre-retirement withdrawals from DC plans and IRAs may be subject 
to tax penalties. 

[E] Although the Internal Revenue Code provides for certain advantages 
for a spouse who is a designated beneficiary, the IRA owner is not 
required to designate his or her spouse as a beneficiary. There is 
also no federal requirement that an IRA owner obtain spousal consent 
before designating a non-spouse beneficiary. However, in states with 
community property laws, the surviving spouse may be entitled to some 
of the IRA assets if he or she had not consented to the alternate 
beneficiary designation. 

[End of table] 

Social Security: Social Security has been instrumental in reducing 
poverty among the elderly. From 1959 to 2007, the poverty rate for 
people age 65 and over decreased from about 35 percent to 9.7 percent, 
according to U.S. Census Bureau estimates. Experts have attributed 
much of this decline to the availability of Social Security benefits, 
which are also available to spouses and widows. Although Social 
Security is not meant to be the sole source of income for retirees, in 
2011 nearly one-quarter of married beneficiaries and 45 percent of 
unmarried elderly beneficiaries age 65 and older relied on Social 
Security for at least 90 percent of their income, according to the 
Social Security Administration. 

Table 2 below describes selected types of Social Security retirement 
benefits, eligibility criteria, and benefit levels for each type of 
benefit.[Footnote 11] In this report, spousal benefits refer to both 
spouse-only and dually-entitled/spouse benefits. Survivor benefits 
refer to both widow(er)-only and dually-entitled/widow(er) benefits. 
Over the course of retirement, individuals can receive different types 
of Social Security retirement benefits. For example, an individual may 
receive retired-worker benefits while their spouse is alive (because 
their retired-worker benefit exceeds 50 percent of their spouse's 
benefit) but when the spouse dies, may receive a dually-entitled 
widow(er) benefit because their own retired-worker benefit does not 
exceed 100 percent of their spouse's benefit.[Footnote 12] 

Table 2: Selected Social Security Retirement Benefit Types and 
Characteristics: 

Benefit Types: Retired worker; 
Eligibility and Benefit Level[A]: A retired worker is eligible for 
benefits based on his or her work record and receives a benefit amount 
based on lifetime earnings, among other factors[A]; 
Basis for Benefit: Own work history. 

Benefit Types: Spouse-only; 
Eligibility and Benefit Level[A]: The spouse of a retired worker 
(including an eligible divorced spouse) is eligible to receive a 
benefit of up to 50% of the retired worker's benefit amount. Spouse-
only beneficiaries do not qualify to receive benefits based on their 
own work record; for example, because they did not work at all or 
because they did not work long enough; 
Basis for Benefit: Spouse's work history. 

Benefit Types: Dually-entitled/spouse; 
Eligibility and Benefit Level[A]: A dually-entitled spouse (including 
an eligible divorced spouse) has a qualifying work history but 
receives a benefit based on the work records of his or her spouse 
because the spouse's benefit is higher. The benefit amount may be up 
to 50% of the retired worker's benefit amount; 
Basis for Benefit: Has own qualifying work history, but benefit based 
on the spouse's work history is higher. 

Benefit Types: Widower-only; 
Eligibility and Benefit Level[A]: The widower of a retired worker 
(including an eligible surviving divorced spouse) is eligible to 
receive a survivor benefit of up to 100% of the deceased retired 
worker's benefit amount. Widower-only beneficiaries do not qualify to 
receive benefits based on their own work record; for example, because 
they did not work at all or because they did not work long enough; 
Basis for Benefit: Spouse's work history. 

Benefit Types: Dually-entitled/widower; 
Eligibility and Benefit Level[A]: A dually-entitled widower (including 
an eligible surviving divorced spouse) has a qualifying work history 
but receives a survivor benefit based on the work records of his or 
her spouse because the spouse's benefit is higher. The benefit amount 
may be up to 100% of the deceased retired worker's benefit amount; 
Basis for Benefit: Has own qualifying work history, but benefit based 
on the spouse's work history is higher. 

Source: GAO analysis of information provided by SSA and applicable 
federal laws and regulations. 

Note: This table lists some of the benefits available under Social 
Security's Old-Age and Survivors Insurance program. For the purposes 
of this report, this table does not describe other Social Security 
program benefits, such as those based on a worker's disability or for 
dependent children. Divorced spouses and divorced survivors may 
qualify to receive all four types of spousal and survivor benefits 
based on a previous marriage to a retired worker if the marriage 
lasted at least 10 years. For a more comprehensive description of 
spouse and survivor benefits, see Alison M. Shelton and Dawn Nuschler, 
Congressional Research Service, Social Security: Revisiting Benefits 
for Spouses and Survivors (Washington, D.C.: Nov. 5, 2010). 

[A] SSA uses a formula to calculate a retired worker's benefits, which 
typically takes into account the worker's highest 35 years of 
earnings, and adjusts or "indexes" each year of earnings to account 
for changes in average wages since the year the earnings were 
received. Other factors, such as the age of the worker and/or the 
spouse or widow(er), may affect the actual benefit amount for each 
beneficiary. 

[End of table] 

Defined benefit plans: DB plans typically offer benefits to a retiree 
in the form of an annuity that provides a monthly payment for life, 
the value of which is typically determined by a formula based on 
particular factors specified by the plan, such as salary and years of 
service. An annuity can help to protect a retiree against risks, 
including the risk of outliving one's assets (longevity risk) and, 
when an inflation-adjusted annuity is provided, the risk of inflation 
diminishing one's purchasing power. The Retirement Equity Act of 1984 
required that DB plans provide benefits in the form of a qualified 
joint and survivor annuity (QJSA) or qualified preretirement survivor 
annuity (QPSA), unless waived by the worker and his or her spouse. 
[Footnote 13] Some DB plans also give participants a choice to take 
lump-sum cash settlements, which could then be rolled over into an 
IRA, instead of taking a lifetime annuity. 

Defined contribution plans: Under DC plans, workers and employers may 
make contributions to individual accounts. The most common type of DC 
plan is the 401(k) plan, which typically allows workers to contribute 
a portion of their pretax compensation to the plan, and may also 
provide for after-tax or employer contributions, depending on the 
plan. In contrast to DB plans, DC plan sponsors are generally not 
required to offer a lifetime annuity and most plans do not offer an 
annuity option, according to a study conducted for the Department of 
Labor.[Footnote 14] Instead, at retirement, DC participants may take a 
lump sum or partial distribution, leave their money in the plan, roll 
their plan savings into an IRA, or purchase an annuity, depending on 
the options available under the plan. 

Individual Retirement Accounts: Individual retirement accounts (IRAs), 
which may allow workers to receive favorable tax treatment for making 
contributions to an individual account,[Footnote 15] and also 
generally accept rollover distributions from 401(k) and other 
retirement plans, are one of the fastest growing types of retirement 
assets in the United States, according to the Investment Company 
Institute, a national association of U.S. investment companies. As of 
the third quarter of 2013, IRA assets totaled approximately $6 
trillion, accounting for 28 percent of U.S. retirement assets, 
according to the Investment Company Institute. 

Same-Sex Marriages: 

In 2013, same-sex spouses became eligible for a number of federal 
benefits for the first time since the first state allowed same-sex 
marriages in 2004. In June 2013, the U.S. Supreme Court struck down a 
provision of the Defense of Marriage Act (DOMA), which had established 
an opposite-sex definition of "marriage" and "spouse" for any act of 
Congress or any ruling, regulation, or interpretation by federal 
bureaus and agencies.[Footnote 16] As a result of DOMA, enacted in 
1996, same-sex couples married under state law were not covered by 
some of the federal requirements described above; for example, 
sponsors of private-sector pensions regulated by ERISA were not 
required to offer survivor benefits to same-sex spouses. At the time 
of this report's issuance, the Executive Branch had issued some 
guidance and was in the process of determining what additional 
guidance to issue, or other actions to take, in response to the U.S. 
Supreme Court ruling.[Footnote 17] 

State of the U.S. Retirement System and Economy: 

Today, working-age households face a host of challenges in saving for 
retirement. First, less than half (46 percent) of all workers age 16 
and older participated in some form of employment-based pension plan 
in 2012,[Footnote 18] raising concerns about the adequacy of 
retirement savings among future retirees. Second, employers have been 
moving from DB to DC plans. Further, many of the remaining DB plans 
now offer lump sums as one of the form-of-payment options under the 
plan;[Footnote 19] participants who elect a lump sum do so in lieu of 
a lifetime annuity. Some DB plan sponsors have also begun offering 
special, one-time lump sum elections to participants who are already 
retired and receiving monthly pension benefits;[Footnote 20] 
participants who accept such offers give up their lifetime annuity. As 
a result of all of these trends and actions, those who do participate 
in an employer-sponsored plan have increasing responsibility to manage 
their pensions and other financial assets so that they have a 
sufficiently adequate income throughout retirement. However, recent 
research suggests that as life expectancy continues to increase, many 
Americans--including baby boomers, who are beginning to retire in 
greater numbers--may outlive their retirement assets. 

The trends in the U.S. economy overall have exacerbated the challenges 
households face in saving for retirement. The 2007-2009 recession 
resulted in major stock indices falling dramatically, reducing 
retirement savings at a critical time for those in or near retirement. 
Further, the recession led to higher rates of unemployment among older 
workers. Prior GAO work has shown that long-term unemployment can 
reduce an older worker's future retirement income in numerous ways, 
including reducing the number of years the worker can accumulate 
savings, prompting workers to claim Social Security retirement 
benefits early, and leading workers to draw down retirement savings to 
pay for living expenses during unemployment.[Footnote 21] 

The Composition and Work Patterns of the Family Have Changed 
Dramatically in the Past 50 Years: 

As Marriage Has Declined, Cohabitation and Single Parenthood Have 
Increased: 

Since 1960, marriage rates in the United States have been declining. 
From 1960 through 2010, the proportion of the population age 15 and 
above that is married declined from 68 to 54 percent (see figure 2). 
Over the same period, the proportion of divorced individuals increased 
from 2 to 10 percent, and the proportion of individuals who were never 
married climbed from 22 to 31 percent.[Footnote 22] These statistics 
represent a snapshot at a particular point in time. Therefore, those 
who were unmarried at one point in time might marry at a later point 
in time and vice versa. One reason for the rising proportion of the 
never-married is that many people have begun getting married later in 
life. According to the Census Bureau, from 1960 through 2011, the 
median age of first marriage increased from an estimated 20 to 27 for 
women and 23 to 29 for men. 

Figure 2: Estimated Percentage of Population Age 15 and Older by 
Marital Status: 1960-2010: 

[Refer to PDF for image: stacked horizontal bar graph] 

Year: 1960; 
Percentage of population 15 years old and over (estimated): 
Not married: 32%: 
Never married: 22%; 
Widowed: 8%; 
Divorced: 2%; 
Currently married: 68%. 

Year: 1970; 
Percentage of population 15 years old and over (estimated): 
Not married: 36%: 
Never married: 25%; 
Widowed: 8%; 
Divorced: 3%; 
Currently married: 64%. 

Year: 1980; 
Percentage of population 15 years old and over (estimated): 
Not married: 39%: 
Never married: 26%; 
Widowed: 7%; 
Divorced: 6%; 
Currently married: 61%. 

Year: 1990; 
Percentage of population 15 years old and over (estimated): 
Not married: 41%: 
Never married: 26%; 
Widowed: 7%; 
Divorced: 8%; 
Currently married: 59%. 

Year: 2000; 
Percentage of population 15 years old and over (estimated): 
Not married: 44%: 
Never married: 28%; 
Widowed: 6%; 
Divorced: 9%; 
Currently married: 56%. 

Year: 2010; 
Percentage of population 15 years old and over (estimated): 
Not married: 46%: 
Never married: 31%; 
Widowed: 6%; 
Divorced: 10%; 
Currently married: 54%. 

Source: GAO analysis of CPS data. 

Note: All estimates in this figure have margins of error at the 95% 
confidence level that are less than or equal to 0.6 percentage points. 
These statistics represent a snap shot at a particular point in time. 
Therefore, those who were unmarried at one point in time might become 
married at a later point in time and vice versa. Percentages within 
the not married sub-categories may not sum to the not married total 
due to rounding. 

[End of figure] 

Over the past two decades it has also become more common for single 
people to live together before they get married. Specifically, between 
1995 and 2006-2010, the percentage of women who cohabited as a first 
union increased from 34 to 48 percent, according to a study by the 
Department of Health and Human Services.[Footnote 23] Overall, the 
number of unmarried cohabiting couples climbed from 2.9 to 7.8 million 
from 1996 through 2012, according to CPS estimates (see figure 3). A 
significant proportion of these households have children under 18 
years old. These households do not include same-sex partnerships. Due 
to limitations in the way these data are collected, both the number of 
same-sex cohabitating couples and same-sex married couples are 
unclear.[Footnote 24] 

Figure 3: Estimated Number of Unmarried Couples (in Thousands): 1996-
2012: 

[Refer to PDF for image: stacked line graph] 

Number of unmarried partners (estimated): 

Year: 1996; 
With children (under age 18): 1,236; 
Without children (under age 18): 1,623; 
Total: 2,859. 

Year: 1997; 
With children (under age 18): 1,300; 
Without children (under age 18): 1,787; 
Total: 3,087. 

Year: 1998; 
With children (under age 18): 1,295; 
Without children (under age 18): 1,844; 
Total: 3,139. 

Year: 1999; 
With children (under age 18): 1,331; 
Without children (under age 18): 2,048; 
Total: 3,379. 

Year: 2000; 
With children (under age 18): 1,563; 
Without children (under age 18): 2,259; 
Total: 3,822. 

Year: 2001; 
With children (under age 18): 1,665; 
Without children (under age 18): 2,435; 
Total: 4,100. 

Year: 2002; 
With children (under age 18): 1,718; 
Without children (under age 18): 2,475; 
Total: 4,193. 

Year: 2003; 
With children (under age 18): 1,877; 
Without children (under age 18): 2,745; 
Total: 4,622. 

Year: 2004; 
With children (under age 18): 1,953; 
Without children (under age 18): 2,724; 
Total: 4,677. 

Year: 2005; 
With children (under age 18): 1,947; 
Without children (under age 18): 2,928; 
Total: 4,875. 

Year: 2006; 
With children (under age 18): 1,971; 
Without children (under age 18): 3,041; 
Total: 5,012. 

Year: 2007; 
With children (under age 18): 2,505; 
Without children (under age 18): 3,940; 
Total: 6,445. 

Year: 2008; 
With children (under age 18): 2,581; 
Without children (under age 18): 4,217; 
Total: 6,798. 

Year: 2009; 
With children (under age 18): 2,558; 
Without children (under age 18): 4,103; 
Total: 6,661. 

Year: 2010; 
With children (under age 18): 2,843; 
Without children (under age 18): 4,686; 
Total: 7,529. 

Year: 2011; 
With children (under age 18): 3,048; 
Without children (under age 18): 4,548; 
Total: 7,596. 

Year: 2012; 
With children (under age 18): 3,166; 
Without children (under age 18): 4,678; 
Total: 7,844. 

Source: Current Population Survey. 

Note: All estimates in this figure have margins of error at the 95% 
confidence level that are less than or equal to 300,000. It should be 
noted that the jump between 2006 and 2007 occurred because of a change 
in measurement: in 2007, the Census Bureau began asking a direct 
question in the CPS Annual Social and Economic Supplement about the 
presence of unmarried partners. According to Census Bureau analyses, 
the increase between 2009 and 2010 appears to reflect a true increase. 

[End of figure] 

Along with the increase in unmarried couples with children over the 
last several decades, the percentage of single-parent families rose. 
In fact, according to CPS estimates, from 1970 through 2012, the 
proportion of single-parent families more than doubled, increasing 
from 13 to 32 percent of all families. The vast majority of single-
parent households were headed by women (see figure 4). 

Figure 4: Among All Families, Estimated Percentage of Single-Parent 
Families by Sex of Head of Household: 1970-2012: 

[Refer to PDF for image: stacked line graph] 

Percentage of all families that are headed by single parents 
(estimated): 

Year: 1970; 
Headed by single women: 11.5%; 
Headed by single men: 1.3%; 
Total: 12.8%. 

No data between 1970 and 1980. 

Year: 1980; 
Headed by single women: 19.4%; 
Headed by single men: 2.1%; 
Total: 21.5%. 

Year: 1981; 
Headed by single women: 20%; 
Headed by single men: 2.4%; 
Total: 22.4%. 

Year: 1982; 
Headed by single women: 22.2%; 
Headed by single men: 2.4%; 
Total: 24.6%. 

Year: 1983; 
Headed by single women: 22.4%; 
Headed by single men: 2.7%; 
Total: 25.1%. 

Year: 1984; 
Headed by single women: 22.9%; 
Headed by single men: 2.8%; 
Total: 25.7%. 

Year: 1985; 
Headed by single women: 23.2%; 
Headed by single men: 3.1%; 
Total: 26.3%. 

Year: 1986; 
Headed by single women: 23.1%; 
Headed by single men: 3.2%; 
Total: 26.3%. 

Year: 1987; 
Headed by single women: 23.7%; 
Headed by single men: 3.2%; 
Total: 26.9%. 

Year: 1988; 
Headed by single women: 23.7%; 
Headed by single men: 3.6%; 
Total: 27.3%. 

Year: 1989; 
Headed by single women: 24%; 
Headed by single men: 3.5%; 
Total: 27.5%. 

Year: 1990; 
Headed by single women: 24.2%; 
Headed by single men: 3.9%; 
Total: 28.1%. 

Year: 1991; 
Headed by single women: 25%; 
Headed by single men: 3.9%; 
Total: 28.9%. 

Year: 1992; 
Headed by single women: 25.5%; 
Headed by single men: 4.2%; 
Total: 29.7%. 

Year: 1993; 
Headed by single women: 25.9%; 
Headed by single men: 4.3%; 
Total: 30.2%. 

Year: 1994; 
Headed by single women: 26.6%; 
Headed by single men: 4.2%; 
Total: 30.8%. 

Year: 1995; 
Headed by single women: 26.5%; 
Headed by single men: 4.6%; 
Total: 31.1%. 

Year: 1996; 
Headed by single women: 26.6%; 
Headed by single men: 5%; 
Total: 31.6%. 

Year: 1997; 
Headed by single women: 26.6%; 
Headed by single men: 5.4%; 
Total: 32%. 

Year: 1998; 
Headed by single women: 26.1%; 
Headed by single men: 5.6%; 
Total: 31.7%. 

Year: 1999; 
Headed by single women: 26.3%; 
Headed by single men: 5.5%; 
Total: 31.8%. 

Year: 2000; 
Headed by single women: 25.8%; 
Headed by single men: 5.5%; 
Total: 31.3%. 

Year: 2001; 
Headed by single women: 25.2%; 
Headed by single men: 5.6%; 
Total: 30.8%. 

Year: 2002; 
Headed by single women: 25.9%; 
Headed by single men: 5.8%; 
Total: 31.7%. 

Year: 2003; 
Headed by single women: 26.1%; 
Headed by single men: 5.8%; 
Total: 31.9%. 

Year: 2004; 
Headed by single women: 26.5%; 
Headed by single men: 5.9%; 
Total: 32.4%. 

Year: 2005; 
Headed by single women: 26.4%; 
Headed by single men: 6.3%; 
Total: 29.7%. 

Year: 2006; 
Headed by single women: 26.4%; 
Headed by single men: 6.4%; 
Total: 32.8%. 

Year: 2007; 
Headed by single women: 24.9%; 
Headed by single men: 4.4%; 
Total: 33.3%. 

Year: 2008; 
Headed by single women: 25%; 
Headed by single men: 4.7%; 
Total: 29.7%. 

Year: 2009; 
Headed by single women: 25.4%; 
Headed by single men: 4.5%; 
Total: 29.9%. 

Year: 2010; 
Headed by single women: 25.6%; 
Headed by single men: 4.5%; 
Total: 30.1%. 

Year: 2011; 
Headed by single women: 26.2%; 
Headed by single men: 4.5%; 
Total: 30.7%. 

Year: 2012; 
Headed by single women: 26.8%; 
Headed by single men: 5.1%; 
Total: 31.9%. 

Source: GAO analysis of Current Population Survey data. 

Note: All estimates in this figure have margins of error at the 95% 
confidence level that are less than or equal to 0.4 percentage points. 
Note that the drop seen in 2007 occurs because of measurement changes 
in the CPS Annual Social and Economic Supplement. Prior to 2007, the 
data show all unmarried parents. Specifically, the data prior to 2007 
include both unmarried parents who are cohabiting with the parent of 
at least one of their children, and "lone parents"--unmarried parents 
who are not cohabiting with the parent of any of their children. 
Starting in 2007, the data identify unmarried parents who are 
cohabiting with the parent of at least one of their children. These 
cohabiting parents were excluded from the data, so only lone parent 
families are shown in the chart starting in 2007. 

[End of figure] 

The Decline in Marriage Is More Pronounced for Certain Demographic 
Groups: 

The decline in marriage has been more pronounced for certain racial 
and ethnic groups as well as income and education levels.[Footnote 25] 
Among racial and ethnic groups, blacks have experienced the greatest 
decline in marriage (see figure 5). From 1960 through 2010, the 
percentage of married black men fell from 61 to 39 and the percentage 
of married black women fell from 60 to 33. Differences in marriage 
rates between blacks and other racial and ethnic categories persisted 
even after accounting for differences in income and education levels, 
and controlling for age.[Footnote 26] Asians were the only group, 
among these racial and ethnic groups, to experience an uptick in 
marriage since 2000. 

Figure 5: Estimated Percentage of Married Individuals by Race and 
Gender over Time: 1960-2010: 

[Refer to PDF for image: 2 multiple line graphs] 

Percentage of married men (estimated): 

Year: 1960; 
White: 70%; 
Black: 61%. 

Year: 1970; 
White: 68%; 
Black: 57%. 

Year: 1980; 
White: 65%; 
Black: 49%. 

Year: 1990; 
White: 63%; 
Black: 45%; 
Hispanic: 57%; 
Asian: 61%. 

Year: 2000; 
White: 60%; 
Black: 43%; 
Hispanic: 55%; 
Asian: 58%. 

Year: 2010; 
White: 57%; 
Black: 39%; 
Hispanic: 49%; 
Asian: 63%. 

Percentage of married women (estimated): 

Year: 1960; 
White: 67%; 
Black: 60%. 

Year: 1970; 
White: 63%; 
Black: 54%. 

Year: 1980; 
White: 61%; 
Black: 45%. 

Year: 1990; 
White: 59%; 
Black: 40%; 
Hispanic: 57%; 
Asian: 63%. 

Year: 2000; 
White: 57%; 
Black: 36%; 
Hispanic: 57%; 
Asian: 59%. 

Year: 2010; 
White: 55%; 
Black: 33%; 
Hispanic: 52%; 
Asian: 63%. 

Source: GAO analysis of CPS data. 

Note: Individuals in the white, black, and Asian race categories may 
be Hispanic. Prior to 2003, individuals in the Current Population 
Survey were required to select only one race. Beginning in 2003, 
individuals were allowed to select multiple race groups. Starting in 
2003, the race categories in this chart include only people who 
reported only a single race group. Percentage estimates in this figure 
have margins of error at the 95% confidence level that are less than 
or equal to 0.6 percentage points for whites; 2.4 percentage points 
for blacks; 4 percentage points for Asians; and 2.2 percentage points 
for Hispanics. 

[End of figure] 

Since the 1970s, the percentage of single-parent families has 
increased dramatically (see figure 6). Specifically, from 1970 through 
2012 (and from 1980 through 2012 in the case of Hispanics), the 
percentage of single-parent families climbed for every race and 
ethnicity category depicted in the figure, with the highest growth 
among whites (an increase of 17 percentage points, for a 170 percent 
increase in the percentage). However, over the entire period, the rate 
of single parenthood was highest among black families and continues to 
be more than double the rate of white families. 

Figure 6: Estimated Percentage of Families Headed by a Single Parent 
by Race: 1970-2012: 

[Refer to PDF for image: multiple line graph] 

Year: 1970; 
White: 10%; 
Black: 36%; 
Hispanic: 26%. 

Note: No data between 1970 and 1980, 

Year: 1980; 
White: 17%; 
Black: 52%; 
Hispanic: 26. 

Year: 1981; 
White: 18%; 
Black: 53%; 
Hispanic: 28. 

Year: 1982; 
White: 20%; 
Black: 55%; 
Hispanic: 29. 

Year: 1983; 
White: 20%; 
Black: 59%; 
Hispanic: 30. 

Year: 1984; 
White: 20%; 
Black: 59%; 
Hispanic: 30%. 

Year: 1985; 
White: 21%; 
Black: 60%; 
Hispanic: 31. 

Year: 1986; 
White: 21%; 
Black: 58%; 
Hispanic: 32. 

Year: 1987; 
White: 22%; 
Black: 59%; 
Hispanic: 33. 

Year: 1988; 
White: 22%; 
Black: 59%; 
Hispanic: 34. 

Year: 1989; 
White: 22%; 
Black: 61%; 
Hispanic: 32. 

Year: 1990; 
White: 23%; 
Black: 61%; 
Hispanic: 33%. 

Year: 1991; 
White: 23%; 
Black: 63%; 
Hispanic: 33%. 

Year: 1992; 
White: 24%; 
Black: 62%; 
Hispanic: 34%. 

Year: 1993; 
White: 25%; 
Black: 63%; 
Hispanic: 35%. 

Year: 1994; 
White: 25%; 
Black: 65%; 
Hispanic: 36%. 

Year: 1995; 
White: 25%; 
Black: 64%; 
Hispanic: 36%. 

Year: 1996; 
White: 26%; 
Black: 64%; 
Hispanic: 37%. 

Year: 1997; 
White: 26%; 
Black: 64%; 
Hispanic: 36%. 

Year: 1998; 
White: 27%; 
Black: 62%; 
Hispanic: 36%. 

Year: 1999; 
White: 27%; 
Black: 63%; 
Hispanic: 35%. 

Year: 2000; 
White: 26%; 
Black: 61%; 
Hispanic: 34%. 

Year: 2001; 
White: 26%; 
Black: 61%; 
Hispanic: 34%. 

Year: 2002; 
White: 27%; 
Black: 61%; 
Hispanic: 35%. 

Year: 2003; 
White: 27%; 
Black: 62%; 
Hispanic: 35%. 

Year: 2004; 
White: 27%; 
Black: 63%; 
Hispanic: 35%. 

Year: 2005; 
White: 28%; 
Black: 62%; 
Hispanic: 36%. 

Year: 2006; 
White: 28%; 
Black: 63%; 
Hispanic: 34%. 

Year: 2007; 
White: 25%; 
Black: 58%; 
Hispanic: 31%. 

Year: 2008; 
White: 25%; 
Black: 60%; 
Hispanic: 30%. 

Year: 2009; 
White: 25%; 
Black: 60%; 
Hispanic: 32%. 

Year: 2010; 
White: 26%; 
Black: 59%; 
Hispanic: 34%. 

Year: 2011; 
White: 26%; 
Black: 60%; 
Hispanic: 33%. 

Year: 2012; 
White: 27%; 
Black: 60%; 
Hispanic: 34%. 

Source: GAO analysis of Current Population Survey data. 

Note: Individuals in the white and black race categories may be 
Hispanic. Prior to 2003, individuals in the Current Population Survey 
were required to select only one race. Beginning in 2003, individuals 
were allowed to select multiple race groups. Starting in 2003, the 
race categories in this chart include only people who reported only a 
single race group. Percentage estimates in this figure have margins of 
error at the 95% confidence level that are less than or equal to 1.8 
percentage points. Note that the drop seen in 2007 occurs because of 
measurement changes in the CPS Annual Social and Economic Supplement. 
Specifically, the data prior to 2007 include both unmarried parents 
who are cohabiting with the parent of at least one of their children, 
and "lone parents"--unmarried parents who are not cohabiting with the 
parent of any of their children. Starting in 2007, the data identify 
unmarried parents who are cohabiting with the parent of at least one 
of their children. These cohabiting parents were excluded from the 
data, so only lone parent families are shown in the chart starting in 
2007. 

[End of figure] 

The decline in marriage and rise in single parenthood over this period 
were also more pronounced among low-income and less-educated 
populations. Over the period from 1960 through 2010 the percentage of 
married, 45-to 54-year-olds in the highest income quintile declined 
slightly for men (see figure 7). In contrast, over the same period, 
the percentage of married, 45-to 54-year-old men and women in the 
lowest income quintile declined markedly. 

Similarly, the percentage of single parents among 45- to 54-year-old 
men and women in the highest income quintile remained flat, while 
there was a steep rise in the percentage of single parents in the 
lowest income quintile during this timeframe. For education, among 
individuals age 18 years and older, the decline in the percentage of 
married individuals was steeper for those with less than a high school 
diploma than for those with 4 or more years of college. Conversely, 
the rise in single parenthood was steeper for women and men without a 
high school diploma in comparison to their counterparts with 4 or more 
years of college (see figure 8). Nonetheless, the rate of single 
parenthood among college-educated women was greater than that of men 
with less than a high school diploma for most of the period, 
reflecting the higher rate of single parenthood among women overall. 

Figure 7: Among People Age 45-54, Estimated Percentage of Married Men 
and Women, and Unmarried Parents, by Highest and Lowest Income 
Quintiles: 1960-2010: 

[Refer to PDF for image: 2 multiple line graphs] 

Individuals who are married (estimated percentage): 

Year: 1960;
Men with the lowest income: 71%;
Women with the lowest income: 51%;
Men with the highest income: 86.6%;
Women with the highest income: 86.3%. 

Year: 1970;
Men with the lowest income: 62.1%;
Women with the lowest income: 47%;
Men with the highest income: 9%;
Women with the highest income: 91%. 

Year: 1980;
Men with the lowest income: 53.3%;
Women with the lowest income: 40%;
Men with the highest income: 91%;
Women with the highest income: 93%. 

Year: 1990;
Men with the lowest income: 45%;
Women with the lowest income: 33%;
Men with the highest income: 89%;
Women with the highest income: 92%. 

Year: 2000;
Men with the lowest income: 35%;
Women with the lowest income: 28.5%;
Men with the highest income: 88%;
Women with the highest income: 91%. 

Year: 2010;
Men with the lowest income: 27%;
Women with the lowest income: 24%;
Men with the highest income: 81.5%;
Women with the highest income: 88%. 

Parents who are currently unmarried (estimated percentage): 

Year: 1960;
Men with the lowest income: 4%;
Women with the lowest income: 33%;
Men with the highest income: 1%;
Women with the highest income: 4%. 

Year: 1970;
Men with the lowest income: 7%;
Women with the lowest income: 48%;
Men with the highest income: 1%;
Women with the highest income: 4%. 

Year: 1980;
Men with the lowest income: 11%;
Women with the lowest income: 55%;
Men with the highest income: 2%;
Women with the highest income: 4%. 

Year: 1990;
Men with the lowest income: 17%;
Women with the lowest income: 62%;
Men with the highest income: 3%;
Women with the highest income: 5%. 

Year: 2000;
Men with the lowest income: 25%;
Women with the lowest income: 63%;
Men with the highest income: 3%;
Women with the highest income: 5%. 

Year: 2010;
Men with the lowest income: 34%;
Women with the lowest income: 68%;
Men with the highest income: 3%;
Women with the highest income: 6%. 

Source: GAO analysis of IPUMS data. 

Note: Percentage estimates in this figure have margins of error at the 
95% confidence level that are less than or equal to 1.3 percentage 
points. "Parents who are currently unmarried" here refers to any adult 
with their own children in the home whose current marital status is 
divorced, separated, widowed, or never married. 

[End of figure] 

Figure 8: Among People Age 18 and Older, Estimated Percentage of 
Married Men and Women, and Unmarried Parents, by Education Level: 1960-
2010: 

[Refer to PDF for image: 2 multiple line graphs] 

Individuals who are married (estimated percentage): 

Year: 1960; 
Men with less than high school diploma: 75%; 
Men with four or more years of college: 82%; 
Women with four or more years of college: 67%; 
Women with less than high school diploma: 67%. 

Year: 1970; 
Men with less than high school diploma: 73%; 
Men with four or more years of college: 81%; 
Women with four or more years of college: 67%; 
Women with less than high school diploma: 60%. 

Year: 1980; 
Men with less than high school diploma: 66%; 
Men with four or more years of college: 73%; 
Women with four or more years of college: 63%; 
Women with less than high school diploma: 53%. 

Year: 1990; 
Men with less than high school diploma: 59%; 
Men with four or more years of college: 71%; 
Women with four or more years of college: 61%; 
Women with less than high school diploma: 45%. 

Year: 2000; 
Men with less than high school diploma: 55%; 
Men with four or more years of college: 70%; 
Women with four or more years of college: 62%; 
Women with less than high school diploma: 44%. 

Year: 2010; 
Men with less than high school diploma: 45%; 
Men with four or more years of college: 68%; 
Women with four or more years of college: 60%; 
Women with less than high school diploma: 39%. 

Parents who are currently unmarried (estimated percentage): 

Year: 1960; 
Men with less than high school diploma: 6%; 
Men with four or more years of college: 1%; 
Women with four or more years of college: 22%; 
Women with less than high school diploma: 8%. 

Year: 1970; 
Men with less than high school diploma: 6%; 
Men with four or more years of college: 1%; 
Women with four or more years of college: 28%; 
Women with less than high school diploma: 10%. 

Year: 1980; 
Men with less than high school diploma: 8%; 
Men with four or more years of college: 2%; 
Women with four or more years of college: 36%; 
Women with less than high school diploma: 13%. 

Year: 1990; 
Men with less than high school diploma: 12%; 
Men with four or more years of college: 4%; 
Women with four or more years of college: 43%; 
Women with less than high school diploma: 15%. 

Year: 2000; 
Men with less than high school diploma: 16%; 
Men with four or more years of college: 5%; 
Women with four or more years of college: 44%; 
Women with less than high school diploma: 17%. 

Year: 2010; 
Men with less than high school diploma: 20%; 
Men with four or more years of college: 6%; 
Women with four or more years of college: 49%; 
Women with less than high school diploma: 19%. 

Source: GAO analysis of IPUMS data. 

[End of figure] 

Married Women Are Participating in the Labor Force at Higher Rates and 
Are Taking Less Time Off for Caregiving: 

Labor force participation among married and single men and women 
changed substantially over the past 50 years (see figure 9). In 1960, 
labor force participation rates among the four groups (married men, 
single men, married women, and single women) ranged from 89 percent 
for married men to 32 percent for married women. The variation in 
workforce participation rates between single men (70 percent) and 
single women (59 percent) was not as wide, but was still substantial. 
Since then, the differences in labor force participation rates for 
these four groups have narrowed. In 2010, the rate was 76 percent for 
married men, 67 percent for single men, 63 percent for single women, 
and 61 percent for married women. However, as figure 9 shows, the 
labor force participation rates of married women have leveled off 
since the mid-nineties, and have declined for single women, single 
men, and married men over the last decade.[Footnote 27] 

Figure 9: Estimated Labor Force Participation Rates by Sex and Marital 
Status: 1960-2010: 

[Refer to PDF for image: multiple line graph] 

Percentage participating in the workforce (estimated): 

Year: 1960; 
Single men: 69.8%; 
Married men: 89.2%; 
Single women: 58.6%; 
Married women: 31.9%. 

Year: 1970; 
Single men: 65.5%; 
Married men: 86.1%; 
Single women: 56.8%; 
Married women: 40.5%. 

Year: 1975; 
Single men: 68.7%; 
Married men: 83%; 
Single women: 59.8%; 
Married women: 44.3%. 

Year: 1980; 
Single men: 72.6%; 
Married men: 80.9%; 
Single women: 64.4%; 
Married women: 49.8%. 

Year: 1985; 
Single men: 73.8%; 
Married men: 78.7%; 
Single women: 66.6%; 
Married women: 53.8%. 

Year: 1990; 
Single men: 74.8%; 
Married men: 78.6%; 
Single women: 66.7%; 
Married women: 58.4%. 

Year: 1995; 
Single men: 73.7%; 
Married men: 77.5%; 
Single women: 66.8%; 
Married women: 61%. 

Year: 2000; 
Single men: 73.6%; 
Married men: 77.3%; 
Single women: 68.9%; 
Married women: 61.1%. 

Year: 2005; 
Single men: 70.1%; 
Married men: 77.2%; 
Single women: 66%; 
Married women: 60.7%. 

Year: 2010; 
Single men: 67.3%; 
Married men: 75.8%; 
Single women: 63.3%; 
Married women: 61%. 

Source: Current Population Survey. 

Note: Percentage estimates in this figure have margins of error at the 
95% confidence level that are less than or equal to 1.2 percentage 
points. The categories for single men and single women include all 
unmarried people, including those who are divorced, widowed, or who 
were never married. 

[End of figure] 

Labor force participation rates for women with children followed a 
similar trajectory to those of married women. From 1975 through 2010, 
labor force participation rates for women with children under age 3 
grew from 34 to 61 percent (see figure 10).[Footnote 28] For women 
with children under age 6, the rate grew from 39 to 64 percent. As in 
the case of married women, labor force participation rates for each of 
these groups have remained relatively level since the mid-1990s. 
[Footnote 29] 

Figure 10: Estimated Labor Force Participation Rates among Women with 
Children of Different Ages: 1975-2010: 

[Refer to PDF for image: multiple line graph] 

Year: 1975; 
Women with children who were ages 6 to 17: 55%; 
Women with children who were younger than 6: 39%; 
Women with children who were younger than 3: 34%. 

Year: 1976; 
Women with children who were ages 6 to 17: 56%; 
Women with children who were younger than 6: 40%; 
Women with children who were younger than 3: 34%. 

Year: 1977; 
Women with children who were ages 6 to 17: 58%; 
Women with children who were younger than 6: 41%; 
Women with children who were younger than 3: 35%. 

Year: 1978; 
Women with children who were ages 6 to 17: 60%; 
Women with children who were younger than 6: 44%; 
Women with children who were younger than 3: 39%. 

Year: 1979; 
Women with children who were ages 6 to 17: 62%; 
Women with children who were younger than 6: 46%; 
Women with children who were younger than 3: 41%. 

Year: 1980; 
Women with children who were ages 6 to 17: 64%; 
Women with children who were younger than 6: 47%; 
Women with children who were younger than 3: 42%. 

Year: 1981; 
Women with children who were ages 6 to 17: 66%; 
Women with children who were younger than 6: 49%; 
Women with children who were younger than 3: 44%. 

Year: 1982; 
Women with children who were ages 6 to 17: 66%; 
Women with children who were younger than 6: 50%; 
Women with children who were younger than 3: 46%. 

Year: 1983; 
Women with children who were ages 6 to 17: 66%; 
Women with children who were younger than 6: 51%; 
Women with children who were younger than 3: 46%. 

Year: 1984; 
Women with children who were ages 6 to 17: 68%; 
Women with children who were younger than 6: 52%; 
Women with children who were younger than 3: 48%. 

Year: 1985; 
Women with children who were ages 6 to 17: 70%; 
Women with children who were younger than 6: 54%; 
Women with children who were younger than 3: 50%. 

Year: 1986; 
Women with children who were ages 6 to 17: 70%; 
Women with children who were younger than 6: 54%; 
Women with children who were younger than 3: 51%. 

Year: 1987; 
Women with children who were ages 6 to 17: 72%; 
Women with children who were younger than 6: 57%; 
Women with children who were younger than 3: 53%. 

Year: 1988; 
Women with children who were ages 6 to 17: 73%; 
Women with children who were younger than 6: 56%; 
Women with children who were younger than 3: 52%. 

Year: 1989; 
Women with children who were ages 6 to 17: 74%; 
Women with children who were younger than 6: 57%; 
Women with children who were younger than 3: 52%. 

Year: 1990; 
Women with children who were ages 6 to 17: 75%; 
Women with children who were younger than 6: 58%; 
Women with children who were younger than 3: 54%. 

Year: 1991; 
Women with children who were ages 6 to 17: 74%; 
Women with children who were younger than 6: 58%; 
Women with children who were younger than 3: 55%. 

Year: 1992; 
Women with children who were ages 6 to 17: 76%; 
Women with children who were younger than 6: 58%; 
Women with children who were younger than 3: 55%. 

Year: 1993; 
Women with children who were ages 6 to 17: 75%; 
Women with children who were younger than 6: 58%; 
Women with children who were younger than 3: 54%. 

Year: 1994[A]; 
Women with children who were ages 6 to 17: 76%; 
Women with children who were younger than 6: 60%; 
Women with children who were younger than 3: 57%. 

Year: 1995; 
Women with children who were ages 6 to 17: 76%; 
Women with children who were younger than 6: 62%; 
Women with children who were younger than 3: 59%. 

Year: 1996; 
Women with children who were ages 6 to 17: 77%; 
Women with children who were younger than 6: 62%; 
Women with children who were younger than 3: 59%. 

Year: 1997; 
Women with children who were ages 6 to 17: 78%; 
Women with children who were younger than 6: 65%; 
Women with children who were younger than 3: 62%. 

Year: 1998; 
Women with children who were ages 6 to 17: 78%; 
Women with children who were younger than 6: 65%; 
Women with children who were younger than 3: 62%. 

Year: 1999; 
Women with children who were ages 6 to 17: 79%; 
Women with children who were younger than 6: 64%; 
Women with children who were younger than 3: 61%. 

Year: 2000; 
Women with children who were ages 6 to 17: 79%; 
Women with children who were younger than 6: 65%; 
Women with children who were younger than 3: 61%. 

Year: 2001; 
Women with children who were ages 6 to 17: 79%; 
Women with children who were younger than 6: 64%; 
Women with children who were younger than 3: 61%. 

Year: 2002; 
Women with children who were ages 6 to 17: 79%; 
Women with children who were younger than 6: 64%; 
Women with children who were younger than 3: 61%. 

Year: 2003; 
Women with children who were ages 6 to 17: 79%; 
Women with children who were younger than 6: 63%; 
Women with children who were younger than 3: 59%. 

Year: 2004; 
Women with children who were ages 6 to 17: 78%; 
Women with children who were younger than 6: 62%; 
Women with children who were younger than 3: 57%. 

Year: 2005; 
Women with children who were ages 6 to 17: 77%; 
Women with children who were younger than 6: 63%; 
Women with children who were younger than 3: 59%. 

Year: 2006; 
Women with children who were ages 6 to 17: 77%; 
Women with children who were younger than 6: 63%; 
Women with children who were younger than 3: 60%. 

Year: 2007; 
Women with children who were ages 6 to 17: 78%; 
Women with children who were younger than 6: 64%; 
Women with children who were younger than 3: 60%. 

Year: 2008; 
Women with children who were ages 6 to 17: 78%; 
Women with children who were younger than 6: 64%; 
Women with children who were younger than 3: 60%. 

Year: 2009; 
Women with children who were ages 6 to 17: 78%; 
Women with children who were younger than 6: 64%; 
Women with children who were younger than 3: 61%. 

Year: 2010; 
Women with children who were ages 6 to 17: 77%; 
Women with children who were younger than 6: 64%; 
Women with children who were younger than 3: 61%. 

Year: 2011; 
Women with children who were ages 6 to 17: 77%; 
Women with children who were younger than 6: 64%; 
Women with children who were younger than 3: 61%. 

Source: Annual Social and Economic Supplements, Current Population 
Survey. 

Note: Children are "own" children and include sons, daughters, 
stepchildren, and adopted children. Not included are nieces, nephews, 
grandchildren, and other related and unrelated children. Percentage 
estimates in this figure have margins of error at the 95% confidence 
level that are less than or equal to 1.8 percentage points. 

[A] Data for 1994 and subsequent years are not directly comparable 
with data for 1993 and earlier because of the introduction of a major 
redesign of the Current Population Survey. 

[End of figure] 

As a result of married women's increasing labor force participation, 
the proportion of married couples with two earners has risen--along 
with the wives' contributions to household income. The proportion of 
households in which both the husband and wife contributed to their 
household's income rose from 44 percent in 1967 to 60 percent in 1994 
(see figure 11). The proportion remained at 60 percent through 2000, 
after which it declined to 54 percent by 2010. Part of this is likely 
explained by the stagnation in married women's labor force 
participation rates over that period, but part of this also may be due 
to the rise in the percentage of married households in which only the 
wife works. Since 1967, the proportion of married households in which 
the wife was the sole earner rose more than threefold--from 2 to 7 
percent. At the same time, the proportion of married households in 
which the husband was the sole earner declined from 36 to 19 percent. 
Women's contributions to family income also grew steadily over this 
period. According to the Bureau of Labor Statistics, from 1970 through 
2010, women's median contribution to household income rose from 27 to 
38 percent. Further, from 1987 through 2010, the percentage of 
households in which the wives' earnings exceeded their husband's rose 
from 24 to 38 percent.[Footnote 30] 

Figure 11: Among Married Households, Estimated Percentage of Married 
Households in Which Both Husband and Wife Have Earnings, Only the 
Husband Has Earnings, and Only the Wife Has Earnings: 1967-2010: 

Percentage of all households (estimated): 

Year: 1967; 
Wife only: 2%; 
Husband only: 36%; 
Dual-earner (both husband and wife): 44%. 

Year: 1968; 
Wife only: 2%; 
Husband only: 35%; 
Dual-earner (both husband and wife): 45%. 

Year: 1969; 
Wife only: 2%; 
Husband only: 34%; 
Dual-earner (both husband and wife): 46%. 

Year: 1970; 
Wife only: 2%; 
Husband only: 33%; 
Dual-earner (both husband and wife): 46%. 

Year: 1971; 
Wife only: 2%; 
Husband only: 34%; 
Dual-earner (both husband and wife): 45%. 

Year: 1972; 
Wife only: 2%; 
Husband only: 33%; 
Dual-earner (both husband and wife): 46%. 

Year: 1973; 
Wife only: 2%; 
Husband only: 31%; 
Dual-earner (both husband and wife): 47%. 

Year: 1974; 
Wife only: 3%; 
Husband only: 30%; 
Dual-earner (both husband and wife): 47%. 

Year: 1975; 
Wife only: 3%; 
Husband only: 30%; 
Dual-earner (both husband and wife): 47%. 

Year: 1976; 
Wife only: 3%; 
Husband only: 28%; 
Dual-earner (both husband and wife): 48%. 

Year: 1977; 
Wife only: 3%; 
Husband only: 27%; 
Dual-earner (both husband and wife): 49%. 

Year: 1978; 
Wife only: 3%; 
Husband only: 26%; 
Dual-earner (both husband and wife): 51%. 

Year: 1979; 
Wife only: 3%; 
Husband only: 24%; 
Dual-earner (both husband and wife): 52%. 

Year: 1980; 
Wife only: 4%; 
Husband only: 24%; 
Dual-earner (both husband and wife): 52%. 

Year: 1981; 
Wife only: 3%; 
Husband only: 23%; 
Dual-earner (both husband and wife): 52%. 

Year: 1982; 
Wife only: 4%; 
Husband only: 23%; 
Dual-earner (both husband and wife): 51%. 

Year: 1983; 
Wife only: 4%; 
Husband only: 22%; 
Dual-earner (both husband and wife): 52%. 

Year: 1984; 
Wife only: 4%; 
Husband only: 21%; 
Dual-earner (both husband and wife): 54%. 

Year: 1985; 
Wife only: 4%; 
Husband only: 20%; 
Dual-earner (both husband and wife): 55%. 

Year: 1986; 
Wife only: 4%; 
Husband only: 19%; 
Dual-earner (both husband and wife): 56%. 

Year: 1987; 
Wife only: 4%; 
Husband only: 19%; 
Dual-earner (both husband and wife): 57%. 

Year: 1988; 
Wife only: 3%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 59%. 

Year: 1989; 
Wife only: 4%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 59%. 

Year: 1990; 
Wife only: 4%; 
Husband only: 17%; 
Dual-earner (both husband and wife): 59%. 

Year: 1991; 
Wife only: 4%; 
Husband only: 17%; 
Dual-earner (both husband and wife): 59%. 

Year: 1992; 
Wife only: 4%; 
Husband only: 17%; 
Dual-earner (both husband and wife): 59%. 

Year: 1993; 
Wife only: 5%; 
Husband only: 16%; 
Dual-earner (both husband and wife): 59%. 

Year: 1994; 
Wife only: 4%; 
Husband only: 16%; 
Dual-earner (both husband and wife): 60%. 

Year: 1995; 
Wife only: 4%; 
Husband only: 17%; 
Dual-earner (both husband and wife): 60%. 

Year: 1996; 
Wife only: 4%; 
Husband only: 16%; 
Dual-earner (both husband and wife): 60%. 

Year: 1997; 
Wife only: 4%; 
Husband only: 16%; 
Dual-earner (both husband and wife): 60%. 

Year: 1998; 
Wife only: 4%; 
Husband only: 17%; 
Dual-earner (both husband and wife): 60%. 

Year: 1999; 
Wife only: 5%; 
Husband only: 16%; 
Dual-earner (both husband and wife): 60%. 

Year: 2000; 
Wife only: 5%; 
Husband only: 17%; 
Dual-earner (both husband and wife): 60%. 

Year: 2001; 
Wife only: 5%; 
Husband only: 17%; 
Dual-earner (both husband and wife): 59%. 

Year: 2002; 
Wife only: 5%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 59%. 

Year: 2003; 
Wife only: 5%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 58%. 

Year: 2004;
Wife only: 5%; 
Husband only: 19%; 
Dual-earner (both husband and wife): 57%. 

Year: 2005; 
Wife only: 5%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 57%. 

Year: 2006; 
Wife only: 6%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 57%. 

Year: 2007; 
Wife only: 6%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 58%. 

Year: 2008; 
Wife only: 6%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 57%. 

Year: 2009; 
Wife only: 7%; 
Husband only: 18%; 
Dual-earner (both husband and wife): 55%. 

Year: 2010; 
Wife only: 7%; 
Husband only: 19%; 
Dual-earner (both husband and wife): 54%. 

Source: Current Population Survey data, Annual Social and Economic 
Supplements. 

Note: Married households in which neither spouse works and married 
households in which additional earners contribute to family income are 
not depicted in this figure, and comprise the remainder of the 100 
percent of all married households. Percentage estimates in this figure 
have margins of error at the 95% confidence level that are less than 
or equal to 0.5 percentage points. 

[End of figure] 

As working women's contributions to household income have grown, women 
have also tended to take less time out of the workforce to raise 
children. Since the 1960s, working women have left the workforce for 
progressively shorter intervals to care for their first born children. 
For children born in the late 1960s, 36 percent of working mothers 
took 4 years or more off to care for their first born and 56 percent 
took less than 6 months off (see figure 12). For children born from 
2000 through 2004, 14 percent of working women took 4 years or more 
off and 65 percent took less than 6 months off.[Footnote 31] 

Figure 12: Among Mothers Who Worked prior to the Birth of Their First 
Child, Length of Time off Taken for Caregiving, by Year First Child 
Was Born: 

[Refer to PDF for image: stacked horizontal bar graph] 

Estimated percentage of mothers: 

First child born 1965-1969: 
Less than 4 years: 64%; 
Less than 6 months: 56%; 
6 to 12 months: 1%; 
1 to 2 years: 3%; 
2 to 3 years: 4%; 
4 or more years: 36%. 

First child born 1970-1974: 
Less than 4 years: 69%; 
Less than 6 months: 56%; 
6 to 12 months: 3%; 
1 to 2 years: 4%; 
2 to 3 years: 6%; 
4 or more years: 31%. 

First child born 1975-1979: 
Less than 4 years: 74%; 
Less than 6 months: 57%; 
6 to 12 months: 2%; 
1 to 2 years: 8%; 
2 to 3 years: 8%; 
4 or more years: 26%. 

First child born 1980-1984: 
Less than 4 years: 76%; 
Less than 6 months: 60%; 
6 to 12 months: 3%; 
1 to 2 years: 6%; 
2 to 3 years: 6%; 
4 or more years: 24%. 

First child born 1985-1989: 
Less than 4 years: 81%; 
Less than 6 months: 64%; 
6 to 12 months: 3%; 
1 to 2 years: 7%; 
2 to 3 years: 7%; 
4 or more years: 19%. 

First child born 1990-1994: 
Less than 4 years: 80%; 
Less than 6 months: 63%; 
6 to 12 months: 3%; 
1 to 2 years: 7%; 
2 to 3 years: 8%; 
4 or more years: 20%. 

First child born 1995-1999: 
Less than 4 years: 84%; 
Less than 6 months: 66%; 
6 to 12 months: 3%; 
1 to 2 years: 8%; 
2 to 3 years: 7%; 
4 or more years: 16%. 

First child born 2000-2004: 
Less than 4 years: 86%; 
Less than 6 months: 65%; 
6 to 12 months: 4%; 
1 to 2 years: 9%; 
2 to 3 years: 8%; 
4 or more years: 14%. 

Source: GAO analysis of 2008 SIPP data. 

Note: Percentages within the "less than 4 years" category may not sum 
to the "less than 4 years" total due to rounding. This figure 
comprises mothers age 20 through 62 who worked prior to the birth of 
their first child. The employment history data were collected in Wave 
1 of the 2008 SIPP, which took place in 2008. There is some potential 
overlap between the categories of spell length exceeding 6 months. The 
SIPP data precisely identify women in the category "Less than 6 
months." However, there is less precision in the other categories. 
This is because if a woman reports that she has taken off 6 months or 
more, she is asked to provide the start and end date for each spell 
out of the workforce; however, the SIPP data only provide the start 
and end year of each spell, so we do not know the month in which each 
spell started and ended. Therefore, there is potential for overlap in 
the categories of timeframes. For example, a woman whose spell started 
in January 2000 and ended in February 2001 - a 13-month spell - would 
be grouped in the "1-2 years" category. But a woman whose spell 
started in December 2000 and ended in January 2002 - also a 13-month 
spell - would be grouped in the "2-3 years" category. Margins of error 
at the 95% confidence level for estimates of the percent of women 
taking off less than 6 months are less than or equal to 5.6 percentage 
points, for each bar. Margins of error for estimates of the percent of 
women taking off 4 or more years are less than or equal to 4.8 
percentage points, for each bar. Margins of error for estimates of the 
percent of women taking off periods between 6 months and 3 years range 
from 0.4 to 2.0 percentage points. Note that these estimates are less 
reliable--the standard errors can be as high as 30 percent of the 
estimates for the earliest childbirth cohort, although they are always 
less than 15 percent of the estimates for estimates from 1980 and 
later. For women who had their first child between 2005 and 2007, 73 
percent took less than 6 months off from work. For these women, we 
could not report on longer periods of time out of the workforce 
because insufficient time had passed between when their first child 
was born and when the data were collected. 

[End of figure] 

It is also noteworthy that the women represented in figure 12--those 
who were working prior to the birth of their first child--comprised a 
growing proportion of new mothers over this period, according to 
analysis by the Census Bureau. Specifically, an estimated 60 percent 
of women who had their first child between 1960 and 1965 had worked 
for at least 6 months or more continuously prior to having children. 
In contrast, an estimated 75 percent of women who had their first 
child between 2000 and 2005 had worked for 6 or more months 
continuously prior to childbirth.[Footnote 32] 

Trends in Marriage and Work Patterns Have Resulted in Fewer Women 
Receiving Social Security Spousal Benefits and Married Women 
Contributing More to Household Savings: 

Fewer Women Are Receiving Social Security Spousal Benefits: 

Fewer women are receiving Social Security spousal benefits today than 
in the past.[Footnote 33] In 1960, nearly one-third of all women over 
age 62 received spousal benefits based exclusively on their spouse's 
work record, according to SSA data (see figure 13). In 2011, 9 percent 
of women received them. Similarly, the proportion of women over age 62 
receiving widow (i.e., survivor) benefits based exclusively on their 
spouse's work record declined from 23 to 16 percent over the same 
period. At the same time, as women's workforce participation and 
earnings have risen, the percentage of women receiving benefits based 
solely on their own work record increased from about 39 to 48 percent. 
However, women's average Social Security retired worker benefit is 
approximately three-quarters of men's average retired worker benefit 
due to women's lower median lifetime earnings relative to men's. 
Specifically, according to SSA, in 2012, the average retired worker 
benefit was $1,417 per month for men and $1,103 per month for women. 

Figure 13: Percentage of All Women Age 62 or Older by Social Security 
Benefit Status over Time: 

[Refer to PDF for image: stacked line graph] 

Percentage of all women aged 62 or older receiving Social Security 
benefits: 

Year: 1960; 
Retired worker: 38.7%; 
Dually-entitled/spouse: 2.4%; 
Spouse only: 32.8%; 
Dually-entitled/widow: 2.1%; 
Widow only: 23.4%. 

Year: 1970; 
Retired worker: 42.1%; 
Dually-entitled/spouse: 3.4%; 
Spouse only: 22.4%; 
Dually-entitled/widow: 5%; 
Widow only: 26.8%. 

Year: 1980; 
Retired worker: 41%; 
Dually-entitled/spouse: 6.2%; 
Spouse only: 17.6%; 
Dually-entitled/widow: 9.6%; 
Widow only: 25.4%. 

Year: 1985; 
Retired worker: 38.5%; 
Dually-entitled/spouse: 8.7%; 
Spouse only: 16.4%; 
Dually-entitled/widow: 11.5%; 
Widow only: 24.9%. 

Year: 1990; 
Retired worker: 36.9%; 
Dually-entitled/spouse: 10.4%; 
Spouse only: 15.3%; 
Dually-entitled/widow: 13%; 
Widow only: 24.3%. 

Year: 1995; 
Retired worker: 36.2%; 
Dually-entitled/spouse: 11.5%; 
Spouse only: 14.3%; 
Dually-entitled/widow: 14.4%; 
Widow only: 23.6%. 

Year: 2000; 
Retired worker: 38%; 
Dually-entitled/spouse: 12%; 
Spouse only: 12.9%; 
Dually-entitled/widow: 15.6%; 
Widow only: 21.5%. 

Year: 2005; 
Retired worker: 41.4%; 
Dually-entitled/spouse: 12%; 
Spouse only: 11.4%; 
Dually-entitled/widow: 16%; 
Widow only: 19.3%. 

Year: 2010; 
Retired worker: 46.3%; 
Dually-entitled/spouse: 12.1%; 
Spouse only: 9.6%; 
Dually-entitled/widow: 15.5%; 
Widow only: 16.4%. 

Year: 2011; 
Retired worker: 47.5%; 
Dually-entitled/spouse: 12%; 
Spouse only: 9.3%; 
Dually-entitled/widow: 15.3%; 
Widow only: 15.9%. 

Source: Annual Statistical Supplement to the Social Security Bulletin, 
2012. 

[End of figure] 

In addition to the effect of rising labor force participation, the 
decline in women's receipt of spousal benefits is partly due to the 
decline in marriage. Specifically, the proportion of women who are not 
eligible to receive Social Security spousal benefits because they were 
either never married, or divorced after less than 10 years of marriage-
-the length of time required for eligibility for Social Security 
divorced spouse benefits--has increased over the last two decades. 
According to analysis by SSA researchers, among women age 50-59, the 
percentage who were not eligible to receive Social Security spouse or 
widow benefits based on marital history doubled from 8 to 16 percent 
from 1990 to 2009 (see figure 14).[Footnote 34] However, this growth 
varied greatly by race. For blacks, the rise in ineligibility for 
spousal or widow benefits has been more dramatic, growing from 13 to 
34 percent from 1990 to 2009.[Footnote 35] In contrast, among whites 
and Hispanics, the rates of ineligibility grew less--by 8 and 6 
percentage points, respectively. 

Figure 14: Estimated Percentage of Women Ineligible for Social 
Security Spouse or Widow Benefits Because of Marital History among 50-
59 Year Olds by Race: 1990 and 2009: 

[Refer to PDF for image: horizontal bar graph] 

Percentage of women ineligible for spousal or widow benefits: 

All: 
1990: 8%; 
2009: 16%. 

White: 
1990: 6%; 
2009: 14%. 

Black: 
1990: 13%; 
2009: 34%. 

Hispanic: 
1990: 11%; 
2009: 17%. 

Source: Howard M. Iams and Christopher R. Tamborini, “The Implications 
of Marital History Change on Women's Eligibility for Social Security 
Wife and Widow Benefits, 1990-2009,” Social Security Bulletin, vol. 
72, no.2 (2012). 

Note: In this figure, women ineligible for Social Security spouse or 
widow benefits because of marital history include women who were never 
married or who were divorced without any 10-year marriage at the time 
of the survey. For each category above, the difference between the 
proportion of women ineligible for Social Security spouse or widow 
benefits because of marital history in 1990 and 2009 was statistically 
significant at the 0.05 level. 

[End of figure] 

The growth in ineligibility also varied by income, and the differences 
in eligibility rates by income grew over time, according to our 
analysis of SIPP data.[Footnote 36] Within the lowest income quintile, 
from 1996 to 2008, the percentage of 50-59 year-old women who were 
married or had been married for at least 10 years declined from an 
estimated 78 to 65 percent. In contrast, among women in the same age 
range, but in the highest income quintile, the percentage who were 
married or had been married for at least 10 years declined from an 
estimated 97 to 94 percent.[Footnote 37] [Footnote 38] 

Today's female spouse-only beneficiaries differ from female 
beneficiaries with a qualifying work record in terms of race, 
education, and family size, but not in terms of their wealth. Spouse-
only beneficiaries are less likely to be black than women receiving 
benefits based exclusively on their own work record (retired worker 
beneficiaries). Dually-entitled beneficiaries, in contrast, are more 
likely to be white than their counterparts receiving retired worker or 
spousal benefits. Spouse-only beneficiaries are also less likely to be 
college educated (see figure 15). In fact, 63 percent of spouse-only 
beneficiaries have less than a college education. At the same time, 39 
and 51 percent of retired worker and dually-entitled beneficiaries 
have less than a college education, respectively. 

Figure 15: Education Levels of Women Receiving Social Security 
Benefits Based on Their Own Work Record (Retired Worker), Their 
Spouse's Work Record (Spouse-only) or Both (Dually-entitled/Spouse): 
2011: 

[Refer to PDF for image: stacked horizontal bar graph] 

Percentage of beneficiaries (estimated): 

Retired worker: 
High school graduate or less: 39%; 
Less than high school: 8%; 
High school graduate: 31%; 
Some college or more: 61%; 
Some college: 34%; 
Bachelor's degree or higher: 27%. 

Spouse-only: 
High school graduate or less: 63%; 
Less than high school: 20%; 
High school graduate: 43%; 
Some college or more: 37%; 
Some college: 19%; 
Bachelor's degree or higher: 18%. 

Dually-entitled: 
High school graduate or less: 51%; 
Less than high school: 8%; 
High school graduate: 43%; 
Some college or more: 49%; 
Some college: 30%; 
Bachelor's degree or higher: 19%. 

Source: GAO analysis of SIPP Gold Standard Data. 

Note: Widows were excluded from the sample of retired workers so that 
we could compare a similar set of female beneficiaries (i.e., all non-
widows). 

[End of figure] 

Among married women, spouse-only beneficiaries have significantly more 
children (an average of 3.1 vs. 2.4 and 2.8 among retired-worker and 
dually-entitled beneficiaries, respectively). In terms of earnings and 
wealth, married spouse-only beneficiaries have lower lifetime 
household earnings than their married counterparts who receive retired-
worker benefits or are dually-entitled, but have comparable levels of 
total net worth, home ownership, and home equity.[Footnote 39] 

Married Women's Contributions to Household Account-Based Retirement 
Savings Approach Men's in Dual-Earner Households: 

With women's rising workforce participation and earnings, married 
women are contributing more to household account-based retirement 
savings, but their share of contributions varies with the work 
patterns of the household. Between 1992 and 2010, married women's 
contributions to household account-based retirement savings 
(accumulated in DC plans, IRAs, and Keoghs) increased from 20 to 38 
percent, on average, according to our analysis of SCF data.[Footnote 
40] However, the proportion of households with any retirement savings 
and the average proportion of the wives' contributions varied 
depending on whether the husband, wife, or both were working. In 2010, 
dual-earner households were the most common type of households among 
55-to 64-year-olds, comprising 45 percent of married households for 
that age group that year (see figure 16). These households were more 
likely to have account-based retirement savings than households with 
only one spouse working. Further, among dual-earner households with 
account-based retirement savings (either in DC plans, IRAs, our 
Keoghs), wives contributed 44 percent of the household's account-based 
retirement savings, on average.[Footnote 41] For households with some 
account-based retirement savings in which wives were the sole earners, 
wives contributed 68 percent of the household's account-based 
retirement savings. In contrast, wives contributed much lower shares 
of retirement savings--14 percent, on average--in households in which 
the husband was the sole earner.[Footnote 42] 

Figure 16: Among Married Households Age 55-64 Years Old, Average 
Proportion of Household Retirement Savings (Defined Contribution Plans 
and Individual Retirement Accounts) Contributed by Each Spouse: 2010. 

[Refer to PDF for image: 9 pie-charts] 

Percentage of married households age 55-64 that are: 
Husband sole earner: 25%; 
Wife sole earner: 15%; 
Dual-earner household: 45%. 

Percentage of those households with retirement savings: 
Husband sole earner: 68%; 
Wife sole earner: 65%; 
Dual-earner household: 76%. 

Wife's contribution to the household's retirement savings: 
Husband sole earner: 14%; 
Wife sole earner: 68%; 
Dual-earner household: 44%. 

Source: GAO analysis of 2010 Survey of Consumer Finances data. 

Note: Married households in which neither spouse works are not 
depicted in this figure (and comprise the remainder of the 100 percent 
of all married households). All estimates in this figure have 95 
percent confidence intervals within +/-8.1 percentage points of the 
estimates themselves. Household retirement savings include the balance 
of all defined contribution, IRA (Roth, traditional, and rollover), 
and Keogh accounts held by all members of the primary economic unit, 
as defined by the Survey of Consumer Finances. DB plans are not 
included in this measure of household savings. 

[End of figure] 

Single-headed households, in contrast, continue to lag behind their 
married counterparts both in terms of having DC plans or IRAs and in 
the level of retirement savings. For the population of all households 
age 50-64, 60 percent have some DC, IRA, or Keogh savings, according 
to our analysis of SCF data. However, this varies widely by marital 
status. Specifically, in 2010, an estimated 71 percent of married 
households age 50-64 years old had DC, IRA, and Keogh savings, while 
only 39 percent and 48 percent of male and female single-headed 
households had retirement savings, respectively (see figure 
17).[Footnote 43] Moreover, the median levels of savings among single-
headed households were much lower than for married households in this 
age category in 2010. For married households with retirement savings, 
the median savings level was $122,560. In contrast, among single men 
with retirement savings, the median savings level was $50,000. Among 
single women with retirement savings, the median level of savings was 
$32,800. 

Figure 17: Among All Households Age 50-64, Percentage of Households 
with Retirement Savings (in Defined Contribution Plans and Individual 
Retirement Accounts) and Among Households with Savings, the Median 
Levels of Savings 2010: 

[Refer to PDF for image: horizontal bar graph] 

Percentage of households with retirement savings: 

Households headed by married couples: 71%; 
Median savings among those with savings: $122,560. 

Households headed by single males: 39%; 
Median savings among those with savings: $50,000. 

Households headed by single females: 48%; 
Median savings among those with savings: $32,800. 

Source: GAO analysis of 2010 Survey of Consumer Finances data. 

Note: Median savings are expressed in 2010 dollars. All percentage 
estimates in this figure have 95 percent confidence intervals within 
+/- 4.8 percentage points of the estimate itself. All median dollar 
estimates in this figure have 95 percent confidence intervals within 
+/- 13,071 dollars of the estimate itself. Household retirement 
savings include the balance of all defined contribution, IRA (Roth, 
traditional, and rollover), and Keogh accounts held by all members of 
the primary economic unit, as defined by the Survey of Consumer 
Finances. DB plans are not included in this measure of household 
savings. 

[End of figure] 

Over One-Third of Married Households Receiving Income from Pensions 
Opted Not to Receive a Survivor Benefit: 

Over the last decade, over one-third of married households receiving 
income from pensions opted not to receive a spousal survivor benefit. 
[Footnote 44] There are many reasons why households might decide to 
opt out of receiving a survivor benefit. For example, a spouse might 
not be expected to outlive the individual receiving the pension (the 
participant). Alternatively, the spouse that is expected to survive 
the spouse receiving the pension might have access to his or her own 
retirement income. According to our analyses of SCF data, the decision 
to opt out differed depending on whether the pension participant was 
the wife or the husband. Specifically, in 2010, 62 percent of men 
opted for a survivor benefit for their spouse while 38 percent of 
women did. Men's higher likelihood of choosing survivor benefits could 
reflect their expectations that their wives are more likely to outlive 
them, or it could reflect that married women are more likely than 
married men to need the security of a survivor benefit because of less 
access to other sources of retirement income. 

Several other factors were related to couples' likelihood of opting 
out of a survivor benefit. Couples with longer marriages were less 
likely to opt out, possibly reflecting a greater sense of financial 
responsibility for the surviving spouse among long-lasting marriages. 
Households in the lowest net worth quintile are more likely to opt out 
of the survivor benefit than wealthier households. This finding may 
reflect the fact that households with limited wealth may be unable to 
afford the benefit reduction that comes with a survivor benefit. 
[Footnote 45] Lastly, households with higher levels of assets in DC 
plans or IRAs were less likely to opt out, perhaps an indicator of the 
household's level of financial literacy and preparedness overall or 
simply that, with other retirement savings, the household did not need 
the higher benefit payment that would result from the couple opting 
out.[Footnote 46] 

The Decline in Receipt of Spousal Benefits Is Likely to Increase Some 
Retirees' Economic Vulnerability: 

Shifts in Marriage and Labor Force Participation Mean Fewer Future 
Retirees Are Expected to Receive Social Security Spousal Benefits: 

Looking forward, fewer women will likely receive Social Security 
spousal benefits because of their declining eligibility based on 
marital history or increased earnings relative to their husbands, 
according to SSA projections.[Footnote 47] More specifically, 
eligibility for spousal benefits is projected to decline for two 
reasons. First, the population of women who never marry is projected 
to more than double in the future. Of women born from 1936 to 1945, 4 
percent are projected to never have married by age 70. In contrast, of 
women born from 1966 to 1975, 10 percent are projected to have never 
married by age 70.[Footnote 48] Second, due to the trend toward 
shorter marriages, fewer elderly divorcees will likely have had a 
marriage that lasted at least 10 years--an eligibility requirement for 
divorced spousal benefits. For women who had their first marriage in 
the early 1960's, 83 percent of marriages lasted at least 10 years, 
whereas 75 percent of comparable marriages that occurred in the early 
1990's lasted this long, according to Census Bureau estimates. 
[Footnote 49] 

Additionally, due to women's increased lifetime earnings relative to 
men, fewer women who are eligible for spousal benefits based on their 
marital history will receive them because they will be eligible for 
retired worker benefits based on their own work record.[Footnote 50] 
Compared to current retirees, more of today's working-age women are 
likely to receive Social Security benefits based exclusively on their 
own work history, known as retired worker benefits. Whereas in 1960, 
39 percent of women age 62 and older received retired worker benefits, 
SSA projects that this will increase from 46 percent in 2012 to 62 
percent in 2040 (see figure 18). This increase will largely be driven 
by the employment and earnings patterns of married and widowed women. 
Over the same period, the proportion of women who are expected to 
receive benefits based exclusively on marriage--known as spouse-only 
benefits--is projected to decline from 8 to 2 percent.[Footnote 51] At 
the same time, the proportion of women who are expected to receive 
benefits based on both work history and marriage--known as dually-
entitled/spouse benefits--is projected to decrease from 15 to 8 
percent. In contrast with today, proportionately more married men are 
projected to receive survivor benefits in the future due to their 
decreased earnings relative to their wives. SSA projects that the 
proportion of men who will receive retired worker benefits will drop 
by 3 percentage points--from 95 to 92 percent--between 2012 and 2040. 
At the same time, the proportion of men who are expected to receive 
survivor benefits based on their marriage to higher-earning wives is 
projected to increase from 3 to 6 percent (see figure 18). 

Figure 18: Projected Distribution of Social Security Beneficiary Type 
by Sex, Age 62 and Older: 2012 to 2040: 

[Refer to PDF for image: vertical bar graph] 

Percentage aged 62 or older receiving Social Security benefits: 

Retired worker: 
2012, Women: 46%; 
2040, Women: 62%; 
2012, Men: 95%; 
2040, Men: 92%. 

Spousal benefits, Dually-entitled/spouse: 
2012, Women: 15%; 
2040, Women: 8%; 
2012, Men: 1%; 
2040, Men: 1%. 

Spousal benefits, Spouse-only: 
2012, Women: 8%; 
2040, Women: 2%; 
2012, Men: 1%; 
2040, Men: 0. 

Survivor benefits, Dually-entitled/widower: 
2012, Women: 25%; 
2040, Women: 26%; 
2012, Men: 3%; 
2040, Men: 6%. 

Survivor benefits, Widower-only: 
2012, Women: 6%; 
2040, Women: 2%; 
2012, Men: 0; 
2040, Men: 0. 

Source: Social Security MINT6 projections. 

Note: Due to limitations of the data, the 2012 projections exclude 
beneficiaries older than 86. As a result, the 2012 data for widows 
differs from the 2010 data shown in figure 13. 

[End of figure] 

Despite these changes, survivor benefits are expected to continue to 
play an important role for many women. Although the proportion of 
widows whose benefits will be based solely on marriage--known as widow-
only--is projected to decrease from 6 to 2 percent by 2040, SSA 
projects that the proportion of women who will be dually-entitled 
widows will remain stable at about one-quarter of all female 
beneficiaries through 2040 (see figure 18). This is in part due to the 
fact that many women who qualify for retired-worker benefits while 
their husbands are alive (because their retired-worker benefit exceeds 
50 percent of their spouse's benefit) will become dually-entitled as 
widows because their own retired-worker benefit does not exceed 100 
percent of their husband's benefit. For example, another analysis 
projects that two-thirds of wives born from 1966 to 1975 will receive 
widow benefits rather than their own retired worker benefit when 
widowed, reflecting women's lower average lifetime earnings relative 
to their husbands.[Footnote 52] 

In general, the trend away from women receiving spousal benefits is 
projected to occur across racial and ethnic groups, with the largest 
shift occurring among black women. According to SSA analyses among 
whites, blacks, and Hispanics, as of 2012, proportionally, black women 
were already the most likely to receive retired worker benefits (see 
figure 19). Furthermore, between 2012 and 2040, black women are 
projected to experience the highest growth in this category. In 
addition, the proportions of dually-entitled/spouse white and black 
women are projected to decrease by about half between 2012 and 2040, 
with a small decrease among Hispanic women. Moreover, the proportion 
of women who are expected to receive widow-only benefits is projected 
to decrease for all three groups and, by 2040, differ by no more than 
two percentage points. Unlike these trends, the projections for dually-
entitled widows diverge by race. More specifically, the proportion of 
white dually-entitled widows is projected to increase by 12 percent, 
while blacks will likely see a 35 percent decrease. These projections 
do not suggest any change among Hispanic dually-entitled widows. 
[Footnote 53] 

Figure 19: Projected Distribution of Social Security Beneficiary Type 
for Women by Race, Age 62 and Older: 2012 to 2040: 

[Refer to PDF for image: vertical bar graph] 

Retired only; 
2012, White: 44%; 
2040, White: 59%; 
2012, Black: 54%; 
2040, Black: 78%; 
2012, Hispanic: 49%; 
2040, Hispanic: 65%. 

Dually entitled/spouse; 
2012, White: 17%; 
2040, White: 9%; 
2012, Black: 7%; 
2040, Black: 3%; 
2012, Hispanic: 9%; 
2040, Hispanic: 8%. 

Spouse only; 
2012, White: 7%; 
2040, White: 1%; 
2012, Black: 6%; 
2040, Black: 1%; 
2012, Hispanic: 12%; 
2040, Hispanic: 4%. 

Dually entitled/widow; 
2012, White: 26%; 
2040, White: 29%; 
2012, Black: 26%; 
2040, Black: 17%; 
2012, Hispanic: 20%; 
2040, Hispanic: 20%. 

Widow only; 
2012, White: 6%; 
2040, White: 2%; 
2012, Black: 7%; 
2040, Black: 1%; 
2012, Hispanic: 10%; 
2040, Hispanic: 3%. 

Source: Social Security MINT6 projections. 

Note: Due to limitations of the data, the 2012 projections exclude 
beneficiaries older than 86. 

[End of figure] 

Current Trends Suggest a Growing Proportion of the Elderly Will Be at 
Risk of Living in Poverty During Retirement: 

As the population that is ineligible for spousal and survivor benefits 
based on marital history--i.e., those who are not married or divorced 
without a marriage of at least 10 years--is projected to increase, the 
proportion of people who will depend entirely on their own earnings 
and savings in retirement is expected to increase as well. For many 
elderly, women in particular, this shift is likely to be positive, 
reflecting their higher earnings and greater capacity to save for 
retirement. However, elderly with low levels of lifetime earnings, who 
have no spouse, or who do not receive a spousal benefit--a group that 
is disproportionately represented by black women--are expected to have 
correspondingly lower retirement benefits.[Footnote 54] As a result, a 
growing proportion of the elderly--especially black women--may be at 
risk of living in poverty in retirement due to the higher cost of 
living and lower income associated with being single. As our past work 
has shown, elderly married individuals have higher levels of median 
household income and lower poverty rates than those who are never 
married, divorced, or widowed.[Footnote 55] Furthermore, in 2040, the 
median income levels of this expanding group of never married elderly 
will continue to be lower than that of their married counterparts, 
according to SSA projections. 

Additionally, even some of those elderly who are eligible to receive 
spousal benefits may be more economically vulnerable after the death 
of one spouse due to the trend toward spouses having more similar 
earnings than in the past. More specifically, the more similar two 
spouses' earnings, the less likely the survivor is to receive a 
widow(er) benefit. As seen in figure 20, three hypothetical couples 
with the same level of preretirement income, but with different 
earnings ratios, will have very different levels of survivor 
benefits.[Footnote 56] The survivor in the one-earner couple receives 
$1,859 per month--or two-thirds of the previous household benefit at 
retirement, as determined by the Social Security benefit 
formula.[Footnote 57] The survivor in the dual-earner couple with 
unequal earnings receives 60 percent of the previous household 
benefit. In contrast, the survivor of the dual-earner couple with 
equal earnings receives just 50 percent of the previous household 
benefit--significantly less than at least one measure of what a single 
person needs to live on relative to what a couple needs.[Footnote 58] 
Moreover, the proportion of dual-earner wives who are projected to not 
receive a survivor benefit--because of their relatively high earnings 
compared to their deceased husbands'--is expected to increase from 18 
percent of wives who were born from 1936 to 1945 to 34 percent for 
wives born from 1966 to 1975.[Footnote 59] The lower level of survivor 
benefits generally may affect women for a longer period of time than 
men, due to women's longer life expectancy. However, working women's 
increased earnings and personal savings may potentially offset the 
lower level of survivor benefit, mitigating this issue to some extent. 

Figure 20: Examples of Social Security Retirement and Survivor Income 
by Household Structure: 

[Refer to PDF for image: series of pie-charts] 

One-earner couple: 
Annual household earnings: $50,000 (Spouse A); 
Monthly household benefit at retirement: $2,789; 
- Spouse A: $1,859; 
- Spouse B: $930; 
Monthly benefit paid to survivor: $1,859; 67% of total benefit when 
both spouses were alive. 

Dual-earner couple (with unequal earnings): 
Annual household earnings: $50,000; 
- Spouse A: $33,300; 
- Spouse B: $16,700; 
Monthly household benefit at retirement: $2,314; 
- Spouse A: $1,390; 
- Spouse B: $924; 
Monthly benefit paid to survivor: $1,390; 60% of total benefit when 
both spouses were alive. 

Dual-earner couple (with equal earnings): 
Annual household earnings: $50,000; 
- Spouse A: $25,000; 
- Spouse B: $25,000; 
Monthly household benefit at retirement: $2,314; 
- Spouse A: $1,157; 
- Spouse B: $1,157; 
Monthly benefit paid to survivor: $1,157; 50% of total benefit when 
both spouses were alive. 

Source: Social Security Administration data and GAO calculations. 

Notes: Estimates assume that (1) the monthly equivalent of these 
annual earnings equals the individual's average indexed monthly 
earnings used to calculate Social Security benefits; (2) all 
individuals retire at age 66 in 2013; and (3) there are no cost of 
living adjustments reflected after the initial benefit is taken at 
retirement. 

[End of figure] 

In addition to the differences in benefit levels among widow(er)s 
mentioned above, figure 20 also illustrates how retirement benefits 
can vary while both spouses are alive, for couples with different 
relative earnings.[Footnote 60] Namely, a one-earner couple receives a 
higher total retirement benefit than a dual-earner couple with the 
same level of household preretirement income. In figure 20, the total 
monthly benefit for the one-earner couple is $2,789. By contrast, the 
dual-earner couples receive $2,314, or 17 percent less. As the 
proportion of one-earner couples continues to decline, however, this 
disparity will become less and less common. 

Some Social Security Policy Options May Offset Effects of Changes in 
Marital Patterns: 

Past GAO work has identified a number of options suggested by experts 
to improve the retirement security of populations vulnerable to 
poverty by changing the Social Security eligibility rules or benefit 
structure, although each option has limitations.[Footnote 61] Some of 
these options focus specifically on modifying benefits for spouses and 
survivors, while others would affect broader populations of vulnerable 
individuals, and may also benefit those affected by changing marital 
patterns. For example, one option would change the eligibility 
requirements for divorced spousal benefits by reducing the required 
length of marriage from 10 to 7 years, with the goal of helping more 
of the divorced elderly remain out of poverty. Another option would 
increase benefit amounts for widow(er)s from the current maximum of 67 
to 75 percent of the combined amount the couple had received, with the 
aim of preventing poverty among widow(er)s.[Footnote 62] This option 
could be targeted specifically toward low-income survivors by capping 
the surviving spouse's benefit amount at the average retired worker 
benefit. However, concerns have been raised that these options may not 
effectively target other economically vulnerable groups like low-
income Social Security retired worker recipients who never married. 
Further, both of these options could provide benefits to higher-income 
elderly who may not be particularly economically vulnerable. 

Other reform options target the vulnerable population of low-income 
people who do not qualify for spousal or survivor benefits, such as 
those who never marry, and who would otherwise receive retired worker 
benefits. Retired worker benefits are generally calculated on the 
basis of a worker's average indexed monthly earnings for the worker's 
highest 35 years of earnings. While Social Security's benefit 
structure is progressive, meaning that its retirement benefits replace 
a higher percentage of earnings for lower-income workers, a lifetime 
low-wage worker would still have correspondingly low benefits. 
Furthermore, the current benefit formula does not distinguish between 
low average wages caused by low lifetime earnings or low average wages 
caused by years of unemployment. One set of proposed options aims to 
increase benefit adequacy for lifetime low wage workers by increasing 
the level of the Special Minimum Primary Insurance Amount (also 
referred to as the "Special Minimum Benefit")--a minimum Social 
Security benefit that is based on the number of years a worker has 
spent in Social Security-covered employment. The Special Minimum 
Benefit is paid to an eligible retired worker if it is higher than the 
benefit amount computed using the typical Social Security benefit 
formula based on the worker's highest 35 years of earnings.[Footnote 
63] Increasing the Special Minimum Benefit could help more elderly 
avoid poverty.[Footnote 64] However, it is difficult to effectively 
target lifetime low earners with this option because it is possible 
for high-wage earners who work part-time or sporadically to accumulate 
11 years of covered employment--the minimum required under current law 
to qualify for the Special Minimum Benefit. 

In addition to revising the Special Minimum Benefit, another option 
targets those who have shorter periods of covered employment because 
they were providing care to a family member and do not qualify for 
spousal benefits. More specifically, this option could change the 
Social Security benefit structure to provide earnings credits to 
caregivers (known as caregiver credits) during years in which they do 
not have earnings because they were providing care to children or 
elderly relatives. However, according to experts, this option may not 
reach its target population either because, for example, low-income 
people are less likely to take time out of the labor market to provide 
care for a family member. Consequently, people who have relatively 
higher incomes may benefit more from the creation of caregiver credits. 

These policy options have costs that would place greater strains on 
the Social Security trust funds, which are already projected to have 
significantly depleted assets by 2033, at which time claimants would 
only receive 77 percent of their entitled benefits. The cost each 
option might impose varies depending on several factors, including: 
(1) the number of additional beneficiaries that would become eligible, 
(2) whether other additional adjustments--such as including a minimum 
work requirement for divorced spouses--would be included, and (3) 
their potential to offset the new costs. 

The administrative burdens associated with implementing such changes 
are also important factors. For example, caregiver credits would be 
complex to administer due to the challenges associated with verifying 
that care was provided to a qualifying person and the amount of care 
that was provided. 

The Shift Away from Defined Benefit Plans Also Increases the Economic 
Vulnerability of Elderly Spouses: 

The transition in private employer plan sponsorship from DB to DC 
plans has increased the vulnerability of spouses due to differences in 
the level of spousal protections between these retirement plans. While 
over half of U.S. workers do not participate in employer-sponsored 
pension plans,[Footnote 65] those with spouses who do participate may 
have different federal protections depending on whether their spouse 
participates in a DB or DC plan. In DB plans, current law requires 
spousal consent if the participant wishes to waive the survivor 
annuity for his or her spouse. In contrast, under DC plans, there is 
generally no federal requirement to provide an annuity option and 
participants do not need spousal consent to withdraw funds from the 
account--either before or at retirement. While spouses are generally 
required to be the default beneficiary of the remaining account 
balance upon the death of the participant, there is no protection 
against the assets being invested unwisely, inadvertently spent down 
too quickly, or simply withdrawn fully upon a job change and rolled 
into an IRA, with a change in beneficiary. In a worst case scenario, 
the spouse who participates in the DC plan could withdraw all the 
assets and spend them in ways that do not provide for the couple's 
retirement security. While the increases in dual-earner couples and 
women's coverage under DC plans have the potential to mitigate these 
concerns by providing alternative sources of retirement income, 
spouses with lower or no earnings may remain vulnerable. 

To address this vulnerability, some policy analysts have suggested 
strengthening spousal protections in DC plans.[Footnote 66] For 
example, requiring spousal consent for a participant to roll over a DC 
plan account to an IRA--which is not required to make the spouse the 
beneficiary of the account balance upon the owner's death--or to cash 
out the account altogether, would allow spouses a level of input 
similar to the legal provisions governing DB plans. However, according 
to some experts, the administrative burden of obtaining spousal 
consent for such modifications to DC plans would impose administrative 
costs on plan sponsors and could slow down the retirement process. For 
example, one expert explained that DC plan benefit elections can be 
done electronically and in a matter of days. In contrast, DB plan 
elections typically take about 30 days, in part because the pen and 
paper process of obtaining notarized spousal consent in order to elect 
something other than a qualified joint and survivor annuity (QJSA) 
holds up the process.[Footnote 67] Additionally, one agency official 
we interviewed questioned the need for DC plans to adopt spousal 
consent features, pointing out that spousal protections under DB plans 
were instituted to protect nonworking spouses at a time when women 
generally did not have the same level of access to and participation 
in retirement plans as they do today.[Footnote 68] 

Despite their relatively greater level of spousal protections, some 
policy advocates have also proposed revising spousal provisions that 
apply to DB plans. First, similar to the concern that the Social 
Security survivor benefit amount is too low, one expert has suggested 
increasing the minimum amount for DB survivor annuities from 50 
percent to at least two-thirds of the prior annuity amount. Such a 
change would help surviving spouses better maintain their standard of 
living. However, such a change would entail lowering the initial 
annuity amount, which may be a hardship for some retirees and could 
deter some participants from electing the QJSA. Another policy option 
would seek to remedy the unequal treatment of widow(er)s under DB 
QJSAs. Typically, in a QJSA, the amount of the annuity that the couple 
receives while both spouses are alive does not change if the worker's 
spouse dies first. In other words, the worker would see no change in 
DB household income if widowed. On the other hand, if the worker dies 
first, the annuity amount that the surviving spouse receives may be 
reduced to as little as 50 percent of the prior amount. To remedy this 
imbalance, two experts have suggested paying the widow(er) a 
predetermined amount, regardless of which spouse dies first. This 
"symmetric" joint and survivor annuity option is currently available 
for plan sponsors to offer, but they generally have not done so. 
Accordingly, one version of this policy option would be to make this 
option the normal form of benefit for married participants, requiring 
spousal consent to opt for a different form of payment. A symmetric 
joint and survivor annuity has the advantage of requiring a smaller 
benefit reduction on account of the survivor benefit than a standard 
joint and survivor annuity, which could lead to more participants 
taking this option. However, one agency official noted that employers 
expressed concern that beneficiaries with this option might not report 
the death of their spouse to avoid incurring any benefit reduction. 

Concluding Observations: 

Over the last 50 years, major societal changes in the United States 
have had important consequences for how Americans live, work, and save 
for retirement. Indeed, unlike in decades past, fewer people are 
getting married, and those who do often marry later in life and for 
shorter periods. These trends have consequences for economic security 
because unmarried individuals are unable to take advantage of the 
economic benefits and efficiencies of marriage and may therefore be at 
greater risk of poverty in old age. Moreover, these trends have 
disproportionately occurred within the nation's most vulnerable 
populations--low-income, less-educated, and some minorities. If these 
trends continue, many Americans' economic security--both in their 
working years and in old age--could further deteriorate. 

While marriage rates overall have declined, the role women play in the 
labor market has expanded. As a result, women have become major 
contributors to household income and household retirement savings. For 
many households, these gains could enhance economic security in old 
age. However, low-wage workers who never marry or were not married 
long enough to qualify for Social Security spousal benefits may not be 
able to accrue sufficient retirement savings to offset the lack of a 
spousal benefit in old age. Even within two-earner households, the 
Social Security survivor benefit declines as earnings for both spouses 
become more similar. Survivors are therefore likely to depend on other 
forms of retirement income to maintain their standard of living prior 
to their spouse's death. Further, the shift to DC plans--which 
typically offer fewer protections for spouses and fewer possibilities 
for participants to annuitize within their plans--will likely result 
in more retirees outliving their assets. 

These trends and their consequences raise important questions for 
existing programs, such as Social Security, and current protections 
for spouses in pension plans. As part of any future reform efforts in 
these areas, it will be important for policy makers to consider these 
trends in order to minimize poverty and ensure retirement security 
among the elderly. 

Agency Comments: 

We provided a draft of this report to the Department of Labor, the 
Department of the Treasury, and the Social Security Administration for 
review and comment. While none of the agencies provided official 
comments, the Department of Labor and the Department of the Treasury 
provided technical comments, which we incorporated as appropriate. 

As agreed with your office, unless you publicly announce its contents 
earlier, we plan no further distribution until 30 days after the date 
of this letter. At that time, we will send copies of this report to 
the Secretary of Labor, the Secretary of the Treasury, and the 
Commissioner of Social Security. In addition, the report will be 
available at no charge on GAO's website at [hyperlink, 
http://www.gao.gov]. If you or your staff have any questions about 
this report, please contact me at (202) 512-7215 or jeszeckc@gao.gov. 
Contact points for our Offices of Congressional Relations and Public 
Affairs may be found on the last page of this report. GAO staff who 
made key contributions to this report are listed in appendix III. 

Sincerely yours, 

Signed by: 

Charles A. Jeszeck: 
Director: 
Education, Workforce, and Income Security: 

[End of section] 

Appendix I: Objectives, Scope, and Methods: 

To analyze trends in marriage and labor force participation and their 
implications for retirement security, we examined: (1) the trends in 
and status of marriage and labor force participation in the American 
household; (2) how those trends have affected spousal benefits and 
retirement savings behavior within households; and (3) what the 
implications of these trends are for retirement security. This 
appendix provides a detailed account of the information and methods we 
used to answer these questions. Section 1 describes the key 
information sources pertaining to analyses we conducted for each 
reporting objective. Sections 2 and 3 provide additional information 
on the characteristics of the individuals and households we analyzed 
for Objective 2. 

Section 1: Information Sources: 

To address our objectives, we obtained information from a variety of 
sources. We obtained and analyzed data from: (1) U.S. Census Bureau 
(Census) and Bureau of Labor Statistics (BLS) reports; (2) the 
Integrated Public Use Microdata Series (IPUMS) data consisting of 
decennial Census data and American Community Survey data; (3) the 
Survey of Income and Program Participation (SIPP) public use and 
restricted Gold Standard data,[Footnote 69] (4) the Current Population 
Survey; (5) the Survey of Consumer Finances (SCF); (6) SSA's Modeling 
Income in the Near Term model; and (7) SSA's actuarial estimates. 
Table 3 summarizes the data sources used to address each objective. To 
inform our data analyses for all three objectives, we obtained 
additional information from (1) prior GAO reports and the academic 
literature on trends in marriage, labor force participation, and 
retirement income; (2) relevant federal laws and regulations, and (3) 
a range of experts in the area of retirement security. This section 
provides a description of the information sources, how we used them, 
and the steps we took to ensure their reliability. 

Table 3: Data Sources Used for Each Reporting Objective: 

Objective: Objective 1: Trends in marriage and labor force 
participation; 
Data sources: 
* Census Bureau and Bureau of Labor Statistics reports; 
* Integrated Public Use Microdata Series (consisting of decennial 
Census data and American Community Survey data); 
* Survey of Income and Program Participation public-use data. 

Objective: Objective 2: Receipt of spousal benefits and household 
retirement savings behavior; 
Data sources: 
* Survey of Income and Program Participation public-use and restricted-
use Gold Standard data; 
* Survey of Consumer Finances. 

Objective: Objective 3: Implications of these trends for future 
retirement security; 
Data sources: 
* Modeling Income in the Near Term model; 
* SSA actuarial estimates. 

Source: GAO. 

[End of table] 

Census and BLS reports: 

To understand trends in marital status, single parenthood, and labor 
force participation over time, which are reported in Objective 1 of 
the report, we compiled and analyzed data from a number of published 
Census and BLS reports. The data in these reports draw upon Census and 
BLS surveys including: the U.S. Census Bureau, Current Population 
Reports; the Survey of Income and Program Participation; and the 
Annual Social and Economic Supplements, Current Population Survey. 

Integrated Public Use Microdata Series (IPUMS): 

To understand the relationship between income and education and trends 
in marriage and single-parenthood from 1960 to 2010, in Objective 1 we 
analyzed the Integrated Public Use Microdata Series (IPUMS). IPUMS 
consists of more than fifty high-precision samples of the American 
population drawn from fifteen federal censuses and from the American 
Community Surveys of 2000-2011. Together these samples constitute a 
rich source of quantitative information on long-term changes in the 
American population. IPUMS assigns uniform codes across all the 
samples and provides documentation for each data set. 

In our analysis of IPUMS data (Figures 7 and 8), the data points for 
years 1960, 1970, 1980, 1990 and 2000 are from the decennial Census, 
and data points for the year 2010 are from the American Community 
Survey. 

Weights and Standard Error Calculations in IPUMS Data: 

For all our analyses of IPUMS data, we used person weights. To 
estimate standard errors, we followed procedures recommended by IPUMS 
methodological guidance papers. We applied the appropriate design 
factors provided in Census Bureau publications for each year of data, 
taking into account the population subsamples on which we conducted 
our analysis. 

Survey of Income and Program Participation: 

To answer parts of Objectives 1 and 2, we analyzed data collected 
through the SIPP, a nationally representative survey conducted by 
Census that collects detailed information on marital history and 
caregiving, among many other areas. Specifically, we used the data to 
analyze trends in (1) the characteristics of women that do not have a 
marriage that would qualify them for Social Security spousal benefits 
and (2) women's time spent out of the workforce providing childcare. 
The SIPP survey is conducted in a series of national panels, each of 
which is collected for multiple years. Within each panel, the data are 
collected in a series of "waves" which take place in 4 month cycles. 
Within each wave, Census administers a core survey consisting of 
questions that are asked at every interview, and several modules 
relating to a particular topic. We used data from the core survey, the 
topical module on marital history, and the topical module on 
employment history from two SIPP panels, the 1996 and 2008 panels. The 
1996 panel, which was collected between April 1996 and March 2000, 
contained approximately 40,000 interviews at the beginning of the 
survey. The 2008 panel, which was collected between September 2008 and 
December 2012, contained approximately 52,000 interviews at the 
beginning of the survey. For both panels, the topical module on 
employment history was administered in Wave 1, and the topical module 
on marital history was administered in Wave 2. In each wave, we 
analyzed data from the fourth interview month. For the 1996 panel, the 
Wave 1 data was collected between March and June of 1996, and the Wave 
2 data was collected between July and October of 1996. For the 2008 
panel, the Wave 1 data was collected between August and November of 
2008, and the Wave 2 data was collected between December 2008 and 
March 2009. 

Weights and Standard Error Calculations in SIPP: 

For all of the SIPP analyses, we used SIPP individual-level weights to 
compute point estimates. Because our analysis of the SIPP public use 
data involves merging data for individuals from multiple waves, we 
chose not to use SIPP replicate weights, as the replicate weight 
analysis becomes complex and computationally burdensome under these 
circumstances. Instead we used Stata's survey data analysis package to 
declare a survey design for the data, using the following survey 
design variables from the SIPP data: GHLFSAM (half-sample code) to 
indicate the primary sampling unit, and GVARSTR (variance stratum 
code) to indicate the strata. GAO internal analysis has shown that 
using these survey design variables yields estimated standard errors 
that are very similar to those produced using the SIPP replicate 
weights. 

SIPP Analysis of Women with Qualifying Marriages: 

To analyze the characteristics of women who had not attained a 
marriage that would qualify them for Social Security spousal benefits, 
we used information from the marital history modules of the 1996 and 
2008 SIPP panels. We classified a woman as attaining a qualifying 
marriage if she either a) was currently married at the time of the 
survey, or b) was not currently married, but had previously been in a 
marriage that lasted for at least 10 years. We used data from the 
marital history module on the start and end date of marriages to 
determine whether a marriage had lasted at least 10 years. Because the 
public use data does not contain the month of marriage, we could not 
know precisely how long each person had been married. For women whose 
marriages were close to the 10-year mark, we had to choose between 
making our classification err on the side of overly inclusive or 
overly exclusive. We chose to define a marriage as lasting at least 10 
years if the difference between the year of marriage termination and 
the year of initiation was at least 11. This means that we did not 
include any women whose marriage was less than 10 years. However, we 
did exclude some women whose marriage was exactly 10 years. For 
example, a woman who married in January 2000 and divorced in February 
2010 would have a 10 year marriage, but our classification excluded 
her. We made this choice to avoid incorrectly including women with 
marriages less than 10 years. For example, if we defined our 
classification using the rule that the difference between the 
termination date and the start date was 10 years, we would include 
women who married in December 2000 and divorced in January 2010--a 
marriage that lasted 9 years and 2 months. 

We analyzed the characteristics of men and women who had attained a 
marriage that would qualify them for Social Security spousal benefits 
by comparing qualifying marriage rates across different demographic 
groups (specifically, across different age, racial and ethnic groups, 
education level, and income categories.) In addition, to examine 
whether the relationship between race and qualifying marriage rates 
remained when controlling for income and education, we also ran 
several multivariate regressions on women age 55-64. We ran probit 
regressions of the probability of attaining a qualifying marriage for 
women on race with controls for income and education. We also ran 
probit regressions of the probability of attaining a qualifying 
marriage on race interacted with income, and on race interacted with 
education. These regressions show the negative relationship between a 
person being in the "black" race category and attaining a qualifying 
marriage remains statistically significant, even with controls for 
income and education, in both 1996 and 2008. Similarly, we ran probit 
regressions of the probability of being currently married on race with 
controls for education and income, and we also ran probit regressions 
of the probability of being currently married on race interacted with 
income, and on race interacted with education, for women age 55-64 in 
1996 and 2008. These regressions show that the negative relationship 
between a person being in the "black" race category and being 
currently married remained significant with controls for education and 
income. 

SIPP Analysis of Time Spent Out of the Workforce to Provide Childcare: 

To analyze women's time spent out of the workforce providing 
childcare, we used data from the 2008 SIPP employment history module, 
combined with data from the fertility history module. We used the 
fertility history module to obtain information about whether an 
individual had ever had a child. This was necessary because we were 
looking at a person's lifetime employment, and many older people who 
have taken time out of the workforce for childcare in the past do not 
currently have children living with them. We restricted this analysis 
to women age 20-62 who had worked for at least 6 months prior to the 
birth of their first child. We set this age range because the SIPP 
"time off for caregiving" questions were only asked for people in this 
age range. We also restricted the analysis to women whose first child 
was born in 2004 or earlier. This restriction was made so that all 
women in the analysis would have given birth at least 4 years before 
the survey interview. We then examined the length of time women took 
out of the workforce, during their first spell out of the workforce. 
We restricted our analysis to the first spell out of the workforce in 
order to mitigate the impact of older women being more likely to have 
had multiple children, and therefore multiple spells out of the 
workforce, compared to younger women. We created five categories for 
time spent out of the workforce. The top category was "4 or more 
years". We made this the top category so that all women in the 
analysis were interviewed at least 4 years since the birth of their 
first child, and therefore would have had the opportunity to have 
taken off 4 or more years. 

SIPP Analysis Comparing Characteristics of Female Beneficiaries: 

For our analysis in Objective 2 comparing the characteristics of women 
who receive Social Security spouse-only benefits with those receiving 
retired-worker benefits or dually-entitled spouse benefits, we used 
the 2008 SIPP restricted-use Gold Standard data. We used the Gold 
Standard SIPP data because, despite its advantages, the SIPP public-
use data has one main limitation for our analysis. As with most survey 
data, SIPP public-use data are self-reported. This can be problematic 
for the reporting of data on income sources and the type of Social 
Security benefits individuals receive. For example, respondents might 
incorrectly report that they receive a spousal benefit, when in fact 
they are receiving a retired-worker benefit. To mitigate this problem, 
we used the 2008 SIPP Gold Standard data. Census created this file by 
extracting variables from the SIPP panels conducted in 2008 and 
merging SSA-provided administrative data including the Summary 
Earnings Records, Detailed Earnings Records, and the Master 
Beneficiary Record. Census refers to these data as the Gold Standard 
because they represent the available confidential micro-data that 
would be used for analysis by an authorized researcher working in a 
restricted-access facility. 

Survey of Consumer Finances: 

To answer part of Objective 2, we analyzed data from the 2010 SCF. 
Specifically, we used the data to analyze (1) the decision to opt out 
of a survivor benefit among married households with pension income, 
and (2) retirement savings behavior among married and single-headed 
households. The SCF is a triennial, nationally representative survey 
from the Board of Governors of the Federal Reserve System (Federal 
Reserve). The 2010 SCF surveyed 6,482 households about their pensions, 
incomes, labor force participation, asset holdings and debts, use of 
financial services, and demographic information. For the purposes of 
this report, a household refers to the primary economic unit within a 
household, which the SCF refers to as a family. To estimate the age, 
marital status, net worth, income, and participation in a DC or DB 
plan, we relied on variable definitions used for Federal Reserve 
publications using the SCF and consulted with an agency 
official.[Footnote 70] 

To analyze the decision to opt out of a survivor benefit, we conducted 
both comparisons of means and cross-tabulations, and ran some simple 
regression analyses using the SCF. Our regression analyses were logit 
models which predicted the probability of a person choosing a survivor 
benefit as a function of net worth and total retirement liquid assets 
(in separate regressions). In our regression of survivor benefit 
choice on net worth, we grouped net worth into quartiles, and 
constructed categorical variables for each quartile. We included 
retirement liquid assets as a continuous variable. 

The SCF is subject to non-sampling error and non-response, which can 
influence the accuracy of point and variance estimates. To address 
this, the Federal Reserve provides a set of implicate data--which 
contain information about imputed variables--and a replicate weight 
file, which permits users of the SCF microdata to obtain more accurate 
variance estimates. We followed Federal Reserve guidance and used both 
the implicate data and the replicate weights when constructing our 
point estimates and standard error estimates, both in our univariate 
analyses and in our regression analyses. 

Modeling Income in the Near Term: 

To project the distribution of Social Security beneficiary types for 
women into the future for Objective 3, officials from the Office of 
Retirement Policy at the Social Security Administration (SSA) provided 
GAO with projections based on simulations they conducted with the 
Modeling Income in the Near Term model, Version 6 (MINT6). This model 
is used to make projections related to individuals' retirement income, 
marital trends, Social Security benefits, and poverty. The MINT6 is 
built from Census survey data and Social Security's own administrative 
records.[Footnote 71] 

Social Security Actuarial Estimates: 

The Office of the Chief Actuary at the Social Security Administration 
provided GAO with the data used in Objective 3, figure 20, which 
presents examples of Social Security Retirement and Survivor Income 
benefit levels for households with different hypothetical earnings 
configurations (i.e. one-earner couples, dual-earner couples with 
unequal earnings, and dual-earner couples with equal earnings). 

Sampling Error: 

Estimates from the surveys described above are subject to sampling 
error because these surveys are probability samples based on random 
selections. Since each possible sample could have provided different 
estimates, we express our confidence in the precision of the sample 
results as 95 percent confidence intervals. These intervals would 
contain the actual population values for 95 percent of the samples 
that could have been drawn. In this report, we report all percentage 
or other numerical estimates along with their 95 percent confidence 
intervals. In addition, any differences between point estimates 
presented in this report are statistically significant at the 95 
percent confidence level unless otherwise noted. 

Data Reliability and Limitations: 

For the IPUMS, SIPP public-use and Gold Standard data, SCF and MINT6 
data described above, we conducted a data reliability assessment of 
selected variables by conducting electronic data tests for 
completeness and accuracy, reviewing documentation on the dataset, or 
interviewing knowledgeable officials about how the data are collected 
and maintained and their appropriate uses. When we learned that 
particular fields were not reliable or were not consistently reported, 
we did not use them. For the purposes of our analysis, we found the 
variables that we ultimately reported on to be sufficiently reliable. 

Nonetheless, our analysis of nationally representative survey data had 
several limitations. The SCF and other surveys that are based on self-
reported data are subject to several other sources of non-sampling 
error, including the inability to get information about all sample 
cases; difficulties of definition; differences in the interpretation 
of questions; respondents' inability or unwillingness to provide 
correct information; and errors made in collecting, recording, coding, 
and processing data. These non-sampling errors can influence the 
accuracy of information presented in the report, although the 
magnitude of their effect is not known. 

We also encountered limitations specific to our study. First, in using 
the 2008 SIPP Gold Standard Data to analyze the characteristics of 
female Social Security beneficiaries, we were unable to analyze 
information for approximately 6 percent of the female respondents age 
62 and over who did not provide their Social Security numbers because 
Census uses the SSN to match the information respondents provide in 
the SIPP survey with the information from administrative files. 
[Footnote 72] Second, in using the SCF to evaluate changes in the 
retirement savings behavior of households, we were unable to analyze 
the decisions of households to opt out of the survivor benefit prior 
to 2001 due to changes in the SCF questionnaire. In addition, in using 
the SCF to analyze near-retirement households, we widened the age 
range we examined from 55-64 to 50-64 in some instances to improve the 
precision of our estimates. 

Literature Review and Interviews: 

To gain an understanding of the trends in marriage and labor force 
participation and their implications for retirement security, we 
conducted an extensive literature review and interviewed a range of 
experts. To identify existing studies, we used online research 
software to search through databases of scholarly and peer-reviewed 
materials for information on these topics, including articles, 
journals, reports, and studies dating back to 2002. In addition, we 
reviewed articles that were collected during the previous GAO study on 
women's retirement security that contained relevant information and 
reviewed articles that were suggested to us by the experts we 
interviewed.[Footnote 73] The articles we cited in the report were 
reviewed for methodological adequacy by at least two GAO analysts with 
expertise in social science methods. 

To supplement the literature review, we conducted interviews with 
experts. To ensure that we obtained a balanced perspective, we 
interviewed 49 experts with a range of perspectives and from 21 
different types of organizations including government, academia, 
advocacy groups, think tanks and industry. We also consulted several 
of these experts on technical issues related to our analysis. We 
identified and interviewed experts who have specialized knowledge 
about spousal protections and retirement security in a specific area 
of expertise (e.g., law or social science), or who were recommended by 
one or more people or groups we had previously interviewed. 
Specifically, we interviewed federal agency officials at the 
Departments of the Treasury and Labor, the Social Security 
Administration, the Census Bureau, and the Office of Management and 
Budget; academic experts and subject-matter experts at Drexel 
University, the Employee Benefit Research Institute, Syracuse 
University, and the Urban Institute; and industry experts and 
advocates from AARP, the American Benefits Council, Global Policy 
Solutions, the Institute for Women's Policy Research, the Investment 
Company Institute, Mercer, the National Women's Law Center, the 
National Gay and Lesbian Task Force, the National Hispanic Council on 
Aging, the Pension Rights Center, and the Women's Institute for a 
Secure Retirement. 

To identify the policy options that may address retirement security 
issues under Social Security and private employer-sponsored plans, we 
reviewed past GAO reports and interviewed experts. For the Social 
Security options, we reviewed GAO's prior work in this area.[Footnote 
74] We selected options that focus on either spousal or survivor 
benefits or, more broadly, on elderly populations that are vulnerable 
to poverty. We identified policy options that would seek to address 
spousal protections in private employer-sponsored DB and DC plans 
through interviews with retirement policy experts representing a range 
of perspectives and from different types of organizations including 
government, academia, advocacy groups, and the private sector. The 
Social Security, DB, and DC policy options listed in this report are 
not intended to be exhaustive and we are not recommending or endorsing 
the adoption of any particular policy option. Rather, we identified 
them as potential options for policymakers to consider, along with 
other relevant proposals, as they address retirement security issues. 
In addition, for this report, we did not conduct an independent 
analysis of the potential legal effects of these policy options. 

Section 2: Analysis of the Characteristics of Social Security Spousal 
Beneficiaries: 

To compare the characteristics of female Social Security beneficiaries 
in Objective 2, we used a sample of beneficiaries from the SIPP Gold 
Standard data set. Table 4 provides descriptive information on the 
population of the beneficiaries we analyzed by categories of 
beneficiaries, including (1) all female Social Security beneficiaries 
(including widows); (2) retired-worker, non-widow beneficiaries; (3) 
spouse-only beneficiaries; and (4) dually-entitled workers. We removed 
the widows from the sample of retired workers so that we could compare 
a similar set of female beneficiaries (i.e. all non-widows). 

Table 5 provides information on the wealth and income characteristics 
of all female Social Security beneficiaries that were married at the 
time of the survey. Again, by restricting the sample to women who are 
in similar life circumstances (i.e. married), we can obtain a more 
accurate picture of the wealth and income levels of spouse-only 
beneficiaries (who are by definition married), relative to their 
married counterparts in the retired-worker and dually-entitled 
categories. 

Table 4: Characteristics of Female Social Security Beneficiaries: 2011: 

Age (years); 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 71.7; 
Standard error: 0.06; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 68.2; 
Standard error: 0.12; 
Spouse-only: 
Mean or Percent: 70.6; 
Standard error: 0.28; 
Dually-entitled spouse: 
Mean or Percent: 69.4; 
Standard error: 0.22. 

Marital status (percent): 

Never married; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 4%; 
Standard error: 0.3%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 10%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 0; 
Standard error: [Empty]; 
Dually-entitled spouse: 
Mean or Percent: 0; 
Standard error: [Empty]. 

Married; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 47%; 
Standard error: 0.5%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 68%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 95%; 
Standard error: 1%; 
Dually-entitled spouse: 
Mean or Percent: 91%; 
Standard error: 1%. 

Divorced; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 13%; 
Standard error: 0.5%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 22%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 5%; 
Standard error: 1%; 
Dually-entitled spouse: 
Mean or Percent: 9%; 
Standard error: 1%. 

Widowed; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 36%; 
Standard error: 0.5%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: [Empty]; 
Standard error: [Empty]; 
Spouse-only: 
Mean or Percent: 0; 
Standard error: [Empty]; 
Dually-entitled spouse: 
Mean or Percent: 0; 
Standard error: [Empty]. 

Number of children; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 2.7; 
Standard error: 0.02; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 2.2; 
Standard error: 0.03; 
Spouse-only: 
Mean or Percent: 3.1; 
Standard error: 0.09; 
Dually-entitled spouse: 
Mean or Percent: 2.8; 
Standard error: 0.06. 

Race and ethnicity (percent): 

White, Non-Hispanic; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 81%; 
Standard error: 0.4%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 79%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 81%; 
Standard error: 2%; 
Dually-entitled spouse: 
Mean or Percent: 90%; 
Standard error: 1%. 

Black, Non-Hispanic; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 9%; 
Standard error: 0.2%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 10%; 
Standard error: 0.5%; 
Spouse-only: 
Mean or Percent: 3%; 
Standard error: 1%; 
Dually-entitled spouse: 
Mean or Percent: 3%; 
Standard error: 1%. 

Hispanic; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 6%; 
Standard error: 0.3%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 7%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 11%; 
Standard error: 2%; 
Dually-entitled spouse: 
Mean or Percent: 4%; 
Standard error: 1%. 

Other, Non-Hispanic; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 4%; 
Standard error: 0.2%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 4%; 
Standard error: 0.4%; 
Spouse-only: 
Mean or Percent: 5%; 
Standard error: 1%; 
Dually-entitled spouse: 
Mean or Percent: 3%; 
Standard error: 1%. 

Education level (percent): 

No high school diploma; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 13%; 
Standard error: 0.5%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 8%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 20%; 
Standard error: 2%; 
Dually-entitled spouse: 
Mean or Percent: 8%; 
Standard error: 1%. 

High school diploma; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 37%; 
Standard error: 0.7%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 31%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 43%; 
Standard error: 2%; 
Dually-entitled spouse: 
Mean or Percent: 43%; 
Standard error: 2%. 

Some college; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 30%; 
Standard error: 0.6%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 34%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 19%; 
Standard error: 2%; 
Dually-entitled spouse: 
Mean or Percent: 30%; 
Standard error: 1%. 

Bachelor's degree or higher; 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 20%; 
Standard error: 0.5%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 27%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 18%; 
Standard error: 2%; 
Dually-entitled spouse: 
Mean or Percent: 19%; 
Standard error: 1%. 

Individual lifetime earnings (2012 $); 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: $721,333; 
Standard error: $7,982; 
Retired workers (excluding widows)[A]: 
Mean or Percent: $1,110,413; 
Standard error: $15,580; 
Spouse-only: 
Mean or Percent: $162,206; 
Standard error: $12,747; 
Dually-entitled spouse: 
Mean or Percent: $395,176; 
Standard error: $11,507. 

Household lifetime earnings (2012 $); 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: $1,737,804; 
Standard error: $42,050; 
Retired workers (excluding widows)[A]: 
Mean or Percent: $2,449,516; 
Standard error: $102,672; 
Spouse-only: 
Mean or Percent: $2,307,574; 
Standard error: $118,324; 
Dually-entitled spouse: 
Mean or Percent: $2,734,215; 
Standard error: $82,496. 

Own home (percent); 
All Female Social Security Beneficiaries (including widows): 
Mean or Percent: 84%; 
Standard error: 0.5%; 
Retired workers (excluding widows)[A]: 
Mean or Percent: 85%; 
Standard error: 1%; 
Spouse-only: 
Mean or Percent: 87%; 
Standard error: 2%; 
Dually-entitled spouse: 
Mean or Percent: 91%; 
Standard error: 1%. 

Observations[B]; 
All Female Social Security Beneficiaries (including widows): 6,701; 
Retired workers (excluding widows)[A]: 2,443; 
Spouse-only: 627; 
Dually-entitled spouse: 853. 

Source: GAO analysis of SIPP Gold Standard data. 

Note: Due to federal confidentiality requirements intended to protect 
the privacy of survey respondents, Census is not permitted to release 
maximum and minimum values of variables from the SIPP restricted Gold 
Standard data. 

[A] We removed the widows from the sample of retired workers so that we 
could compare a similar set of female beneficiaries (i.e. all non 
widows) since the spouse-only and dually-entitled spouse beneficiary 
categories by definition do not include widows. 

[B] Some variables had a lower number of observations, due to lower 
response rates for particular segments of the SIPP survey. 

[End of table] 

Table 5: Comparison of Wealth and Income Characteristics of Married 
Female Social Security Beneficiaries: 2011: 

Own home; 
Married Retired workers (excluding widows): 
Percent: 94; 
Standard error: 1; 
Married Spouse-only: [Empty]; 
Percent: 89; 
Standard error: 2; 
Married Dually-entitled-spouse: 
Percent: 93; 
Standard error: 1. 

Total net worth categories: 

Less than or equal to $150,000; 
Married Retired workers (excluding widows): 
Percent: 28; 
Standard error: 1; 
Married Spouse-only: 
Percent: 32; 
Standard error: 2; 
Married Dually-entitled-spouse: 
Percent: 23; 
Standard error: 2. 

Greater than $150,000 through $300,000; 
Married Retired workers (excluding widows): 
Percent: 21; 
Standard error: 1; 
Married Spouse-only: 
Percent: 17; 
Standard error: 1; 
Married Dually-entitled-spouse: 
Percent: 20; 
Standard error: 2. 

Greater than $300,000 through $450,000; 
Married Retired workers (excluding widows): 
Percent: 13; 
Standard error: 1; 
Married Spouse-only: 
Percent: 15; 
Standard error: 1; 
Married Dually-entitled-spouse: 
Percent: 15; 
Standard error: 1. 

Greater than $450,000 through $600,000; 
Married Retired workers (excluding widows): 
Percent: 8; 
Standard error: 1; 
Married Spouse-only: 
Percent: 7; 
Standard error: 1; 
Married Dually-entitled-spouse: 
Percent: 8; 
Standard error: 1. 

Greater than $600,000; 
Married Retired workers (excluding widows): 
Percent: 30; 
Standard error: 1; 
Married Spouse-only: 
Percent: 29; 
Standard error: 2; 
Married Dually-entitled-spouse: 
Percent: 34; 
Standard error: 2. 

Home equity categories: 

Less than or equal to $75,000; 
Married Retired workers (excluding widows): 
Percent: 27; 
Standard error: 1; 
Married Spouse-only: 
Percent: 32; 
Standard error: 2; 
Married Dually-entitled-spouse: 
Percent: 22; 
Standard error: 1. 

Greater than $75,000 through $150,000; 
Married Retired workers (excluding widows): 
Percent: 24; 
Standard error: 1; 
Married Spouse-only: 
Percent: 22; 
Standard error: 2; 
Married Dually-entitled-spouse:
Percent: 26; 
Standard error: 2. 

Greater than $150,000 through $225,000; 
Married Retired workers (excluding widows): 
Percent: 15; 
Standard error: 1; 
Married Spouse-only: 
Percent: 12; 
Standard error: 2; 
Married Dually-entitled-spouse:
Percent: 15; 
Standard error: 1. 

Greater than $225,000 through $300,000; 
Married Retired workers (excluding widows): 
Percent: 11; 
Standard error: 1; 
Married Spouse-only: 
Percent: 10; 
Standard error: 1; 
Married Dually-entitled-spouse: 
Percent: 12; 
Standard error: 1. 

Greater than $300,000; 
Married Retired workers (excluding widows): 
Percent: 23; 
Standard error: 1; 
Married Spouse-only: [Empty]; 
Percent: 23; 
Standard error: 2; 
Married Dually-entitled-spouse: 
Percent: 25; 
Standard error: 2. 

Household lifetime earnings categories: 

Less than or equal to $750,000; 
Married Retired workers (excluding widows): 
Percent: 5; 
Standard error: 1; 
Married Spouse-only: 
Percent: 11; 
Standard error: 1; 
Married Dually-entitled-spouse: 
Percent: 4; 
Standard error: 1. 

Greater than $750,000 through $1,500,000; 
Married Retired workers (excluding widows): 
Percent: 10; 
Standard error: 1; 
Married Spouse-only: 
Percent: 21; 
Standard error: 2; 
Married Dually-entitled-spouse: 
Percent: 7; 
Standard error: 1. 

Greater than $1,500,000 through $2,250,000; 
Married Retired workers (excluding widows): 
Percent: 17; 
Standard error: 1; 
Married Spouse-only: 
Percent: 26; 
Standard error: 2; 
Married Dually-entitled-spouse: 
Percent: 22; 
Standard error: 2. 

Greater than $2,250,000; 
Married Retired workers (excluding widows): 
Percent: 68; 
Standard error: 1; 
Married Spouse-only: 
Percent: 42; 
Standard error: 02; 
Married Dually-entitled-spouse: 
Percent: 68; 
Standard error: 2. 

Observations; 
Married Retired workers (excluding widows): 1,594; 
Married Spouse-only: 580; 
Married Dually-entitled-spouse: 764. 

Source: GAO analysis of SIPP Gold Standard data. 

Note: Due to federal confidentiality requirements intended to protect 
the privacy of survey respondents, Census is not permitted to release 
maximum and minimum values of variables from the SIPP Gold Standard 
data. Some variables had a lower number of observations, due to lower 
response rates for particular segments of the SIPP survey. Some within 
group percentages do not total 100 due to rounding. 

[End of table] 

Section 3: Analysis of the Characteristics of Married Households that 
Opt Out of Survivor Benefits: 

To understand the characteristics of married households that opt not 
to receive a survivor benefit for Objective 2, we analyzed a sample of 
married households currently receiving pension income from the SCF 
data. To identify these households, we used three questions from the 
SCF survey. Specifically, we selected households that: 

* Responded to "Is anyone in the household receiving pension benefits 
currently?" with "Yes"; 

* Responded to "Is this pension currently an account plan, such as a 
401(k), where you could take the whole balance as one payment if you 
wanted to?" with "No;" and: 

* Responded to "Is this a payment or account from a current job, past 
job, a disability or military benefit, former spouse's pension, or 
something else?" with either "Current job pension of respondent or 
spouse/partner," "Past job pension of respondent or spouse/partner 
(except military)" or "Disability." 

Among the households that are currently receiving pension benefits, we 
identified households that opted for a survivor benefit by using an 
additional question from the SCF survey. Specifically, we selected 
households that: 

* Responded to "If you were to die, what percent of this current 
retirement payment would your spouse/partner continue to receive?" 
with "Start at a later time, no indication of reduced benefit," "Start 
at a later time, indication of reduced benefit," "Lump sum payout," 
"Full amount for limited period," "Full amount, no time limit or time 
period not specified," or "other." 

To ensure that we categorized these households appropriately, we 
consulted an official at the Federal Reserve. 

Table 6 presents descriptive information on the characteristics of 
married households currently receiving pension income (presented in 
the third column) as well as on all households currently receiving 
pension income (presented in the middle column). 

Table 6: Characteristics of Households Currently Receiving Pension 
Income: 

Age of household head: Under 45 years old; 
Percentage of all households: 2%; 
Percentage of all married households: 2%. 

Age of household head: 45-54 years old; 
Percentage of all households: 5%; 
Percentage of all married households: 5%. 

Age of household head: 55-64 years old; 
Percentage of all households: 22%; 
Percentage of all married households: 26%. 

Age of household head: 65-69 years old; 
Percentage of all households: 17%; 
Percentage of all married households: 22%. 

Age of household head: 70-79 years old; 
Percentage of all households: 33%; 
Percentage of all married households: 30%. 

Age of household head: 80 years and older; 
Percentage of all households: 21%; 
Percentage of all married households: 15%. 

Marital status: Married; 
Percentage of all households: 54%; 
Percentage of all married households: 100%. 

Marital status: Living with partner; 
Percentage of all households: 2%; 
Percentage of all married households: 0. 

Marital status: Separated; 
Percentage of all households: 1%; 
Percentage of all married households: 0. 

Marital status: Divorced; 
Percentage of all households: 13%; 
Percentage of all married households: 0. 

Marital status: Widowed; 
Percentage of all households: 26%; 
Percentage of all married households: 0. 

Marital status: Never married; 
Percentage of all households: 4%; 
Percentage of all married households: 0. 

Current length of marriage: 0-10 years; 
Percentage of all households: NA; 
Percentage of all married households: 12%. 

Current length of marriage: 11-20 years; 
Percentage of all households: NA; 
Percentage of all married households: 10%. 

Current length of marriage: 21-40 years; 
Percentage of all households: NA; 
Percentage of all married households: 35%. 

Current length of marriage: Over 40 years; 
Percentage of all households: NA; 
Percentage of all married households: 42%. 

Age of spouse/partner: 18-34 years old; 
Percentage of all households: NA; 
Percentage of all married households: 1%. 

Age of spouse/partner: 35-44 years old; 
Percentage of all households: NA; 
Percentage of all married households: 2%. 

Age of spouse/partner: 45-54 years old; 
Percentage of all households: NA; 
Percentage of all married households: 12%. 

Age of spouse/partner: 55-64 years old; 
Percentage of all households: NA; 
Percentage of all married households: 32%. 

Age of spouse/partner: 65-69 years old; 
Percentage of all households: NA; 
Percentage of all married households: 20%. 

Age of spouse/partner: 70-79 years old; 
Percentage of all households: NA; 
Percentage of all married households: 25%. 

Age of spouse/partner: 80 years and older; 
Percentage of all 
households: NA; 
Percentage of all married households: 7%. 

Race/ethnicity of household head: White non-Hispanic; 
Percentage of all households: 83%; 
Percentage of all married households: 85%. 

Race/ethnicity of household head: Black; 
Percentage of all households: 11%; 
Percentage of all married households: 8%. 

Race/ethnicity of household head: Hispanic; 
Percentage of all households: 2%; 
Percentage of all married households: 3%. 

Race/ethnicity of household head: Other; 
Percentage of all households: 3%; 
Percentage of all married households: 3%. 

Education of household head: No high school diploma/GED; 
Percentage of all households: 10%; 
Percentage of all married households: 9%. 

Education of household head: High school diploma or GED; 
Percentage of all households: 36%; 
Percentage of all married households: 32%. 

Education of household head: Some college; 
Percentage of all households: 17%; 
Percentage of all married households: 18%. 

Education of household head: College degree or more; 
Percentage of all households: 38%; 
Percentage of all married households: 41%. 

Household income bracket: less than $25,000; 
Percentage of all households: 21%; 

Percentage of all married households: 7%. 

Household income bracket: $25,000-$49,999; 
Percentage of all households: 32%; 
Percentage of all married households: 30%. 

Household income bracket: $50,000-$74,999; 
Percentage of all households: 21%; 
Percentage of all married households: 24%. 

Household income bracket: $75,000-$99,999; 
Percentage of all households: 10%; 
Percentage of all married households: 13%. 

Household income bracket: $100,000 - $249,999; 
Percentage of all households: 14%; 
Percentage of all married households: 22%. 

Household income bracket: $250,000 or greater; 
Percentage of all households: 2%; 
Percentage of all married households: 4%. 

Household net worth bracket: less than $50,000; 
Percentage of all households: 18%; 
Percentage of all married households: 10%. 

Household net worth bracket: $50,000 - $199,999; 
Percentage of all households: 28%; 
Percentage of all married households: 25%. 

Household net worth bracket: $200,000 - $399,999; 
Percentage of all households: 20%; 
Percentage of all married households: 23%. 

Household net worth bracket: $400,000 - $999,999; 
Percentage of all households: 21%; 
Percentage of all married households: 26%. 

Household net worth bracket: $1,000,000 - $1,999,999; 
Percentage of all households: 9%; 
Percentage of all married households: 11%. 

Household net worth bracket: $2,000,000 or greater; 
Percentage of all households: 4%; 
Percentage of all married households: 5%. 

Source: GAO analysis of 2010 Survey of Consumer Finances data. 

Table notes: Due to limitations in the SCF data, households in the 
above analysis include both households that are currently receiving 
retirement income from a defined benefit plan or have chosen to 
annuitize their defined contribution benefits. The estimates presented 
in this table have 95 percent confidence intervals that are within +/- 
2.10 percentage points of the estimate itself. Some within group 
percentages do not total 100 due to rounding. 

[End of table] 

To understand the savings behavior of working-age households for 
Objective 2, we analyzed a sample of working-age households from the 
SCF data. Table 7 presents descriptive information on all of the 
working-age households and married households in this sample. 

Table 7: Characteristics of Working-age Households: 

Age of household head: 18-34 years old; 
Percentage of all households: 27%; 
Percentage of all married households: 19%. 

Age of household head: 35-44 years old; 
Percentage of all households: 23%; 
Percentage of all married households: 26%. 

Age of household head: 45-54 years old; 
Percentage of all households: 27%; 
Percentage of all married households: 31%. 

Age of household head: 55-64 years old; 
Percentage of all households: 23%; 
Percentage of all married households: 25%. 

Marital status: Married; 
Percentage of all households: 51%; 
Percentage of all married households: 100%. 

Marital status: Living with partner; 
Percentage of all households: 9%. 

Marital status: Separated; 
Percentage of all households: 3%. 

Marital status: Divorced; 
Percentage of all households: 14%. 

Marital status: Widowed; 
Percentage of all households: 3%. 

Marital status: Never married; 
Percentage of all households: 19%. 

Age of spouse/partner: 18-34 years old; 
Percentage of all households: NA; 
Percentage of all married households: 24%. 

Age of spouse/partner: 35-44 years old; 
Percentage of all households: NA; 
Percentage of all married households: 27%. 

Age of spouse/partner: 45-54 years old; 
Percentage of all households: NA; 
Percentage of all married households: 30%. 

Age of spouse/partner: 55-64 years old; 
Percentage of all households: NA; 
Percentage of all married households: 17%. 

Age of spouse/partner: 65-69 years old; 
Percentage of all households: NA; 
Percentage of all married households: 1%. 

Age of spouse/partner: 70-79 years or older; 
Percentage of all households: NA; 
Percentage of all married households: 0%. 

Race/ethnicity of household head: White non-Hispanic; 
Percentage of all households: 68%; 
Percentage of all married households: 74%. 

Race/ethnicity of household head: Black; 
Percentage of all households: 15%; 
Percentage of all married households: 8%. 

Race/ethnicity of household head: Hispanic; 
Percentage of all households: 13%; 
Percentage of all married households: 13%. 

Race/ethnicity of household head: Other; 
Percentage of all households: 5%; 
Percentage of all married households: 5%. 

Education of household head: No high school diploma/GED; 
Percentage of all households: 10%; 
Percentage of all married households: 9%. 

Education of household head: High school diploma or GED; 
Percentage of all households: 31%; 
Percentage of all married households: 30%. 

Education of household head: Some college; 
Percentage of all households: 20%; 
Percentage of all married households: 18%. 

Education of household head: College degree or more; 
Percentage of all households: 39%; 
Percentage of all married households: 43%. 

Household type: Married; 
Percentage of all households: 51%; 
Percentage of all married households: 100%. 

Household type: Living with partner; 
Percentage of all households: 9%. 

Household type: Female single-headed; 
Percentage of all households: 24%. 

Household type: Male single-headed; 
Percentage of all households: 16%. 

Labor force participation[A]: Male 1-earner household; 
Percentage of all households: 29%; 
Percentage of all married households: 27%. 

Labor force participation[A]: Female 1-earner household; 
Percentage of all households: 23%; 
Percentage of all married households: 6%. 

Labor force participation[A]: 2-earner household; 
Percentage of all households: 39%; 
Percentage of all married households: 64%. 

Labor force participation[A]: 0-earner household; 
Percentage of all households: 9%; 
Percentage of all married households: 4%. 

Household income bracket: less than $25,000; 
Percentage of all households: 24%; 
Percentage of all married households: 10%. 

Household income bracket: $25,000-$49,999; 
Percentage of all households: 27%; 
Percentage of all married households: 20%. 

Household income bracket: $50,000-$74,999; 
Percentage of all households: 18%; 
Percentage of all married households: 21%. 

Household income bracket: $75,000-$99,999; 
Percentage of all households: 11%; 
Percentage of all married households: 16%. 

Household income bracket: $100,000 - $249,999; 
Percentage of all households: 16%; 
Percentage of all married households: 26%. 

Household income bracket: $250,000 or greater; 
Percentage of all households: 4%; 
Percentage of all married households: 7%. 

Household net worth bracket: less than $50,000; 
Percentage of all households: 49%; 
Percentage of all married households: 34%. 

Household net worth bracket: $50,000 - $199,999; 
Percentage of all households: 23%; 
Percentage of all married households: 26%. 

Household net worth bracket: $200,000 - $399,999; 
Percentage of all households: 11%; 
Percentage of all married households: 14%. 

Household net worth bracket: $400,000 - $999,999; 
Percentage of all households: 9%; 
Percentage of all married households: 13%. 

Household net worth bracket: $1,000,000 - $1,999,999; 
Percentage of all households: 4%; 
Percentage of all married households: 6%. 

Household net worth bracket: $2,000,000 or greater; 
Percentage of all households: 4%; 
Percentage of all married households: 7%. 

Balance of retirement savings accounts[B]: less than $25,000; 
Percentage of all households: 68%; 
Percentage of all married households: 55%. 

Balance of retirement savings accounts[B]: $25,000 - $49,999; 
Percentage of all households: 7%; 
Percentage of all married households: 9%. 

Balance of retirement savings accounts[B]: $50,000 - $99,999; 
Percentage of all households: 7%; 
Percentage of all married households: 10%. 

Balance of retirement savings accounts[B]: $100,000 - $249,999; 
Percentage of all households: 8%; 
Percentage of all married households: 13%. 

Balance of retirement savings accounts[B]: $250,000 - $499,999; 
Percentage of all households: 4%; 
Percentage of all married households: 7%. 

Balance of retirement savings accounts[B]: $500,000 - $999,999; 
Percentage of all households: 2%; 
Percentage of all married households: 4%. 

Balance of retirement savings accounts[B]: $1,000,000 or more; 
Percentage of all households: 1%; 
Percentage of all married households: 2%. 

Source: GAO analysis of 2010 Survey of Consumer Finances data. 

Table notes: The estimates presented in this table have 95 percent 
confidence intervals that are within +/-1.06 percentage points of the 
estimate itself. Some within group percentages do not total 100 due to 
rounding. 

[A] The type of households (male sole earner, female sole earner, and 
dual-earner) reflects the employment status of both household members 
at a particular point in time. Therefore, it is not surprising that we 
observe a portion of retirement savings contributions from the non-
working spouse, which could reflect his or her labor force 
participation in prior years. 

[B] Includes the balance of all defined contribution, IRA (Roth, 
regular, and rollover), and Keogh accounts held by all members of the 
primary economic unit, as defined by the Survey of Consumer Finances. 

[End of table] 

[End of section] 

Appendix II: Selected Social Security Policy Options That May Offset 
Effects of Changes in Marital Patterns: 

This appendix provides a summary of Social Security policy options, 
suggested by experts and cited in past GAO reports, to improve the 
retirement security of vulnerable populations.[Footnote 75] The 
options listed are not intended to be exhaustive, and each selected 
policy option may include multiple proposals. In this report, we are 
not recommending or endorsing the adoption of any particular policy 
option. Rather, we identify them as potential options for policymakers 
to consider, along with other relevant proposals, as they address 
retirement security issues. Table 8 presents options that focus 
specifically on modifying benefits for spouses and survivors. Table 9 
presents other options that would affect broader populations of 
vulnerable individuals, including those who do not qualify for spousal 
or survivor benefits, such as the never-married. 

Table 8: Selected Policy Options That Target Spouses and Survivors: 

Selected policy options: See GAO reports; 
Reduce the minimum marriage duration required for divorced spouse and 
survivor benefits: 
[hyperlink, http://www.gao.gov/products/GAO-08-105]; 
[hyperlink, http://www.gao.gov/products/GAO-10-101R]; 
[hyperlink, http://www.gao.gov/products/GAO-12-699]; 
Increase spousal benefits for divorced spouses: 
[hyperlink, http://www.gao.gov/products/GAO-12-699]; 
Increase survivor benefits: 
[hyperlink, http://www.gao.gov/products/GAO-08-105], 
[hyperlink, http://www.gao.gov/products/GAO-10-101R], 
Reduce spousal benefits & increase survivor benefits: 
[hyperlink, http://www.gao.gov/products/GAO-12-699]; 
Increase benefits for disabled surviving spouses: 
[hyperlink, http://www.gao.gov/products/GAO-08-105]; 
[hyperlink, http://www.gao.gov/products/GAO-12-699]; 
Adopt earnings sharing: 
[hyperlink, http://www.gao.gov/products/GAO-10-101R]. 

Selected policy options: Targeted group; 
Reduce the minimum marriage duration required for divorced spouse and 
survivor benefits: This option targets divorced spouses whose 
marriages lasted less than 10 years and therefore do not qualify for 
divorced spouse and survivor benefits; 
Increase spousal benefits for divorced spouses: This option targets 
divorced spouses who were married for at least 10 years and who qualify 
for divorced spouse and survivor benefits; 
Increase survivor benefits: This option targets widowers; 
Reduce spousal benefits & increase survivor benefits: This option 
targets widowers; 
Increase benefits for disabled surviving spouses: This option targets 
disabled surviving spouses; 
Adopt earnings sharing: This option targets divorced spouses whose 
marriages lasted less than 10 years and therefore do not qualify for 
spouse or survivor benefits, and whose incomes while married were lower 
than their spouses. An earnings sharing approach is often proposed as 
an alternative to existing spousal and survivor benefits. 

Selected policy options: What happens now; 
Reduce the minimum marriage duration required for divorced spouse and 
survivor benefits: Currently, a divorced spouse can receive benefits 
based on a retired worker's earnings record if the marriage lasted at 
least 10 years; 
Increase spousal benefits for divorced spouses: Currently, divorced 
spouses who qualify for spousal benefits receive a benefit equal to up 
to 50 percent of the worker's benefits. Divorced spouses who qualify 
for widow/widower benefits receive a benefit of up to 100 percent of 
the deceased worker's benefits; 
Increase survivor benefits: Currently, a surviving spouse at full 
retirement age or older typically receives 100 percent of the worker's 
basic benefit amount. This decreases household income upon the death of 
one spouse by about one-third if the couple's retirement benefits had 
been based on one spouse's work history, and up to 50 percent if both 
spouses had been receiving retired worker benefits; 
Reduce spousal benefits & increase survivor benefits: Currently, a 
surviving spouse at full retirement age or older typically receives 100 
percent of the worker's basic benefit amount. This decreases household 
income upon the death of one spouse by about one-third if the couple's 
retirement benefits had been based on one spouse's work history and up 
to 50 percent if both spouses had been receiving retired worker 
benefits. Dual-earner couples experience a proportionally greater 
decrease in benefits upon the death of a spouse than single-earner 
couples experience. Researchers have expressed concern about whether 
this decline in total household benefits is too large to maintain the 
survivor's previous standard of living; 
Increase benefits for disabled surviving spouses: Currently, to 
qualify for disabled surviving spouse benefits, disabled surviving 
spouses must be at least age 50 and have become disabled not later 
than 7 years after the spouse's death or 7 years after last being 
eligible for mothers or father's benefits or widower's benefits based 
upon a disability. In addition, disabled surviving spouses younger 
than the full retirement age generally receive lower benefits than 
those who wait to receive their benefits until the full retirement age; 
Adopt earnings sharing: Under the current system, a spouse who has not 
worked or who has low earnings can be entitled to a benefit equal to 
as much as one-half of the retired worker's full benefit. The total 
benefit received by the couple would be 150 percent of the worker's 
benefit. If the spouse is divorced, he or she can still get benefits 
based on a retired worker's earnings record if the marriage lasted at 
least 10 years, and the spouse is unmarried and at least 62 years old. 

Selected policy options: Selected reform options; 
Reduce the minimum marriage duration required for divorced spouse and 
survivor benefits: Experts have recommended expanding eligibility for 
divorced spousal benefits to require a minimum of 7 years of marriage. 
Additionally, some experts have suggested marriage years could also be 
accumulated across multiple marriages. One Social Security reform 
proposal suggests that reducing the required marriage duration could 
be combined with a minimum work requirement for the divorced spouse. 
Combining at least 7 years of marriage with a minimum of 3 years of 
work would mimic the standard 10-year work requirement for Social 
Security retirement benefits; 
Increase spousal benefits for divorced spouses: Experts have suggested 
raising benefits for divorced spouses to 75 percent of the former 
spouse's benefit while the former spouse is still alive. Upon the 
death of the former spouse, the surviving divorced spouse would 
receive the full widower's benefit of 100 percent; 
Increase survivor benefits: Experts have proposed calculating this new 
benefit in different ways. For example, the surviving spouse could 
receive 75 percent of the couple's retired-workers benefit but the 
benefit would be capped at the maximum earner's benefit or at the 
benefit of the "lifelong average earner." This option could also be 
targeted specifically toward low-income survivors, for example, by 
including a cap; 
Reduce spousal benefits & increase survivor benefits: Some experts 
have proposed an increase in survivor benefits be paired with a 
decrease in spousal benefits, from one-half the retired worker's 
benefit to one-third; 
Increase benefits for disabled surviving spouses: Experts have 
proposed raising benefits for disabled surviving spouses to 100 
percent of the deceased spouse's benefit. It would also remove the 7 
year limitation and the age 50 requirement. Lastly, it would make 
divorced spouses who are disabled eligible for benefits on the same 
basis as disabled surviving spouses; 
Adopt earnings sharing: Earnings sharing, as proposed by experts, 
would combine married individuals' annual earnings and evenly divide 
them between the two spouses for each year of marriage when 
calculating individuals' Social Security retirement benefits. Each 
spouse would accrue an individual benefit, even if only one of them 
worked. Divorced spouses whose marriages lasted less than 10 years 
would be entitled to the individual benefits accrued during the 
marriage. This option is also seen as a way to equalize benefits 
received by dual-earner married couples with those of single-earner 
couples. 

Selected policy options: Some potential effects on the elderly; 
Reduce the minimum marriage duration required for divorced spouse and 
survivor benefits: More divorced elderly would qualify for spousal 
benefits. For example, one study estimated that lowering the marriage-
duration requirement from 10 to 7 years would increase benefits for 
about 8 percent of all divorced women age 62 and over in the year 
2030. Proponents of this option note that reducing the marriage 
requirement from 10 to 7 years would reflect current trends for 
shorter marriages. However, this option could benefit higher-income 
elderly who are not economically vulnerable; 
Increase spousal benefits for divorced spouses: Divorce can result in 
a substantial loss of assets and income. Some experts argue that a 50 
percent benefit is not enough to keep divorced spouses from falling 
into poverty. It has been estimated that increasing the benefit rate 
for divorced spouses to 75 percent would lower the poverty rate among 
divorced spouses from 30 percent to 11 percent; 
Increase survivor benefits: Increasing survivor benefits could help 
widowers maintain their standard of living. However this option would 
not address benefit adequacy for those who do not qualify for spousal 
or survivor benefits; 
Reduce spousal benefits & increase survivor benefits: Increasing 
survivor benefits could help widowers maintain their standard of 
living. It would also smooth household benefit levels before and after 
widowhood. However, decreasing spousal benefits could have a negative 
impact on the couple while both spouses are alive and would not 
address benefit adequacy for those who do not qualify for spousal or 
survivor benefits; 
Increase benefits for disabled surviving spouses: This option would 
increase benefits for a very vulnerable population. Both divorce and 
widowhood can result in a decrease in retirement security. Further, 
disabled surviving spouses, including those who have been divorced, 
cannot work and may have no partner to depend on for support. In 
addition, disability issues affect a high number of divorced spouses, 
making them more vulnerable to income insecurity. One study estimated 
that more than one-fifth of all divorced spouses had health problems 
that meet disability criteria established by SSA; 
Adopt earnings sharing: Earnings sharing could increase benefits for 
divorced elderly. However, this option would not do much to improve 
benefits for economically vulnerable beneficiaries, in part, because it 
is not well targeted. For example, SSA's simulations found that 
earnings sharing would decrease benefits for the majority of future 
retirees, although benefits for some would increase. 

Selected policy options: Other possible implications; 
Reduce the minimum marriage duration required for divorced spouse and 
survivor benefits: The extent to which this option would increase 
Social Security costs depends on how many people would become eligible 
with a shorter marriage requirement and the way the option is 
designed. For example, not including a corresponding work requirement 
would increase costs more because people who have no work history 
would also be eligible; 
Increase spousal benefits for divorced spouses: This benefit increase 
would increase costs, thereby decreasing solvency; 
Increase survivor benefits: Increasing benefits would increase costs 
and decrease solvency. The extent to which this option increases costs 
depends on how much greater the benefit amount is across all eligible 
survivors. Capping the amount of the increase based on income could 
help limit costs. This option could also be complex to administer, in 
part because it uses a "couple's benefit" as a baseline for 
calculating survivor benefits. Since such a benefit does not currently 
exist in the Social Security system this could be problematic, for 
example, in cases where one of the spouses dies before retiring. In 
addition, there are many complicated rules for survivors because of an 
existing provision, called the widower's limit, that caps benefit 
amounts for some survivors. Benefit increases expected under this 
option could be negated by this provision; 
Reduce spousal benefits & increase survivor benefits: This option could 
provide savings to the Social Security system to partially or fully 
offset the costs of the increased survivor benefits; 
Increase benefits for disabled surviving spouses: This benefit 
increase would decrease solvency because it would increase costs. See 
above; 
Adopt earnings sharing: Because earnings sharing would increase 
benefits for some but decrease them for others, its net impact on 
Social Security's solvency is unclear. The extent to which this option 
increases SSA's workload depends on the number of newly-eligible 
people who would receive benefits. Some additional administrative 
effort and cost would also be required to transition from the current 
system's spousal benefit to an earnings sharing approach, in part 
because of the need to verify marriage and divorce data. 

Source: GAO analysis of prior GAO reports. 

[End of table] 

Table 9: Selected Policy Options That Target All Elderly, Regardless 
of Marital Status: 

Selected policy options: See GAO reports; 
Increase the minimum benefit: 
[hyperlink, http://www.gao.gov/products/GAO-08-105]; 
[hyperlink, http://www.gao.gov/products/GAO-10-101R], 
[hyperlink, http://www.gao.gov/products/GAO-12-699]; 
Provide caregiver credits: 
[hyperlink, http://www.gao.gov/products/GAO-08-105], 
[hyperlink, http://www.gao.gov/products/GAO-10-101R], 
[hyperlink, http://www.gao.gov/products/GAO-12-699]; 
Reduce work requirements for eligibility: 
[hyperlink, http://www.gao.gov/products/GAO-10-101R]; 
Supplement benefits for low-income single workers: 
[hyperlink, http://www.gao.gov/products/GAO-10-101R]; 
Provide an additional Social Security benefit to the oldest old: 
[hyperlink, http://www.gao.gov/products/GAO-10-101R], 
[hyperlink, http://www.gao.gov/products/GAO-12-699]. 

Selected policy options: Targeted group; 
Increase the minimum benefit: The guaranteed minimum benefit option 
targets lifetime low earners; 
Provide caregiver credits: Caregiver credits seek to improve benefit 
adequacy for workers, primarily women, who have shorter earnings 
records because they spent time providing care for children or elderly 
relatives and do not qualify for spousal benefits; 
Reduce work requirements for eligibility: This option targets workers 
with low lifetime earnings due to short work histories, as opposed to 
those with long histories of low earnings; 
Supplement benefits for low-income single workers: This option targets 
low-income workers who never married or were not married long enough 
to qualify for spousal benefits; 
Provide an additional Social Security benefit to the oldest old: 
Providing longevity insurance targets the oldest Social Security 
beneficiaries. 

Selected policy options: What happens now; 
Increase the minimum benefit: A "special minimum" benefit provision 
was enacted in 1972 for workers with 11 or more covered years of 
employment. However, because the minimum benefit amount has not kept 
pace with wage growth, few people still qualify for the benefit. For 
workers not eligible for the special minimum benefit, benefits are 
generally calculated on the basis of a worker's average indexed monthly 
earnings during the 35 years in which they were the highest. While the 
benefit structure is progressive, a lifetime minimum-or low-wage worker 
would still have correspondingly low benefits. The current benefit 
formula does not distinguish between low average wages caused by low 
lifetime earnings or low average wages caused by years of unemployment; 
Provide caregiver credits: Under the current system, Social Security 
eligibility and benefit amounts for retired workers depend on the 
amount of time a worker spends in covered employment. Time spent out 
of covered employment as a caregiver may reduce benefits for workers, 
and others may not work enough to earn the required 40 credits 
generally required to be eligible for benefits; 
Reduce work requirements for eligibility: Under current law, workers 
generally must accrue 40 credits--about 10 years of earnings--in 
covered employment to be eligible for Social Security retirement 
benefits based on their own work record; 
Supplement benefits for low-income single workers: In 2009, the 
benefit formula replaced 90 percent of the first $744 of a worker's 
average indexed monthly earnings (AIME), 32 percent of the AIME 
between $745 and $4,483, and 15 percent of the AIME above $4,483, up 
to a cap. A worker's AIME is calculated based on a worker's highest 35 
years' earnings, after earnings have been indexed for wage growth over 
time; 
Provide an additional Social Security benefit to the oldest old: As 
people grow older, they risk outliving their other resources, become 
less able to work, and become more dependent on Social Security 
benefits for their income. While Social Security benefits are intended 
to replace lost wages and are adjusted annually to reflect price 
inflation, they are not meant to be the sole source of retirement 
income. However, the value of other income sources, such as pensions 
and annuities, may be eroded by inflation over time. 

Selected policy options: Selected reform options; 
Increase the minimum benefit: Experts have suggested several options 
for increasing the minimum benefit. The amount and structure of the 
benefit varies among proposals, but most minimum benefit options are 
designed to address benefit adequacy by providing a retirement benefit 
equal to some multiple of the federal poverty level, with the multiple 
based on years worked in covered employment. For example, one option 
would provide a minimum benefit equal to 120 percent of the federal 
poverty level for a minimum-wage earner who had worked for 30 years. 
Another option would provide a minimum benefit equal to 100 percent of 
the federal poverty level for a 30-year worker and 111 percent of the 
poverty level for a 40-year worker; 
Provide caregiver credits: Experts have suggested a caregiver credit 
option, which can be designed in different ways. One design allows a 
specified amount of caregiving time, such as 3 or 4 years, to count as 
covered employment, and assigns a wage to that time. For example, an 
average wage for all workers could be assigned or a wage linked to an 
individual beneficiary's prior earnings could be used. Another design 
would exclude a limited number of caregiving years from the benefit 
calculation so that instead of averaging earnings over 35 years, 
earnings would be averaged over fewer years. A final design supplements 
caregivers' retired worker benefits directly, regardless of whether 
they took time out of the workforce for caregiving. For example, an 
income-tested supplement could be given to increase retired worker 
benefits by 75 percent for those who have one child and 80 percent for 
those with two or more children. Both parents of a child would be 
eligible for this supplement, as long as the total household income did 
not exceed 125 percent of the federal poverty level; 
Reduce work requirements for eligibility: Experts have suggested 
reducing the work requirements for Social Security retirement benefit 
eligibility, which would enable people who have shorter earnings 
histories to receive benefits. Benefit amounts would be calculated 
under the existing formula, which uses the worker's average indexed 
monthly earnings during the 35 years in which he or she earned the 
most, even if there were no earnings from covered employment during 
some of those years; 
Supplement benefits for low-income single workers: Experts have 
suggested supplementing benefits for low-income single workers by 
adjusting the formula used to calculate Social Security retirement 
benefits. In one proposal, the first threshold in the benefit formula 
would be adjusted or supplemented so that it increased by one-half, 
from $744 to $1,116 in 2009, for eligible beneficiaries. The benefit 
amount would be capped to prevent eligible workers from receiving 
higher benefits than those who just miss qualifying for the supplement. 
To be eligible for the supplement, a worker's AIME must be lower than 
a multiple of the existing formula's first threshold, such as 150 
percent or 300 percent. For example, if the multiple were set at 300 
percent, a worker whose AIME was less than $2,232 (3 x $744) in 2009 
would qualify. To receive the supplement, a worker must have at least 
30 years of covered employment and the worker cannot be eligible for 
spousal benefits, nor can anyone else claim spousal benefits based on 
that worker's earnings record; 
Provide an additional Social Security benefit to the oldest old: 
Experts have suggested that providing longevity insurance addresses 
concerns about benefit adequacy by increasing Social Security 
retirement benefits for beneficiaries who reach an advanced age. For 
example, Social Security recipients over the age of 80 or 85 could 
receive an additional benefit, such as an extra 5 percent on top of 
their regular benefit. Longevity insurance seeks to reduce the risk 
that the very old fall into poverty by increasing their Social 
Security benefits. This option could be targeted specifically toward 
low-income beneficiaries, or provided to all those who reach an 
advanced age. Work history could be an additional condition for 
eligibility. For example, one longevity insurance proposal would 
increase benefits for people who have low benefits at age 82 and have 
at least 20 years of covered employment. It would provide a minimum 
benefit equal to 70 percent of the federal poverty level for a 20-year 
worker and increases the benefit for each additional year of work. 
Another proposal would increase benefits by 10 percent at age 85 for 
30-year workers whose benefits are lower than 75 percent of the 
average benefit all workers receive. 

Selected policy options: Some potential effects on the elderly; 
Increase the minimum benefit: A guaranteed minimum benefit could help 
keep more elderly, lifetime low-earners out of poverty by increasing 
their monthly income. However, it is difficult to target lifetime low 
earners effectively. For example, requiring a long work history is 
problematic because low earners often have recurring periods of 
unemployment and cannot satisfy such a requirement. Thus, the target 
population may not be reached if a lifetime of work is required to 
earn the benefit. However, if a lifetime of work is not required, some 
people outside the target population would also benefit, e.g., higher-
wage workers who worked for a short period of time; 
Provide caregiver credits: Caregiver credits seek to improve benefit 
adequacy for workers, primarily women, who have shorter earnings 
records because they spent time providing care for children or elderly 
relatives and do not qualify for spousal benefits because they never 
married or were not married long enough to qualify for them. For 
various reasons, implementing this option may not reach the target 
population. For example, low-income people are less likely to be able 
to take time off from work. Therefore, people who have relatively 
higher incomes may benefit more from the creation of caregiver 
credits; 
Reduce work requirements for eligibility: Reducing the Social Security 
work requirement is an option that targets workers with low lifetime 
earnings due to short work histories, as opposed to those with long 
histories of low earnings. According to SSA officials, many people 
fall just short of the 40 credits requirement because they have 
intermittent work histories. However, officials also said many of 
those people may already be eligible for spousal benefits, resulting 
in few people benefiting from this option. In addition, agency 
officials and experts said benefits based on such short work histories 
are likely to be very low and questioned the effectiveness of this 
option in addressing benefit adequacy; 
Supplement benefits for low-income single workers: The benefit 
supplement option targets lifetime low earners who never married or 
were not married long enough to qualify for spousal benefits. Low-
income single and divorced women are expected to benefit most from 
this option; 
Provide an additional Social Security benefit to the oldest old: 
Women, who tend to live longer than men, would be more likely to 
receive this extra benefit. Older women may need extra benefits as 
income and assets may have been used to care for a deceased spouse or 
to pay for increasing medical costs. An additional benefit may be 
particularly helpful for low-income women. While this could be an 
effective option for addressing concerns about benefit adequacy for 
the very old, unless this option is specifically targeted toward low-
income beneficiaries, most of the benefits would accrue to higher-
income people because they tend to live longer. In addition, this 
option could create disincentives to save for retirement or incentives 
to spend down resources before beneficiaries become old enough to 
qualify for the longevity increase. 

Selected policy options: Other possible implications; 
Increase the minimum benefit: Cost implications of this option depend 
on the number of work years required for eligibility, since that 
requirement will directly influence the number of people who would 
qualify for benefit increases. A shorter work requirement will result 
in more people being eligible, and thus costs will be higher. 
Additionally, the multiple of the federal poverty level used can have 
a significant impact on cost; 
Provide caregiver credits: Because caregiver credits increase 
benefits, they have cost implications for Social Security's solvency. 
The extent to which this option affects solvency depends largely on 
who would be eligible to receive the credit: one or both parents, all 
caregivers, or just those who have low incomes. Extending eligibility 
to a greater number of people will increase costs. In addition, the 
number of years that credits may be received and the wage assigned to 
those years will impact costs. Caregiver credits would be complex to 
administer. A key issue is how to verify that care was provided to a 
qualifying person. While a birth certificate could be used to document 
child care, elder care would be more burdensome to document. Measuring 
time off and verifying that caregiving actually occurred would also be 
difficult; 
Reduce work requirements for eligibility: Because this option 
increases the number of people receiving benefits, it has cost 
implications for Social Security's solvency. The number of credits 
required will directly influence the number of people who would be 
newly eligible for benefits. A shorter work requirement will result in 
more people being eligible. However, because few people are actually 
expected to receive benefits under this option, and those who do are 
expected to receive modest benefits, the impact of a reduced work 
requirement on program solvency is unlikely to be very large. 
Furthermore, because few people are expected to gain eligibility under 
this option, the impact on SSA's workload is likely to be small; 
Supplement benefits for low-income single workers: The extent to which 
this option would decrease Social Security's solvency would depend 
largely on the number of people who would be eligible for it. A key 
factor that directly influences the number of eligible beneficiaries 
is the multiple that would be applied to a worker's AIME, ranging from 
150 percent to 300 percent. Another factor that could influence cost 
is the way "single" is defined for purposes of determining eligibility. 
Moreover, determining an individual's single status could be 
administratively complex because people's marital statuses change over 
time and could change after an initial determination is made, for 
example, from single to married; 
Provide an additional Social Security benefit to the oldest old: 
Providing longevity insurance would increase Social Security program 
costs. Key factors that influence costs include the age at which the 
benefit increases, the amount of the increase, and whether all 
beneficiaries or only low-income ones are eligible to receive the 
benefit. In addition, costs could increase if life expectancy 
continues to increase in the future. While this option would be easy 
to administer, adding measures to improve targeting would increase 
administrative complexity. 

Source: GAO analysis of prior GAO reports. 

[End of table] 

[End of section] 

Appendix III: GAO Contact and Staff Acknowledgments: 

GAO Contact: 

Charles Jeszeck, (202) 512-7215 or jeszeckc@gao.gov: 

Staff Acknowledgments: 

Michael Collins, Assistant Director; Jennifer Cook, Erin M. Godtland, 
Rhiannon Patterson, and Ryan Siegel made significant contributions to 
the report. In addition, James Bennett, Benjamin Bolitzer, Michelle 
Bracy, Amy Buck, David Chrisinger, Sarah Cornetto, Hedieh Fusfield, 
Maria Gaona, Mark Glickman, Jennifer Gregory, Kathy Leslie, Gene 
Kuehneman, Courtney LaFountain, Avani Locke, Grant Mallie, Ashley 
McCall, Sheila McCoy, Thomas McCool, Liam O'Laughlin, Carol D. 
Petersen, Mark Ramage, Marylynn Sergent, Frank Todisco, and Shana 
Wallace made valuable contributions. 

[End of section] 

Footnotes: 

[1] Relevant federal laws governing these benefits include the Social 
Security Act, which governs the Social Security Old-Age and Survivors 
Insurance program (referred to in this report as Social Security 
retirement benefits), as well as the Employee Retirement Income 
Security Act of 1974 (ERISA) and the Internal Revenue Code (IRC), 
which govern certain employer-sponsored pension plans. 

[2] DB plans typically offer retirement benefits to a retiree in the 
form of an annuity that provides a monthly payment for life, the value 
of which is typically determined by a formula based on particular 
factors specified by the plan, such as salary or years of service. 
ERISA generally requires that covered DB plans pay benefits in the 
form of a qualified joint and survivor annuity (QJSA), unless certain 
waiver requirements are met, including the spouse's written consent. 

[3] Under DC plans, such as 401(k) plans, workers and employers may 
make contributions to individual accounts. In contrast to DB plans, DC 
plans are not generally required to offer a lifetime annuity and DC 
plans more often provide participants with a lump sum distribution as 
the only option. 

[4] The analysis of the restricted SIPP data was made possible through 
the use of the SIPP Synthetic Beta data, which was funded by the 
Census Bureau and SSA, with additional funding from NSF Grants 
#0427889 and #0339191. The Synthetic Data Server is funded through NSF 
grant SES-1042181. 

[5] Because these data sources are probability samples, our results 
based on these data sources are estimates and subject to sampling 
error. In this report, sampling errors for estimates are presented 
along with their associated margins of error at the 95% confidence 
level. With the exception of the MINT projections, for which SSA does 
not typically produce confidence intervals, any differences between 
point estimates discussed in this report are statistically significant 
at the 95% confidence level unless otherwise noted. Please refer to 
appendix I for additional information on sampling error. 

[6] For additional details on our scope and methods, see appendix I. 

[7] Carmen DeNavas-Walt, Bernadette D. Proctor, and Jessica C. Smith, 
U. S. Census Bureau, Income, Poverty, and Health Insurance Coverage in 
the United States: 2012, Current Population Reports, P60-245 
(Washington, D.C.: September 2013). 

[8] See Karen C. Holden and Angela Fontes, "Economic Security in 
Retirement: How Changes in Employment and Marriage Have Altered 
Retirement-Related Economic Risks for Women," Journal of Women, 
Politics, & Policy, vol. 30 (2009), 173-197. 

[9] Under the Internal Revenue Code, the federal income tax liability 
of a married couple may not be the same as that of two similarly-
situated single taxpayers. Depending on the circumstances, there could 
be either a marriage penalty (married couple owes more tax) or a 
marriage bonus (married couple owes less tax). 

[10] This report focuses on the retirement income of private sector 
workers and their spouses; the retirement income of public sector 
(federal, state, and local government) workers was beyond the scope of 
this review. 

[11] See generally 42 U.S.C. § 401 et seq. 

[12] Social Security spousal and survivor benefits may be offset in 
certain circumstances due to the Social Security Windfall Elimination 
Provision or the Government Pension Offset. For more information on 
these provisions, see GAO, Opportunities to Reduce Potential 
Duplication in Government Programs, Save Tax Dollars, and Enhance 
Revenue, [hyperlink, http://www.gao.gov/products/GAO-11-318SP] 
(Washington, D.C.: March 1, 2011). 

[13] See 29 U.S.C. § 1055. 

[14] See Michael J. Brien and Constantijn W.A. Panis, Annuities in the 
Context of Defined Contribution Plans, a study prepared for the U.S. 
Department of Labor, November 2011. In some limited circumstances, DC 
plans may be required to offer a QJSA or QPSA. 29 U.S.C. § 
1055(b)(1)(B)-(C). 

[15] See 26 U.S.C. § 408. 

[16] United States v. Windsor, 133 S. Ct. 2675 (2013). 

[17] For example, on September 18, 2013, the Department of Labor 
issued guidance explaining its interpretation of the definitions of 
"spouse" and "marriage" under ERISA, in light of the U.S. Supreme 
Court decision. See Technical Release No. 2013-04. On August 29, 2013, 
the Department of the Treasury and the IRS ruled that same-sex 
couples, legally married in jurisdictions that recognize their 
marriages, will be treated as married for federal tax purposes. The 
ruling applies regardless of whether the couple lives in a 
jurisdiction that recognizes same-sex marriage or a jurisdiction that 
does not recognize same-sex marriage. Rev. Rul. 2013-17, 2013-38 
I.R.B. 201. In addition, according to SSA officials, the agency is 
processing some types of claims for members of same-sex married 
couples and paying benefits when they are due. SSA is also working 
with the Department of Justice to develop and implement additional 
policies and processing instructions related to the Windsor decision. 

[18] Craig Copeland, Retirement Plan Participation: Survey of Income 
and Program Participation (SIPP) Data, 2012, EBRI Notes, vol. 34, no. 
8, Employee Benefit Research Institute (Washington, D.C.: August 2013). 

[19] Sudipto Banerjee, "Annuity and Lump-Sum Decisions in Defined 
Benefit Plans: The Role of Plan Rules," EBRI Issue Brief, no. 381 
(January 2013). 

[20] JPMorgan Chase & Co., Pension Pulse (New York, N.Y.: Fall 2012). 

[21] GAO, Unemployed Older Workers: Many Experience Challenges 
Regaining Employment and Face Reduced Retirement Security, [hyperlink, 
http://www.gao.gov/products/GAO-12-445] (Washington, D.C.: Apr. 25, 
2012). 

[22] Despite the growth in the proportion of the divorced population, 
the divorce rate has actually leveled off in recent years. According 
to data from the National Center for Health Statistics, the annual 
divorce rate for married women increased from 15 to 20 divorces per 
1,000 married women between 1970 and 1975 and remained at that level 
through the mid-1990s. However, for individuals 50 and older, the 
divorce rate between 1990 and 2010 more than doubled. See Rose M. 
Kreider and Renee Ellis, U.S. Census Bureau, Number, Timing, and 
Duration of Marriages and Divorces: 2009, Current Population Reports, 
Household Economic Studies, P70-125 (Washington, D.C.: May 2011), and 
Susan L. Brown and I-Fen Lin, "The Gray Divorce Revolution: Rising 
Divorce Among Middle-Aged and Older Adults, 1990-2010" Journals of 
Gerontology Series B: Psychological Sciences and Social Sciences, vol. 
67 no. 6, 731-741 (2012). 

[23] Casey E. Copen, Kimberly Daniels, and William D. Mosher, U.S. 
Department of Health and Human Services, Centers for Disease Control 
and Prevention, National Center for Health Statistics, "First 
Premarital Cohabitation in the United States: 2006-2010 National 
Survey of Family Growth," National Health Statistics Reports, Number 
64 (Atlanta, GA: April 4, 2013). 

[24] A federal interagency group, headed by the Office of Management 
and Budget, is investigating new methodologies for gathering and 
utilizing household relationship data in federal surveys, with a 
primary goal of improving statistics on same-sex couple households. 
While there is no established timetable, new question wording and 
survey formats are being assessed in large-scale field tests. If 
successful, they may be implemented more broadly. 

[25] Unless otherwise noted, throughout this report individuals in the 
white, black, and Asian racial and ethnic categories are non-Hispanic. 

[26] See appendix I for further details on the data and our analyses. 

[27] While this report focuses on the effects of rising labor force 
participation among married women, policy analysts are increasingly 
concerned about the steady decline in men's labor force participation 
overall, which has been more pronounced among unmarried men in recent 
years. 

[28] We obtained data for labor force participation rates among women 
with children of varying ages from the BLS report, Women in the Labor 
Force: A Databook, U.S. Bureau of Labor Statistics, Report 1040 
(February 2013). This report presented data on labor force 
participation rates of mothers with children starting in 1975. 

[29] Several legislative changes related to maternal employment may 
have played a role in the rising labor force participation of mothers 
in the 1970s through the 1990s. These included 1976 changes to the 
federal tax code that permitted working families with a dependent 
child to take a tax credit for child care costs, a 1978 amendment to 
Title VII of the Civil Rights Act of 1964 to prohibit sex 
discrimination on the basis of pregnancy or childbirth, and the Family 
and Medical Leave Act of 1993, which requires covered employers to 
provide, at minimum, up to 12 weeks of unpaid parental leave for 
eligible employees. See Lynda Laughlin, U.S. Census Bureau, "Maternity 
Leave and Employment Patterns of First-Time Mothers: 1961-2008," 
Current Population Reports, Household Economic Studies, P70-128 
(October 2011). 

[30] U.S. Bureau of Labor Statistics, Women in the Labor Force: A 
Databook, BLS Reports, Report 1040 (Washington, D.C.: February 2013). 

[31] In addition to caring for children, many women also bear 
responsibility for caring for an elderly person or a disabled adult, 
which can affect women's participation in the workforce. Some 
researchers have argued that today's generation of middle-aged women 
faces a significant caretaking burden, providing financial and 
practical support for elderly parents at an age when many workers 
should be contributing heavily to their own retirement funds. Due to 
changes in the SIPP survey questions about caretaking 
responsibilities, we were unable to examine trends over time in the 
percentage of women who take time out from the workforce to care for 
elderly or disabled adults. However, our analysis of the 2008 SIPP 
shows that in 2008, approximately 5 percent of women age 50-62 had 
taken at least 6 months off of work to care for an elderly or disabled 
adult. 

[32] Lynda Laughlin, U.S. Census Bureau, "Maternity Leave and 
Employment Patterns of First-Time Mothers: 1961-2008," Current 
Population Reports, Household Economic Studies, P70-128 (October 2011). 

[33] This section primarily focuses on women's receipt of spousal 
benefits because only a small percentage of men (less than 1 percent 
as of 2012) currently receive Social Security spousal benefits. If 
women's wages continue to rise relative to men's, men's receipt of 
spousal benefits may rise in the future. See Objective 3 for further 
discussion of projected receipt of future benefits by sex and benefit 
type. All statistics in this section of the report pertaining to the 
Social Security beneficiary categories of spouse-only, dually-entitled 
spouse, widow-only, and dually-entitled widow include divorced spouses 
who are eligible for spousal and survivor benefits because they were 
married for at least 10 years. 

[34] Howard M. Iams and Christopher R. Tamborini, "The Implications of 
Marital History Change on Women's Eligibility for Social Security Wife 
and Widow Benefits, 1990-2009," Social Security Bulletin, vol. 72, no. 
2 (2012). 

[35] ibid. 

[36] Despite the high correlation between race, education and income, 
the differences in eligibility between blacks and other racial and 
ethnic categories persisted even after accounting for differences in 
income and education levels, according to our analysis of SIPP data. 

[37] Here we are presenting data for women age 50-59 for comparability 
with the data reported by Iams et al. Similar trends are seen among 
women age 55-64. Among women age 55-64 in the lowest quintile of 
income, the percentage who were married or had been married for at 
least 10 years declined from an estimated 85 percent in 1996 to an 
estimated 73 percent in 2008. Among women age 55-64 in the highest 
quintile of income, the percentage who were married or had been 
married for at least 10 years declined from an estimated 97 to 95 
percent. 

[38] These estimates represent a lower bound of the number of women 
who have been in a marriage for at least 10 years. The SIPP marital 
history module provides the year in which each marriage starts and--
for those marriages that have ended in death or divorce--the year in 
which each marriage terminates. Because the month of the marriage 
start/end date is not provided, the length of the marital spell is 
imprecise. For example, consider a person who married in the year 2000 
and divorced in the year 2010. The estimated length of the marriage 
spell is 10 years. However, if the person married in the second half 
of the start year (say, December 2000) and divorced in the first half 
of the termination year (say January 2010), the actual length of the 
marriage spell would be less than 10 years - in this example, 9 years 
and two months. To avoid this problem, we defined a 10-year marriage 
as one in which the termination date minus the start date is at least 
11 years. With this more conservative definition, we avoid including 
any marriages that fall short of 10 years--but we also exclude some 
marriages that are exactly 10 years. These estimates are therefore a 
lower bound of the number of people who have 10-year marriages. 

[39] The lack of statistically significant differences may be the 
result of data limitations associated with the small sample size 
available in the SIPP Gold Standard Data. See appendix I for more 
details on our data and the analyses. 

[40] Household DC plan and IRA assets include the balance of all 
defined contribution, IRA (Roth, traditional, and rollover), and Keogh 
accounts held by all members of the primary economic unit, as defined 
by the Survey of Consumer Finances. DB plans are not included in this 
measure of household savings. 

[41] This finding is consistent with recent results by Irena Dushi and 
Howard Iams. Using SIPP data from 2009, they find that in households 
in which both spouses contributed to a DC plan, the wife's 
contribution comprised around 42 percent of the total family 
contribution. See Irena Dushi and Howard M. Iams, "Pension Plan 
Participation Among Married Couples," Social Security Bulletin, vol. 
73, no. 3 (2013). 

[42] The type of household (male sole earner, female sole earner, and 
dual-earner) reflects the employment status of both household members 
at a particular point in time. Therefore, it is not surprising that we 
observe a portion of retirement savings contributions from the non-
working spouse, which could reflect their labor force participation in 
prior years. 

[43] Due to limitations with our sample size, we expanded the analysis 
of individuals age 55-64 to include individuals in households age 50-
54. See appendix I for more details on our analyses and data 
limitations. 

[44] For purposes of this section of the report, pension income is 
defined based on the structure of the SCF data on current benefits 
from pensions. Specifically it refers to any income from a pension--
from a current or past job or a disability benefit--that is not an 
account plan, such as a 401(k) plan, where the participant can take 
the whole balance as one payment. In addition, a survivor benefit 
includes any type of spousal benefit provided after the participant's 
death, regardless of the duration of the benefit received, amount of 
benefit received, or payment method (lump sum or annuity). Survivor 
benefits from tax-qualified defined benefit plans are required to be 
in the form of a qualified joint and survivor annuity, unless the 
participant (with the spouse's consent) chooses to receive an 
alternate form of benefit, such as a single life annuity. For the 
purposes of this report, due to limitations in the SCF data, a 
survivor benefit may also include households that have chosen to 
annuitize their defined contribution plan savings or households that 
are receiving an annuity from a divorce settlement under a qualified 
domestic relations order (QDRO). According to our estimates, between 
2001 and 2010, the proportion of married households that opted out of 
a survivor benefit ranged from 35 to 44 percent. However, the 
differences over this period were not statistically significant. 

[45] Typically, a joint and survivor annuity provides a smaller 
monthly payment while the participant is alive, to allow for payments 
to the spouse upon the participant's death. A single life annuity can 
provide a higher monthly payment because it lasts only for the life of 
the participant. 

[46] This finding was significant at the 90 percent confidence level. 
These results were generally consistent with what has been found in 
past work by outside researchers. See, for example, Richard W. 
Johnson, Cori E. Uccello, and Joshua H. Goldwyn, "Who Forgoes Survivor 
Protection in Employer-Sponsored Pension Annuities?" The 
Gerontologist, vol. 45, no 1, 26-35 (2005). 

[47] This section primarily focuses on women's receipt of spousal 
benefits because only a small percentage of men (less than 1 percent 
as of 2012) currently receive Social Security spousal benefits. If 
women's wages continue to rise relative to men's, men's receipt of 
spousal benefits may rise in the future. Additionally, the Social 
Security benefit projections presented in this report are based on a 
model that was developed prior to the U.S. Supreme Court's June 2013 
decision holding a provision of the Defense of Marriage Act to be 
unconstitutional. See United States v. Windsor, 133 S. Ct. 2675 
(2013). Consequently, the estimates are based on heterosexual 
marriages exclusively and do not incorporate trends among same-sex 
married couples. 

[48] Barbara A. Butrica and Karen E. Smith, "The Retirement Prospects 
of Divorced Women", Social Security Bulletin, vol. 72, no. 1 (2012). 

[49] Rose M. Kreider and Renee Ellis, U.S. Census Bureau, Number, 
Timing, and Duration of Marriages and Divorces: 2009, Current 
Population Reports, Household Economic Studies, P70-125, (Washington, 
D.C.: May 2011). 

[50] The overall labor force participation rate is actually projected 
to decline gradually for both men and women. According to the U.S. 
Bureau of Labor Statistics, women's labor force participation is 
expected to decline from its peak of 60 percent in 1999 to 53 percent 
in 2040. Men's labor force participation is expected to fall to 64.8 
percent in 2040. See U.S. Bureau of Labor Statistics, Women in the 
Labor Force: A Databook, Report 1040 (Washington, D.C.: February 2013) 
for historical data. The U.S. Bureau of Labor Statistics provided the 
2040 projected data to GAO in a separate analysis. However, the pay 
gap between men and women is also expected to continue to narrow, 
further decreasing the gap between women's and men's lifetime 
earnings. For example, according to projections by the Center for 
Retirement Research, wives born from 1966 to 1975 are expected to earn 
about 68 percent of their husbands' earnings, more than double the 
ratio for wives born from 1931 to 1935. See April Yanyuan Wu, Nadia S. 
Karamcheva, Alicia H. Munnell, and Patrick Purcell, "How Does Women 
Working Affect Social Security Replacement Rates?" Center for 
Retirement Research at Boston College, Issue in Brief, no. 13-10 
(Chestnut Hill, MA: June 2013). 

[51] For all Social Security beneficiary projections in this section 
of the report, the categories of spouse-only, dually-entitled spouse, 
widow(er)-only, and dually-entitled widow(er) include divorced spouses 
who are eligible for spousal and survivor benefits because they were 
married for at least 10 years. 

[52] See Barbara A. Butrica and Karen E. Smith, "The Impact of Changes 
in Couples' Earnings on Married Women's Social Security Benefits," 
Social Security Bulletin, vol. 72, no. 1 (2012). However, as explained 
in greater detail below, as couples' earnings become more similar, the 
Social Security survivor benefit replaces a smaller proportion of the 
couples' Social Security income received by the household when both 
spouses were alive. 

[53] There are no substantial differences by race in the type of 
benefits men received in 2012 or are projected to receive in 2040. 

[54] For example, the median household income in 2010 for women 65 and 
older was $27,610 for black women, $32,820 for Hispanic women, $33,590 
for white women, and $41,960 for Asian women. For men the incomes 
ranged from $35,310 for black men to $47,530 for Asian men. See GAO, 
Retirement Security: Women Still Face Challenges, [hyperlink, 
http://www.gao.gov/products/GAO-12-699] (Washington, D.C.: July 19, 
2012). 

[55] The median household income in 2010 for married women 65 and 
older was $46,990. The next highest income level for women was 
$25,320, for divorced women. For married men the median income was 
$49,720 and the next highest level was $35,080 for separated men. See 
[hyperlink, http://www.gao.gov/products/GAO-12-699]. 

[56] The preretirement income used in this example is the annual 
equivalent of the average indexed monthly earnings used to calculate 
Social Security benefits. 

[57] In general, the widow(er) of a retired worker is eligible to 
receive a survivor benefit that equals 100% of the deceased retired 
worker's benefit amount, assuming that the widow(er) claims this 
benefit at the full retirement age--which varies depending on the 
worker's date of birth. Widow(er)s may claim benefits as early as age 
60, although the amount is reduced accordingly. 

[58] DeNavas-Walt et al., Income, Poverty, and Health Insurance 
Coverage. 

[59] Barbara A. Butrica and Karen E. Smith, "The Impact of Changes in 
Couples' Earnings on Married Women's Social Security Benefits," Social 
Security Bulletin, vol. 72, no. 1 (2012). 

[60] GAO previously reported on this topic in GAO, Social Security: 
Issues Involving Benefit Equity for Working Women, [hyperlink, 
http://www.gao.gov/products/GAO/HEHS-96-55] (Washington, D.C.: Apr. 
10, 1996). 

[61] Appendix II provides a detailed summary of these policy options. 
We reviewed GAO's prior work and selected these options from among a 
larger list of policy options presented in three prior GAO reports. We 
selected these policy options based on either their specific focus on 
spousal or survivor benefits or their broader focus on elderly 
populations that are vulnerable to poverty. See also [hyperlink, 
http://www.gao.gov/products/GAO-12-699]; GAO, Social Security: Options 
to Protect Benefits for Vulnerable Groups When Addressing Program 
Solvency, [hyperlink, http://www.gao.gov/products/GAO-10-101R] 
(Washington, D.C.: Dec. 7, 2009); and GAO, Retirement Security: Women 
Face Challenges in Ensuring Financial Security in Retirement, 
[hyperlink, http://www.gao.gov/products/GAO-08-105] (Washington, D.C.: 
Oct. 11, 2007). The listed options are not intended to be exhaustive. 
In this report, we are not recommending or endorsing the adoption of 
any particular policy option. Rather, we identify them as potential 
options for policymakers to consider, along with other relevant 
proposals, as they address retirement security issues. 

[62] The current maximum of 67 percent applies to the survivor of a 
couple that had previously received a retired-worker benefit plus 
spouse-only or plus a dually-entitled spouse benefit, i.e., 150 
percent of the retired worker's benefit. Upon the death of one spouse, 
the surviving spouse receives 100 percent of the retired-worker 
benefit, which equals 67 percent of the prior total household benefit. 
As an example, see the one-earner couple in figure 20 above. 

[63] The Special Minimum Primary Insurance Amount provision was added 
by the Social Security Amendments of 1972, Pub. L. No. 92-603, § 101, 
86 Stat. 1329, 1333-35 (codified as amended at 42 U.S.C. § 
415(a)(1)(C)(i)). 

[64] Currently, fewer beneficiaries qualify for the Special Minimum 
Benefit every year because it is indexed to price inflation, which has 
risen less than regular initial Social Security benefits, which are 
indexed to wage inflation. As a result, the Special Minimum Benefit 
does not necessarily ensure that individuals who have spent many years 
in Social Security-covered employment will have an income above the 
poverty line. As of 2012, the maximum possible Special Minimum Benefit 
was $9,487 annually. In 2012, the federal poverty guideline was 
$11,170 for a single person. See Alison M. Shelton, Congressional 
Research Service, Social Security: The Minimum Benefit Provision, 
R41518 (Washington, D.C.: May 24, 2012). 

[65] Copeland, "Retirement Plan Participation". 

[66] We identified these policy options for DB and DC plans through 
interviews and consultations with retirement policy experts 
representing a range of perspectives and from different types of 
organizations including government, academia, advocacy groups, and the 
private sector. The listed options are not intended to be exhaustive 
and we are not recommending or endorsing the adoption of any 
particular policy option. Rather, we identify them as potential 
options for policymakers to consider, along with other relevant 
proposals, as they address retirement security issues. 

[67] In order to help ease the administrative burden of documenting 
spousal consent in DB plans, Internal Revenue Service regulations 
allow electronic acknowledgment or notarization but still require the 
signature of the spouse to be witnessed in the physical presence of 
the plan representative or notary public to avert coercion and fraud. 
However, one expert we interviewed believes that notaries have not yet 
begun using this method to document spousal consent because they have 
not yet found a viable way to implement it. See also GAO, Private 
Pensions: Revised Electronic Disclosure Rules Could Broaden Use and 
Better Protect Participant Choice, [hyperlink, 
http://www.gao.gov/products/GAO-13-594] (Washington, D.C.: Sept. 13, 
2013). 

[68] However, while women are closing the pension gap and saving more 
for retirement, they still generally lag behind men. See [hyperlink, 
http://www.gao.gov/products/GAO-12-699]. 

[69] The analysis of the Gold Standard SIPP data was made possible 
through the use of the SIPP Synthetic Beta data, which was funded by 
the Census Bureau and the Social Security Administration (SSA), with 
additional funding from NSF Grants #0427889 and #0339191.The Synthetic 
Data Server is funded through National Science Foundation (NSF) grant 
SES-1042181. 

[70] See Jess Bricker, et al., "Changes in U.S. Family Finances from 
2007 to 2010: Evidence from the Survey of Consumer Finances," Federal 
Reserve Bulletin, vol. 98, no. 2 (June 2012). 

[71] For more information on the MINT6, see Karen E. Smith, Melissa M. 
Favreault, Barbara A. Butrica, and Philip Issa, Final Report: Modeling 
Income in the Near Term Version 6, The Urban Institute (December 2010). 

[72] An additional 1,091 of the 42,803 women in the SIPP data did not 
have SSNs. These women, however, did not have a birth date in the SIPP 
data, so we do not know what proportion of them were age 62 and over 
and might have been receiving Social Security benefits and therefore 
excluded from our analysis. To the extent that women who did not have 
SSNs differ systematically from women included in the analysis, this 
represents a potential source of bias. 

[73] GAO, Retirement Security: Women Still Face Challenges, 
[hyperlink, http://www.gao.gov/products/GAO-12-699] (Washington, D.C.: 
July 2012). 

[74] See GAO, Retirement Security: Women Still Face Challenges, 
[hyperlink, http://www.gao.gov/products/GAO-12-699] (Washington, D.C.: 
July 2012); GAO, Social Security: Options to Protect Benefits for 
Vulnerable Groups When Addressing Program Solvency, [hyperlink, 
http://www.gao.gov/products/GAO-10-101R] (Washington, D.C.: December 
2009); and GAO, Retirement Security: Women Face Challenges in Ensuring 
Financial Security in Retirement, [hyperlink, 
http://www.gao.gov/products/GAO-08-105] (Washington, D.C.: October 
2007). 

[75] The policy options described in this appendix have been suggested 
by experts including agency officials and retirement security 
researchers and were compiled directly from prior GAO reports. For 
this report, we did not conduct an independent analysis of the 
possible legal effects of each policy option. 

[End of section] 

GAO's Mission: 

The Government Accountability Office, the audit, evaluation, and 
investigative arm of Congress, exists to support Congress in meeting 
its constitutional responsibilities and to help improve the 
performance and accountability of the federal government for the 
American people. GAO examines the use of public funds; evaluates 
federal programs and policies; and provides analyses, recommendations, 
and other assistance to help Congress make informed oversight, policy, 
and funding decisions. GAO's commitment to good government is 
reflected in its core values of accountability, integrity, and 
reliability. 

Obtaining Copies of GAO Reports and Testimony: 

The fastest and easiest way to obtain copies of GAO documents at no 
cost is through GAO's website [hyperlink, http://www.gao.gov]. Each 
weekday afternoon, GAO posts on its website newly released reports, 
testimony, and correspondence. To have GAO e-mail you a list of newly 
posted products, go to [hyperlink, http://www.gao.gov] and select 
“E-mail Updates.” 

Order by Phone: 

The price of each GAO publication reflects GAO's actual cost of 
production and distribution and depends on the number of pages in the 
publication and whether the publication is printed in color or black 
and white. Pricing and ordering information is posted on GAO's 
website, [hyperlink, http://www.gao.gov/ordering.htm]. 

Place orders by calling (202) 512-6000, toll free (866) 801-7077, or 
TDD (202) 512-2537. 

Orders may be paid for using American Express, Discover Card, 
MasterCard, Visa, check, or money order. Call for additional 
information. 

Connect with GAO: 

Connect with GAO on facebook, flickr, twitter, and YouTube.
Subscribe to our RSS Feeds or E mail Updates. Listen to our Podcasts.
Visit GAO on the web at [hyperlink, http://www.gao.gov]. 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 
Website: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]; 
E-mail: fraudnet@gao.gov; 
Automated answering system: (800) 424-5454 or (202) 512-7470. 

Congressional Relations: 

Katherine Siggerud, Managing Director, siggerudk@gao.gov: 
(202) 512-4400: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7125: 
Washington, DC 20548. 

Public Affairs: 

Chuck Young, Managing Director, youngc1@gao.gov: 
(202) 512-4800: 
U.S. Government Accountability Office: 
441 G Street NW, Room 7149: 
Washington, DC 20548. 

[End of document]