Retirement Security: Women Face Challenges in Ensuring Financial 
Security in Retirement (11-OCT-07, GAO-08-105). 		 
                                                                 
Women aged 65 and over will account for a growing segment of the 
U.S. population over the next several decades. Despite increases 
in women's workforce behavior in the past 65 years, elderly women
have persistently high rates of poverty. Thus, it is important to
understand the differences between men's and women's retirement  
income, and how women may fare given future reforms to Social	 
Security and pensions. GAO was asked to examine (1) how women's  
retirement income compares with men's and the reasons for	 
differences; (2) how certain life events such as divorce,	 
widowhood, and workforce interruptions affect women's retirement 
income; and (3) the possible effect on women's retirement income 
of certain changes to Social Security and pensions that seek to  
mitigate the effects of differences in workforce participation	 
patterns. To address these objectives, GAO reviewed the relevant 
literature, interviewed academics and other retirement experts,  
and used a microsimulation model to project future retirement	 
income. GAO provided a draft of this report to the departments of
Labor and Treasury, the Internal Revenue Service, and the Social 
Security Administration. Cognizant agency officials provided	 
technical comments which were incorporated as appropriate. GAO is
making no recommendations.					 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-08-105 					        
    ACCNO:   A77267						        
  TITLE:     Retirement Security: Women Face Challenges in Ensuring   
Financial Security in Retirement				 
     DATE:   10/11/2007 
  SUBJECT:   Financial analysis 				 
	     Income statistics					 
	     Labor force					 
	     Labor statistics					 
	     Pensions						 
	     Retirement 					 
	     Retirement age					 
	     Retirement benefits				 
	     Retirement income					 
	     Social security benefits				 
	     Statistical data					 
	     Women						 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-08-105

   

     * [1]Results in Brief
     * [2]Background
     * [3]Women Have Less Retirement Income than Men Largely because o

          * [4]Women Have Less in Total Retirement Income than Men
          * [5]Women Have More Intermittent Work Histories and Lower Earnin

     * [6]Certain Life Events May Reduce Women's Retirement Resources

          * [7]Divorce May Reduce Women's Retirement Income
          * [8]Family Caregiving, Which Can Reduce Lifetime Earnings, Is Mo
          * [9]Women Are More Likely to Experience Widowhood, Which Puts Th
          * [10]Health Care Costs May Deplete Elderly Women's Retirement Res

     * [11]Specific Changes to Social Security and Employer-Sponsored P

          * [12]System Modifications Designed to Increase Social Security Be

     * [13]Implement Dependent Care Credits

          * [14]GAO Simulation of a Dependent Care Credit

     * [15]Increase Minimum Benefit

          * [16]GAO Simulation of an Increased Minimum Benefit

     * [17]Increase Survivor Benefits

          * [18]GAO Simulation of an Increased Survivor Benefit

     * [19]Reduce Spousal Benefits and Increase Survivor Benefits

          * [20]GAO Simulation of a Reduction in Spousal Benefits with Incre

     * [21]Reduce Duration of Marriage Requirement for Divorced Spouse

          * [22]GAO Simulation of a Reduced Marriage Requirement
          * [23]Pension Modifications That Address the Changing Pension Land

     * [24]Lowering Vesting Requirements

          * [25]GAO Simulation of Lowered Vesting Requirements for Employer-

     * [26]Automatic Rollover upon Leaving Employment Prior to Retireme

          * [27]GAO Simulation of 100 Percent Automatic Rollover for Employe

     * [28]Concluding Observations
     * [29]Agency Comments
     * [30]Assumptions and Limitations

          * [31]2007 Social Security Trustees' Assumptions
          * [32]Family Equivalence Scale

     * [33]Description of Simulated Social Security Modifications

          * [34]Dependent Care Credit--Analyzing Impact of Inserting a "Credi
          * [35]Decrease Spousal Benefit/Increase Survivor Benefit
          * [36]Increased Survivor Benefit Only
          * [37]Reduce 10-Year Marriage Requirement
          * [38]Increase/Strengthen Minimum Benefit

     * [39]Description of Simulated Pension Modifications

          * [40]Decrease Vesting Requirements
          * [41]Automatic Rollover

     * [42]Data Reliability
     * [43]Benchmark Policy Scenarios

          * [44]Criteria
          * [45]Tax-Increase-Only or "Promised Benefits," Benchmark Policies
          * [46]Benefit-Reduction-Only, or "Funded Benefits," Benchmark Poli

               * [47]Phase-in Period
               * [48]Defining the PIA Formula Factor Reductions

     * [49]GAO Contact
     * [50]Acknowledgments
     * [51]GAO's Mission
     * [52]Obtaining Copies of GAO Reports and Testimony

          * [53]Order by Mail or Phone

     * [54]To Report Fraud, Waste, and Abuse in Federal Programs
     * [55]Congressional Relations
     * [56]Public Affairs

Report to the Ranking Member, Special Committee on Aging, U.S. Senate

United States Government Accountability Office

GAO

October 2007

RETIREMENT SECURITY

Women Face Challenges in Ensuring Financial Security in Retirement

GAO-08-105

Contents

Letter 1

Results in Brief 4
Background 6
Women Have Less Retirement Income than Men Largely because of Differences
in Labor Force Participation and Lifetime Earnings 12
Certain Life Events May Reduce Women's Retirement Resources More Than
Men's 21
Specific Changes to Social Security and Employer-Sponsored Pensions Will
Affect Women Differently than Men Because of Differences in Lifetime Work
Histories 29
Implement Dependent Care Credits 30
Increase Minimum Benefit 33
Increase Survivor Benefits 37
Reduce Spousal Benefits and Increase Survivor Benefits 39
Reduce Duration of Marriage Requirement for Divorced Spouse Benefit
Eligibility 41
Lowering Vesting Requirements 44
Automatic Rollover upon Leaving Employment Prior to Retirement Age 46
Concluding Observations 48
Agency Comments 50
Appendix I Methodology 51
Assumptions and Limitations 54
Description of Simulated Social Security Modifications 56
Description of Simulated Pension Modifications 58
Data Reliability 59
Benchmark Policy Scenarios 59
Appendix II Simulation Results for Social Security Modifications 66
Appendix III Low Benefit Avoidance Rates 82
Appendix IV Effect of Simulated Reform on Social Security System Solvency
84
Appendix V GAO Contact and Staff Acknowledgments 85

Tables

Table 1: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Addition of the Dependent Care Credit--Promised
Benefits Benchmark 33
Table 2: Percentage of Total Simulation Population Whose Benefits Changed
after Addition of the Dependent Care Credit--Promised Benefits Benchmark
33
Table 3: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Addition of the Increased Minimum Benefit--Promised
Benefits Benchmark 35
Table 4: Percentage of Total Simulation Population Whose Benefits Changed
after Addition of the Increased Minimum Benefit--Promised Benefits
Benchmark 36
Table 5: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Addition of Increased Survivor Benefits--Promised
Benefits Benchmark 38
Table 6: Percentage of Total Simulation Population Whose Benefits Changed
after Addition of Increased Survivor Benefits--Promised Benefits Benchmark
38
Table 7: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Addition of Decreased Spousal Benefits Paired with
Increased Survivor Benefits--Promised Benefits Benchmark 40
Table 8: Percentage of Total Simulation Population Whose Benefits Changed
after Addition of Decreased Spousal Benefits Paired with Increased
Survivor Benefits--Promised Benefits Benchmark 41
Table 9: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Reduction in Marriage Requirement from 10 to 7
Years--Promised Benefits Benchmark 43
Table 10: Percentage of Total Simulation Population Whose Benefits Changed
after Reduction in Marriage Requirement from 10 to 7 Years--Promised
Benefits Benchmark 43
Table 11: Median Percentage Change in Pension Benefits for Individuals
Born in 1985 Whose Benefits Changed After Lowered Vesting Schedules 46
Table 12: Percentage of Total Simulation Population Whose Benefits Changed
after Implementation of Lowered Vesting Schedules 46
Table 13: Median Percentage Change in Pension Benefits for Individuals
Born in 1985 Whose Benefits Changed after Implementation of 100 Percent
Automatic Rollover 48
Table 14: Percentage of Total Simulation Population Whose Benefits Changed
after Implementation of 100 Percent Automatic Rollover 48
Table 15: Summary of Benchmark Policy Scenarios 60
Table 16: Summary of Benchmark Policy Scenario Parameters 65
Table 17: Simulation Results of the Dependent Care Credit under
Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts 67
Table 18: Simulation Results for Dependent Care Credit by Income Quintile
under Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts
68
Table 19: Simulation Results for Dependent Care Credit by Marital Status
under Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts
69
Table 20: Simulation Results for Increased Survivor Benefit Only under
Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts 70
Table 21: Simulation Results for Increased Survivor Benefit Only by Income
Quintile under Alternative Benchmark Scenarios, for the 1950 and 1985
Birth Cohorts 71
Table 22: Simulation Results for Increased Survivor Benefit Only by
Marital Status under Alternative Benchmark Scenarios, for the 1950 and
1985 Birth Cohorts 72
Table 23: Simulation Results for Increased Survivor Benefit and Decreased
Spouse Benefit under Alternative Benchmark Scenarios, for the 1950 and
1985 Birth Cohorts 73
Table 24: Simulation Results for Increased Survivor Benefit and Decreased
Spouse Benefit by Income Quintile under Alternative Benchmark Scenarios,
for the 1950 and 1985 Birth Cohorts 74
Table 25: Simulation Results for Increased Survivor Benefit and Decreased
Spouse Benefit by Marital Status under Alternative Benchmark Scenarios,
for the 1950 and 1985 Birth Cohorts 75
Table 26: Simulation Results for Reduced Marriage Requirement for Divorced
Spouse Benefits under Alternative Benchmark Scenarios, for the 1950 and
1985 Birth Cohorts 76
Table 27: Simulation Results for Reduced Marriage Requirement for Divorced
Spouse Benefits by Income Quintile under Alternative Benchmark Scenarios,
for the 1950 and 1985 Birth Cohorts 77
Table 28: Simulation Results for Reduced Marriage Requirement for Divorced
Spouse Benefits by Marital Status under Alternative Benchmark Scenarios,
for the 1950 and 1985 Birth Cohorts 78
Table 29: Simulation Results for Increased Minimum Benefit under
Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts 79
Table 30: Simulation Results for Increased Minimum Benefit by Income
Quintiles under Alternative Benchmark Scenarios, for the 1950 and 1985
Birth Cohorts 80
Table 31: Simulation Results for Increased Minimum Benefit by Marital
Status under Alternative Benchmark Scenarios, for the 1950 and 1985 Birth
Cohorts 81
Table 32: Average Low Benefit Avoidance Rates Before and After
Modifications for Individuals with Less Than 100 Percent Low Benefit
Avoidance Pre-Modification 82
Table 33: Changes in the 75-year Actuarial Balance as a Percentage of
Taxable Payroll resulting from Program Modifications, after Achieving
75-Year Solvency with Benchmark Scenarios 84

Figures

Figure 1: Life Expectancy at Age 65, 1940 to 2006 and Projected 2007 to
2085 7
Figure 2: Men and Women Aged 65 and Over: Number and Percent of the Total
U.S. Population, Projections 2000 to 2050 8
Figure 3: Percentage of the Population Age 65 and Older Receiving Income
from Various Sources, 2004 13
Figure 4: Median Annual Income of the Population Age 65 and Older, 2004 15
Figure 5: Poverty Rates among People Age 65 and Older by Marital Status,
2004 16
Figure 6: Overall Labor Force Participation Rate of Men and Women Age 16
and Older, 1950-2006 17
Figure 7: Women's Median Earnings For Full-time, Year-Round Work as a
Percentage of Men's, 1960 to 2005 19
Figure 8: Marital Status of the Population Age 65 and Older, 2005 26

Abbreviations

AIME Average Indexed Monthly Earnings
CSSS Commission to Strengthen Social Security
DB defined benefit
DC defined contribution
ERISA Employee Retirement Income Security Act of 1974
GEMINI Genuine Microsimulation of Social Security and Accounts
IRA individual retirement account
MTR Maintain Tax Rates
OASDI Old-Age, Survivors, and Disability Insurance
OCACT Social Security Administration's Office of the Chief Actuary 
OLC overlapping cohorts
PENSIM Pension Simulator
PIA Primary Insurance Amount
PSG Policy Simulation Group
QDRO qualified domestic relations order
RCS representative cohort sample
REA Retirement Equity Act
SPIA Special Primary Insurance Amount
SSA Social Security Administration
SSASIM Social Security and Accounts Simulator
SSI Supplemental Security Income

This is a work of the U.S. government and is not subject to copyright
protection in the United States. The published product may be reproduced
and distributed in its entirety without further permission from GAO.
However, 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
Washington, DC 20548

October 11, 2007

The Honorable Gordon Smith
Ranking Member
Senate Special Committee on Aging
United States Senate

Dear Senator Smith:

Over the next 40 years women aged 65 and over will account for a growing
segment of the U.S. population. In 2000, there were about 20 million women
aged 65 and over, more than 7 percent of the U.S. population; by 2050 that
number is estimated to grow by nearly 28 million to about 12 percent of
the population. Elderly women have persistently high rates of poverty, and
the major source of income for many retired women is Social Security.
However, the Social Security system is affected by the decrease in the
rate of growth of the working age population. Under current law, the
Social Security Trustees project that by 2041 the Social Security Trust
Funds could be insufficient to pay full benefits.^1

Demographics as well as rising health care costs are profoundly affecting
not only the Social Security system, but also Medicare, private pension
and health benefits, and personal savings in ways that will likely present
serious challenges to ensuring financial security for future retirees and,
ultimately, the economic security of the nation. In recent years, many
proposed reforms of the Social Security system have focused on long-term
solvency and financing issues, many of which could result in decreased
benefits for individuals. Alternatively, some Social Security proposals
developed over the past several decades include elements that seek to
modify the program and address its limitations when applied to
nontraditional-family or earnings structures. These limitations may be due
both to the evolving nature of families and to changes in women's labor
force participation that have emerged since Social Security's creation.^2
These elements often address the needs of two-earner families as well as
retirement benefits after divorce. Additionally, recent and ongoing
changes in employer-sponsored pension plans, most notably the shift from
defined benefit pension plans to defined contribution plans, may
complement changes in workforce patterns, but also place greater
responsibility for prudent savings and investment decisions on workers.
Given the existing differences in men's and women's incomes, and the
changes in women's workforce behavior in the later half of the 20th
century, any future changes to Social Security as well as both proposed
and ongoing changes to employer-provided pensions could have different
impacts on women and men. It is important to understand how each will fare
under various proposals.

^1The Board of Trustees, Federal Old-Age and Survivors Insurance and
Disability Insurance Trust Funds, The 2007 Annual Report of the Board of
Trustees of the Federal Old-Age and Survivors Insurance and Disability
Insurance Trust Funds (Washington, D.C.: Apr. 23, 2007).

You asked us to help clarify what drives the gap in retirement income
between men and women and to provide information on the implications of
different modifications to both Social Security as well as
employer-provided pensions. Our objectives were to examine (1) how women's
retirement income compares with men's and describe the reasons for
differences; (2) how certain life events such as divorce, widowhood, and
workforce interruptions affect women's retirement income, as compared with
men's; and (3) the possible effect on women's retirement income of certain
changes to Social Security and pensions that seek to mitigate the effects
of differences in workforce participation patterns.

To address these objectives, we reviewed the relevant literature and
federal laws, interviewed academics and other retirement experts, and used
a retirement-income microsimulation model to project women's future
retirement income.^3 Specifically, we reviewed and summarized government
and academic research on women's retirement income, life events, and
poverty as well as proposed changes to Social Security and
employer-provided pensions. We used the Policy Simulation Group's
retirement income microsimulation model to illustrate differences in
benefit levels by differences in workforce attachment and marital status
of a simulated population, born in 1985. We also used the microsimulation
model to project changes in Social Security benefit levels for a sample of
workers at age 70, from the 1950 and 1985 birth cohorts, under a variety
of possible modifications. The results of our analyses reflect outcomes
for individuals in the simulated populations and do not attempt to
estimate outcomes for an actual population. Unlike some of our prior work,
rather than evaluating Social Security reform packages that seek to
achieve sustainable solvency, we evaluated proposed individual changes
targeted to enhance benefits for certain groups. We also used the model to
project changes in pension benefit levels for a sample of workers at age
70, from the 1985 birth cohort.^4 We used two cohorts in order to identify
differences in the effects of the changes that could be due to variations
in labor force participation across generations. For some of our analyses,
we used a measure of income that adjusts to account for household size and
economies of scale. The adjustment is made by dividing household benefit
levels by a "family equivalence scale."^5 We did this to facilitate
comparisons between non-married persons and married persons, whose
household income includes income from both spouses that can vary
significantly between them.^6 We also evaluated the effect of
modifications on individual benefit levels. In addition to evaluating
changes in benefit levels resulting from each modification, we assessed
changes in a variable in the model that serves as a proxy for poverty
avoidance.^7 Consistent with our past work on Social Security reform, when
simulating benefits we compared benefits under each reform to two
hypothetical benchmark policy scenarios that would achieve 75-year
solvency, one by only increasing payroll taxes (which simulates "promised
benefits") and the other by only reducing benefits (which simulates
"funded benefits").^8 However, unlike prior GAO work, for the purposes of
this study, we evaluated certain specific individual modifications, rather
than comprehensive reform packages. We did this in order to focus on
modifications that account for more recent changes in family structure and
labor force composition. Additionally, to facilitate comparisons across
cohorts, we generally report the effect of each modification in terms of
percent changes in benefit levels. Using such a measure, the outcomes from
each benchmark are largely similar. In the body of the report, in most
cases, we present output from simulations using the "promised benefits"
benchmark. Detailed output from both benchmarks is presented in appendix
II. For this report, we focused on examining the distribution of benefits
and did not assess equity measures. While our simulations provide
estimates of future retirement income, there is a considerable amount of
uncertainty involved with these estimates. Since these estimates could
change significantly, depending on assumptions used and behavioral
responses, they should not be considered predictions.

^2Family structure changed substantially during the later part of the 20th
century. According to the Census Bureau, in 1970, about 70 percent of all
households were composed of married couple families. By 2003, this had
fallen to less than 52 percent. In addition, the proportion of families
with children that were headed by a single parent increased from 13
percent in 1970 to 32 percent in 2003.

^3We used the GEMINI model under a license from the Policy Simulation
Group, a private contractor. GEMINI estimates individual effects of policy
scenarios for a representative sample of future beneficiaries. GEMINI can
simulate different reform features for their effects on the level and
distribution of benefits. See appendix I for more detail on the modeling
analysis, including a discussion of our assessment of the data reliability
of the model.

^4We did not simulate pension benefits for the 1950 birth cohort because
the current version of PENSIM does not have a realistic characterization
of pre-1996 employer pension offerings, and therefore should not be used
to simulate lifetime pension accumulation for cohorts born before 1975.

^5There are both advantages and disadvantages of using such measures. For
additional information on the development, use, and limitations of
equivalence scales see, Constance F. Citro and Robert T. Michael (eds.),
Measuring Poverty: A New Approach, Washington, DC: National Academy Press,
1995 and GAO, Social Security: Program's Role in Helping Ensure Income
Adequacy, [57]GAO-02-62 (Washington, D.C.: Nov 30, 2001).

^6For more information on our adjustment of income, see appendix I.

^7This variable is called the "low benefit avoidance rate." It is produced
by the model and is expressed as a percentage of retirement years in which
Social Security benefits (plus any earnings) are above a predetermined low
benefit threshold. The threshold is measured separately for married
couples and for unmarried individuals. It does not include pension income
or savings, and so cannot be called a true poverty-avoidance measure.

We conducted our work between January 2006 and August 2007 in accordance
with generally accepted government auditing standards. A more detailed
discussion of our scope and methodology appears in appendix I.

Results in Brief

Generally, women have less retirement income than men, largely because of
women's lower labor force attachment and lower earnings, on average. Fewer
women than men have income from most major retirement sources, and those
women who do receive income from these sources receive less than men.
Women's median Social Security benefit is approximately 70 percent of the
median benefit that men receive. Meanwhile, fewer women than men have
pension incomes, and the median value of their pensions is about half that
of men's. While only a small proportion of men and women aged 65 and over
are engaged in the paid labor force, among those who are, women earn just
over half of what men earn. While there is less distinction between the
income of men and women from assets such as interest, dividends, rents,
and royalties, women earn somewhat less than men from these sources as
well. Not surprisingly, older women are more often poor than men. Among
those 65 and over, 12 percent of women are in poverty, compared to 7
percent of men. Although women's work outside the home has increased
substantially in the last century--with the labor force participation rate
of married women aged 16 and over increasing from approximately 32 percent
in 1960 to 61 percent in 2006--they spend fewer years in the labor force
than men and they more often work part-time. Additionally, they tend to
earn less than men during their working years, earning only 77 percent of
what men earned for full-time, year-round work in 2005. Although work
patterns are key in earnings differences, in prior work we found that even
after accounting for these and other behavioral differences--such as
educational attainment--women still earn less than men.

^8See appendix I for a complete description of our benchmark policy
scenarios.

Certain life events--including changes in marital status, labor force
interruptions, and long-term care needs-- can significantly reduce the
amount of pension income and Social Security benefits for both men and
women. However, because of women's lower earnings and labor force
participation, these events may increase the probability women will enter
retirement with fewer financial resources than men. Divorce often results
in economic loss for both men and women, but women tend to experience more
economic loss than men. Further, at retirement, Social Security
divorced-spouse benefits are available only if the marriage lasted at
least 10 years. Women's role as primary family caregiver for children and
elderly relatives can also reduce their career earnings. For example, one
study documented that almost half of women who worked during pregnancy
with their first child took unpaid leave and one-quarter quit their jobs.
Because women tend to live longer than men, they are more likely than men
to experience widowhood. Social Security income is reduced at the
household level upon the death of a spouse, and widows do not often retain
all of their husbands' pension benefits. In part because of their longer
average life spans, with age, women are also more likely than men to
become disabled and need long-term care, further increasing demand upon
their retirement resources.

Our simulations of some proposed changes to the Social Security system and
the employer-sponsored pension system resulted in different effects on
women and men, and among different subgroups of women, because of
differences in lifetime work histories. Some of the proposed changes to
Social Security that we analyzed are in fact designed to increase the
benefits of targeted groups by taking advantage of differences in
workforce participation patterns. On one hand, our model results showed
that modifications that compensate for low earnings or time spent out of
the workforce for caregiving tend to increase benefits for beneficiaries
overall, and particularly those in lower income quintiles. For example,
when we simulated a dependent care credit to compensate for zero or low
earnings when children are young, benefits increased across the board for
women in all marital statuses and in all income quintiles. On the other
hand, the changes that focus on shifts in family structure, such as
increases in two-earner couples and increased incidence of divorce, tend
to increase the benefits of groups targeted by the change, but produce
mixed results for others. For example, our simulation of a reduction in
Social Security's marriage eligibility rule had a narrowly focused impact
on a very small number of women and almost no men. Some pension rule
changes that have been proposed or passed into law in the past several
years take into account changes in the labor force and the changing norms
of employer-provided retirement plans; while these changes are
gender-neutral, they may provide important new opportunities for some
women to increase their retirement income. For example, when we simulated
a pension change that would ensure that 100 percent of retirement account
balances would roll over to another qualified account when individuals
switched jobs (rather than allowing some or all of the balance to be
withdrawn or spent), among those affected, women had larger median
percentage changes than men, and among women, never-married and divorced
women had the largest median percentage changes in pension income.

The departments of Labor and the Treasury and the Social Security
Administration provided technical comments which we incorporated as
appropriate.

Background

The retirement outlook for both men and women in the United States has
changed significantly in the last 30 years. Like many industrialized
countries, the United States is undergoing a significant demographic shift
toward an aging population and is experiencing the increased pressures on
the social insurance, medical, and private pension systems that this shift
creates.

While life expectancy in the United States has steadily increased over the
last 50 years, birthrates have declined, and both have led to rapid growth
in the proportion of the population comprised of elderly people: in 1950,
those aged 65 or older made up 8 percent of the population; in 2000, this
proportion rose to 12 percent and is projected to rise to almost 20
percent by 2030. Also, between 1940 and 1980, women's life expectancy
generally increased faster than men's. During the same time period, the
difference in life expectancy at age 65 for women and men grew from 1.5
years in 1940 to 4.4 years in 1980. (See fig. 1.)

Figure 1: Life Expectancy at Age 65, 1940 to 2006 and Projected 2007 to
2085

Note: Data for 2004 through 2006 are preliminary or estimated. The period
life expectancy at a given age for a given year represents the average
number of years of life remaining if a group of persons at that age were
to experience the mortality rates for that year over the course of their
remaining lives.

The difference in men's and women's longevity has decreased over the past
25 years, and that difference is expected to remain stable throughout much
of the 21st century. Nevertheless, the ratio of elderly women to elderly
men increased substantially in the post-World War II era, and elderly
women will continue to outnumber elderly men both in numbers and as a
percent of the population for the foreseeable future. (See fig. 2.) As a
result of these trends, women can expect on average to spend more years in
retirement than men.

Figure 2: Men and Women Aged 65 and Over: Number and Percent of the Total
U.S. Population, Projections 2000 to 2050

Note: Percentage values over each bar show the total U.S. population
represented by each bar.

Traditionally, the financial resources that provide retirement security
have been characterized as a three-legged stool: Social Security,
pensions, and savings, although increasingly, earnings are also a
significant source of income for the elderly. Overall, women have been
more likely than men to rely on Social Security to finance their
retirement. Moreover, some aspects of the Social Security system
particularly benefit women. For instance, because women tend to have lower
lifetime taxable earnings than men, they benefit from the Social Security
system's progressive benefit formula, which replaces a larger portion of
lifetime earnings for people with low earnings than for people with high
earnings. In addition, Social Security is designed specifically to
accommodate both low- or non-earning spouses, often women, by providing
them with a dependent benefit based upon their spouses' earnings. Social
Security was created based upon the model of a single-earner married
couple family structure, and while many women still never enter the paid
workforce or choose to reduce their workforce participation, at least in
part to care for children or other family members, the single-earner
family model no longer describes the typical American household.
Nevertheless, this structure has been and continues to be extremely
beneficial to some women.

However, Social Security faces a long-term financing shortfall resulting
largely from longer life spans and lower birthrates. According to 2007
Social Security projections,^9 absent policy changes, Social Security tax
revenue is expected to fall short of benefit payments for the first time
in 2017; by 2041 the system may have inadequate resources to pay full
benefits. As a result, in the future, Social Security's role could change.
Reductions in scheduled benefits and/or increases in program revenues will
be needed to restore the long-term solvency and sustainability of the
program. Within the program's current structure, possible benefit changes
might include changes to the benefit formula or reductions in
cost-of-living increases, among other options; revenue increases might
include increases in payroll taxes or transfers from the Treasury's
general fund.^10 In addition, many proposals have been put forth over the
past several decades to address the adequacy of Social Security benefits
for different kinds of workers and their families. These proposals often
address the needs of spouses, survivors, and low earners as well as those
with significant workforce interruptions.

Additionally, many workers bear greater risk and responsibility for their
retirement savings than in the past. About half of U.S. workers do not
have a pension plan through their employer, and those who do are less
likely than in the past to be covered by defined benefit (DB) plans. Among
those who offer plans, employers have increasingly shifted from
traditional DB to defined contribution (DC) plans, such as 401(k)s, which
are based on contributions to and investment returns on individuals'
accounts. While private sector DB plans must offer a guaranteed lifetime
income in the form of an annuity, DC plans more often provide the
beneficiary with a lump sum as the only option.^11 While individuals could
take the proceeds of their lump sum and purchase an annuity, the cost of
purchasing a private annuity may make this option unattractive to many
households. In addition, the private equity market charges women a higher
premium for a life annuity than it charges men of the same age, because on
average women live longer than men.^12

^9The Board of Trustees, Federal Old-Age and Survivors Insurance and
Disability Insurance Trust Funds, The 2007 Annual Report of the Board of
Trustees of the Federal Old-Age and Survivors Insurance and Disability
Insurance Trust Funds.

^10Also, some proposals would change the structure of the program to
incorporate a system of individual accounts. Many such proposals would
reduce benefits under the current system and make up for those reductions
to some degree with income from the individual accounts.

^11Although DC plans generally are not required to have an annuity option,
certain DC plans, known as money purchase plans, are required to make an
annuity payout option available to participants. In addition, lump sum
distributions from DB plans are becoming more common.

Personal savings have traditionally been an important source of retirement
income. Unfortunately, despite the challenges facing both public and
private benefit systems for the elderly, relatively few Americans are
currently saving, and some research suggests that women have less in
savings than men. According to some measures, America has the lowest
overall saving rate of any major industrialized nation. The U.S personal
saving rate as a percentage of disposable personal income has recently
reached levels not seen since the Great Depression, falling below 1
percent in 2005 and 2006.^13 A variety of proposals seeking to encourage
more individuals to save have been introduced in the past several years.
Many of these proposals target low- and moderate-income workers who are
least likely to have access to employer-sponsored pension plans. Some of
these proposals create added incentives or ease access for individuals to
save through existing savings vehicles, such as 401(k) plans or individual
retirement accounts (IRA). Other plans would create new vehicles for
savings, such as 401(k) type plans for those not currently covered by a
plan. One mechanism that already exists to encourage individuals to save
is the so-called spousal IRA offered to non-earning spouses, who are most
often women, as a way to build retirement income. A spousal IRA, allows
non-earning spouses to accumulate retirement savings in their own
retirement accounts.^14

^12The difference in annuity benefits for men and women exists only for
private annuities. In 1983, the U.S. Supreme Court held that an employer's
use of sex-segregated actuarial tables to calculate retirement benefits is
unlawful, whether or not the tables reflect an accurate prediction of the
longevity of women as a class. Arizona Governing Comm. for Tax Deferred
Annuity and Deferred Compensation Plans v. Norris, 463 U.S. 1073, 1084
(1983).

^13GAO analysis of National Income and Product Accounts (NIPA) data from
the Bureau of Economic Analysis (BEA). Personal saving, as measured in the
NIPA, does not include capital gains on existing assets because capital
gains reflect a revaluation of the nation's existing capital stock and do
not provide resources for financing investment that adds to the capital
stock. In other words, although an individual household can tap its wealth
by selling assets to finance consumption or accumulate other assets, the
sale of an existing asset merely transfers ownership; it does not generate
new economic output.

Finally, health care coverage and rising health costs have added to the
financial burden for retirees. Retired Americans often rely on
employer-sponsored health benefits to provide health coverage until they
become eligible for Medicare or to supplement their Medicare coverage. In
2005 about 37 percent of retirees were covered by such plans.^15 However,
retirees are paying more for these benefits, and the number of private
employers offering them has declined considerably.^16 The rate of growth
of health care costs has generally outpaced the rate of U.S. economic
growth, and this trend is likely to continue, jeopardizing the
availability of employer-sponsored insurance for many. Rapidly rising
health care costs may be particularly burdensome for retirees with limited
financial resources. Additionally, the odds of having a disability or
chronic illness increase with age; since, on average, women live longer,
dealing with the cost of declining health may be a particular concern for
women.

^14In 2006 and 2007, an individual may make a contribution of up to $4,000
annually to a spousal IRA. The contribution amount increases to $5,000 in
2008 and will be adjusted based on inflation after that. In addition,
those over age 50 are permitted to contribute an additional $1,000 per
year.

^15The Kaiser/HRET Employer Health Benefits 2001 to 2006 Annual Surveys
found that between 2001 and 2006 the share of employers with 200 or more
workers offering retiree health benefits remained relatively steady, with
about 35 percent offering retiree health benefits in 2006. Survey data
also show that retiree health benefits are most likely offered by large or
unionized firms.

^16People aged 85 or more are much more likely to be covered only by
Medicare than those in the 65-74 age category.

Women Have Less Retirement Income than Men Largely because of Differences in
Labor Force Participation and Lifetime Earnings

Generally, women have less retirement income than men, largely because, on
average, women have lower labor force attachment and lower earnings than
men. While about 90 percent of men and women aged 65 and older receive
Social Security benefits, fewer women than men have income from most other
major sources of retirement income,^17 and they receive less than men from
those sources, according to a Congressional Research Service analysis of
Census Bureau data.^18 Additionally, women aged 65 and older have higher
rates of poverty than men of the same age. While women's labor force
participation has increased substantially in the last half century, it has
flattened out in recent years and remains more intermittent than men's.
Women also tend to earn less than men during their working years.

Women Have Less in Total Retirement Income than Men

While Social Security provides retirement income to almost 90 percent of
all elderly people, a smaller percentage of women than men age 65 and
older have additional income from pensions, assets--such as interest or
dividends from lifetime savings, or earnings, according to the
Congressional Research Service analysis of Census Bureau data.^19 For
example, in 2004, the percentage of men with income from pensions was
almost twice that of women and 44 percent more men than women had wage and
salary income. (See fig. 3.)

^17However, women make up the majority of the poor elderly recipients of
the Supplemental Security Income program--a joint federal-state poverty
program designed to help the elderly (and the blind and disabled of all
ages), who have little or no income, meet their basic needs for food,
clothing, and shelter.

^18Patrick Purcell, Topics in Aging: Income of Americans Age 65 and Older,
1969 to 2004 (Washington, D.C.: Congressional Research Service, 2006).

^19The Census Bureau's Current Population Survey, measures the sources and
amount of income people receive. It does not, however, measure a person's
wealth, which would include the total amount of lifetime savings.
Consequently, in this report, asset income refers to income received from
interest or dividends earned on savings, as well as rents and royalties
from other types of property.

Figure 3: Percentage of the Population Age 65 and Older Receiving Income
from Various Sources, 2004

Moreover, women's median incomes from each of the various retirement
sources are lower than men's. As shown in figure 4, men's median annual
Social Security income was $12,583 in 2004, while women's was $8,799.
Nevertheless, Social Security is an important financial resource for
women, many of whom receive spousal, divorced spousal, or survivor
benefits. According to the Social Security Administration (SSA), while
Social Security is the largest single source of income for most of the
elderly population age 65 and older, it represents 53 percent of total
income for elderly unmarried women--including divorced, widowed, and never
married women, compared to 38 percent for unmarried men. Moreover, Social
Security was nearly the only source of income for close to half of all
elderly unmarried women who received it in 2004, compared to a little more
than a third of elderly unmarried men. Importantly, Social Security can
become a growing fraction of total retirement income over time since it is
indexed to offset the effects of inflation. In contrast, private pensions
and income from assets are rarely indexed. Unlike Social Security, pension
income may end upon the death of the spouse if the retired worker elects
to receive a single life annuity,^20 in which payments cease at the time
of the worker's death.^21

The difference between men's and women's pension income is larger than for
Social Security. In 2004, men's median pension income was nearly twice
women's, $12,000 and $6,141, respectively. Asset income is relatively low
for both women and men compared to Social Security and pensions.
Nevertheless, women's median asset income was $750, while men's was
slightly higher at $964. Finally, while less than a fifth of the elderly
had wage and salary earnings in 2004, men's median earnings were $20,800,
while women's were $12,000. (See fig. 4.)

^20If the worker is covered by a DB plan or by certain DC plans, such as a
money purchase plan, the pension plan is required to obtain the written
consent of the worker's spouse if the worker declines the qualified joint
and survivor annuity option.

^21A joint and survivor annuity provides income to the surviving spouse
should the retired worker die first. However, one study found that 28
percent of married men and 69 percent of married women opted for single
life annuities instead of joint and survivor annuities. See Richard W.
Johnson, Cori E. Uccello, and Joshua H. Goldwyn, Single Life vs. Joint and
Survivor Pension Payout Options: How Do Married Retirees Choose? The Urban
Institute, September 2003.

Figure 4: Median Annual Income of the Population Age 65 and Older, 2004

Additionally, rates of poverty among those 65 and over are higher for
women than for men.^22 Over time, overall poverty rates among the elderly
have declined from 35 percent in 1959 to 10 percent in 2005, according to
Census Bureau data. This is in part due to Social Security benefits for
the aged. Here, too, gender differences remain. In 2004, 12 percent of
women and 7 percent of men age 65 and older had incomes below the federal
poverty level, with more pronounced variation among individuals of
different marital status. For example, never-married elderly men and women
had the highest rates of poverty, while the next highest rates were among
divorced and widowed elderly women. Married couples had significantly
lower rates than all other marital statuses. (See fig. 5.) Moreover,
almost 21 percent of women age 65 and older who lived alone were poor, in
comparison to almost 15 percent of men who lived alone.

^22People and families are classified as poor if their income is less than
the federal poverty level. The official weighted average poverty threshold
in 2005 for a single person age 65 or older was $9,367. For a two-person
household in which at least one member was at least 65 years old, the
poverty threshold was $11,815.

Figure 5: Poverty Rates among People Age 65 and Older by Marital Status,
2004

Women Have More Intermittent Work Histories and Lower Earnings than Men

Women's labor force participation increased substantially in the latter
half of the 20th century, although women continue to work fewer total
years than men and more often work part-time. While women's participation
in the labor force increased from the mid-1960s through the late 1990s,
men's labor force participation has steadily decreased, most significantly
between the mid-1950s and the early-1970s.^23 (See fig. 6.)

^23While the decline in men's labor force participation occurred in most
age groups, it was more rapid among those aged 55 years and older.

Figure 6: Overall Labor Force Participation Rate of Men and Women Age 16
and Older, 1950-2006

Much of the change in women's participation rates is due to higher labor
force participation rates among married women. According to Census Bureau
statistics, between 1960 and 1995 married women's labor force
participation increased from almost 32 to 61 percent and has not changed
significantly in the past decade. As a result, there are now more married
couple households with two earners than when Social Security was first
established. These overall trends have recently stabilized, and in 2006
the Bureau of Labor Statistics predicted that women's labor force
participation rate will not change significantly in the near future.^24

Despite the overall increases in women's labor force participation, women
continue to have more intermittent labor force participation than men. As
we reported in 2003, women have fewer years of work experience, work fewer
hours per year, are less likely to work a full-time schedule, and leave
the labor force for longer periods of time than men.^25 For example, 25
percent of women and almost 11 percent of men age 16 and older usually
worked part-time in 2005.^26

24Mitra Toossi, "A New Look at Long-Term Labor Force Projections to 2050,"
Monthly Labor Review, vol. 129, no.11 (November 2006).

In addition to their spending less time in the workforce overall, women
earn less than men when they are working.^27 Although women's earnings
have risen relative to men's over time, women nevertheless continue to
earn less than men. According to Census Bureau data, in 2005, women earned
77 percent of what men earned for full-time, year-round work. (See fig.
7.)

^25GAO, Women's Earnings: Work Patterns Partially Explain Difference
between Men's and Women's Earnings, [58]GAO-04-35 (Washington, D.C.: Oct.
31, 2003).

^26U.S. Department of Labor, Bureau of Labor Statistics, Women in the
Labor Force: A Databook, Report 996 (Washington, D.C.: September 2006).

^27U.S. Department of Labor, Bureau of Labor Statistics, Women in the
Labor Force: A Databook.

Figure 7: Women's Median Earnings For Full-time, Year-Round Work as a
Percentage of Men's, 1960 to 2005

Note: Based on median earnings of full-time, year-round workers 15 years
old and over. Before 1989 earnings are for civilian workers only.

This difference may be due, in part, to the fact that women continue to
take primary responsibility for family care and those who work outside the
home may trade some career advancement for schedule flexibility. In fact,
in prior work we found that work patterns are a key factor in explaining
the differences in men's and women's earnings. However, even after
accounting for these and other behavioral factors--such as educational
attainment--unexplained differences remained.^28

Changes in women's labor force participation have also increased their
participation in employer-provided pension plans, according to one
study,^29 though, as noted earlier, their overall rates of participation
are still lower than men's. Women who worked full-time throughout the year
actually had higher pension participation rates than men with similar work
schedules in 2005,^30 but women's overall rates remain lower because, in
part, of their lower rate of full-time work and lower earnings, according
to the Employee Benefit Research Institute. While the increase suggests
that a larger share of women in younger cohorts will likely qualify for
pensions based on their own earnings, many women may continue to receive
spousal or survivor benefits through their husbands' pensions. In
addition, the general shift from DB to DC plans may have both positive and
negative consequences for women. Women may especially benefit from the
greater portability afforded by DC plans because of their more
intermittent labor force participation. However, another consequence of
this general shift is that with many DC plans, individuals have a greater
responsibility to make prudent investment decisions and to make their
retirement savings last over their lifetimes, which for women, on average,
are longer than for men.

^28 [59]GAO-04-35 .

^29Alicia Munnell and Pamela Perun, "An Update on Private Pensions," Issue
Brief No. 50 (Center for Retirement Research, Boston College: August
2006).

Changes in women's labor force participation have also increased the
percentage of women who are insured under Social Security based on their
own work history, even though many women continue to receive dependent
benefits as spouses. According to SSA, women who were eligible to receive
benefits based on their own work records increased from 22 percent to 84
percent between 1950 and 2006. Nevertheless, in December of 2005,
approximately 60 percent of retired women received Social Security
benefits based, at least in part, on their marital history.^31 Moreover,
nearly all spousal and survivor beneficiaries were women in 2005. Further,
as women's labor force participation increases, many will find that
benefits based on their own work records are more generous than the
spousal benefit. However, when many of these same women become widows,
they will likely begin to collect benefits based on their marital status,
as the survivor's benefit, at 100 percent of their deceased spouse's
benefit, is likely to be greater than their own.

Data on current retirees reflect the fact that those retirees comprise
older generations of workers, in which women's labor force participation
rates were lower than those of current workers. In the future, data that
include later generations of women, with greater labor force participation
rates, may show greater percentages of women eligible for and collecting
benefits based on their own work records. Nevertheless, the key factor
contributing to the difference in men's and women's Social Security income
levels will continue to be the difference in their lifetime work histories
and earnings. Women's continued intermittent labor force participation and
lower median earnings than men's result in lower benefit amounts, even
though Social Security replaces a greater percentage of preretirement
earnings for lower-wage workers.^32

^30Employee Benefit Research Institute, "Employment-Based Retirement Plan
Participation: Geographic Differences and Trends, 2005," Issue Brief No.
299, November 2006.

^31More than 19 million women aged 65 and older received benefits for
December 2005. About 39 percent were entitled solely to a retired worker
benefit. Almost 30 percent were dually entitled to a retired worker
benefit and a wife's or widow's benefit. About 31 percent were receiving
wife's or widow's benefits only.

Certain Life Events May Reduce Women's Retirement Resources More Than Men's

Certain life events-- including changes in marital status, labor force
interruptions, and long-term care needs-- can significantly reduce the
amount of pension income and Social Security benefits for both men and
women. However, because of women's lower earnings and labor force
participation, these events may exacerbate the deficiency of women's
financial resources in retirement. Divorce often results in economic loss
for both men and women, but women tend to experience more economic loss
than men. In addition, women are most often the family members who provide
unpaid care, which can reduce their career earnings as well. The death of
a spouse can also reduce retirement income for the survivor, and because
they generally live longer, women have higher rates of widowhood than men
at older ages. While declining health at older ages has significant
implications for both men's and women's financial security, because of
life expectancy differences, women more often require costly long-term
care assistance.

^32Benefit levels are determined by averaging the highest 35 years of
indexed covered earnings. Years spent out of the labor force are
represented by zeros. Consequently, an intermittent work history and lower
wages result in a lower benefit level.

Divorce May Reduce Women's Retirement Income

Research has shown that married couples generally have greater household
wealth than nonmarried men and women and that marital disruption
negatively affects both men's and women's economic statuses.^33 While
divorce may result in a reduced standard of living for both men and women,
divorced women, as a group, experience more economic loss than divorced
men.^34 For example, the Census Bureau reported that in 2001, 23 percent
of recently divorced women, in comparison to nearly 8 percent of recently
divorced men, had income below the poverty level.^35 One study found that
marital disruption, including divorce, resulted in a substantial drop in
women's income and loss of assets.^36 Another study projected in 2000 that
most divorced women are more likely than never-married, married, and
widowed women to be in the bottom 40 percent of the income distribution at
age 67.^37 As shown in figure 5, elderly divorced women have higher rates
of poverty, at over 20 percent, than elderly divorced men, at 12 percent.

In retirement, divorce has the potential to reduce Social Security
benefits because Social Security's eligibility rules require that the
marriage last at least 10 years for a divorced spouse to claim benefits
from an ex-spouse's earnings record.^38 However, Census Bureau data from
2001 show that more than half of first and second marriages that ended in
divorce lasted less than 10 years.^39

^33Janet Wilmoth and Gregor Koso, "Does Marital History Matter? Marital
Status and Wealth Outcomes among Pre-Retirement Adults," Journal of
Marriage and Family 64 (February 2002): 254-268; Patricia A. McManus and
Thomas A. DiPrete, "Losers and Winners: The Financial Consequences of
Separation and Divorce for Men," American Sociological Review 66, no. 2
(April 2001): 246-268; Richard W. Johnson, Gordon B.T. Mermin, and Cori E.
Uccello, When the Nest Egg Cracks: Financial Consequences of Health
Problems, Marital Status Changes, and Job Layoffs at Older Ages, (Urban
Institute: January 2006).

^34Rose M. Kreider, "Number, Timing, and Duration of Marriages and
Divorces: 2001," Current Population Reports P70-97, (U.S. Census Bureau,
Washington, D.C.: February 2005); Richard Peterson, "A Re-Evaluation of
the Economic Consequences of Divorce," American Sociological Review 61,
no. 3 (June 1996): 528-536; Jay D. Teachman and Kathleen M. Paasch,
"Financial Impact of Divorce on Children and Their Families," The Future
of Children 4, no. 1, (Spring 1994): 63-83; Karen C. Holden and Pamela J.
Smock, "The Economic Costs of Marital Dissolution: Why Do Women Bear a
Disproportionate Cost?" Annual Review of Sociology 17 (1991): 51-78.

^35Kreider, "Number, Timing, and Duration of Marriages and Divorces:
2001," 13.

^36Jacqueline L. Angel, Cynthia J. Buckley, Ronald J. Angel and Maren A.
Jimenez, The Economic Consequences of Marital Disruption for
Pre-Retirement Age: African-American, Hispanic and Non-Hispanic White
Women, University of Texas at Austin. Paper presented at the Population
Association of America Annual Meeting, Minneapolis, Minnesota (May 2003).

^37Barbara A. Butrica and Howard M. Iams, "Divorced Women at Retirement:
Projections of Economic Well-Being in the Near Future," Social Security
Bulletin 63, no. 3 (2000): 8.

Unlike Social Security benefits, divorced spouses can, under certain
circumstances, receive all or part of their former spouses' private
pension benefits, regardless of the marriage's duration. Although the
Employee Retirement Income Security Act of 1974 (ERISA)^40 generally does
not allow workers to assign their benefits to another person in this way,
Congress amended the law in 1984 through the Retirement Equity Act
(REA)^41 to permit the payment of pension benefits to a worker's former
spouse under a qualified domestic relations order (QDRO). A QDRO, which
meets certain statutory requirements, including approval by a court and
the plan administrator, may be used to satisfy certain obligations, such
as child support, alimony, or the division of marital property.^42
However, the worker's pension benefits may be reduced. Additionally, women
often forgo the protection provided by QDRO's. This may happen for a
variety of reasons, in some cases women may be unaware that their spouses
are covered by a pension, while others may not know that they can receive
benefits while their spouses are alive.

^38While the majority of women receive Social Security retirement benefits
based, at least in part, on their own work record, among women age 65 and
older who received Social Security benefits in December 2005, 31 percent
received benefits based exclusively on their marital history, and about 30
percent were "dually entitled"; i.e., benefits were based on both their
own work record and their marital history. In comparison, the percentage
of women who received divorced spousal benefits was relatively small.
Whereas close to 15 million women received benefits as a retired worker in
December 2005, less than 500,000 received benefits as either a divorced
spouse or divorced widow. See Social Security Administration, Annual
Statistical Supplement to the Social Security Bulletin, 2006, SSA
Publication No. 13-11700, Washington, D.C., June 2007, pp. 2, 5.3, 5.10,
and 5.17.

^39Kreider, "Number, Timing, and Duration of Marriages and Divorces:
2001," 9.

^40The Employee Retirement Income Security Act of 1974, as amended,
governs areas such as pension coverage, vesting periods, benefit accrual
and distribution, and survivor's benefits.

^41Pub. L. No. 98-397.

^4229 U.S.C. S 1056(d)(3).

Family Caregiving, Which Can Reduce Lifetime Earnings, Is More Common for Women
Than for Men

Family caregiving, which encompasses important child care and elder care
responsibilities, is more often provided by women. In order to meet these
family needs, some caregivers reduce work hours or leave the labor force
altogether. For example, one Census Bureau study shows that 45 percent of
women who worked during pregnancy with their first child between 1996 and
2000 took unpaid leave and one-quarter quit their job.^43 Bureau of Labor
Statistics data indicate that, among parents with children under age 6,
almost 92 percent of fathers, compared to over 58 percent of mothers, were
employed in 2005.^44 Research shows that in addition to caring for
children, women provide unpaid care for a family member or friend more
often than men. In 2002, daughters or daughters-in-law provided care to
frail, older adults living in the community more often than sons or
sons-in-law, according to one study.^45 Similarly, another study found
that employed women were more likely than employed men to provide care for
a child, spouse, or partner with a disability.^46 Finally, one study
reported that wives tend to reduce their work hours when a husband
experiences a severe health shock, such as a stroke.^47

Caregiving can negatively affect the provider's career earnings and,
consequently, retirement income. Although many caregivers are employed,
research shows that caregivers can experience substantial losses in career
development and workforce earnings as well as significant out-of-pocket
expenses.^48 For example, one study showed that women age 46 and older who
began caregiving for elderly relatives between 1987 and 1992 experienced
an average of over $3,000 loss in annual earnings.^49 Another study showed
that over half of caregivers who worked while providing care reported that
this role required them to adjust their work schedules, such as arriving
late, or even quit work. Furthermore, years spent out of the paid labor
force can reduce a worker's Social Security benefit amount. Moreover, one
study found that caregiving for adult parents can raise women's risk of
poverty in later years.^50

^43Unpaid leave includes all unpaid maternity, sick, and vacation leave,
and other unpaid leave. See Julia Overturf Johnson and Barbara Downs,
"Maternity Leave and Employment Patterns of First-Time Mothers:
1961-2000," Current Population Reports P70-103, U.S. Census Bureau,
Washington, D.C., October 2005, p. 9.

^44U.S. Department of Labor, Bureau of Labor Statistics, Women in the
Labor Force: A Databook, Report 996 (Washington, D.C.: September 2006).

^45Richard W. Johnson and Joshua M. Wiener, "A Profile of Frail Older
Americans and Their Caregivers," Occasional Paper 8, Urban Institute:
(February 2006).

^46Institute for Women's Policy Research, "The Widening Gap: A New Book on
the Struggle to Balance Work and Caregiving," Research in Brief C349
(October 2001).

^47Courtney C. Coile, "Health Shocks and Couples' Labor Supply Decisions,"
National Bureau of Economic Research Working Paper 10810 (September 2004).

^48Mature Market Institute at Metropolitan Life Insurance Company, MetLife
Juggling Act Study: Balancing Caregiving with Work and the Costs Involved.
Findings from a National Study by the National Alliance for Caregiving and
the National Center on Women and Aging at Brandeis University (November
1999); National Alliance for Caregiving and AARP, Caregiving in the U.S.
(April 2004); Genworth Financial, The Impact of Long Term Care on Women-An
Analysis of Women as Care Providers and Care Recipients, 2006.

Women Are More Likely to Experience Widowhood, Which Puts Them at Risk for
Poverty

Older women are several times more likely than older men to experience
widowhood. For example, in 2004 women age 65 and older were as likely to
be widowed as married, while men were 5.5 times more likely to be married
than widowed. (See fig. 8.)

^49Chizuko Wakabayashi and Katharine M. Donato, "The Consequences of
Caregiving: Effects on Women's Employment and Earnings," Population
Research and Policy Review 24, no. 5, (October 2005): 482.

^50 Other factors, such as education, marital status, and race, appear to
be correlated with whether the female caregivers live in poverty. See
Chizuko Wakabayashi and Katherine M. Donato, "Does Caregiving Increase
Poverty in Later Life? Evidence from the Health and Retirement Survey."
Journal of Health and Social Behavior 47 (September 2006): 264.

Figure 8: Marital Status of the Population Age 65 and Older, 2005

Note: "Married" refers to now married, except separated.

Despite changes to the Social Security system in the 1970s that improved
widows' financial outcomes, research shows that widows continue to be at
risk for poverty in old age.^51 As noted earlier, widows age 65 and older
had over three times the poverty rate of married women or men in 2004.
Widowhood may cause Social Security and pension income to decrease at the
household level, which can be a hardship if certain fixed costs, such as
housing, remain the same. More specifically, household Social Security
income is reduced by one-third if the couple's benefits had been based on
one spouse's work history and by up to 50 percent if both spouses had been
receiving retired worker benefits.^52 In addition, pension income is
likely to be reduced for the surviving spouse.

^51Richard W. Johnson, Gordon B.T. Mermin, and Cori E. Uccello, "How
secure are retirement nest eggs?" Issue in Brief No. 45, Center for
Retirement Research at Boston College (April 2006); Purvi Sevak, David R.
Weir, and Robert J. Willis, "The Economic Consequences of a Husband's
Death: Evidence from the HRS and AHEAD," Social Security Bulletin 65, no.
3 (2003/2004): 31-44; and Catherine D. Zick and Karen Holden, "An
Assessment of the Wealth Holdings of Recent Widows," Journal of
Gerontology: Social Sciences 55B, no. 2 (2000): S90-S97. Note: While
research has shown that widowhood increases the incidence of poverty among
women who were not poor when married, some research indicates that many
widows in poverty also had poor economic status in marriage as well. (See,
Sevak, Weir, and Willis, "The Economic Consequences of a Husband's Death";
Zick and Holden, "An Assessment of the Wealth Holdings of Recent Widows.")

The REA helped protect spouses and widows by requiring employers to obtain
a spouse's written consent in order for a worker to decline the joint and
survivor annuity default option. A common default for the joint and
survivor annuity provides a 50 percent benefit level to the surviving
spouse. However, the shift from DB pension plans to DC pension plans has
the potential to provide even less income security for widows who rely on
a spouse's pension income. Although DC plans generally must provide that
an employee's vested account balance is payable in full on death to the
surviving spouse, the employee may, during his or her lifetime, make
withdrawals from the account or roll over the balance into an IRA without
spousal consent.

^52A widow or widower who meets eligibility requirements is entitled to
receive a percentage of the deceased spouse's Social Security benefits
("survivor benefits") or benefits based on his or her own work
history--whichever is greater. Generally, the survivor is entitled to a
benefit in the amount of the deceased spouse's PIA. However, the
survivor's benefit amount may be reduced if the deceased spouse retired
before reaching full retirement age, or increased if the deceased spouse
delayed retirement beyond the full retirement age.

Health Care Costs May Deplete Elderly Women's Retirement Resources

Declining health at older ages has significant implications for women's
financial security. At least in part because women have longer average
life spans than men, women are more likely than men to become disabled and
need long-term care as they age. In 2003, among Medicare enrollees age 65
and older, more women than men reported an inability to perform at least
one of five certain physical functions, such as the ability to walk two to
three blocks.^53 Women are significantly more likely than men to develop
severe disabilities, and one study estimated that women age 65 have a 44
percent chance of entering a nursing home, compared to 27 percent for
men.^54 Women represented 72 percent of all nursing home residents in 1999
and 70 percent of home care consumers in 2000.^55 Research shows that
nursing home entry has important financial consequences for the elderly,
especially for unmarried women.^56 In 2006, Genworth Financial estimated
that the average annual cost for nursing home care was $70,912 and for
assisted living facilities, $32,294.^57 In addition, out-of-pocket medical
costs during the last years of a spouse's life can deplete the couple's
resources substantially and contribute to poverty among surviving spouses,
who are most often widows.^58

^53Federal Interagency Forum on Aging-Related Statistics. Older Americans
Update 2006: Key Indicators of Well-Being. Federal Interagency Forum on
Aging-Related Statistics. Washington, D.C.: U.S. Government Printing
Office, May 2006, p. 29.

^54Jeffrey R. Brown and Amy Finkelstein, "Supply or Demand: Why is the
Market for Long-Term Care Insurance So Small?" National Bureau of Economic
Research, Working Paper 10782 (September 2004).

^55Adrienne Jones, The National Nursing Home Survey: 1999 summary.
National Center for Health Statistics. Vital Health Stat 13(152). 2002;
Current Home Health Care Patients. National Center for Health Care
Statistics, Centers for Disease Control and Prevention, February 2004.
[60]http://www.cdc.gov/nchs/about/major/nhhcsd/nhhcshomecare3.htm

^56Kathleen McGarry and Robert F. Schoeni, "Medicare Gaps and Widow
Poverty," Social Security Bulletin 66, no. 1, (2005) and Genworth
Financial, The Impact of Long Term Care on Women; Johnson, Mermin, and
Uccello, "When the Nest Egg Cracks."

^57Genworth Financial, The Impact of Long Term Care on Women.

^58McGarry and Schoeni, "Medicare Gaps and Widow Poverty."

Specific Changes to Social Security and Employer-Sponsored Pensions Will Affect
Women Differently than Men Because of Differences in Lifetime Work Histories

The specific changes to Social Security and pensions that we modeled had
different effects on women and men, and among different subgroups of
women, because of differences in lifetime work histories. Some of the
proposed modifications to Social Security that we analyzed are in fact
designed to increase the benefits of targeted groups by accounting for
differences in workforce participation patterns. On one hand, our model
results showed that modifications that compensate for low earnings or time
spent out of the workforce for caregiving tend to increase benefits for
beneficiaries overall, and particularly those in lower income quintiles.
On the other hand, our results showed that modifications that focus on
changes in family structure, such as more two-earner couples and an
increased incidence of divorce, tend to increase the benefits of groups
targeted by the change, but produce mixed results for others. A number of
pension modifications proposed in the last several years take into account
changes in the labor force and the changing norms of employer-provided
retirement plans; while these reforms are gender-neutral, they may provide
important new opportunities for women to increase their retirement income.

While the costs associated with each of the Social Security program
modifications modeled in this report vary, all but one^59 would have a
negative effect on trust fund solvency regardless of the benchmark used.
Because we have analyzed each modification in isolation, we present the
solvency impact for each modification in isolation; as part of a larger
package of reforms, the solvency impact of each change may vary. For a
summary of the solvency impact of each simulated change under both the
"promised benefits" and "funded benefits" benchmarks, see appendix IV. For
more information on both the "promised benefits" and the "funded benefits"
benchmark, see appendix I.

^59Under the "Promised Benefits" benchmark, the "increased survivor
benefits with decreased spousal benefits" modification resulted in a
positive impact on the 75 year actuarial balance. For all solvency
results, see appendix IV.

System Modifications Designed to Increase Social Security Benefits for Specific
Populations Do So, but Sometimes Only Marginally

Using the Policy Simulation Group's GEMINI and SSASIM models,^60 we
assessed the effects of certain specific modifications that were designed
to enhance Social Security benefits for specific subgroups of
beneficiaries at age 70.^61 We used both a family equivalence scale and
individual level analysis to be able to both compare between married and
unmarried individuals and to assess the impact of modifications on
individuals alone. Each of the changes discussed below increases benefits
for the targeted group, but the size of the increase projected varies with
the number of people affected; generally, the smaller the population that
was targeted by the reform, the larger the change in benefits and vice
versa. While some of the individual modifications modeled below have no
benefit impact outside of the targeted group, for those that do have
broader effects, the impact on other beneficiaries can vary. The results
of our analyses reflect outcomes for individuals in the simulated
populations and do not attempt to estimate outcomes for an actual
population.

Implement Dependent Care Credits

Dependent care credit proposals seek to compensate those who spend some
years out of the workforce to care for dependents or those with low or
reduced earnings while attending to caregiving responsibilities. Fewer
years in the workforce or reduced earnings during caregiving years tend to
lower a caregiver's average lifetime earnings and, thus, his or her Social
Security benefit levels in retirement. Many advocates have proposed
dependent care proposals that either modify Social Security's traditional
benefit formula, which uses a 35-year span of time to average lifetime
earnings or credit caregivers with additional earnings. These proposals
vary in design and do not necessarily produce similar results. Either
option would require data collection and computation beyond SSA's current
practices, a potential administrative complication.

           o Specified earnings credit: This approach would assign to the
           caregiver one half of average earnings for each year in which
           there was a child in care and the actual earning was zero or
           reduced from prior earnings for a fixed number of years. Those who
           already earn the credit amount would not benefit. Research has
           shown that a caregiver credit model based on crediting earnings
           does a better job of targeting lower earners than excluding care
           years from earning averages.^62

           o Excluding care years from earning averages: This approach would
           reduce the standard 35-year basis for determining a worker's
           average indexed monthly earnings by subtracting the number of
           years spent providing care. It generally also specifies a limit
           for the number of years that can be dropped from the calculation
           of average earnings. Because the design specifies years of zero
           earnings, this approach may not target financially needy
           populations who lack sufficient resources to take full years off
           from work. High-income caregivers may be more likely to benefit
           from such an approach if economic necessity drives low-income
           caregivers back to the workforce, while those in high-income
           families are able to stay out of the workforce longer.

^60For more information on the Policy Simulation Group's models, see
appendix I.

^61Outcomes from each reform could be different for individuals at ages
other than 70, particularly as the incidence of widowhood in the
population increases at higher ages.

GAO Simulation of a Dependent Care Credit

We simulated the effect of a dependent care credit similar to one
evaluated by the Urban Institute^63 that would credit a worker's earnings
record with one-half of average wages in years in which there was a child
under 5 years of age^64 in the household and the worker's earnings were
less than one half of average wages.^65 We evaluated the impact on the
Social Security benefits of two simulated populations--one cohort born in
1950 and another born in 1985.^66

^62Melissa Faverault and Eugene Steuerle, Social Security Spouse and
Survivor Benefits for the Modern Family (Washington DC: Urban Institute,
2006).

^63Melissa Favreault and Frank Sammartino, The Impact of Social Security
Reform on Low-Income and Older Women (Washington D.C.: Urban Institute,
2002).

^64Some dependent care proposals might also provide enhanced benefits to
individuals caring for dependent adults. However, for ease of modeling we
simulated a proposal in which the benefits were limited to those with
young dependent children. Proposals that do include benefits for adult
caregivers would likely affect a larger number of beneficiaries than the
version we simulated.

^65Individuals in the sample population that met these criteria were
credited with one-half of average wages (as measured by the Average Wage
Index used by the Social Security Administration) for up to 5 years. If
both parents meet these criteria in any given year, only the lower-earning
parent would receive the credit; if parents had identical earnings, the
credit is split evenly. The credit would not contribute to quarters of
coverage for eligibility purposes. The design of this reform effectively
would provide a limited minimum benefit for workers with children under 5.

^66We used the 1950 and 1985 simulated birth cohorts in all of the
simulations of Social Security reform elements presented in this report.

Using the GEMINI model, we found that this particular dependent care
credit resulted in positive median benefit changes for both women and men,
though it provided a larger increase in median benefits for women. The
credit was more beneficial for those born in 1950 than in 1985. As shown
in table 1, its outcome was progressive for those affected by the change,
with women in the lowest earnings quintile showing the highest
proportionate gain in their Social Security benefits. In both cohorts,
never-married women had the largest median change in benefits.^67 The
effect of this change was also broad; for both birth cohorts, it affected
the largest number of beneficiaries, both men and women, of any of our
simulations.

^67This reform would reduce the 75-year solvency achieved by the "promised
benefits" benchmark by 0.19 percent. The "promised benefits" benchmark
achieves 75-year solvency by increasing payroll taxes by the amount of
Social Security's actuarial deficit as a percentage of payroll. For more
information on this and the "funded benefits" benchmark see appendix I.
For a summary of the solvency impact of all simulated modification under
both the "promised benefits" and the "funded benefits" benchmarks see
appendix IV.

Table 1: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Addition of the Dependent Care Credit--Promised
Benefits Benchmark

Numbers in percent
Birth cohort                                 All men             All women
1950                                            2.60                  3.24
1985                                            2.06                  2.63
                            Women by income quintile
Birth                                         Middle               Highest 
cohort       Lowest quintile   Quintile 2   quintile Quintile 4   quintile 
1950                    7.96         4.16       2.80       1.68       0.92 
1985                    6.88         3.50       2.12       1.14       0.72 
                            Women by marital status
Birth cohort          Never married    Divorced     Married        Widowed
1950                           7.95        5.05        2.64           2.87
1985                           6.69        4.18        2.19           2.43

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I). Income quintiles
are based on the distribution of the present value of family lifetime
earnings for the whole population (male and female). For percentage of
population affected and results using the "Funded benefits" benchmark, see
appendix II.

Table 2: Percentage of Total Simulation Population Whose Benefits Changed
after Addition of the Dependent Care Credit--Promised Benefits Benchmark

Numbers in percent              
                      Birth Cohort
                        1950  1985 
All men             28.43 35.06 
All women           30.93 30.99 

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I).

Increase Minimum Benefit

Minimum benefit proposals aim to ensure Social Security benefit adequacy
for low earners. Under current law, Social Security includes a Special
Primary Insurance Amount (also referred to as the Special Minimum Benefit)
intended to reduce poverty among retired lifetime low-wage workers. The
Special Primary Insurance Amount targets retirees with a low benefit based
on a steady, long-time, low wage work record, rather than on those with a
low benefit based on intermittent workforce attachment;^68 the Special
Primary Insurance Amount makes this distinction by basing the calculation
on years worked rather than earnings level. Very few people currently
receive benefits based on the Special Primary Insurance Amount; however,
the majority of those who do are women. Additionally, the benefit provided
by the Special Primary Insurance Amount is less than the official poverty
level for aged persons, and because the benefit is indexed to price
inflation rather than wage growth, it has provided a less generous benefit
over time relative to the traditional wage-indexed Social Security
benefit. Because of this, SSA has projected that the Special Primary
Insurance Amount will phase out as early as 2013. A newly designed minimum
benefit could expand benefits for low earners across all demographic
groups, including women, who are more likely than men to be at the bottom
of the income distribution. Expanded minimum benefits may also be of
renewed importance to benefit adequacy as part of a broad Social Security
reform scenario that reduces benefits for all beneficiaries, including low
earners.

Several Social Security reform proposals include a new minimum benefit
that would guarantee a benefit equal to a set percentage of the poverty
level, dependent on the number of years worked across a lifetime. For
example, Estimated OASDI Financial Effects of the "Bipartisan Retirement
Security Act of 2005"--legislation introduced as H.R. 440 by
Representative Jim Kolbe and Representative Allen Boyd would provide low
earners who had 40 years of minimum wage earnings a benefit equal to 120
percent of the federal poverty level when fully phased in.^69

GAO Simulation of an Increased Minimum Benefit

To simulate the effect of an increased minimum benefit, we modeled a
change to set a minimum benefit of 120 percent of the federal poverty
level for 30-year workers, linearly phased to zero for workers with 20
years or less of covered employment. Slightly fewer men than women in the
simulation were affected by the change, and those who were had a slightly
lower median benefit changes than women affected by the change. As
expected, the outcome of this change was largely progressive. In both
cohorts, a larger share of women in the lower two income quintiles had
benefit changes resulting from this modification than women in the upper
three income quintiles.^70 Moreover, the median percentage changes in
benefits for women in the bottom quintile was much larger than those for
women in the higher quintiles. (See table 3.)

^68Previous to the Special Primary Insurance Amount, Social Security had a
Minimum Benefit that did not target lifetime low earners; this benefit was
criticized for providing windfall benefits for workers with only a minimal
attachment to the Social Security system.

^69See Office of the Chief Actuary, Social Security Administration,
Estimated OASDI Financial Effects of the "Bipartisan Retirement Security
Act of 2005" (Nov. 4, 2005) at
http://www.ssa.gov/OACT/solvency/Kolbe_20051104.pdf.

Table 3: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Addition of the Increased Minimum Benefit--Promised
Benefits Benchmark

Numbers in percent
Birth cohort                                 All men             All women
1950                                            8.47                  9.89
1985                                            6.20                  6.70
                            Women by income quintile
Birth                                         Middle               Highest 
cohort       Lowest quintile   Quintile 2   quintile Quintile 4   quintile 
1950                   22.83         9.65       5.21       3.51       2.55 
1985                    8.26         2.46       2.27       0.09          0 
                            Women by marital status
Birth cohort          Never married    Divorced     Married        Widowed
1950                          22.80       17.13        7.43           8.34
1985                          19.01       14.50        4.55           6.48

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I). Income quintiles
are based on the distribution of the present value of family lifetime
earnings for the whole population (male and female). For percentage of
population affected and results using the "Funded benefits" benchmark, see
appendix II.

^70For data, see appendix II, table 30.

Table 4: Percentage of Total Simulation Population Whose Benefits Changed
after Addition of the Increased Minimum Benefit--Promised Benefits
Benchmark

Numbers in percent              
                      Birth Cohort
                        1950  1985 
All men              7.71  8.77 
All women            1.18  1.28 

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I).

While the share of women affected by the minimum benefit was fairly
similar across marital statuses (never-married, divorced, married and
widowed) in each cohort (for data, see app. II, table 30), never-married
and divorced women had much larger percent changes in median benefits. For
never married women affected by the modification, the percent change in
median benefits was more than double under the minimum benefit than under
any other modification.^71

The impact of the minimum benefit is larger in the 1950 cohort versus the
1985 cohort because the minimum benefit is linked to the poverty line,
which is indexed to prices, while Social Security initial benefits are
indexed to wages; over time, this results in a lower minimum benefit
relative to Social Security benefits. This effect could be reduced by
indexing the poverty level to wages rather than prices, as was done in the
Kolbe-Boyd proposal.

A minimum benefit following parameters such as we simulated may increase
benefits for both part-time and full-time workers. Because Social Security
only tracks annual earnings rather than wages or hours worked, a
higher-earning, part-time worker could receive the same benefit as a
full-year, low-income worker; enhanced benefits may therefore also be
provided to individuals who work part-time by choice.^72

^71The increased minimum benefit, as we've modeled it, would reduce the
75-year actuarial balance achieved by the "Promised benefits" benchmark by
0.05 percent. For more information on this and the "Funded benefits"
benchmark, see appendix I. For a summary of the solvency impact of all
simulated reforms under both the "Promised benefits" and the "Funded
benefits" benchmarks see appendix IV.

Increase Survivor Benefits

One way that some proposals address the vulnerability of widows or
widowers to poverty is by raising survivor benefits to a set percentage of
a married couple's prior combined benefit (for example, two-thirds or
three-fourths of the level of benefits received by the couple while both
were living). Under current law, a survivor receives the larger of his or
her own benefit or the benefit of the deceased spouse. Thus, survivor
benefits for many dually entitled women or women receiving only spousal
benefits would replace approximately 67 percent of the couple's prior
total benefit level. Researchers have expressed concern about whether this
decline in total household benefits is too large to maintain the
survivor's previous standard of living. On the other hand, for survivors
of two-earner couples where both spouses received retired worker benefits
on their own record, a widow's benefit under current law may range between
50 percent and 67 percent of the couple's prior total benefits upon the
death of a spouse, causing an even greater decline in total household
benefit income.

GAO Simulation of an Increased Survivor Benefit

GAO modeled a survivor benefit that would provide a surviving spouse with
the higher of 75 percent of the couple's previous combined benefit level,
capped at the average benefit level for all new retirees, or the current
law survivor benefit. We did not simulate the effect of the current
provision ensuring surviving spouses a minimum of 82.5 percent of the
deceased worker's PIA; had this been included, results may be slightly
higher. This simulation resulted in increased benefits for both men and
women. (See table 5.) While about three times the number of women as men
were affected, the magnitude of the benefit change was larger for men who
were affected by the program modification; their median percentage change
in benefits was nearly 29 percent in both cohorts. This is attributable to
the increased survivor benefit modification compared to current law.
Current law allows survivors the greater of their own benefit or their
spouse's benefit. As most men receive a larger benefit than their spouses,
a survivor benefit of the larger of 75 percent of the couple's combined
benefit (capped at the average benefit level for all new retirees) or the
husband's benefit would provide a higher benefit level than current law to
lower-earning men who outlive their wives.

^72A potential unintended consequence of a more generous minimum benefit
is that higher benefits may disqualify certain individuals from
Supplemental Security Income (SSI) eligibility. This is significant for
some beneficiaries living in states that provide automatic Medicaid
eligibility for SSI recipients.

Table 5: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Addition of Increased Survivor Benefits--Promised
Benefits Benchmark

Numbers in percent
Birth cohort                                 All men             All women
1950                                           28.78                 15.71
1985                                           28.98                 18.36
                            Women by income quintile
Birth                                         Middle               Highest 
cohort       Lowest quintile   Quintile 2   quintile Quintile 4   quintile 
1950                   14.81        16.28      15.77      13.69      24.44 
1985                   19.39        18.36      16.00      16.00       8.67 
                            Women by marital status
Birth cohort          Never married    Divorced     Married        Widowed
1950                              0       16.67           0          15.58
1985                              0       18.42           0          18.35

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I). Income quintiles
are based on the distribution of the present value of family lifetime
earnings for the whole population (male and female). For percentage of
population affected and results using the "Funded benefits" benchmark, see
appendix II.

Table 6: Percentage of Total Simulation Population Whose Benefits Changed
after Addition of Increased Survivor Benefits--Promised Benefits Benchmark

Numbers in percent              
                      Birth Cohort
                        1950  1985 
All men              0.78  0.46 
All women            2.11  1.13 

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I).

In both the 1985 and 1950 cohorts, the increased survivor benefit
modification increased the number of women who never fell below the
microsimulation model's low benefit threshold by about 6 percentage points
versus current law. For those women affected by the benefit in our
simulation, the median percentage change in benefits was about 16 percent
in the 1950 cohort and about 18 percent in the 1985 cohort.^73 As
expected, for both cohorts, the majority of women with benefit changes
resulting from this change are widows and divorced women. Additionally, in
terms of number of people affected, the impacts of this change were
concentrated primarily in the bottom two income quintiles.^74

By capping this program modification at the level of average benefits,
this modification targets the increased survivor's benefit to
lower-earning widows and widowers. Although exempting wealthier
beneficiaries from this benefit enhancement creates fiscal savings, it may
limit many survivors from two-earner couples from benefiting from the
change.

Reduce Spousal Benefits and Increase Survivor Benefits

An increase in survivor benefits is sometimes paired with a decrease in
spousal benefits, from one-half the retired worker's benefit to one-third.
This pairing provides nearly the same average percentage change in
benefits to widows as the modification above, but contains costs by
reducing the spousal benefit while the worker is still living.

GAO Simulation of a Reduction in Spousal Benefits with Increased Survivor
Benefits

Our simulation of this modification resulted in nearly the same benefit
changes for widows as the "Increase survivor benefits" projection
described above, while the benefits of affected married women and divorced
women--recipients of spousal benefits--had a generally negative change.
When all women affected by this modification are sorted into quintiles by
lifetime household income, the median percentage change in benefit levels
is similar across quintiles, and is in all cases negative. (See table 7.)
Despite this, the low benefit avoidance rates were quite similar to those
produced by the survivor benefit increase modification for all cohorts,
indicating that both modifications have nearly the same positive impact on
our proxy for poverty avoidance (see app. III).

^73For the 1985 cohort, this reform had an impact on twice the number of
widows in the "Funded benefits" framework than in the "Promised benefits"
framework; this is because the "Funded benefits" results in lower benefit
levels and more survivors have incomes below the cap of average PIA.

^74As simulated, this change would reduce the 75-year solvency achieved by
the "Promised benefits" benchmark by 0.07 percent. For more information on
this and the "Funded benefits" benchmark see appendix I. For a summary of
the solvency impact of all simulated changes under both the "Promised
benefits" and the "Funded benefits" benchmarks see appendix IV.

Table 7: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Addition of Decreased Spousal Benefits Paired with
Increased Survivor Benefits--Promised Benefits Benchmark

Numbers in percent
Birth cohort                                 All men             All women
1950                                           -2.81                 -2.66
1985                                           -9.14                 -8.78
                            Women by income quintile
Birth                                         Middle               Highest 
cohort       Lowest quintile   Quintile 2   quintile Quintile 4   quintile 
1950                   -2.55        -2.55      -2.66      -2.67      -2.67 
1985                   -9.82        -8.06      -8.21      -8.45      -9.27 
                            Women by marital status
Birth cohort          Never married    Divorced     Married        Widowed
1950                              0       -8.00       -2.67          15.58
1985                              0      -24.11       -9.01          18.91

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I). Income quintiles
are based on the distribution of the present value of family lifetime
earnings for the whole population (male and female). For percentage of
population affected and results using the "Funded benefits" benchmark, see
appendix II.

Table 8: Percentage of Total Simulation Population Whose Benefits Changed
after Addition of Decreased Spousal Benefits Paired with Increased
Survivor Benefits--Promised Benefits Benchmark

Numbers in percent              
                      Birth Cohort
                        1950  1985 
All men             11.45 11.58 
All women           11.78 11.31 

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I).

This proposal would both smooth household benefit levels before and after
widowhood and provide savings to the Social Security system to offset
costs of the increased survivor benefit.^75

Reduce Duration of Marriage Requirement for Divorced Spouse Benefit Eligibility

Proposals to shorten the current requirement for 10-year marriage duration
to be eligible for divorced spouse and survivor benefits would expand
eligibility for benefits to divorced spouses from marriages that do not
meet the 10-year milestone.^76 Timing of divorce can have a large impact
on retirement benefits, as an individual divorced one day before the 10
year anniversary would not be eligible for benefits, while another
individual who waited one more day would be eligible for a full spousal or
survivor benefit. Reducing the 10-year marriage requirement would make
more divorced individuals eligible for divorced spouse and survivor
benefits but would also increase the probability that an individual with
several former spouses could have several spouses receive benefits on one
worker's earnings record.

^75Based on our simulation, this change would improve the 75-year
actuarial balance achieved by the "Promised benefits" benchmark by 0.02
percent. The "Funded benefits" benchmark does not fully offset; the
solvency impact of this reform on the "Funded benefits" benchmark is
-0.06%

^76In 1995, 30 percent of marriages ended in divorce within the first 10
years.

GAO Simulation of a Reduced Marriage Requirement

GAO simulated a modification that would reduce the duration of marriage
requirement for receiving divorced spouse benefits from 10 to 7 years.^77
In the 1950 and 1985 cohorts, among women who had a benefit change due to
the reduced marriage requirement, the median percentage change in benefits
was about 65 percent and 45 percent respectively (see table 9), the
largest median change in benefits for women among all reforms modeled. The
scope of impact, however, was extremely small: In both cohorts, those
affected made up less than 1 percent of the model sample.

The changes in this simulation also resulted in a handful of newly
eligible beneficiaries: In the 1950 cohort, three individuals who were not
previously eligible for Social Security benefits became eligible under the
reform scenario, and in the 1985 cohort, 43 individuals became eligible.
Among the seven simulations that we ran, this was the only one that
resulted in new beneficiaries. This is because individuals who were not
eligible on their record became eligible as spouses or survivors under the
shorter duration of marriage requirement. These newly eligible
beneficiaries are not included in the median percent change measures.

^77Based on our simulation, this change would reduce the 75-year solvency
achieved by the "Promised benefits" benchmark by 0.02 percent of payroll.

Table 9: Median Percentage Change in Benefits for Individuals Whose
Benefits Changed after Reduction in Marriage Requirement from 10 to 7
Years--Promised Benefits Benchmark

Numbers in percent
Birth cohort                                 All men             All women
1950                                            0.85                 65.32
1985                                           37.22                 45.05
                            Women by income quintile
Birth                                         Middle               Highest 
cohort       Lowest quintile   Quintile 2   quintile Quintile 4   quintile 
1950                   85.18        54.66     162.36      33.33          0 
1985                   53.99        34.41      33.33      36.31      25.14 
                            Women by marital status
Birth cohort          Never married    Divorced     Married        Widowed
1950                              0       65.32           0              0
1985                              0       45.05           0              0

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I). Income quintiles
are based on the distribution of the present value of family lifetime
earnings for the whole population (male and female). For percentage of
population affected and results using the "Funded benefits" benchmark, see
appendix II.

Table 10: Percentage of Total Simulation Population Whose Benefits Changed
after Reduction in Marriage Requirement from 10 to 7 Years--Promised
Benefits Benchmark

Numbers in percent              
                      Birth Cohort
                        1950  1985 
All men                 0  0.03 
All women            0.04  0.51 

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I).

Pension Modifications That Address the Changing Pension Landscape and Changing
Workforce Patterns May Serve Women Better than Traditional Pension Models

In addition to the Social Security proposal elements above, we simulated
two pension modifications that address challenges related to the shift to
DC plans and changing workforce patterns.^78 Both modifications modeled by
GAO generally resulted in higher pension benefits, and address issues that
may be of particular concern to women. In particular, in DC plans,^79
contributing early and maintaining an account over time has a significant
positive impact on the balance of that account and the resulting
retirement benefit. Decreasing vesting requirements would allow workers
who change jobs more frequently to attain increased benefits from pension
plans. Automatically rolling over accounts at a job's end increases the
probability that accrued retirement balances will in fact be saved for
retirement. The GEMINI/PENSIM models do not account for behavioral
responses to program changes; therefore, data do not take into account
possible employer or employee responses to the program modifications
below.

Lowering Vesting Requirements

Under current law, eligible employees must be allowed to participate in a
plan as of age 21 and after completing 1 year of service, subject to
certain exceptions. An employee's own contributions to their pension plan
are nonforfeitable, as are employer contributions once an employee's
benefits have vested. ERISA, as amended, requires cliff vesting^80 in DBs
within 5 years and full vesting under a graduated vesting^81 schedule
within 7 years. cash balance plans ("hybrid" plans) generally will require
vesting within 3 years. Beginning in 2008 employer contributions to DC
plans^82 must vest in either a 3-year cliff or 6-year phased schedule
(includes service prior to 2007).^83

^78Pension reforms were modeled using only the 1985 cohort, as PENSIM model
data are not valid for cohorts born before 1975 (per PENSIM
documentation).

^79Defined benefit plans promise to provide a benefit that is generally
based on an employee's salary and years of service. Defined contribution
plans have individual accounts to which the employer, employees, or both
make periodic contributions. For more information, see GAO, Answers to Key
Questions about Private Pension Plans, [61]GAO-02-745SP (Washington, D.C.:
Sept.18, 2002).

^80Plans with cliff vesting have a specified point at which participants
have a right to benefits accrued to date and benefits accrued thereafter.
For more information, see [62]GAO-02-745SP .

^81Plans with graduated vesting give participants a right to an increasing
percentage of their total accrued benefit over time. For more information,
see [63]GAO-02-745SP .

^82The Pension Protection Act of 2006 accelerated some vesting
requirements for DC plans.

GAO Simulation of Lowered Vesting Requirements for Employer-Provided Pensions

Using the PENSIM microsimulation model, we projected the impact of a
reduced vesting schedule on retirement benefits. For DB plans we specified
2- year cliff vesting, and for DC plans we specified 2-year cliff, and
3-year graduated vesting schedules. For the 1985 cohort, the median
percentage change in benefit levels for women who were affected by the
change was an increase of 6.29 percent. Similarly, men's median percentage
change in benefits for those affected was of 5.74 percent.^84 (See table
11.) While the number of women affected by this change was fairly evenly
distributed across the top four income quintiles with fewer in the lowest,
the median percentage change in benefits is much larger for the women in
the lowest quintile than in higher quintiles; for those in the lowest
quintile, the median percentage change in benefit levels was more than
four times the change for women in the highest quintile and nearly twice
that of women in the second lowest income quintile. This suggests that
increases in pension benefits gained as a result of this change represent
a larger portion of total pension accumulation for less affluent women.
Similarly, while the number of married women affected was larger than the
number affected in other marital classifications, the median percentage
change in benefits for never married and divorced women was almost twice
the median percentage change in benefits for married and widowed women.

^83One survey found that the most frequently cited reason for not
participating in a retirement plan when a plan was offered by employers
was an insufficient period of employment (39 percent of male employees
versus 35 percent of female employees).

^84We simulated pension income only for the 1985 birth cohort because the
current version of PENSIM does not have a realistic characterization of
pre-1996 employer pension offerings and therefore should not be used to
simulate lifetime pension accumulation for cohorts born before 1975.

Table 11: Median Percentage Change in Pension Benefits for Individuals
Born in 1985 Whose Benefits Changed After Lowered Vesting Schedules

Numbers in percent
All men                                                        All women
5.74                                                                6.29
                           Women by income quintile
Lowest quintile   Quintile 2 Middle quintile Quintile 4 Highest quintile 
16.37                   8.90            6.18       5.11             3.51 
                           Women by marital status
Never married                Divorced      Married               Widowed
11.14                           11.30         5.35                  6.81

Source: GAO analysis based on the PENSIM model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I). Income quintiles
are based on the distribution of the present value of family lifetime
earnings for the whole population (male and female). For percentage of
population affected and results using the "Funded benefits" benchmark, see
appendix II.

Table 12: Percentage of Total Simulation Population Whose Benefits Changed
after Implementation of Lowered Vesting Schedules

Numbers in percent              
                      Birth Cohort
                              1985 
All men                   14.34 
All women                 12.62 

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I).

Automatic Rollover upon Leaving Employment Prior to Retirement Age

"Automatic rollover" proposals would maintain the level of accrued
retirement benefits in DC plans when an individual switches jobs before
retirement by automatically contributing retirement balances to a
qualified retirement savings account. According to research conducted by
the Employee Benefit Research Institute in 2003, under 50 percent of
recipients of lump sum distributions between the ages of 30 and 50
reported using the entire portion for reinvestment into a qualified
account.^85 For those aged 21 to 30, the percentage using the entire
distribution for tax-qualified financial savings drops to under 35. This
modification would provide greater retirement income for men and women.
However, some research shows that women roll over a lower percentage of
their accrued balances than men do;, because of this, requiring automatic
rollover may have a larger effect on women overall.

GAO Simulation of 100 Percent Automatic Rollover for Employer-Provided Pensions

GAO used the GEMINI microsimulation model to determine what the impact
could be on retirement benefit levels if 100 percent of accrued retirement
balances were reinvested into qualified accounts after every job change
until retirement. For the 1985 cohort, the median percentage change in
benefits for those affected was quite similar for men and women, 7.3
percent and 7.63 percent, respectively. (See table 13.) This assumes that
those affected would not make any changes in their savings or spending
behavior to offset the requirement. Among women who were affected by the
change, those who were never married or who were divorced had the largest
percentage median increases in benefits. Additionally, while the number of
women affected by this change again was fairly evenly distributed across
the top four income quintiles with fewer in the lowest, those in the
lowest two income quintiles had substantially larger median percentage
changes in benefits than those in the highest two quintiles.

^85"Lump-Sum Distributions," EBRI Notes, Volume 16, no.12 (Employee
Benefit Research Institute, December 2005).

Table 13: Median Percentage Change in Pension Benefits for Individuals
Born in 1985 Whose Benefits Changed after Implementation of 100 Percent
Automatic Rollover

Numbers in percent
All men                                                        All women
7.30                                                                7.63
                           Women by income quintile
Lowest quintile   Quintile 2 Middle quintile Quintile 4 Highest quintile 
16.94                  11.85            8.50       5.97             3.62 
                           Women by marital status
Never married                Divorced      Married               Widowed
15.04                           12.09         6.74                  7.48

Source: GAO analysis based on the PENSIM model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I). Income quintiles
are based on the distribution of the present value of family lifetime
earnings for the whole population (male and female). For percentage of
population affected and results using the "Funded benefits" benchmark, see
appendix II.

Table 14: Percentage of Total Simulation Population Whose Benefits Changed
after Implementation of 100 Percent Automatic Rollover

Numbers in percent              
                      Birth Cohort
                              1985 
All men                   12.26 
All women                 11.21 

Source: GAO analysis based on the GEMINI model.

Note: Simulations calculated benefits for individuals at age 70 and were
based on benefits adjusted for household size (for more information, see
discussion of Family Equivalence Scale in appendix I).

Concluding Observations

Despite the increases in women's labor force participation over the past
55 years, certain groups of women will continue to be vulnerable to
economic insecurity in retirement. While women are working more than in
the past, they remain the primary source of family caregiving and are more
likely than men either to reduce their workforce participation or never to
enter the paid workforce. Consequently, despite elements of the Social
Security and employer-sponsored pension systems that provide retirement
income for low- or non-earning spouses, the remaining gaps between women's
and men's labor force participation, earnings, and pension participation
will continue to leave many women with fewer financial resources in
retirement than men. In addition to the choices many women make to stay
out of the workforce or to reduce the amount of their work, certain life
events (such as divorce and widowhood) are likely to exacerbate this
disparity. Ultimately, women's roles in the workplace and within a family
may hinder them from building sufficient retirement resources, leaving
them at greater risk of poverty in old age.

Other trends can exacerbate the vulnerability of women in retirement. Many
proposed reforms for Social Security and employer-provided pensions have
focused on long-term solvency and financing issues. Research has shown
that many of these types of reforms have the potential to reduce
retirement income from levels scheduled in current law for a large number
of beneficiaries. This is of particular concern for women because of their
reliance on Social Security as a main source of retirement income. Changes
in the structure of employer-sponsored pensions, which can have some
benefits for women, may also have some negative consequences. Although the
shift to DC type plans can have a positive impact on people who change
jobs frequently or work intermittently, such plans can transfer more of
the responsibility to make prudent investment decisions and to manage
longevity risk to individuals. While this is true for both men and women,
it is of particular concern for women because of their greater longevity.
Moreover, the original design of the Social Security system was based on a
particular household structure--single earner families--and that structure
is no longer the norm in America. If policy makers wish to design a system
that adequately and equitably compensates all retirees, then it will be
necessary to design a system that reflects the diversity of employment
patterns and family structure within the population it serves. In fact,
several past reform proposals included modernization elements that would
target benefit enhancements to various subgroups.

In contemplating modifications to Social Security or employer-provided
pensions, it is helpful to understand all possible effects, including the
impact on Social Security solvency or costs to employers. Each of the
Social Security changes that we modeled would have small, but negative,
effects on program solvency. Small effects such as these, when included in
a larger package of reforms, could be overwhelmed by the effects of other
changes. Nevertheless, the trade-offs between enhanced benefits and costs
are always important to consider. It is also helpful to understand how
changes may affect different types of individuals with different work and
earnings histories--for example, women who never enter the workforce or
choose to reduce their work, possibly to care for children or other family
members. Changes to benefit structures may also have different effects on
individuals within different family structures, such as single-earner
married couples, dual-earner households, or unmarried heads of household.
Recognizing these differences is important not only in terms of improving
adequacy and equity in the benefit structure, but also in understanding
how different benefit structures might affect the choices individuals make
regarding their own workforce attachment. With such knowledge, policy
makers have the potential to mitigate both existing disparities in
retirement income as well as the differential effects of reforms.

Agency Comments

We provided a draft of this report to the departments of Labor and the
Treasury, the Internal Revenue Service, and the Social Security
Administration. The departments of Labor and the Treasury and the Social
Security Administration provided technical comments, which we incorporated
where appropriate.

We are sending copies of this report to the Secretary of Labor, the
Commissioner of Internal Revenue, the Commissioner of Social Security, the
Secretary of the Treasury, and appropriate congressional committees, and
other interested parties. We will also make copies available to others
upon request. In addition, the report will be available at no charge on
GAO's Web site at http://www.gao.gov.

If you or your staff have any questions about this report, please contact
me at (202) 512-7215 or at [email protected]. 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 major contributions to this
report are listed in appendix V.

Barbara D. Bovbjerg, Director
Education, Workforce, and Income Security Issues

Appendix I: Methodology

To identify the effects of individual reform elements on Social Security
and pension benefit levels for women, we used the Policy Simulation
Group's (PSG) microsimulation models to simulate Social Security benefits
and pension income.

For our simulations, we used PSG's Social Security and Accounts Simulator
(SSASIM), Genuine Microsimulation of Social Security Accounts (GEMINI),
and Pension Simulator (PENSIM) simulation models. GEMINI simulates Social
Security benefits and taxes for large representative samples of people
born in the same year. GEMINI simulates all types of Social Security
benefits including retired workers', spouses', survivors', and disability
benefits. It can be used to model a variety of changes to Social Security.
GEMINI uses inputs from SSASIM, which has been used in numerous GAO
reports, and PENSIM, which was developed for the Department of Labor.
GEMINI relies on SSASIM for economic and demographic projections and
relies on PENSIM for simulated life histories of large representative
samples of people born in the same year and their spouses.^1 Life
histories include educational attainment, labor force participation,
earnings, job mobility, marriage, disability, childbirth, retirement, and
death. Life histories are validated against data from the Survey of Income
and Program Participation, the Current Population Survey, Modeling Income
in the Near Term (MINT3),^2 and the Panel Study of Income Dynamics.
Additionally, any projected statistics (such as life expectancy,
employment patterns, and marital status at age 60) are, where possible,
consistent with intermediate cost projections from the Social Security
Administration's Office of the Chief Actuary (OCACT). At their best, such
models can provide only very rough estimates of future incomes. However,
these estimates may be useful for comparing future incomes across
alternative policy scenarios and over time.

GEMINI can be operated as a free-standing model or it can operate as a
SSASIM add-on. When operating as an add-on, GEMINI is started
automatically by SSASIM for one of two purposes. GEMINI can enable the
SSASIM macro model to operate in the Overlapping Cohorts (OLC) mode or it
can enable the SSASIM micro model to operate in the Representative Cohort
Sample (RCS) mode. The SSASIM OLC mode requests GEMINI to produce samples
for each cohort born after 1934 in order to build up aggregate payroll tax
revenues and (Old-Age, Survivors, and Disability Insurance) OASDI benefit
expenditures for each calendar year, which are used by SSASIM to calculate
standard trust fund financial statistics. In either mode, GEMINI operates
with the same logic, but typically with smaller cohort sample sizes in OLC
mode than in the RCS or stand-alone-model mode.

^1While these models use sample data, our report, like others using these
models, does not address the issue of sampling errors. The results of the
analysis reflect outcomes for individuals in the simulated populations and
do not attempt to estimate outcomes for an actual population.

^2MINT3 is a detailed microsimulation model developed jointly by the
Social Security Administration, the Brookings Institution, RAND, and the
Urban Institute to project the distribution of income in retirement for
the 1931 to 1960 birth cohorts.

Using the GEMINI model, we estimated Social Security benefits at age 70
for approximately 2 percent of individuals born in each of two
illustrative birth cohorts, 1950 (resulting in a sample of 63,813
individuals) and 1985 (resulting in a sample of 78,857 individuals). We
also used the PENSIM model to estimate pension income for those born in
1985.^3 We simulated Social Security benefits for two cohorts in order to
identify differences in the effects of modifications that could be due to
variations in labor force participation across generations. We also used
the microsimulation models to simulate Social Security benefits, pension
income, and the earnings of spouses not yet retired, in order to explore
the relationships between benefit levels and workforce attachment and
marital status. For this analysis we simulated benefit levels at ages 67
and 85 for the 1985 birth cohort. These models do not include measures of
personal savings, earnings in retirement, health benefits, or income from
other income support programs.

Additionally, we evaluated the effect of Social Security modifications on
a "low benefit avoidance rate," a measure produced by the model that
proxies for poverty avoidance. However, it does not include pension income
or savings, and so cannot be called a poverty avoidance measure. The low
benefit avoidance rate is expressed as the percentage of retirement years
in which an individual's Social Security benefits (plus any earnings) are
above a low-benefit threshold in the GEMINI model (the thresholds are
$9,669 for individuals and $12,186 for couples, in 2007 dollars). Both
income and the threshold are based on individual data when unmarried and
on couple data when married.

^3We did not simulate pension benefits for the 1950 birth cohort because
the current version of PENSIM does not have a realistic characterization
of pre-1996 employer pension offerings, and therefore, should not be used
to simulate lifetime pension accumulation for cohorts born before 1975.

Benefits and taxes were simulated under our tax increase only (promised
benefits) and proportional benefit reduction (funded benefits) benchmarks
(described below) and certain specific individual programmatic changes.
These simulations are based on the Social Security Trustees' 2007
intermediate economic and actuarial assumptions. While our simulations
provide projections of future retirement income, there is a considerable
amount of uncertainty involved with these estimates. Since these estimates
could change significantly, depending on assumptions used and behavior
responses, they should not be considered predictions. Furthermore, because
simulations are sensitive to economic and demographic assumptions, it is
generally more appropriate to compare benefits across the scenarios than
to focus on the actual estimates themselves. Therefore, to avoid
inappropriate comparisons, we evaluated the effects of modifications based
on the changes in benefit levels rather than comparing actual benefit
levels.

In general, GAO has suggested that policy makers should consider three
basic criteria when evaluating reform proposals^4

           o the extent to which the proposal achieves sustainable solvency
           and how the proposal would affect the economy and the federal
           budget;
           o the balance struck between the goals of individual equity^5
           (rates of return on individual contributions) and income
           adequacy^6 (level and certainty of monthly benefits); and
           o how readily such changes could be implemented, administered, and
           explained to the public.

Moreover, changes to the system should be evaluated as packages that
strike a balance among the individual elements of the proposal and the
interactions among these elements. The overall evaluation of any
particular reform proposal depends on the weight individual policy makers
place on each of the above criteria.

^4See GAO, Social Security: Criteria for Evaluating Reform Proposals,
[64]GAO/T-HEHS-99-94 (Washington, D.C.: Mar. 25, 1999), and GAO, Social
Security: Evaluating Reform Proposals, [65]GAO/AIMD/HEHS-00-29
(Washington, D.C.: Nov. 4, 1999).

^5For a discussion of individual equity issues, see GAO, Social Security:
Issues in Comparing Rates of Return with Market Investments,
[66]GAO/HEHS-99-110 (Washington, D.C.: Aug. 5, 1999).

^6 [67]GAO-02-62 .

However, for the purposes of this study we evaluated only specific
individual modifications. We looked at certain specific changes in order
to focus on those that account for more recent shifts in family structure
and labor force composition. In particular, we wanted to identify the
direction of the impact of modifications that might be used to mitigate
the effects of broad reform packages likely to reduce benefits.
Nevertheless, we recognize that there could be important interaction
effects with any set of reforms and maintain the importance of considering
all possible effects of any reform package as a whole. The solvency impact
of any single programmatic change may be marginal and, as part of a
package, could be overwhelmed by other changes. Nevertheless, in appendix
IV, we have provided the impact of each change on the Social Security
Trust Fund balance, after achieving 75-year solvency with the benchmark
scenarios. Additionally, because we simulated programmatic changes in
isolation we could not calculate traditional equity measures, which relate
benefits received to taxes paid. Because we simulated programmatic changes
in isolation on top of solvent benchmark scenarios, the results did not
achieve long-term solvency. We did not speculate on how the changes would
be paid for in the context of overall reform. Without information on total
contributions or benefits under each simulation, traditional equity
measures would not be meaningful. Finally, given the limited scope of the
changes we simulated, we did not address issues of implementation,
administration, or public comprehension.

Assumptions and Limitations

Simulating retirement income almost 50 years into the future requires many
assumptions and simplifications, and consequently, our simulations have a
number of limitations. A primary limitation of our analysis is that it
does not include important components of retirement income such as
personal savings, earnings in retirement, health benefits, and other
public assistance programs such as Supplemental Security Income (SSI). To
facilitate our modeling analysis, we made a variety of assumptions
regarding economic and demographic trends. In choosing our assumptions, we
focused our analysis to illustrate relevant points about distributional
effects and hold equal as much as possible any variables that were either
not relevant to or would unduly complicate that focus. As a result of
these assumptions, as well as issues inherent in any modeling effort, our
analysis has some key limitations.

2007 Social Security Trustees' Assumptions

The simulations are based on economic and demographic assumptions from the
2007 Social Security Trustees' report.^7 We used the Trustees'
intermediate assumptions for inflation, real wage growth, mortality
decline, immigration, labor force participation, and interest rates.

Family Equivalence Scale

For some of our analyses, we used a measure of income adjusted to account
for household size and economies of scale. We did this to facilitate
comparisons between nonmarried persons and married persons whose household
income includes income from both spouses that can vary significantly
between them. For instance, although a married couple may need
approximately twice as much for food and clothing as a single person,
other needs, such as housing and transportation, are not additive in the
same way. However, the effect of using data adjusted for household size on
a reform targeted at married couples, such as a change in spousal
benefits, is that the change in benefits resulting from the program
modification is shared by both the husband and the wife. Thus, population
data based on the adjusted measure describe the number of people whose
household had a benefit change resulting from the modification. For
example, in the "Decreased Spousal Benefit and Increased Survivor Benefit"
modification, the family equivalence data indicate that a nearly equal
percent of men and women are affected by the modification because the
effect of the benefit change is shared by both spouses. Therefore, in
order to identify just the percent of men and women who had changes to
their own benefit as a result of the program change, percentages are also
calculated based on benefits not adjusted for household size.

The adjustment is made by dividing household benefit levels by a "family
equivalence scale."^8 This equivalence scale reflects both differences in
consumption by adults and children under 18 and the economies of scale
that benefit families. The family equivalence scale in the GEMINI model
(shown below) and its default parameters are based on the recommendations
of the National Academy of Sciences' Panel on Poverty and Family
Assistance.

^7The Board of Trustees, Federal Old-Age and Survivors Insurance and
Disability Insurance Trust Funds, The 2007 Annual Report of the Board of
Trustees of the Federal Old-Age and Survivors Insurance and Federal
Disability Insurance Trust Funds (Washington, D.C.: Apr. 23, 2007).

^8There are both advantages and disadvantages of using such measures. For
additional information on the development, use, and limitations of
equivalence scales see, Constance F. Citro and Robert T. Michael (eds.),
Measuring Poverty: A New Approach, Washington, DC: National Academy Press,
1995, and GAO, Social Security: Program's Role in Helping Ensure Income
Adequacy, [68]GAO-02-62 (Washington, D.C.: Nov 30, 2001).

Family equivalence scale = (A + P*k)^f

where A is the number of adults in the family,

k is the number of children, each of whom is treated as a proportion P of
an adult, and

f is the scale economy factor.

Thus, the formula calculates the number of adult equivalents (A + P*k) and
raises the result to a power f that reflects economies of scale for
families. We used the default parameters in the model, so that both P and
f are 0.70.

Description of Simulated Social Security Modifications

Dependent Care Credit--Analyzing Impact of Inserting a "Credit"

To simulate a dependent care credit we provided a "credit" in the OASDI
work record that would top off a year's earnings to the level of half of
average wages. In a given year, the OASDI work records of individuals with
a child 5 years of age or under and who had earnings less than one-half of
average wages were credited with one half of average wages (as measured by
the Average Wage Index used by the Social Security Administration). A
lifetime maximum of five credits was allowed. If both parents met these
criteria in any given year, only the lower-earning parent received the
credit; if both parent's earnings were identical in any given year, the
credit was split evenly. Credits were not counted toward benefit
eligibility. In our simulation the reform went into effect in 2010.

This change as specified has not been scored publicly by OCACT. Using the
OLC mode of SSASIM that mimics the intermediate assumptions of the 2007
Trustees' report, we estimated this modification as increasing the size of
the long-range actuarial deficit by 0.19 percent under the "promised
benefits" benchmark and by 0.17 under the "funded benefits" benchmark.

Decrease Spousal Benefit/Increase Survivor Benefit

To simulate the effects of a decrease in spousal benefits paired with an
increase in survivor benefits, we reduced spousal benefits from one-half
to one-third of the retired worker's benefit. We also increased the
survivor benefit to 75 percent of the couple's previous combined benefit
level, if higher than the survivor benefit available under current law.
The benefit of this reform is capped at the average PIA for all new
retirees. Our simulation of the reform went into effect in 2010.

These changes as specified have not been scored publicly by OCACT.^9 Using
the OLC mode of SSASIM that mimics the intermediate assumptions of the
2007 Trustees' report, we estimated this modification as decreasing the
size of the long-range actuarial deficit by 0.02 percent under the
"promised benefits" benchmark and increasing the size of the long range
actuarial deficit by 0.06 under the "funded benefits" benchmark.

Increased Survivor Benefit Only

We also simulated the effects of an increase in survivor benefits
independently. To do this, as above, we increased the survivor benefit to
75 percent of the couple's previous combined benefit level, if higher than
the survivor benefit available under current law. The benefit of this
reform is capped at the average PIA for all new retirees. Our simulation
of the reform went into effect in 2010.

This change is similar to one scored by OCACT as part of the OASDI
Financial Effects of the Social Security Guarantee Plus Act of 2005 (May
12, 2005), which estimated a 0.08 percent increase to the long-range
actuarial deficit, and Estimated OASDI Financial Effects of the Bipartisan
Retirement Security Act of 2005 (November 4, 2005), which also estimated a
0.08 percent increase to the long-range actuarial. Using the OLC mode of
SSASIM that mimics the intermediate assumptions of the 2007 Trustees'
report, we estimated this modification as increasing the size of the
long-range actuarial deficit by 0.07 percent under the "promised benefits"
benchmark and by 0.14 percent under the "funded benefits" benchmark.

^9This change is similar to one scored by OCACT as part of the 1994-96
Advisory Council Report, which estimates a 0.32 percent increase to the
long-range actuarial deficit. However, the OCACT scoring of the similar
change in the 1994-96 Advisory Council Report did not include a cap at the
level of average benefits, so the impact on long-range solvency was
greater. A capped version of an increase in survivors' benefits was
presented as part of Estimates of Financial Effects for Three Models
Developed by the President's Commission to Strengthen Social Security in
January 2002; for these outcomes, please see the description of "increase
survivor benefit" change.

Reduce 10-Year Marriage Requirement

To simulate the effects of a reduced marriage requirement to qualify for
divorced spouse benefits, we reduced duration of marriage requirements
from 10 to 7 years. In our simulation, this reform took effect in 2010.

This change as specified has not been scored publicly by OCACT. Using the
OLC mode of SSASIM that mimics the intermediate assumptions of the 2007
Trustees' report, we estimated this change as increasing the size of the
long-range actuarial deficit by 0.02 percent under the "promised benefits"
benchmark and increasing the size of the long-range actuarial deficit by
0.06 under the "funded benefits" benchmark.

Increase/Strengthen Minimum Benefit

To simulate an increased minimum benefit, we provided a partial benefit
enhancement for workers with more than 80 quarters of coverage and provide
a full benefit enhancement to workers with 120 quarters of coverage. It
would equal 120 percent of the aged poverty threshold for workers with 120
quarters of coverage and be linearly prorated to zero for workers with 80
quarters of coverage. These provisions would also apply in determining the
PIA levels used for calculating auxiliary benefits and DI benefits. The
first year of full implementation is 2010.

This modification as specified has not been scored publicly by OCACT.
Using the OLC mode of SSASIM that mimics the intermediate assumptions of
the 2007 Trustees' report, we estimated this change as increasing the size
of the long-range actuarial deficit by 0.05 percent under the "promised
benefits" benchmark and increasing the size of the long range actuarial
deficit by 0.13 percent under the "funded benefits" benchmark as well.

Description of Simulated Pension Modifications

Decrease Vesting Requirements

In our modeling of this modification, we reduced maximum allowable vesting
periods to 2 years for all pension programs with cliff vesting and 3 years
for all pension programs with graduated vesting. If plans within the model
already used shorter vesting schedules, those vesting schedules remained
unchanged. In our simulation, this change took effect in 2010.

Automatic Rollover

In order to consider the potential upper bound of impact for a
modification that would ensure full rollover of retirement assets at
preretirement job terminations, we modeled a full rollover of retirement
balances into qualified accounts. In our simulation, this change took
effect in 2010.

Data Reliability

To assess the reliability of simulated data from GEMINI, we reviewed PSG's
published validation checks, examined the data for reasonableness and
consistency, and compared our solvency estimates, where applicable, with
published results from the actuaries at the Social Security
Administration.

PSG has published a number of validation checks of its simulated life
histories. For example, simulated life expectancy is compared with
projections from the Social Security Trustees; simulated benefits at age
62 are compared with administrative data from SSA; and simulated
educational attainment, labor force participation rates, and job tenure
are compared with values from the Current Population Survey. We found that
simulated statistics for the life histories were reasonably close to the
validation targets.

Benchmark Policy Scenarios

According to current projections of the Social Security Trustees for the
next 75 years, revenues will not be adequate to pay full benefits as
defined by the current benefit formula. Therefore, estimating future
Social Security benefits should reflect that actuarial deficit and account
for the fact that some combination of benefit reductions and revenue
increases will be necessary to restore long-term solvency.

To illustrate a full range of possible outcomes, we developed hypothetical
benchmark policy scenarios that would achieve 75-year solvency either by
only increasing payroll taxes or by only reducing benefits.^10 In
developing these benchmarks, we identified criteria to use to guide their
design and selection. Our tax-increase-only benchmark simulates "promised
benefits," or those benefits promised by the current benefit formula,
while our benefit-reduction-only benchmark simulates "funded benefits," or
those benefits for which currently scheduled revenues are projected to be
sufficient. Under the latter policy scenario, the benefit reductions would
be phased in between 2010 and 2040 to strike a balance between the size of
the incremental reductions each year and the size of the ultimate
reduction.

^10These benchmarks were first developed for our report [69]GAO-02-62 . We
have since used them in other studies, including [70]GAO-03-310 ; GAO,
Social Security Reform: Analysis of a Trust Fund Exhaustion Scenario,
[71]GAO-03-907 (Washington, D.C.: July 29, 2003); GAO, Social Security and
Minorities: Earnings, Disability Incidence, and Mortality Are Key Factors
That Influence Taxes Paid and Benefits Received, [72]GAO-03-387
(Washington, D.C.: Apr. 23, 2003); [73]GAO-04-747 ; and GAO, Social
Security Reform: Implications of Different Indexing Choices,
[74]GAO-06-804 (Washington, D.C.: Sept. 14, 2006).

Social Security Administration (SSA) actuaries scored our original 2001
benchmark policies and determined the parameters for each that would
achieve 75-year solvency.^11 Table 8 summarizes our benchmark policy
scenarios. For our benefit reduction scenario, the actuaries determined
these parameters assuming that disabled and survivor benefits would be
reduced on the same basis as retired worker and dependent benefits. If
disabled and survivor benefits were not reduced at all, reductions in
other benefits would be greater than shown in this analysis.

Table 15: Summary of Benchmark Policy Scenarios

                                                                 Ultimate new 
                                                                      benefit 
Benchmark policy                                    Phase-in  reductions^a 
scenario          Description                       period       (percent) 
Tax increase only Increases payroll taxes in 2008   Immediate            0 
(promised         by amount necessary to achieve                           
benefits)         75-year solvency (0.98 percent of                        
                     payroll each for employees and                           
                     employers)                                               
Proportional      Reduces benefit formula factors   2013-2043           27 
benefit reduction proportionally across all                                
(funded benefits) earnings levels                                          

Source: GAO.

aThese benefit reduction amounts do not reflect the implicit reductions
resulting from the gradual increase in the full retirement age that has
already been enacted.

^11The Social Security actuaries provided these scorings for a previous
report and used assumptions from the 2001 trustees' report. The actuaries
did not believe it was necessary to provide new scorings using updated
assumptions for the purposes of our study, since the assumptions and the
estimates of actuarial balance on which they are based have changed little
from the 2001 report. In particular, they did not believe that the
differences in assumptions would materially affect the shape of the
distribution of benefits, which is the focus of our analysis. All
estimates related to the indexing scenarios and benchmark policy scenarios
were simulated using the SSASIM OLC mode.

Criteria

According to our analysis, appropriate benchmark policies should ideally
be evaluated against the following criteria:

           1. Distributional neutrality: The benchmark should reflect the
           current system as closely as possible while still restoring
           solvency. In particular, it should try to reflect the goals and
           effects of the current system with respect to redistribution of
           income. However, there are many possible ways to interpret what
           this means, such as

                        a. producing a distribution of benefit levels with a
                        shape similar to the distribution under the current
                        benefit formula (as measured by coefficients of
                        variation, skewness, kurtosis, and so forth);
                        b. maintaining a proportional level of income
                        transfers in dollars;
                        c. maintaining proportional replacement rates; and
                        d. maintaining proportional rates of return.

           2. Demarcating upper and lower bounds: These would be the bounds
           within which the effects of alternative proposals would fall. For
           example, one benchmark would reflect restoring solvency solely by
           increasing payroll taxes and therefore maximizing benefit levels,
           while another would solely reduce benefits and therefore minimize
           payroll tax rates.
           3. Ability to model: The benchmark should lend itself to being
           modeled within the GEMINI model.
           4. Plausibility: The benchmark should serve as a reasonable
           alternative within the current debate; otherwise, the benchmark
           could be perceived as an invalid basis for comparison.
           5. Transparency: The benchmark should be readily explainable to
           the reader.

Tax-Increase-Only or "Promised Benefits," Benchmark Policies

Our tax-increase-only benchmark would raise payroll taxes once and
immediately by the amount of Social Security's actuarial deficit as a
percentage of payroll. It results in the smallest ultimate tax rate of
those we considered and spreads the tax burden most evenly across
generations; this is the primary basis for our selection. The later that
taxes are increased, the higher the ultimate tax rate needed to achieve
solvency, and in turn the higher the tax burden on later taxpayers and
lower on earlier taxpayers. Still, any policy scenario that achieves
75-year solvency only by increasing revenues would have the same effect on
the adequacy of future benefits in that promised benefits would not be
reduced. Nevertheless, alternative approaches to increasing revenues could
have very different effects on individual equity.

Benefit-Reduction-Only, or "Funded Benefits," Benchmark Policies

We developed alternative benefit reduction benchmarks for our analysis.
For ease of modeling, all benefit reduction benchmarks take the form of
reductions in the benefit formula factors; they differ in the relative
size of those reductions across the three factors, which are 90, 32, and
15 percent under the current formula. Each benchmark has three dimensions
of specification: scope, phase-in period, and the factor changes
themselves. For our analysis, we apply benefit reductions in our
benchmarks very generally to all types of benefits, including disability
and survivors' benefits as well as old-age benefits. Our objective is to
find policies that achieve solvency while reflecting the distributional
effects of the current program as closely as possible. Therefore, it would
not be appropriate to reduce some benefits and not others. If disabled and
survivor benefits were not reduced at all, reductions in other benefits
would be deeper than shown in this analysis.

  Phase-in Period

We selected a phase-in period that begins with those becoming initially
entitled in 2013 and continues for 30 years. We chose this phase-in period
to achieve a balance between two competing objectives: (1) minimizing the
size of the ultimate benefit reduction and (2) minimizing the size of each
year's incremental reduction to avoid "notches," or unduly large
incremental reductions. Notches create marked inequities between
beneficiaries close in age to each other. Later birth cohorts are
generally agreed to experience lower rates of return on their
contributions already under the current system. Therefore, minimizing the
size of the ultimate benefit reduction would also minimize further
reductions in rates of return for later cohorts. The smaller each year's
reduction, the longer it will take for benefit reductions to achieve
solvency, and in turn the greater the eventual reductions will have to be.
However, the smallest possible ultimate reduction would be achieved by
reducing benefits immediately for all new retirees by 13 percent; this
would create a notch.

In addition, we feel it is appropriate to delay the first year of the
benefit reductions for a few years because those within a few years of
retirement would not have adequate time to adjust their retirement
planning if the reductions applied immediately. The Maintain Tax Rates
(MTR) benchmark in the 1994-1996 Advisory Council report also provided for
a similar delay.^12

Finally, the timing of any policy changes in a benchmark scenario should
be consistent with the proposals against which the benchmark is compared.
The analysis of any proposal assumes that the proposal is enacted, usually
within a few years. Consistency requires that any benchmark also assumes
enactment of the benchmark policy in the same time frame. Some analysts
have suggested using a benchmark scenario in which Congress does not act
at all and the trust funds become exhausted.^13 However, such a benchmark
assumes that no action is taken, while the proposals against which it is
compared assume that action is taken, which is inconsistent. It also seems
unlikely that a policy enacted over the next few years would wait to
reduce benefits until the trust funds are exhausted; such a policy would
result in a sudden large benefit reduction and create substantial
inequities across generations.

  Defining the PIA Formula Factor Reductions

When workers retire, become disabled, or die, Social Security uses their
lifetime earnings records to determine each worker's PIA, on which the
initial benefit and auxiliary benefits are based. The PIA is the result of
two elements--the Average Indexed Monthly Earnings (AIME) and the benefit
formula. The AIME is determined by taking the lifetime earnings record,
indexing it, and taking the average of the highest 35 years of indexed
wages.^14 To determine the PIA, the AIME is then applied to a step-like
formula, shown here for 2007.

PIA = 90% (AIME[1] =< $680) + 32% (AIME[2] > $680 and =< $4100) + 15%
(AIME[3] > $4100)

where AIMEi is the applicable portion of AIME.

^12Advisory Council on Social Security. Report of the 1994-1996 Advisory
Council on Social Security, Vols. 1 and 2. Washington, D.C.: Jan. 1997.

^13See [75]GAO-03-907 , in which we analyzed such a policy scenario under
a congressional request.

^14The highest 35 years of salary are used in the calculation of a retired
worker benefit. The disabled worker benefit is calculated using the number
of years between the age of entitlement and age 21, divided by 5.

All of our benefit-reduction benchmarks are variations of changes in PIA
formula factors.

Proportional reduction: Each formula factor is reduced annually by
subtracting a constant proportion of that factor's value under current
law, resulting in a constant percentage reduction of currently promised
benefits for everyone. That is,

F^i[t+1] = F^i[t] - (F^i[2008] x)

where

F^i[t] represents the three PIA formula factors in year t and

x = constant proportional formula factor reduction.

The value of x is calculated to achieve 75-year solvency, given the chosen
phase-in period and scope of reductions.

The formula for this reduction specifies that the proportional reduction
is always taken as a proportion of the current law factors rather than the
factors for each preceding year. This maintains a constant rate of benefit
reduction from year to year. In contrast, taking the reduction as a
proportion of each preceding year's factors implies a decelerating of the
benefit reduction over time because each preceding year's factors get
smaller with each reduction. To achieve the same level of 75-year
solvency, this would require a greater proportional reduction in earlier
years because of the smaller reductions in later years.

The proportional reduction hits lower earners harder than higher earners
because the constant x percent of the higher formula factors results in a
larger percentage reduction over the lower earnings segments of the
formula. For example, in a year when the cumulative size of the
proportional reduction has reached 10 percent, the 90 percent factor would
then have been reduced by 9 percentage points, the 32 percent factor by
3.2 percentage points, and the 15 percent factor by 1.5 percentage points.
As a result, earnings in the first segment of the benefit formula would be
replaced at 9 percentage points less than the current formula, while
earnings in the third segment of the formula would be replaced at only 1.5
percentage points less than the current formula.^15

Table 9 summarizes the features of our benchmarks.

Table 16: Summary of Benchmark Policy Scenario Parameters

                                 Annual PIA factor            
                               reduction (percentage    Ultimate PIA factor
                                      point)             (2043) (percent)
                                   90      32      15      90      32      15 
Benchmark policy Phase-in  percent percent percent percent percent percent 
scenario         period     factor  factor  factor  factor  factor  factor 
Tax increase     2008            0       0       0   90.00   32.00   15.00 
only (promised                                                             
benefits)                                                                  
Proportional     2013-2043    0.80    0.28    0.13   65.28   23.21   10.88 
benefit                                                                    
reduction                                                                  
(funded                                                                    
benefits)                                                                  

Source: GAO's analysis as scored by SSA actuaries.

Note: Annual PIA factor reductions rounded to the nearest hundredth of a
percent.

^15 Other analyses have addressed the concern about the effect of the
proportional reduction on low earners by modifying that offset to apply
only to the 32 and 15 percent formula factors. The MTR policy in the 1994
to 1996 Advisory Council report used this approach, which in turn was
based on the individual account (IA) proposal in that report. However, the
MTR policy also reflected other changes in addition to PIA formula
changes.

Appendix II: Simulation Results for Social Security Modifications

To account for differences in household size and economies of scale
associated with larger households, we based our analyses on benefit levels
adjusted for household size (see app. I for more information). However,
the effect of using data adjusted for household size on a reform targeted
at married couples, such as a change in spousal benefits, is that the
change in benefits resulting from the program modification is shared by
both the husband and the wife. Thus, population data based on the adjusted
measure describe the number of people whose household had a benefit change
resulting from the modification. For example, in the "Decreased Spousal
Benefit and Increased Survivor Benefit" modification, the family
equivalence data indicate that a nearly equal percent of men and women are
affected by the modification because the effect of the benefit change is
shared by both spouses. Therefore, in order to identify only the percent
of men and women who had changes to their own benefit as a result of the
program change, percentages are also calculated based on benefits not
adjusted for household size.

Table 17: Simulation Results of the Dependent Care Credit under
Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts

                             Dependent care credit
Benchmark                          Cohort             All men    All women 
                     Percentage change in median benefits for individuals
                     whose benefits changed, benefits adjusted for household
                     size      
Promised benefits                    1950                2.60         3.24 
                          1985          2.06                             2.63
Funded benefits                      1950                2.58         3.24 
                          1985          2.07                             2.61
                     Percentage of specified subpopulation of simulated
                     sample whose benefits changed, benefits adjusted for
                     household size
Promised benefits                    1950               60.71        57.98 
                          1985         57.05                            60.52
Funded benefits                      1950               60.47        58.04 
                          1985         56.94                            60.25
                     Percentage of total simulation population whose benefits
                     changed, benefits adjusted for household size
Promised benefits                    1950               28.43        30.83 
                          1985         35.06                            30.99
Funded benefits                      1950               28.32        30.86 
                          1985         27.79                            30.84
                     Percentage of specified subpopulation of simulated
                     sample whose benefits changed, individual benefits not
                     adjusted for household size
Promised benefits                    1950               47.85        52.56 
                          1985         44.21                            55.59
Funded benefits                      1950               47.83        52.60 
                          1985         44.13                            55.27
                     Percentage of total simulation population whose benefits
                     changed, individual benefits not adjusted for household
                     size      
Promised benefits                    1950               22.11        28.27 
                          1985         21.38                            28.71
Funded benefits                      1950               22.11        28.29 
                          1985         21.34                            28.54

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 18: Simulation Results for Dependent Care Credit by Income Quintile
under Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts

                                    Women only
                                   Quintile                Quintile                 
Benchmark    Cohort Lowestquintile        2 Middlequintile        4 Highestquintile 
          Percentage change in median benefits for individuals whose benefits
          changed, benefits adjusted for household size
Promised       1950           7.96     4.16           2.80     1.68            0.92 
benefits  1985      6.88        3.50   2.12                1.14                0.72
Funded         1950           7.96     4.16           2.82     1.68            0.93 
benefits  1985      6.85        3.51   2.10                1.13                0.72
          Percentage of women in each income quintile whose benefits changed,
          benefits adjusted for household size
Promised       1950          65.03    64.71          59.36    52.63           43.16 
benefits  1985     73.41       68.15  58.93               54.00               43.31
Funded         1950          65.06    64.75          59.39    52.72           43.28 
benefits  1985     72.78       67.84  58.76               53.95               43.14
          Percentage of total simulation population whose benefits changed,
          benefits adjusted for household size
Promised       1950           7.25     8.13           6.52     5.30            3.62 
benefits  1985      8.09        7.63   6.26                5.33                3.68
Funded         1950           7.25     8.14           6.53     5.31            3.63 
benefits  1985      8.02        7.59   6.24                5.32                3.66

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70. Income
quintiles are based on the distribution of the present value of family
lifetime earnings for the whole population (male and female).

Table 19: Simulation Results for Dependent Care Credit by Marital Status
under Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts

                                Women only
Benchmark    Cohort  Never married    Divorced    Married    Widowed 
             Percentage change in median benefits for women whose
             benefits changed, benefits adjusted for household size
Promised       1950           7.95        5.05       2.64       2.87 
benefits       1985                       6.69       4.18       2.19  2.43 
Funded         1950           7.95        5.05       2.65       2.87 
benefits       1985                       6.69       4.15       2.18  2.37 
             Percentage of women in each marital status whose benefits
             changed, benefits adjusted for household size
Promised       1950          21.06       65.71      64.09      52.39 
benefits       1985                      21.81      67.24      66.85 59.62 
Funded         1950          21.06       65.75      64.21      52.39 
benefits       1985                      21.81      67.08      66.61 58.85 
             Percentage of total simulation population whose benefits
             changed, benefits adjusted for household size
Promised       1950           0.97        8.21      15.04       6.61 
benefits       1985                       1.30       6.89      17.96  4.83 
Funded         1950           0.97        8.21      15.07       6.61 
benefits       1985                       1.30       6.87      17.90  4.77 

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 20: Simulation Results for Increased Survivor Benefit Only under
Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts

                        Increased survivor benefit only
Benchmark                    Cohort          All men             All women
                     Percentage change in median benefits for individuals
                     whose benefits changed, benefits adjusted for household
                     size       
Promised benefits              1950            28.78                 15.71
                           1985           28.98           18.36
Funded benefits                1950            28.36                 15.71
                           1985           28.57           20.09
                     Percentage of specified subpopulation of simulated
                     sample whose benefits changed, benefits adjusted for
                     household size
Promised benefits              1950             1.66                  3.97
                           1985            0.95            2.20
Funded benefits                1950             1.67                  3.97
                           1985            1.76            3.96
                     Percentage of total simulation population whose benefits
                     changed, benefits adjusted for household size
Promised benefits              1950             0.78                  2.11
                           1985            0.46            1.13
Funded benefits                1950             0.78                  2.11
                           1985            0.86            2.03
                     Percentage of specified subpopulation of simulated
                     sample whose benefits changed, individual benefits not
                     adjusted for household size
Promised benefits              1950             1.59                  3.71
                           1985            0.93            2.10
Funded benefits                1950             1.60                  3.71
                           1985            1.71            3.79
                     Percentage of total simulation population whose benefits
                     changed, individual benefits not adjusted for household
                     size       
Promised benefits              1950             0.74                  2.00
                           1985            0.45            1.09
Funded benefits                1950             0.74                  2.00
                           1985            0.83            1.96

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 21: Simulation Results for Increased Survivor Benefit Only by Income
Quintile under Alternative Benchmark Scenarios, for the 1950 and 1985
Birth Cohorts

                                   Women only
                  Cohort   Lowest              Middle Quintile        Highest 
Benchmark             quintile Quintile 2 quintile        4       quintile 
             Percentage change in median benefits for individuals whose
             benefits changed, benefits adjusted for household size
Promised         1950    14.81      16.28    15.77    13.69          24.44 
benefits   1985     19.39  18.40    16.00         16.00               8.67
Funded           1950    14.81      16.46    15.77    14.59          24.44 
benefits   1985     19.64  22.08    21.03         20.75              11.40
             Percentage of women in each income quintile whose benefits
             changed, benefits adjusted for household size
Promised         1950     7.48       6.80     2.96     0.72           0.26 
benefits   1985      5.02   3.51     1.33          0.31               0.09
Funded           1950     7.48       6.80     2.96     0.72           0.26 
benefits   1985      5.63   5.88     4.09          2.49               0.84
             Percentage of total simulation population whose benefits
             changed, benefits adjusted for household size
Promised         1950     0.83       0.85     0.33     0.07           0.02 
benefits   1985      0.55   0.39     0.14          0.03               0.01
Funded           1950     0.83       0.85     0.33     0.07           0.02 
benefits   1985      0.62   0.66     0.43          0.25               0.07

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70. Income
quintiles are based on the distribution of the present value of family
lifetime earnings for the whole population (male and female).

Table 22: Simulation Results for Increased Survivor Benefit Only by
Marital Status under Alternative Benchmark Scenarios, for the 1950 and
1985 Birth Cohorts

                                  Women only
Benchmark        Cohort Never married     Divorced     Married    Widowed
                 Percentage change in median benefits for women whose
                 benefits changed, benefits adjusted for household size
Promised           1950             0        16.67           0      15.58
benefits                                                             
                      1985             0        18.42           0      18.35
Funded             1950             0        16.67           0      15.58
benefits                                                             
                      1985             0        21.22           0      19.65
                 Percentage of women in each marital status whose benefits
                 changed, benefits adjusted for household size
Promised           1950             0         2.62           0      14.12
benefits                                                             
                      1985             0         1.60           0      11.86
Funded             1950             0         2.62           0      14.12
benefits                                                             
                      1985             0         2.73           0      21.58
                 Percentage of total simulation population whose benefits
                 changed, benefits adjusted for household size
Promised           1950             0         0.33           0       1.78
benefits           1985                    0        0.16           0  0.96
Funded             1950             0         0.33           0       1.78
benefits           1985                    0        0.28           0  1.75

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 23: Simulation Results for Increased Survivor Benefit and Decreased
Spouse Benefit under Alternative Benchmark Scenarios, for the 1950 and
1985 Birth Cohorts

            Increased survivor benefit and decreased spousal benefit
Benchmark                    Cohort          All men             All women
                     Percentage change in median benefits for individuals
                     whose benefits changed, benefits adjusted for household
                     size       
Promised benefits              1950            -2.81                 -2.66
                           1985           -9.14           -8.78
Funded benefits                1950            -2.86                 -2.66
                           1985           -8.77           -7.98
                     Percentage of specified subpopulation of simulated
                     sample whose benefits changed, benefits adjusted for
                     household size
Promised benefits              1950            24.45                 22.16
                           1985           23.73           22.09
Funded benefits                1950            24.83                 22.01
                           1985           24.35           23.65
                     Percentage of total simulation population whose benefits
                     changed, benefits adjusted for household size
Promised benefits              1950            11.45                 11.78
                           1985           11.58           11.31
Funded benefits                1950            11.63                 11.70
                           1985           11.89           12.11
                     Percentage of specified subpopulation of simulated
                     sample whose benefits changed, individual benefits not
                     adjusted for household size
Promised benefits              1950             4.82                 19.28
                           1985            5.22           16.81
Funded benefits                1950             4.71                 19.10
                           1985            5.83           18.57
                     Percentage of total simulation population whose benefits
                     changed, individual benefits not adjusted for household
                     size       
Promised benefits              1950             2.23                 10.37
                           1985            2.53            8.68
Funded benefits                1950             2.18                 10.27
                           1985            2.82            9.59

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 24: Simulation Results for Increased Survivor Benefit and Decreased
Spouse Benefit by Income Quintile under Alternative Benchmark Scenarios,
for the 1950 and 1985 Birth Cohorts

                                   Women only
                                     Quintile   Middle Quintile       Highest 
Benchmark   Cohort Lowestquintile        2 quintile        4      quintile 
             Percentage change in median benefits for individuals whose
             benefits changed, benefits adjusted for household size
Promised      1950          -2.55    -2.55    -2.66    -2.67         -2.67 
benefits  1985   -9.82      -8.06       -8.21      -8.45             -9.27
Funded        1950          -2.55    -2.54    -2.66    -2.67         -2.67 
benefits  1985   -9.52      -6.71       -6.99      -7.73             -9.09
             Percentage of women in each income quintile whose benefits
             changed, benefits adjusted for household size
Promised      1950          24.40    22.56    20.51    21.37         21.72 
benefits  1985   23.25      21.44       22.50      22.75             20.14
Funded        1950          24.33    22.36    20.28    21.22         21.64 
benefits  1985   24.31      23.41       24.83      24.74             20.39
             Percentage of total simulation population whose benefits
             changed, benefits adjusted for household size
Promised      1950           2.72     2.84     2.25     2.15          1.82 
benefits  1985    2.56       2.40        2.39       2.24              1.71
Funded        1950           2.71     2.81     2.23     2.14          1.81 
benefits  1985    2.68       2.62        2.64       2.44              1.73

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70. Income
quintiles are based on the distribution of the present value of family
lifetime earnings for the whole population (male and female).

Table 25: Simulation Results for Increased Survivor Benefit and Decreased
Spouse Benefit by Marital Status under Alternative Benchmark Scenarios,
for the 1950 and 1985 Birth Cohorts

Benchmark        Cohort Never married        Divorced    Married   Widowed 
                 Percentage change in median benefits for women whose
                 benefits changed, benefits adjusted for household size
Promised           1950             0           -8.00      -2.67     15.58 
benefits           1985                 0      -24.11    -9.01       18.91
Funded             1950             0           -8.00      -2.67     15.58 
benefits           1985                 0      -21.31    -8.98       20.75
                 Percentage of women in each marital status whose benefits
                 changed, benefits adjusted for household size
Promised           1950             0           12.84      35.89     13.92 
benefits           1985                 0       11.91    34.19       11.11
Funded             1950             0           12.80      35.57     13.92 
benefits           1985                 0       13.13    33.96       20.22
                 Percentage of total simulation population whose benefits
                 changed, benefits adjusted for household size
Promised           1950             0            1.60       8.42      1.76 
benefits           1985                 0        1.22     9.19        0.90
Funded             1950             0            1.60       8.35      1.76 
benefits           1985                 0        1.34     9.13        1.64

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 26: Simulation Results for Reduced Marriage Requirement for Divorced
Spouse Benefits under Alternative Benchmark Scenarios, for the 1950 and
1985 Birth Cohorts

           Reduced marriage requirement for divorced spouse benefits
Benchmark                     Cohort           All men           All women
                     Median percentage change in median benefits for
                     individuals whose benefits changed, benefits adjusted
                     for household size
Promised benefits               1950              0.85               65.32
                           1985             37.22           45.05
Funded benefits                 1950                 0               65.32
                           1985             40.51           47.97
                     Percentage of specified subpopulation of simulated
                     sample whose benefits changed, benefits adjusted for
                     household size
Promised benefits               1950                 0                0.07
                           1985              0.06            1.00
Funded benefits                 1950                 0                0.07
                           1985              0.06            1.01
                     Percentage of total simulation population whose benefits
                     changed, benefits adjusted for household size
Promised benefits               1950                 0                0.04
                           1985              0.03            0.51
Funded benefits                 1950                 0                0.04
                           1985              0.03            0.52
                     Percentage of specified subpopulation of simulated
                     sample whose benefits changed, individual benefits not
                     adjusted for household size
Promised benefits               1950                 0                0.06
                           1985              0.06            0.96
Funded benefits                 1950                 0                0.06
                           1985              0.06            0.97
                     Percentage of total simulation population whose benefits
                     changed, individual benefits not adjusted for household
                     size       
Promised benefits               1950                 0                0.03
                           1985              0.03            0.49
Funded benefits                 1950                 0                0.03
                           1985              0.03            0.50

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 27: Simulation Results for Reduced Marriage Requirement for Divorced
Spouse Benefits by Income Quintile under Alternative Benchmark Scenarios,
for the 1950 and 1985 Birth Cohorts

                                   Women only
                           Lowest              Middle Quintile        Highest 
Benchmark      Cohort quintile Quintile 2 quintile        4       quintile 
             Percentage change in median benefits for individuals whose
             benefits changed, benefits adjusted for household size
Promised         1950    85.18      54.66   162.36    33.33              0 
benefits   1985    53.99 34.41         33.33      36.31              25.14
Funded           1950    85.18      54.66   162.36    33.33              0 
benefits   1985    58.00 37.43         39.34      35.08              25.14
             Percentage of women in each income quintile whose benefits
             changed, benefits adjusted for household size
Promised         1950     0.07       0.13     0.06     0.05              0 
benefits   1985     2.46  1.13          0.70       0.33               0.08
Funded           1950     0.07       0.13     0.06     0.05              0 
benefits   1985     2.44  1.17          0.70       0.35               0.08
             Percentage of total simulation population whose benefits
             changed, benefits adjusted for household size
Promised         1950     0.01       0.02     0.01        0              0 
benefits   1985     0.27  0.13          0.07       0.03               0.01
Funded           1950     0.01       0.02     0.01        0              0 
benefits   1985     0.27  0.13          0.07       0.03               0.01

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70. Income
quintiles are based on the distribution of the present value of family
lifetime earnings for the whole population (male and female).

Table 28: Simulation Results for Reduced Marriage Requirement for Divorced
Spouse Benefits by Marital Status under Alternative Benchmark Scenarios,
for the 1950 and 1985 Birth Cohorts

                                  Women only
Benchmark        Cohort Never married        Divorced   Married   Widowed
                 Percentage change in median benefits for women whose
                 benefits changed, benefits adjusted for household size
Promised           1950             0           65.32         0         0
benefits           1985                     0        45.05         0     0
Funded             1950             0           65.32         0         0
benefits           1985                     0        47.97         0     0
                 Percentage of women in each marital status whose benefits
                 changed, benefits adjusted for household size
Promised           1950             0            0.29         0         0
benefits           1985                     0         4.99         0      
Funded             1950             0            0.29         0         0
benefits           1985                     0         5.04         0     0
                 Percentage of total simulation population whose benefits
                 changed, benefits adjusted for household size
Promised           1950             0            0.04         0         0
benefits           1985                     0         0.51         0     0
Funded             1950             0            0.04         0         0
benefits           1985                     0         0.52         0     0

Retirement Security Retirement Security Retirement Security Retirement
Security Retirement Security Retirement Security Retirement Security
Retirement Security

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 29: Simulation Results for Increased Minimum Benefit under
Alternative Benchmark Scenarios, for the 1950 and 1985 Birth Cohorts

                           Increased minimum benefit
Benchmark                   Cohort           All Men             All Women
                   Percentage change in median benefits for individuals whose
                   benefits changed, benefits adjusted for household size
Promised                      1950              8.47                  9.89
benefits              1985             6.20             6.70
Funded benefits               1950              8.58                  9.94
                         1985             8.31             8.62
                   Percentage of specified subpopulation of simulated sample
                   whose benefits changed, benefits adjusted for household
                   size       
Promised                      1950             16.68                 16.30
benefits              1985             2.43             2.48
Funded benefits               1950             17.02                 16.45
                         1985            13.66            12.57
                   Percentage of total simulation population whose benefits
                   changed, benefits adjusted for household size
Promised                      1950              7.71                  8.77
benefits              1985             1.18             1.28
Funded benefits               1950              7.87                  8.85
                         1985             6.61             6.49
                   Percentage of specified subpopulation of simulated sample
                   whose benefits changed, individual benefits not adjusted
                   for household size
Promised                      1950             10.60                 14.47
benefits              1985             1.79             1.79
Funded benefits               1950             10.63                 14.54
                         1985            10.28             9.03
                   Percentage of total simulation population whose benefits
                   changed, individual benefits not adjusted for household
                   size       
Promised                      1950              4.90                  7.78
benefits              1985             0.86             0.92
Funded benefits               1950              4.91                  7.82
                         1985             4.97             4.67

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Table 30: Simulation Results for Increased Minimum Benefit by Income
Quintiles under Alternative Benchmark Scenarios, for the 1950 and 1985
Birth Cohorts

                                   Women only
                           Lowest              Middle Quintile        Highest 
Benchmark      Cohort quintile Quintile 2 quintile        4       quintile 
             Percentage change in median benefits for individuals whose
             benefits changed, benefits adjusted for household size
Promised         1950    22.83       9.65     5.21     3.51           2.55 
benefits    1985     8.26    2.46    2.27          0.09                  0
Funded           1950    22.83       9.70     5.29     3.57           2.55 
benefits    1985    14.59    5.33    2.98          1.70              3.70*
             Percentage of women in each income quintile whose benefits
             changed, benefits adjusted for household size
Promised         1950    20.68      28.14    15.74     7.18           2.62 
benefits    1985     8.17    1.56    0.46          0.15               0.15
Funded           1950    20.80      28.36    15.92     7.36           2.62 
benefits    1985    28.46   18.65    6.37          1.67               0.46
             Percentage of total simulation population whose benefits
             changed, benefits adjusted for household size
Promised         1950     2.85       3.38     1.65     0.69           0.21 
benefits    1985     1.03    0.17    0.05          0.01               0.01
Funded           1950     2.87       3.40     1.66     0.71           0.21 
benefits    1985     3.60    2.03    0.66          0.16               0.04

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70. Income
quintiles are based on the distribution of the present value of family
lifetime earnings for the whole population (male and female).

*Contrary to the pattern for the 1950 cohort, the median change for the
highest income quintile of the 1985 cohort among women who had a simulated
benefit change is larger than the median change for some of the other
income quintiles. The median change in this quintile, however, is
consistent with our assessment from both cohorts that higher median
changes occur in lower income quintiles. Also, it is worth noting that in
the highest income quintile, women who had simulated benefit changes
accounted for less than 0.25 percent of the population in each cohort.

Table 31: Simulation Results for Increased Minimum Benefit by Marital
Status under Alternative Benchmark Scenarios, for the 1950 and 1985 Birth
Cohorts

                                   Women only
Benchmark     Cohort      Never married     Divorced    Married    Widowed 
             Percentage change in median benefits for women whose benefits
             changed, benefits adjusted for household size
Promised        1950              22.80        17.13       7.43       8.34 
benefits                                                                   
                   1985              19.01        14.50       4.55       6.48 
Funded          1950              22.80        17.14       7.46       8.40 
benefits                                                                   
                   1985              20.53        15.21       6.87       8.34 
             Percentage of women in each marital status whose benefits
             changed, benefits adjusted for household size
Promised        1950              16.97        14.87      17.52      15.17 
benefits                                                                   
                   1985               2.93         2.00       2.82       1.59 
Funded          1950              16.97        14.88      17.83      15.21 
benefits                                                                   
                   1985               9.84         9.45      15.01      10.48 
             Percentage of total simulation population whose benefits
             changed, benefits adjusted for household size
Promised        1950               0.81         1.91       4.19       1.86 
benefits                                                                   
                   1985               0.18         0.21       0.76       0.13 
Funded          1950               0.81         1.91       4.27       1.87 
benefits                                                                   
                   1985               0.61         0.99       4.06       0.84 

Source: GAO analysis of GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Appendix III: Low Benefit Avoidance Rates

The low benefit avoidance rate is expressed as the percent of retirement
years in which an individual's Social Security benefits (plus any
earnings) are above a low-benefit threshold (set at $9,669 for individuals
and $12,186 for couples, in 2007 dollars). Both income and the threshold
are based on individual data when unmarried and on couple data when
married.

Table 32: Average Low Benefit Avoidance Rates Before and After
Modifications for Individuals with Less Than 100 Percent Low Benefit
Avoidance Pre-Modification

                               Promised benefits (tax increase only) benchmark
                     1950 cohort                        1985 cohort                      
Social                                                                                           
Security       All   Never                                  All   Never                          
modification women married Divorced Married Widowed       women married Divorced Married Widowed 
Current law  17.05    0.06     9.77   37.41   11.77 19.21     0    8.87      42.18 12.53         
(no reform)                                                                                      
Dependent    21.18    1.33    13.50   42.27   17.70 24.49  1.50   13.32      49.46 20.15         
care credit                                                                                      
Increased    23.11    0.06    12.22   47.35   23.16 25.59     0   11.47      53.40 24.79         
survivor                                                                                         
benefit only                                                                                     
Decreased    22.76    0.06    11.91   46.75   22.89 24.10     0    9.95      50.83 24.01         
spousal                                                                                          
benefit with                                                                                     
increased                                                                                        
survivor                                                                                         
benefit                                                                                          
Decreased    17.14    0.06    10.03   37.41   11.77 20.72     0   13.57      42.18 12.53         
marriage                                                                                         
requirement                                                                                      
Increased    23.18    6.88    15.29   42.80   19.73 19.66     0    8.87      43.36 12.80         
minimum                                                                                          
benefit                                                                                          

                       Funded benefits (proportional benefit reduction only) benchmark
                     1950 cohort                            1985 cohort
Social                                                                                           
Security       All   Never                                  All   Never                          
modification women married Divorced Married Widowed       women married Divorced Married Widowed 
Current law  17.00    0.06     9.63   37.32   11.79 14.82     0    5.83      30.51  7.84         
(no reform)                                                                                      
Dependent    21.17    1.33    13.38   42.28   17.70 17.72  0.42    7.52      35.44 10.59         
care credit                                                                                      
Increased    23.09    0.06    12.12   47.30   23.17 21.02     0    8.82      40.26 18.01         
survivor                                                                                         
benefit only                                                                                     
Decreased    22.67    0.06    11.68   46.60   22.90 20.12     0    8.55      38.37 17.48         
spousal                                                                                          
benefit with                                                                                     
increased                                                                                        
survivor                                                                                         
benefit                                                                                          
Decreased    17.10    0.06     9.89   37.32   11.79 15.51     0    8.23      30.51  7.84         
marriage                                                                                         
requirement                                                                                      
Increased    23.20    6.88    15.18   42.88   19.78 16.58     0    5.84      34.85  8.23         
minimum                                                                                          
benefit                                                                                          

Source: GAO analysis based on GEMINI model.

Note: Simulations calculated benefits for individuals at age 70.

Appendix IV: Effect of Simulated Reform on Social Security System Solvency

Table 33: Changes in the 75-year Actuarial Balance as a Percentage of
Taxable Payroll resulting from Program Modifications, after Achieving
75-Year Solvency with Benchmark Scenarios

                                                              Funded benefits 
                                      Promised benefits (proportional benefit 
Modification                     (tax increase only)       reduction only) 
Dependent care credit                          -0.17                 -0.19 
Increased survivor benefit only                -0.07                 -0.14 
Increased survivor benefit with                 0.02                 -0.06 
decreased spousal benefit                                                  
Reduced marriage requirement                   -0.02                 -0.06 
(from 10 years to 7 years)                                                 
Increased minimum benefit                      -0.05                 -0.13 

Source: GAO analysis of SSASIM model.

0 (130556)

Appendix V: GAO Contact and Staff Acknowledgments

GAO Contact

Barbara D. Bovbjerg, Director (202) 512-7215

Acknowledgments

Alicia Puente Cackley, Assistant Director; Mindy Bowman,
Analyst-in-Charge; Jennifer Cook, Analyst; and Meaghan Muldoon Mann,
Senior Analyst, made significant contributions to all phases of this
report. In addition, Melinda Cordero, Nagla'a El-Hodiri, and Walter Vance
provided data analysis; Joseph Applebaum, Michael Collins, Chuck Ford,
Gene Kuehneman and Ken Stockbridge provided methodological assistance;
Sheila McCoy provided legal assistance; Marc Goldwein and Emily Pickrell
assisted with data collection and analysis; and Sue Bernstein, Kim
Granger, Kevin Kumanga, Lise Levie, and Beth Morrison assisted in report
development.

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 Web site ( [76]www.gao.gov ). Each weekday, GAO posts
newly released reports, testimony, and correspondence on its Web site. To
have GAO e-mail you a list of newly posted products every afternoon, go to
[77]www.gao.gov and select "E-mail Updates."

Order by Mail or Phone

The first copy of each printed report is free. Additional copies are $2
each. A check or money order should be made out to the Superintendent of
Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more
copies mailed to a single address are discounted 25 percent. Orders should
be sent to:

U.S. Government Accountability Office 441 G Street NW, Room LM Washington,
DC 20548

To order by Phone: Voice: (202) 512-6000
TDD: (202) 512-2537
Fax: (202)
512-6061

To Report Fraud, Waste, and Abuse in Federal Programs

Contact:

Web site: [78]www.gao.gov/fraudnet/fraudnet.htm
E-mail: [79][email protected]
Automated answering system: (800) 424-5454 or (202) 512-7470

Congressional Relations

Gloria Jarmon, Managing Director, [80][email protected] , (202) 512-4400
U.S. Government Accountability Office, 441 G Street NW, Room 7125
Washington, DC 20548

Public Affairs

Susan Becker, Acting Manager, [81][email protected] , (202) 512-4800 U.S.
Government Accountability Office, 441 G Street NW, Room 7149 Washington,
DC 20548

To view the full product, including the scope
and methodology, click on [82]GAO-08-105 .

For more information, contact Barbara Bovbjerg at (202) 512-7215 or
[email protected].

Highlights of [83]GAO-08-105 , a report to Ranking Member, Special
Committee on Aging, U.S. Senate

October 2007

RETIREMENT SECURITY

Women Face Challenges in Ensuring Financial Security in Retirement

Women aged 65 and over will account for a growing segment of the U.S.
population over the next several decades. Despite increases in women's
workforce behavior in the past 65 years, elderly women have persistently
high rates of poverty. Thus, it is important to understand the differences
between men's and women's retirement income, and how women may fare given
future reforms to Social Security and pensions. GAO was asked to examine
(1) how women's retirement income compares with men's and the reasons for
differences; (2) how certain life events such as divorce, widowhood, and
workforce interruptions affect women's retirement income; and (3) the
possible effect on women's retirement income of certain changes to Social
Security and pensions that seek to mitigate the effects of differences in
workforce participation patterns.

To address these objectives, GAO reviewed the relevant literature,
interviewed academics and other retirement experts, and used a
microsimulation model to project future retirement income. GAO provided a
draft of this report to the departments of Labor and Treasury, the
Internal Revenue Service, and the Social Security Administration.
Cognizant agency officials provided technical comments which were
incorporated as appropriate.

GAO is making no recommendations.

In general, women have less retirement income than men, largely because
ofwomen's lower labor force attachment and lower earnings, on average.
Fewer women than men have income from most major retirement sources, and
women have less income from these sources. Women's median Social Security
income is 70 percent of men's. Also, fewer women than men have pensions.
Among the population age 65 and over who continue to work, women earn just
over half of what men earn. Women also have somewhat smaller income than
men from assets, such as interest and dividends. Accordingly, rates of
poverty among those 65 and over are substantially higher for women than
for men. Although their participation has increased substantially in the
last century, women still spend fewer years in the labor force than men,
and they more often work part-time. Also, women tend to earn less than
men, despite increases in their wages over time relative to men. Although
work patterns are key in earnings differences, in prior work, we found
that even after accounting for behavioral differences such as education or
labor force participation, women still earn less than men.

Certain life events--including changes in marital status, labor force
interruptions, and long-term care needs--can significantly reduce the
amount of pension income and Social Security benefits women receive--and
leave women with fewer financial resources at retirement than men. Social
Security divorced spousal benefits are available only if the marriage
lasted at least 10 years. Furthermore, pension benefits are available to a
divorced spouse only under certain circumstances. Women's role as primary
family caregiver for children and elderly relatives can reduce their
career earnings, on which retirement income is based. Because women tend
to live longer than men, widowhood and costly long-term care assistance
may further reduce their retirement resources.

GAO's simulations of some Social Security changes that would compensate
for low earnings or time out of the workforce showed that those changes
tend to increase benefits for beneficiaries overall, and particularly
those in lower income quintiles. Alternatively, changes that focus on
shifts in family structure, such as increases in two-earner couples and
increased incidence of divorce, tend to increase the benefits of groups
targeted by the change, but produce mixed results for others. Some pension
changes that have been proposed in the past several years take into
account the changing labor force and norms of employer-provided retirement
plans; while these changes are gender-neutral, they may provide important
new opportunities for women to increase their retirement income. For
example, decreased vesting requirements may provide additional pension
income to those with intermittent workforce participation who would not
qualify for pension benefits under a longer vesting schedule.

References

Visible links
  57. http://www.gao.gov/cgi-bin/getrpt?GAO-02-62
  58. http://www.gao.gov/cgi-bin/getrpt?GAO-04-35
  59. http://www.gao.gov/cgi-bin/getrpt?GAO-04-35
  60. http://www.cdc.gov/nchs/about/major/nhhcsd/nhhcshomecare3.htm
  61. http://www.gao.gov/cgi-bin/getrpt?GAO-02-745SP
  62. http://www.gao.gov/cgi-bin/getrpt?GAO-02-745SP
  63. http://www.gao.gov/cgi-bin/getrpt?GAO-02-745SP
  64. http://www.gao.gov/cgi-bin/getrpt?GAO/T-HEHS-99-94
  65. http://www.gao.gov/cgi-bin/getrpt?GAO/AIMD/HEHS-00-29
  66. http://www.gao.gov/cgi-bin/getrpt?GAO/HEHS-99-110
  67. http://www.gao.gov/cgi-bin/getrpt?GAO-02-62
  68. http://www.gao.gov/cgi-bin/getrpt?GAO-02-62
  69. http://www.gao.gov/cgi-bin/getrpt?GAO-02-62
  70. http://www.gao.gov/cgi-bin/getrpt?GAO-03-310
  71. http://www.gao.gov/cgi-bin/getrpt?GAO-03-907
  72. http://www.gao.gov/cgi-bin/getrpt?GAO-03-387
  73. http://www.gao.gov/cgi-bin/getrpt?GAO-04-747
  74. http://www.gao.gov/cgi-bin/getrpt?GAO-06-804
  75. http://www.gao.gov/cgi-bin/getrpt?GAO-03-907
  76. http://www.gao.gov/
  77. http://www.gao.gov/
  78. http://www.gao.gov/fraudnet/fraudnet.htm
  79. mailto:[email protected]
  80. mailto:[email protected]
  81. mailto:[email protected]
  82. http://www.gao.gov/cgi-bin/getrpt?GAO-08-105
  83. http://www.gao.gov/cgi-bin/getrpt?GAO-08-105
*** End of document. ***