This is the accessible text file for GAO report number GAO-08-8 
entitled 'Private Pensions: Low Defined Contribution Plan Savings May 
Pose Challenges to Retirement Security, Especially for Many Low-Income 
Workers' which was released on December 11, 2007. 

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

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

Report to the Chairman, Committee on Education and Labor, House of 
Representatives: 

United States Government Accountability Office: 

GAO: 

November 2007: 

Private Pensions: 

Low Defined Contribution Plan Savings May Pose Challenges to Retirement 
Security, Especially for Low-Income Workers: 

Private Pensions: 

GAO-08-8: 

GAO Highlights: 

Highlights of GAO-08-8, a report to the Chairman, Committee on 
Education and Labor, House of Representatives. 

Why GAO Did This Study: 

Over the last 25 years, pension coverage has shifted primarily from 
“traditional” defined benefit (DB) plans, in which workers accrue 
benefits based on years of service and earnings, toward defined 
contribution (DC) plans, in which participants accumulate retirement 
balances in individual accounts. DC plans provide greater portability 
of benefits, but shift the responsibility of saving for retirement from 
employers to employees. This report addresses the following issues: (1) 
What percentage of workers participate in DC plans, and how much have 
they saved in them? (2) How much are workers likely to have saved in DC 
plans over their careers and to what degree do key individual decisions 
and plan features affect plan saving? (3) What options have been 
recently proposed to increase DC plan coverage, participation, and 
savings? GAO analyzed data from the Federal Reserve Board’s 2004 Survey 
of Consumer Finances (SCF), the latest available, utilized a computer 
simulation model to project DC plan balances at retirement, reviewed 
academic studies, and interviewed experts. 

What GAO Found: 

GAO’s analysis of 2004 SCF data found that only 36 percent of workers 
participated in a current DC plan. For all workers with a current or 
former DC plan, including rolled-over retirement funds, the total 
median account balance was $22,800. Among workers aged 55 to 64, the 
median account balance were $50,000, and those aged 60 to 64 had 
$60,600 (see figure below). Low-income workers had less opportunity to 
participate in DC plans than the average worker, and when offered an 
opportunity to participate in a plan, they were less likely to do so. 
Modest balances might be expected, given the relatively recent 
prominence of 401(k) plans. 

Projections of DC plan savings over a career for workers born in 1990 
indicate that DC plans could on average replace about 22 percent of 
annualized career earnings at retirement for all workers, but projected 
“replacement rates” vary widely across income groups and with changes 
in assumptions. Projections show almost 37 percent of workers reaching 
retirement with zero plan savings. Projections also show that workers 
in the lowest income quartile have projected replacement rates of 10.3 
percent on average, with 63 percent of these workers having no plan 
savings at retirement, while highest-income workers have average 
replacement rates of 34 percent. Assuming that workers offered a plan 
always participate raises projected overall savings and reduces the 
number of workers with zero savings substantially, particularly among 
lower-income workers. 

Recent regulatory and legislative changes and proposals could have 
positive effects on DC plan coverage, participation, and savings, some 
by facilitating the adoption of automatic enrollment and escalation 
features. Some options focus on encouraging plan sponsorship, while 
others would create accounts for people not covered by an employer 
plan. Our findings indicate that DC plans can provide a meaningful 
contribution to retirement security for some workers but may not ensure 
the retirement security of lower-income workers. 

Figure: Total DC Balances for Workers with a Current or Former DC Plan, 
Including Rolled-Over Retirement Funds, by Age Group, 2004 SCF: 

This figure is a bar chart showing total DC balances for workers with a 
current or former DC plan, including rolled-over retirement funds, by 
age group, 2004 SCF. The X axis represents the age group, and the Y 
axis represents the total DC balances including rollovers in thousands 
of dollars. 

Age group 18-29: Mean: 11.9; 
Age group 18-29: Median: 6.2. 

Age group 30-39: Mean 34.9; 
Age group 30-39: Median: 14.6. 

Age group 40-49: Mean: 64.1; 
Age group 40-49: Median: 31.6. 

Age group 50-59: Mean: 126.3; 
Age group 50-59: Median: 43.2. 

Age group 60-64: Mean: 135.3; 
Age group 60-64: Median: 60.5. 

Source: GAO analysis of 2004 Survey of Consumer Finances. 

[End of figure] 

What GAO Recommends: 

GAO is not making any recommendations. 

To view the full product, including the scope and methodology, click on 
[hyperlink, http://www.GAO-08-8]. For more information, contact Barbara 
Bovbjerg at (202) 512-7215 or bovbjergb@gao.gov. 

[End of section] 

[See PDF for image] 

[End of figure] 

Contents: 

Letter: 

Results in Brief : 

Background 1: 

Many Workers Have No Plan Coverage, and Most DC Plan Participants 
Currently Have Modest Account Balances 1: 

Projected DC Plan Savings Are Small for Most Workers, but Could Vary 
Widely Depending on Several Factors 1: 

Recent Changes and Proposals Could Have Positive Effects on DC Plan 
Coverage, Participation, and Savings 1: 

Concluding Observations 1: 

Agency Comments 1: 

Appendix I: Scope and Methodology 84: 

Methodology and Assumptions Using Survey of Consumer Finances Data 1: 

Methodology and Assumptions Using PENSIM Microsimulation Model 1: 

Appendix II Comparison of DC Plan Projections Based on PENSIM to Other 
Studies 97: 

Appendix III GAO Contact and Staff Acknowledgments 104: 

Tables: 

Table 1: Key Characteristics of Defined Contribution and Defined 
Benefit Plans 1: 

Table 2: Estimated Social Security Replacement Rates for Workers 
Turning 65 in 2007 and in 2055, Percent of Career-Average Earnings 1: 

Table 3: Projected Average Annuity Equivalents and Replacement Rates 
from DC Plan Balances at Retirement, by Income, under Baseline 
Assumptions 1: 

Table 4: Projected Average Annuity Equivalents and Replacement Rates 
from DC Plan Balances at Retirement, by Income, under Different Model 
Assumptions 1: 

Table 5: Projected Average Annuity Equivalents and Replacement Rates 
from DC Plan Balances at Retirement, by Income, Under Different Model 
Assumptions 1: 

Table 6: Sampling Errors Greater than 4 Percentage Points for 
Percentage Estimates at the 95 Percent Confidence Interval 1: 

Table 7: Sampling Errors Greater Than 25 Percent for Numerical 
Estimates at the 95 Percent Confidence Interval 1: 

Table 8: Summary Statistics, PENSIM 1990 Cohort 1: 

Table 9: Sample Summary Statistics, PENSIM 1990 Cohort, Medians 1: 

Table 10: Cross-Sectional Pension Characteristics of Sample 1: 

Table 11: Retirement Savings and Income Replacement Rates for Unmarried 
Householders, Annual Total Contributions Equal to 8 Percent of 
Household Earnings and 50th Percentile of Returns (2004 Dollars) 1: 

Table 12: Retirement Savings and Income Replacement Rates for Unmarried 
Householders, Annual Total Contributions Equal to 6 Percent of 
Household Earnings and 50th Percentile of Returns (2004 dollars) 1: 

Table 13: Median Replacement Rates from DC Plan Balances for Workers 
Turning 65 between 2030 and 2039, by Income Quartile 1: 

Table 14: Mean Projected DC Plan Assets for Cohorts Retiring in 2000, 
2010, 2020, 2030, and 2040, by Lifetime Earnings Deciles (in 2000 
dollars) 1: 

Table 15: Mean Projected DC Plan Assets for Cohorts Retiring in 2000, 
2010, 2020, 2030, and 2040, Assuming Rate of Return on Equities is 300 
Basis Points Less than Historic Rate, by Lifetime Earnings Deciles (in 
2000 dollars) 1: 

Table 16: Replacement Rates by Income 1: 

Figures: 

Figure 1: Mechanics of Accumulating Retirement Savings in DC Plans 1: 

Figure 2: Percentage of Working Individuals Participating in Current DC 
Plans by Age Group, 2004 1: 

Figure 3: Total DC Balances for Working Individuals with a Current or 
Former DC Plan, by Age Group, 2004 1: 

Figure 4: Portion of Lump-Sum Distribution Recipients Using the Entire 
Portion of Their Most Recent Distribution, by Use, 1993, and 2003 1: 

Figure 5: Total DC Plan Balances for Working Individuals with a Current 
or Former DC Plan, by Household Wealth Quartiles, 2004 1: 

Abbreviations: 

AWI: average wage index: 

BLS: Bureau of Labor Statistics: 

CODA: cash or deferred arrangement: 

CPS: Current Population Survey: 

CRS: Congressional Research Service: 

DB: defined benefit: 

DC: defined contribution: 

EBRI: Employee Benefit Research Institute: 

EBSA: Employee Benefits Security Administration: 

ERISA: Employee Retirement and Income Security Act of 1974: 

ESOP: employee stock ownership plan: 

HRS: Health and Retirement Study: 

ICI: Investment Company Institute: 

IRA: individual retirement account: 

MINT3: Modeling Income in the Near Term: 

OCACT: Office of the Chief Actuary: 

PPA: Pension Protection Act of 2006: 

PSG: Policy Simulation Group: 

PSID: Panel Study of Income Dynamics: 

SCF: Survey of Consumer Finances: 

SEP: Simplified Employee Pension Plan: 

SIMPLE: Savings Incentive Match Plan for Employees of Small Employers: 

SIPP: Survey of Income and Program Participation: 

SSA: Social Security Administration: 

UA: universal account: 

United States Government Accountability Office: 

Washington, DC 20548: 

November 29, 2007: 

The Honorable George Miller: 
Chairman: 
Committee on Education and Labor: 
House of Representatives: 

Dear Mr. Chairman: 

Employer-sponsored pensions represent an important component of 
retirement income. Since the early 1980s, while the percentage of 
workers participating in a pension plan has remained around 50 percent 
of the private sector work force, pension coverage has seen a 
noticeable shift away from "traditional" defined benefit (DB) plans, in 
which workers typically accrue benefits based on years of service and 
earnings, toward defined contribution (DC) plans, in which participants 
accumulate balances in personal accounts. DC plans provide participants 
tax-preferred savings vehicles, portability, and the transparency of 
known account balances. However, they shift the responsibility of 
saving for retirement, and certain key risks, from employers to 
employees. Under such plans, workers often manage the investment of 
plan assets throughout their lives. However, workers may receive 
limited or no contributions from their employers, spend accumulated 
savings prior to retirement, or choose not to participate in a pension 
plan at all, ultimately arriving at retirement with insufficient 
savings to support themselves. Further, retirees usually must manage 
their DC plan savings to make them last throughout retirement. Possible 
reforms to Social Security to address that program's long-term solvency 
could reduce benefits for future retirees, possibly increasing the 
future role of DC plans, as well as other personal savings, in 
providing retirement income.[Footnote 1] 

Based on these concerns, this report addresses the following questions: 
(1) What percentage of workers participate in DC plans, and how much 
have they saved in them? (2) How much are workers likely to have saved 
in DC plans by the time they reach retirement and to what degree do 
certain key individual decisions and plan features affect plan saving? 
(3) What options have been recently proposed to increase DC plan 
coverage, participation, and savings? 

To analyze participation and savings in DC plans, we examined data from 
the 2004 version of the Federal Reserve Board's Survey of Consumer 
Finances (SCF), the latest available SCF, and data published by other 
government agencies, by private organizations, and from academic 
studies. [Footnote 2] To analyze how much Americans can expect to save 
in DC plans by the time they retire and the factors that affect these 
savings, we utilized a microsimulation model, PENSIM, that simulates 
demographic, working, and pension patterns for a constructed sample 
over their lives. (See appendix I for further details of our 
projections using PENSIM.) We also reviewed similar studies to compare 
their methodologies and conclusions (see app. II). To analyze policy 
options to increase DC plan coverage, participation, and savings, we 
synthesized information gathered from interviews of plan practitioners, 
financial managers, and public policy experts, as well as from academic 
and policy studies on DC plan participation and savings. We also 
researched current government initiatives and policy proposals to 
broaden participation in account-based pension plans and increase 
retirement savings. We conducted our work from July 2006 to October 
2007 in accordance with generally accepted government auditing 
standards. 

Results in Brief: 

Regardless of the age of the individual, and at most income levels, DC 
account participation is low, and the account balances of workers 
participating in DC plans are modest. According to our analysis of the 
2004 SCF, shows that only 36 percent of workers were participating in a 
DC plan with their current employer. For all workers with a current or 
former DC plan, including funds rolled over into a new plan or an IRA, 
the median account balances measured $22,800. Among workers aged 55 to 
64 with a current or former DC plan, including rolled over retirement 
funds, the median account balance was $50,000, which if converted into 
an annuity at age 65 would represent about $4,400 per year for life. 
While we might expect older workers to have limited balances in DC 
plans because relatively few employers started offering such plans 
until the 1980s, thus not giving older workers the opportunity to save 
in them for their whole careers, it is notable that DC plan savings 
will be only a limited component of retirement income for this group. 
Leakage, or the cashing out of lump-sum distributions for non- 
retirement purposes, could adversely affect account accumulation for 
some plan participants. From our analysis of the 2004 SCF, of the 21 
percent of households reporting that they had previously received lump- 
sum distributions from previous jobs' retirement plans, about 47 
percent cashed out all the funds, 4 percent cashed out some of the 
funds, and 50 percent rolled over all the funds into another retirement 
account. Low-income workers had the opportunity to participate in DC 
plans less frequently than the average worker, and when they were 
offered a plan, they were less likely to do so. As a result, only 8 
percent of workers in the lowest income quartile participated in DC 
plans with their current employer. 

Simulations of future workers' DC plan savings over an entire working 
career indicate that DC plans could replace, on average, about 22 
percent of annualized career earnings at retirement, but with projected 
replacement rates varying widely across income groups and with changes 
in certain assumptions. These projections show that individuals in this 
cohort would accumulate enough DC plan savings over their careers to 
produce average annuitized retirement income of $18,784 (in 2007 
dollars) per year, but also that about 37 percent of the sample 
population would have zero savings from DC plans when they retire. 
Workers in the lowest income quartile have projected replacement rates 
of 10.3 percent on average, but 63 percent of these workers are 
projected to have no DC savings at retirement. Highest-income workers, 
in contrast, have average projected replacement rates of almost 34 
percent from DC plans. Workers who are eligible to participate in a 
plan for at least 15 years have an average projected replacement rate 
of 33.5 percent, but about 16 percent of these workers still have no 
projected savings at retirement. Some changes in assumptions in our 
projections indicate that certain key individual behaviors or plan 
features have a significant impact on plan savings, especially for low- 
income workers, and on the number of workers with zero savings at 
retirement. Assuming that all workers who are offered a plan always 
participate in it would raise average retirement income from DC plans 
by about 40 percent, with particularly high increases to lowest-income 
workers and a reduction in the number of workers who do not save at 
all. Assuming that workers do not withdraw money from their accounts 
while they are working--that is, no leakage occurs--raises overall 
average annuity income from DC plans by about 11 percent and reduces 
the percentage of those with no DC savings at retirement by over 25 
percent. Other scenarios, such as working longer or raising 
contribution limits, raised savings primarily for higher-income 
workers, and reduced the number of workers with zero savings at 
retirement only slightly. 

Recent regulatory and legislative changes and proposals could have 
positive effects on DC plan coverage, participation, and savings. Some 
of these have facilitated plan sponsors' adoption of automatic 
enrollment and automatic escalation of contributions, which some 
studies indicate may increase DC participation and savings among 
workers who already have access to a plan. Other proposals focus on 
encouraging more employers to sponsor plans in order to increase plan 
participation and savings. In one, the "State-K" proposal, states would 
collaborate with private financial institutions to offer employers the 
option of adopting a state-designed low-cost plan. Broader options, 
such as the automatic individual retirement account (IRA) or universal 
accounts proposals, would seek to extend retirement account coverage by 
facilitating savings in IRAs or creating retirement savings vehicles 
for people not covered by a voluntary employer based retirement plan. 
Another would expand the saver's credit by making it refundable to 
workers who pay little or no federal income tax. It is important to 
note that Social Security benefits provide the bulk of retirement 
benefits for most households; evaluations of income security should 
consider total retirement income from all sources, not just DC plans. 

Background: 

Employer-sponsored pensions fall into two major categories: defined 
benefit (DB) and defined contribution (DC) plans. In DB, or 
traditional, plans, benefits are typically set by formula, with workers 
receiving benefits upon retirement based on the number of years worked 
for a firm and earnings in years prior to retirement.[Footnote 3] In DC 
plans, workers accumulate savings through contributions to an 
individual account. These accounts are tax-advantaged in that 
contributions are typically excluded from current income, and earnings 
on balances grow tax-deferred until they are withdrawn.[Footnote 4] An 
employer may also make contributions, either by matching employee's 
contributions up to plan or legal limits, or on a non-contingent basis. 

Like DB plans, DC plans operate in a voluntary system with tax 
incentives for employers to offer a plan and for employees to 
participate. Contributions to and earnings on DC plan accounts are not 
taxed until the participant withdraws the money, although participants 
making withdrawals prior to age 59 ˝ may incur an additional 10 percent 
tax.[Footnote 5] In 2006, the pension tax expenditure for DC plans 
amounted to $54 billion.[Footnote 6] In addition, a nonrefundable tax 
credit to qualifying low-and middle-income workers who make 
contributions, the saver's credit, accounted for less than 2 percent of 
the 2006 tax expenditure on account-based retirement plans.[Footnote 7] 

DC plans offer workers more control over their retirement asset 
management, but also shift some of the responsibility and certain risks 
onto workers. Workers generally must elect to participate in a plan and 
make regular contributions into their plans over their careers. 
Participants typically choose how to invest plan assets from a range of 
options provided under their plan, and accordingly face investment 
risk.[Footnote 8] Savings in DC plans are portable in the sense that a 
participant may keep plan balances in a tax-protected account upon 
leaving a job, either by rolling over plan balances into a new plan or 
an IRA, or in some cases leaving money in an old plan.[Footnote 9] 
Workers may have access to plan savings prior to retirement, either 
through loans or withdrawals; participants may find such features 
desirable, but pre-retirement access may also lead to lower retirement 
savings (sometimes referred to as leakage) and possible tax penalties. 
Workers who receive DC distributions in lump-sum form must manage 
account withdrawals such that their savings last throughout retirement. 
In contrast, a formula, often based on preretirement average pay and 
years of service, determines DB plan benefits, and workers are usually 
automatically enrolled in a plan. The employer has the responsibility 
to ensure that the plan has sufficient funding to pay promised 
benefits, although the sponsor can choose to terminate the 
plan.[Footnote 10] DB plans also typically offer the option to take 
benefits as a lifetime annuity, or periodic benefits until death. An 
annuity provides longevity insurance against outliving one's savings, 
but may lose purchasing power if benefits do not rise with inflation. 
Table 1 summarizes some of the primary differences between DC and DB 
plans. 

Table 1: Key Characteristics of Defined Contribution and Defined 
Benefit Plans: 

What determines the level of benefits?; 
Defined contribution plans: Contributions into a personal account and 
the return on assets; 
Defined benefit plans: A formula, typically based on years of service 
and salary history. 

What does the employee have to do to participate and earn benefits in 
the plan?; 
Defined contribution plans: May require waiting for eligibility and 
sign-up by employee. Participants may need to work up to 6 years to 
fully vest in employer matching contributions; 
Defined benefit plans: Eligibility and participation are typically 
automatic. Workers working at least 1,000 hours per year earn years of 
service toward benefits. Participants may need to work for up to 7 
years to fully vest in plan benefits. 

How are contributions made?; 
Defined contribution plans: Typically, employee decides how much to 
contribute from current wages; 
employer may also contribute; 
Defined benefit plans: Typically by employer only, except in some 
public sector plans. 

Who manages the assets and assumes the risks of investing it?; 
Defined contribution plans: Employee, in most plans; 
Defined benefit plans: Plan sponsor; 
benefits are government-insured up to certain limits. 

What happens to the benefits when the employee leaves the job?; 
Defined contribution plans: Can be left in plan, rolled over to an IRA, 
or cashed out (often with a penalty if done before age 59 ˝); 
Defined benefit plans: Sometimes unavailable until beneficiary reaches 
specified retirement age. 

How are benefits taken in retirement and what are the major risks they 
pose?; 
Defined contribution plans: Typically by withdrawing from total 
balances, and must be managed to last throughout retirement; 
Defined benefit plans: Typically payable as life annuities, but plan 
may offer lump sum option. Annuities lose purchasing power over time if 
not indexed to inflation. 

Source: GAO analysis. 

[End of table] 

Over the past 25 years, DC plans have become the dominant type of 
private sector employee pension. In 1980, private DB plans had 38 
million participants, while DC plans had 20 million. As of 2004, 64.6 
million participants had DC plans, while 41.7 million had DB plans. 
Further, over 80 percent of private sector DC participants in 2004 were 
active participants (in a plan with their current employer), while 
about half of DB participants had separated from their sponsoring 
employer or retired. According to the Employee Benefit Research 
Institute (EBRI), while overall pension coverage among families 
remained around 40 percent between 1992 and 2001, 38 percent of 
families with a pension relied exclusively on a DC plan for retirement 
coverage in 1992, while 62 percent had a DB plan. In 2001, 58 percent 
of pension-participating families had only a DC plan, while 42 percent 
had a DB plan.[Footnote 11] Assets in all DB plans exceeded total DC 
assets as recently as 1995. As of 2006, DC plans had almost $3.3 
trillion in assets while DB plans had almost $2.3 trillion. In 
addition, assets in IRAs, accounts that are also tax protected and 
include assets from rolled-over balances from employer-sponsored plans, 
measured over $4.2 trillion in 2006. 

There are several different categories of DC plans. Most of these plans 
are types of cash or deferred arrangements (CODA), in which employees 
can direct pre-tax dollars, along with any employer contributions, into 
an account, with assets growing tax deferred until withdrawal. The 
401(k) plan is the most common, covering over 85 percent of active DC 
participants. Certain types of tax-exempt employers may offer plans, 
such as 403(b) or 457 plans, which have many features similar to 401(k) 
plans. Many employers match employee contributions, generally based on 
a specified percentage of the employee's salary and the rate at which 
the participant contributes.[Footnote 12] Small business owners may 
offer employees a Savings Incentive Match Plan for Employees of Small 
Employers (SIMPLE) or a Simplified Employee Pension Plan (SEP), two 
types of DC plans that have reduced regulatory requirements for 
sponsors. Other types of DC plans keep the basic individual account 
structure of the 401(k), but with different requirements and employer 
practices. Some are designed primarily for employer contributions. 
These include money purchase plans, which specify fixed annual employer 
contributions; profit sharing plans, in which the employer decides 
annual contributions, perhaps based on profits, into the plan, and 
allocations of these to each participant; and employee stock ownership 
plans (ESOPs), in which contributions are primarily invested in company 
stock.[Footnote 13] 

Building up retirement savings in DC plans rests on factors that are, 
to some degree, outside of the control of the individual worker, as 
well as behaviors an individual does control (see fig. 1). Factors 
outside the individual's direct control include the following: 

* Plan sponsorship--the employer's decision to sponsor a plan, as well 
as participation eligibility rules. 

* Employer contributions--whether the sponsor makes matching or 
noncontingent contributions. 

* Investment options--the plan sponsor's decisions about investment 
options to offer to participants under the plan. 

* Market returns on plan assets--market performance of plan assets. 

Key individual decisions and behaviors that may affect retirement 
savings include the following: 

* Employee contributions--deposits into the plan account, typically out 
of current wages. 

* Investment decisions--how to invest plan assets given investment 
options offered under the plan. 

* Withdrawals/loans--pre-retirement withdrawals from account balances, 
which usually incur a tax penalty. Similarly, taking out a loan from a 
plan, if allowed, may reduce future balances if the loan is not repaid 
in full and treated as a withdrawal, or by lowering investment returns. 

* Rollover--upon separation from a job, a participant may transfer the 
plan account balance to an IRA, which maintains most of the same tax 
preferences on the balances, move it to a new tax-qualified plan, or 
leave the money in the old plan. Alternatively, any cash withdrawal 
would likely be subject to income tax and penalties. 

* Age at retirement--the decision as to when to retire determines how 
many years the worker has to accumulate plan balances and how long the 
money has to last in retirement. 

Figure 1: Mechanics of Accumulating Retirement Savings in DC Plans: 

This figure is a chart showing mechanics of accumulating retirement 
savings in DC plans. 

[See PDF for image] 

Source: GAO analysis. 

[End of figure] 

There is little consensus about how much constitutes "enough" savings 
to have going into retirement. We may define retirement income adequacy 
relative to a standard of minimum needs, such as the poverty rate, or 
to the consumption spending that households experienced during working 
years.[Footnote 14] Some economists and financial advisors consider 
retirement income adequate if the ratio of retirement income to pre- 
retirement income--or replacement rate--is between 65 and 85 percent. 
Retirees may not need 100 percent of pre-retirement income to maintain 
living standards for several reasons. Retirees will no longer need to 
save for retirement, retirees' payroll and income tax liability will 
likely fall, work expenses will no longer be required, and mortgages 
and children's education and other costs may have been paid 
off.[Footnote 15] However, some researchers cite uncertainties about 
future health care costs and future Social Security benefit levels as 
reasons to suggest that a higher replacement rate, perhaps above 100 
percent or higher, would be considered adequate.[Footnote 16] 

To achieve adequate replacement rate levels, retirees depend on 
different sources of income to support themselves in retirement. Social 
Security benefits provide the bulk of retirement benefits for most 
households. As of 2004, annuitized pension benefits provided almost 20 
percent of total income to households with someone age 65 or older, 
while Social Security benefits provided 39 percent.[Footnote 17] Social 
Security benefits compose over 50 percent of total income for two- 
thirds of households with someone age 65 or older, and at least 90 
percent of income for one-third of such households. Table 2 shows 
estimated replacement rates from Social Security benefits for low and 
high earners retiring in 2007 and 2055, as well as the remaining amount 
of pre-retirement income necessary to achieve a 75 percent replacement 
rate.[Footnote 18] These figures give rough guidelines for how much 
retirement income workers might need from other sources, such as 
employer-sponsored pensions, as well as earnings and income from other 
savings or assets. 

Table 2: Estimated Social Security Replacement Rates for Workers 
Turning 65 in 2007 and in 2055, Percent of Career-Average Earnings: 

Source of replacement rate income: Social Security; 
Year in which a 65 year old retires: 2007: Low earner: 54.2; 
Year in which a 65 year old retires: 2007: High earner: 33.5; 
Year in which a 65 year old retires: 2055: Low earner: 49.0; 
Year in which a 65 year old retires: 2055: High earner: 30.1. 

Source of replacement rate income: Replacement from other sources to 
achieve 75 percent replacement rate; 
Year in which a 65 year old retires: 2007: Low earner: 20.8; 
Year in which a 65 year old retires: 2007: High earner: 41.5; 
Year in which a 65 year old retires: 2055: Low earner: 26.0; 
Year in which a 65 year old retires: 2055: High earner: 44.9. 

Source: The 2007 Annual Report of the Board of Trustees of the Federal 
Old-Age and Survivors Insurance and Federal Disability Insurance Trust 
Funds, Table VI.F10. 

Note: Based on scheduled benefits under intermediate assumptions of 
Social Security projections. Replacement rates represent benefits as a 
percentage of career-average earnings for low and high earners. 

[End of table] 

It is important to keep certain economic principles in mind when 
evaluating the effectiveness of retirement accounts, or any pensions, 
in providing retirement income security. First, balances accumulated in 
a DC plan may not represent new saving; individuals may have saved in 
another type of account in the absence of a DC plan or its tax 
preferences. Second, evaluating worker income security should consider 
total compensation, not just employer contributions to DC plans. All 
else equal, we should generally expect more generous employer-sponsored 
pension benefits to lower cash wages and that the split between current 
wages and deferred compensation is largely a reflection of labor market 
conditions, tax provisions, and worker and employer preferences. 

Many Workers Have No Plan Coverage, and Most DC Plan Participants 
Currently Have Modest Account Balances: 

Many workers do not have DC plans, and median savings levels among 
participants show modest balances. While it is worth noting that for 
workers nearing retirement age, DC plans were not considered primary 
pension plans for a significant portion of their working careers, 
participation rates and median balances in such plans are low across 
all ages. Only 36 percent of working individuals were actively 
participating in a DC plan, according to data from the 2004 
SCF.[Footnote 19] Further, workers aged 55 to 64 had median balances 
totaling $50,000 in account-based retirement savings vehicles, 
including DC plans and rollover accounts.[Footnote 20] Leakage, when 
workers withdraw DC savings before retirement age, can also reduce 
balances; almost half of those taking lump-sum distributions upon 
leaving a job reported cashing out their balances for non-retirement 
purposes. Participation among lower-income workers was particularly 
limited, and those who did have accounts had very low balances. 

Most Workers Do Not Have a Current DC Plan, and Participants Have 
Modest Plan Balances: 

The majority of workers, in all age groups, are not participating in DC 
plans with their current employers. Employers do not always offer 
retirement plans, and when they do, plans may have eligibility 
restrictions initially, and some eligible workers do not choose to 
participate.[Footnote 21] According to our analysis of the 2004 SCF, 
only 62 percent of workers were offered a retirement plan by their 
employer, and 84 percent of those offered a retirement plan 
participated.[Footnote 22] Only 36 percent of working individuals 
participated in a DC plan with their current employer (see fig. 2). 
Data indicated similar participation rates for working households, as 
42 percent of households had at least one member with a current DC 
plan. 

Figure 2: Percentage of Working Individuals Participating in Current DC 
Plans by Age Group, 2004: 

This figure is a bar chart showing percentage of working individuals 
participating in current DC plans by age group, 2004. The X axis 
represents the age group, and the Y axis represents the percent. 

Age group: 18-29: 24. 

Age group: 30-39: 36. 

Age group: 40-49: 41. 

Age group: 50-59: 43. 

Age group: 60-64: 45. 

[See PDF for image] 

Source: GAO analysis of 2004 Survey of Consumer Finances. 

Note: Most SCF percentage estimates have 95 percent confidence 
intervals between plus-or-minus 4 percentage points of the percentage 
itself. Exceptions to this rule are presented in app. I. 

[End of figure] 

For many workers who participated in a plan, overall balances in DC 
plans were modest, suggesting a potentially small contribution toward 
retirement security for most plan participants and their households. 
However, since DC plans were less common before the 1980s, older 
workers would not have had access to these plans their whole careers. 
In order to approximate lifetime DC balances when discussing mean and 
median DC balances in this report, our analysis of the 2004 SCF 
aggregates the "total balances" of DC plans with a current employer, DC 
plans with former employers that have been left with the former 
employer, and any retirement plans with former employers that have been 
rolled over into a new plan or an IRA.[Footnote 23] Workers with a 
"current or former DC plan" refers to current workers with one or more 
of those three components. For all workers with a current or former DC 
plan, the median total balance was $22,800. For all households with a 
current or former DC plan, the median total balance was $27,940 (see 
fig. 3). 

Figure 3: Total DC Balances for Working Individuals with a Current or 
Former DC Plan, by Age Group, 2004: 

Age group 18-29: Mean: 11.9; 
Age group 18-29: Median: 6.2. 

Age group 30-39: Mean 34.9; 
Age group 30-39: Median: 14.6. 

Age group 40-49: Mean: 64.1; 
Age group 40-49: Median: 31.6. 

Age group 50-59: Mean: 126.3; 
Age group 50-59: Median: 43.2. 

Age group 60-64: Mean: 135.3; 
Age group 60-64: Median: 60.5. 

Source: GAO analysis of 2004 Survey of Consumer Finances. 

Note: The majority of SCF estimates of medians and means have 95 
percent confidence intervals within plus-or-minus 25 percent of the 
estimate itself. Exceptions to this rule are summarized in app. I. 

[End of figure] 

For individuals nearing retirement age, total DC plan balances are 
still low. Given trends in coverage since the 1980's, older workers 
close to retirement age are more likely than younger ones to have 
accrued retirement benefits in a DB plan. However, older workers who 
will rely on DC plans for retirement income may not have time to 
substantially increase their total savings without extending their 
working careers, perhaps for several years. Among all workers aged 55 
to 64 with a current or former DC plan, the median balance according to 
the 2004 SCF was $50,000, which would provide an income of about $4,400 
a year, replacing about 9 percent of income for the average worker in 
this group.[Footnote 24] Among all workers aged 60 to 64 with a current 
or former DC plan, the median balance was $60,600 for their accounts. 
Markedly higher values for mean balances versus median balances in 
figure 3 illustrate that some individuals in every age group are 
successfully saving far more than the typical individual, increasing 
the mean savings. These are primarily individuals at the highest levels 
of income. 

Leakage, or cashing out accumulated retirement savings for non- 
retirement purposes, adversely affects account accumulation for some of 
those with accounts, particularly for lower-income workers with small 
account balances. Participants who withdraw money from a DC plan before 
age 59 ˝ generally pay ordinary income taxes on the 
distributions,[Footnote 25] plus an additional 10 percent tax in most 
circumstances.[Footnote 26] Participants may roll their DC plan 
balances into another tax-preferred account when they leave a job, and 
employers are required, in the absence of participant direction, to 
automatically roll DC account distributions greater than $1,000 but not 
greater than $5,000 into an IRA, or to leave the money in the plan. As 
of 2004, 21 percent of households in which the head of household was 
under 59, had ever received lump-sum distributions from previous jobs' 
retirement plans. Among these households that received lump-sum 
distributions, 47 percent had cashed out all the funds, 4 percent 
cashed out some of the funds, and 50 percent preserved all the funds by 
rolling them over into another retirement account.[Footnote 27] Workers 
were more likely to roll over funds when the balances are greater. 
Among households that had cashed out all retirement plans with former 
employers, the median total value of those funds was $6,800. For 
households that had rolled over all retirement plans with former 
employers, the median total value of rolled-over funds was 
$24,200.[Footnote 28] 

Some evidence suggests that pre-retirement withdrawals may be 
decreasing. One study finds that those receiving lump-sum distributions 
are more likely to preserve funds in tax-qualified accounts than they 
were in the past.[Footnote 29] For example, data show that in 1993, 19 
percent of lump-sum distributions recipients preserved all of their 
savings by rolling them into a tax-qualified account, compared to 43 
percent in 2003. Further, 23 percent used all of their distribution for 
consumption in 1993, declining to 15 percent in 2003 (see fig. 4). 
According to the same study, age and size of the distribution are major 
determinants of whether or not the distribution is preserved in a tax- 
qualified account. For example, the authors found 55.5 percent of 
recipients aged 51 to 60 rolled their entire distribution in a tax- 
qualified account compared with 32.7 percent of recipients 21 to 30. 
Additionally, 19.9 percent of distributions from $1 to $499 were rolled 
over in tax-qualified accounts, as opposed to 68.1 percent of 
distributions of $50,000 or more. 

Figure 4: Portion of Lump-Sum Distribution Recipients Using the Entire 
Portion of Their Most Recent Distribution, by Use, 1993, and 2003: 

This figure is a bar chart showing portion of lump-sum distribution 
recipients using the entire portion of their most recent distribution, 
by use, 1993, and 2003. The X axis represents distribution use, and the 
Y axis represents percentage use. 

Tax qualified financial savings: 1993: 19; 
Tax qualified financial savings: 2003: 43. 

Non tax-qualified financial savings: 1993: 8; 
Non tax-qualified financial savings: 2003: 6. 

Debts, business, and home: 1993: 18; 
Debts, business, and home: 2003: 22. 

Education expenses: 1993: 1; 
Education expenses: 2003: 1. 

Consumption: 1993: 23; 
Consumption: 2003: 15. 

[See PDF for image] 

Source: "Lump Sum Distributions" in Employee Benefit Research Institute 
Notes, Volume 26 Number 12. 

Note: EBRI estimates from 2001 Panel of the Survey of Income and 
Program Participation Topical Module 7, 1996; Panel of the Survey of 
Income and Program Participation Topical Module 7, and April, 1993 
Employee Benefits Supplement to the Current Population Survey. 

[End of figure] 

Additionally, some participants take loans from their DC plan, which 
may reduce plan savings. One survey found that in 2005, 85.2 percent of 
employers surveyed offered a loan option.[Footnote 30] Most eligible 
participants do not take loans, and one analysis finds that at the year 
end 2006, loans amounted to 12 percent of account balances for those 
who had loans.[Footnote 31] Individuals may prefer to take out pension 
loans in lieu of other lines of credit because pension loans require no 
approval and have low or no transaction costs. Borrowers also pay the 
loan principal and interest back to their own accounts. However, 
someone borrowing from a DC plan may still lose money if the interest 
on the loan paid back to the account is less than the account balance 
would have earned if the loan had not been taken. Further, loans not 
paid back in time, or not paid back before the employee leaves the job, 
may be subject to early withdrawal penalties. No data have been 
reported on the rate of loan defaults, but it is expected to be much 
lower where repayments are made by payroll withholding. However, a loan 
feature may also have a positive effect on participation, as some 
workers may choose to participate who otherwise might not, precisely 
because they can borrow from their accounts for non-retirement purposes 
at relatively low interest rates.[Footnote 32] 

Low-Income Workers Have Particularly Low DC Plan Coverage and Plan 
Balances: 

Among workers in the lowest income quartile, only 8 percent 
participated in a current DC plan, a result of markedly lower access as 
well as lower participation than the average worker (see fig. 5). Only 
25 percent of workers in the lowest income quartile were offered any 
type of retirement plan by their employer, and among those offered a 
retirement plan, 60 percent elected to participate, compared with 84 
percent among workers of all income levels. Workers in the lower half 
of the income distribution with either current or former DC plans had 
total median balances of $9,420. 

Figure 5: Total DC Plan Balances for Working Individuals with a Current 
or Former DC Plan, by Household Wealth Quartiles, 2004: 

This figure is a bar chart showing total DC plan balances for working 
individuals with a current or former DC plan, by household wealth 
quartiles, 2004. The X axis represents the household wealth quartile, 
and the Y axis represents the total DC balances in thousands of 
dollars. 

1: Mean: 13.3; 
1: Median: 6.4. 

2: Mean: 25.6; 
2: Median: 9.9. 

3: Mean: 54.8; 
3: Median: 23.6. 

4: Mean: 141.3; 
4: Median: 56.2. 

[See PDF for image] 

Source: GAO analysis of 2004 Survey of Consumer Finances. 

[End of figure] 

Older workers who were less wealthy also had limited retirement 
savings. Workers with a current or former DC plan, aged 50-59 and at or 
below the median level of wealth, had median total savings of only 
$13,800.[Footnote 33] Workers with a current or former DC plan, aged 60-
64 and at or below the median level of wealth, had median total savings 
of $18,000, a level that could provide at best only a limited 
supplement to retirement income. If converted into a single life 
annuity at age 65, this balance would provide only $132 per month-- 
about $1,600 per year.[Footnote 34] 

Notably, workers with low DC balances were actually less likely to have 
a DB pension to fall back on than workers with higher DC balances. 
Among all workers participating in current or former DC plans, only 17 
percent of those in the bottom quartile for total plan savings also 
were covered by a current DB plan. In contrast, 32 percent of those in 
the top quartile for total DC savings also had DB coverage. Among all 
workers with a current or former DC plan, the plan balances for those 
with DB coverage were higher than for those without DB coverage. The 
median DC balance for workers with a DB account was $31,560, while the 
median DC balance for someone without a DB account was $20,820. 

Projected DC Plan Savings Are Small for Most Workers, but Could Vary 
Widely Depending on Several Factors: 

Simulations of projected retirement savings in DC plans suggest that a 
large percentage of workers may accumulate enough over their careers to 
replace only a small fraction of their working income, although results 
vary widely by income levels and depend on model assumptions. Projected 
savings allow us to analyze how much workers might save over a full 
working career under a variety of conditions in a way that analyzing 
current plan balances cannot, since DC plans have become primary 
employer-sponsored plans only relatively recently. Baseline simulations 
of projected retirement savings for a hypothetical 1990 birth cohort 
indicate that DC plan savings would on average replace about 22 percent 
of annualized career earnings,[Footnote 35] but provide no savings to 
almost 37 percent of the working population, perhaps because of 
different factors --working for employers who do not offer a plan, 
choosing not to participate, or withdrawing any accumulated plan 
savings prior to retirement.[Footnote 36] Further, projected DC account 
balances vary widely by income quartile, with workers in the lowest-
income quartile saving enough for about a 10 percent replacement rate, 
while those in the highest quartile saving enough for a 34 percent 
replacement rate, on average. Assuming changes in certain plan 
features, individual behavior, or market assumptions, such as increased 
participation or account rollover rates, increased projected average 
savings and increased the number of workers who had some DC plan 
savings at retirement, especially for low-income workers. Other 
scenarios, such as assuming higher contribution limits or delaying 
retirement, raised average replacement rates, but with more of the 
positive impact on higher-income workers and having little effect on 
reducing the number of workers with no savings at retirement.[Footnote 
37] 

Projected DC Plan Balances Vary Widely by Income, with Many Workers 
Having No Plan Savings at Retirement: 

Our projections, based on a sample of workers born in 1990, show that 
workers would save enough in their DC plans over their careers to 
produce, when converted to a lifetime annuity at the time of 
retirement, an average of $18,784 per year in 2007 dollars (see table 
3).[Footnote 38] The projections assume that all workers fully 
annuitize all accumulated DC plans balances at retirement, which occurs 
sometime between age 62 and 70. Participants are assumed to always 
invest all plan assets in life cycle funds, and stocks earn an average 
real annual return of 6.4 percent. This $18,784 annuity would replace, 
on average, 22.2 percent of annualized career earnings for workers in 
the cohort. Savings and replacement rates vary widely across income 
groups. Almost 37 percent of workers in this cohort have no projected 
DC plan savings at retirement, which brings down overall average 
replacement rates. Workers in the lowest income quartile accumulate DC 
plan savings equivalent to an annuity of about $1,850 per year, or a 
10.3 percent replacement rate, and 63 percent of this group have no 
plan savings by the time they retire. In contrast, highest income 
quartile workers save enough to receive about $50,000 per year in 
annuity income, enough for a 33.8 percent replacement rate. Even in 
this highest-income group, over 16 percent of workers have zero plan 
savings at retirement. In all cases, our replacement rates include 
projected savings only in DC plans. Retirees may also receive benefits 
from DB plans, as well as from Social Security, which typically 
replaces a higher percentage of earnings for lower-income workers. 

Projected household-level plan savings show a higher average 
replacement rate of 33.8 percent, with about 29 percent of households 
having no plan savings at retirement. When we assume that plan assets 
earn a lower average real annual return of 2.9 percent, average 
replacement rates from DC plan savings fall to about 16 percent for the 
sample.[Footnote 39] Under this assumption, workers in the lowest- 
income quartile receive an average 7.1 percent replacement rate from DC 
plans, while highest-income quartile workers receive an average 25 
percent replacement rate. Lower rates of return affect the percentage 
of workers with no accumulated DC plan savings only slightly, perhaps 
because on the margins some participants might choose (or have their 
employers choose) to cash out lower balances. 

Table 3: Projected Average Annuity Equivalents and Replacement Rates 
from DC Plan Balances at Retirement, by Income, under Baseline 
Assumptions: 

Individual-level results: Annuity equivalent (per year, 2007 dollars); 
Overall: 18,784; 
By income quartile: 1: 1,850; 
By income quartile: 2: 6,554; 
By income quartile: 3: 16,635; 
By income quartile: 4: 50,098. 

Individual-level results: Replacement rate (percent); 
Overall: 22.2; 
By income quartile: 1: 10.3; 
By income quartile: 2: 18.2; 
By income quartile: 3: 26.3; 
By income quartile: 4: 33.8. 

Individual-level results: Percent of workers with no DC savings; 
Overall: 36.8; 
By income quartile: 1: 63.0; 
By income quartile: 2: 39.8; 
By income quartile: 3: 27.9; 
By income quartile: 4: 16.4. 

Household-level results: Annuity equivalent (per year, 2007 dollars); 
Overall: 24,664; 
By income quartile: 1: 4,176; 
By income quartile: 2: 11,918; 
By income quartile: 3: 25,560; 
By income quartile: 4: 57,000. 

Household-level results: Replacement rate (percent); 
Overall: 33.8; 
By income quartile: 1: 18.7; 
By income quartile: 2: 30.3; 
By income quartile: 3: 40.9; 
By income quartile: 4: 45.5. 

Household-level results: Percent of workers with no DC savings; 
Overall: 28.8; 
By income quartile: 1: 48.1; 
By income quartile: 2: 30.7; 
By income quartile: 3: 21.8; 
By income quartile: 4: 14.5. 

Only workers eligible for a DC plan for 15+ years: Annuity equivalent 
(per year, 2007 dollars); 
Overall: 29,844; 
By income quartile: 1: 5,133; 
By income quartile: 2: 13,629; 
By income quartile: 3: 30,178; 
By income quartile: 4: 70,437. 

Only workers eligible for a DC plan for 15+ years: Replacement rate 
(percent); 
Overall: 33.5; 
By income quartile: 1: 21.7; 
By income quartile: 2: 30.2; 
By income quartile: 3: 39.7; 
By income quartile: 4: 42.3. 

Only workers eligible for a DC plan for 15+ years: Percent of workers 
with no DC savings; 
Overall: 15.6; 
By income quartile: 1: 32.6; 
By income quartile: 2: 16.6; 
By income quartile: 3: 9.1; 
By income quartile: 4: 4.1. 

Only those working 25+ years full-time: Annuity equivalent (per year, 
2007 dollars); 
Overall: 25,533; 
By income quartile: 1: 4,447; 
By income quartile: 2: 11,407; 
By income quartile: 3: 25,610; 
By income quartile: 4: 60,668. 

Only those working 25+ years full-time: Replacement rate (percent); 
Overall: 26.5; 
By income quartile: 1: 16.3; 
By income quartile: 2: 23.3; 
By income quartile: 3: 31.7; 
By income quartile: 4: 34.9. 

Only those working 25+ years full-time: Percent of workers with no DC 
savings; 
Overall: 28.8; 
By income quartile: 1: 46.7; 
By income quartile: 2: 31.8; 
By income quartile: 3: 22.8; 
By income quartile: 4: 14.5. 

Assuming 2.9 percent real annual return on stocks: Annuity equivalent 
(per year, 2007 dollars); 
Overall: 13,803; 
By income quartile: 1: 1,277; 
By income quartile: 2: 4,687; 
By income quartile: 3: 12,145; 
By income quartile: 4: 37,100. 

Assuming 2.9 percent real annual return on stocks: Replacement rate 
(percent); 
Overall: 16.1; 
By income quartile: 1: 7.1; 
By income quartile: 2: 13.0; 
By income quartile: 3: 19.2; 
By income quartile: 4: 25.1. 

Assuming 2.9 percent real annual return on stocks: Percent of workers 
with no DC savings; 
Overall: 37.2; 
By income quartile: 1: 63.3; 
By income quartile: 2: 40.3; 
By income quartile: 3: 28.3; 
By income quartile: 4: 16.7. 

Source: GAO projections using PENSIM model. 

Note: All results are individual level, except as indicated. Model 
assumptions include the following: 1) workers fully annuitize all 
accumulated DC plan balances at retirement, between age 62 and 70; 2) 
participants invest all plan assets in life cycle funds; 3) stocks earn 
an average annual 6.4 percent real return, except where specified. 
Replacement rates equal annuitized income from lifetime DC plan savings 
divided by annualized career earnings. See app. I for more details. 

[End of table] 

Table 3 also shows savings statistics for sub-samples of the cohort who 
have a better chance of accumulating significant DC plan savings, such 
as those workers who have long-term eligibility to participate in a 
plan or who work for many years. As expected, these groups have higher 
projected savings; replacement rates also show more even distribution 
across income groups, compared to those in the full sample. However, we 
still see a significant portion of the workers with no DC savings at 
retirement. First, we limit the sample only to those workers who are 
eligible to participate in a plan for at least 15 years over their 
careers. Average replacement rates for this group measure 33.5 percent, 
with rates ranging from 21.7 percent for lowest income quartile workers 
to 42.3 percent for the highest quartile.[Footnote 40] Even with such 
long-term eligibility for plan coverage, however, 15.6 percent of these 
workers, and almost one-third of lowest-income workers, have nothing 
saved in DC plans at the time they retire. This could result from 
workers choosing not to participate or from cashing out plan balances 
prior to retirement. 

We also analyze the prospects of workers with long-term attachment to 
the labor market, for which we use people who work full-time for at 
least 25 years, without regard to plan coverage or participation. Among 
these workers, average DC plan savings at retirement account for a 26.5 
percent replacement rate. Still, almost 29 percent of these workers 
have no projected savings. This suggests that while DC plans have the 
potential to provide significant retirement income, saving may be 
difficult for some workers who work for many years, even among those 
whose employers offer a plan. 

Universal Participation in Sponsored Plans and Universal Account 
Rollover Raise Projected DC Plan Savings Substantially for Lower-Income 
Workers: 

Our simulations indicate that increasing participation and reducing 
leakage out of DC plans may have a particularly significant impact on 
overall savings, especially for lower-income workers. Of the changes in 
the model assumptions that we simulated, these had the broadest effect 
on savings because they not only raised average savings for the entire 
sample, but had a relatively strong impact on workers in the lowest 
income quartile and on the number of workers with no DC plan savings at 
retirement. While these assumptions represent stylized scenarios, they 
illustrate the potential effect of such changes on savings. 

We project DC plan savings assuming that all employees of a firm that 
sponsors a DC plan participate immediately, rather than having to wait 
for eligibility or choosing not to participate.[Footnote 41] In our 
baseline projections, 6 percent of workers whose employers sponsor a 
plan are ineligible to participate, and 33 percent of those eligible do 
not choose to participate; therefore, this assumption significantly 
raises plan participation rates among workers. Accordingly, average DC 
savings rise by almost 40 percent, raising average replacement rates to 
35 percent, and the percentage of the population with no savings at 
retirement drops by half, down to 17.7 percent (see table 4). 

Table 4: Projected Average Annuity Equivalents and Replacement Rates 
from DC Plan Balances at Retirement, by Income, under Different Model 
Assumptions: 

Baseline results, individual-level: Annuity equivalent (per year, 2007 
dollars); 
Overall: 18,784; 
By income quartile: 1: 1,850; 
By income quartile: 2: 6,554; 
By income quartile: 3: 16,635; 
By income quartile: 4: 50,098. 

Baseline results, individual-level: Replacement rate (percent); 
Overall: 22.2; 
By income quartile: 1: 10.3; 
By income quartile: 2: 18.2; 
By income quartile: 3: 26.3; 
By income quartile: 4: 33.8. 

Baseline results, individual-level: Percent of workers with no DC 
savings; 
Overall: 36.8; 
By income quartile: 1: 63.0; 
By income quartile: 2: 39.8; 
By income quartile: 3: 27.9; 
By income quartile: 4: 16.4. 

Instant eligibility/participation: Annuity equivalent (per year, 2007 
dollars); 
Overall: 26,265; 
By income quartile: 1: 4,243; 
By income quartile: 2: 11,142; 
By income quartile: 3: 24,370; 
By income quartile: 4: 65,305. 

Instant eligibility/participation: Replacement rate (percent); 
Overall: 35.0; 
By income quartile: 1: 25.4; 
By income quartile: 2: 31.3; 
By income quartile: 3: 38.8; 
By income quartile: 4: 44.7. 

Instant eligibility/participation: Percent of workers with no DC 
savings; 
Overall: 17.7; 
By income quartile: 1: 30.0; 
By income quartile: 2: 18.4; 
By income quartile: 3: 13.7; 
By income quartile: 4: 8.6. 

Participants always roll over balances upon job separation: Annuity 
equivalent (per year, 2007 dollars); 
Overall: 20,797; 
By income quartile: 1: 2,428; 
By income quartile: 2: 7,892; 
By income quartile: 3: 18,949; 
By income quartile: 4: 53,918. 

Participants always roll over balances upon job separation: Replacement 
rate (percent); 
Overall: 25.6; 
By income quartile: 1: 13.8; 
By income quartile: 2: 22.0; 
By income quartile: 3: 30.1; 
By income quartile: 4: 36.6. 

Participants always roll over balances upon job separation: Percent of 
workers with no DC savings; 
Overall: 27.0; 
By income quartile: 1: 48.8; 
By income quartile: 2: 28.1; 
By income quartile: 3: 19.3; 
By income quartile: 4: 11.6. 

Source: GAO projections using PENSIM model. 

Note: All results are individual level. Model assumptions include the 
following: 1) workers fully annuitize all accumulated DC plan balances 
at retirement, between age 62 and 70; 2) participants invest all plan 
assets in life cycle funds; 3) stocks earn an average annual 6.4 
percent real return, except where specified. Replacement rates equal 
annuitized income from lifetime DC plan savings divided by annualized 
career earnings. See app. I for more details. 

[End of table] 

Assuming automatic eligibility and participation raises projected plan 
savings significantly for lower-wage workers, more than doubling the 
annuity equivalent of retirement savings for the lowest-income 
quartile. Workers in the highest income group also increase savings 
under this scenario, with plan savings rising by 30 percent. This 
change in projected savings suggests that automatically enrolling new 
employees in plans as a default could have a significant positive 
impact on DC balances, especially for low-income workers whose jobs 
offer a plan, although this stylized scenario likely describes a more 
extreme change in eligibility and participation than plans are likely 
to implement under automatic enrollment, and that higher participation 
and savings would raise employer's pension costs, perhaps leading to a 
reduction in benefits or coverage. 

Another stylized scenario we model assumes that all workers who have a 
DC plan balance always keep the money in a tax-preferred account upon 
leaving a job, either by keeping the money in the plan, transferring it 
to a new employer plan, or rolling it into an IRA, rather than cashing 
out any accumulated savings.[Footnote 42] Eliminating this source of 
leakage raises average annuity income from DC plans by almost 11 
percent and average replacement rates from 22.2 percent in the baseline 
to 25.6 percent; it also reduces the percentage of the cohort with no 
DC savings at retirement by over 25 percent. As with the instant 
participation scenario, "universal rollover" raises annuity savings and 
reduces the number of retirees with zero plan savings by the biggest 
percentages among lower-income workers, suggesting that cashing out 
accumulate plan savings prior to retirement may be a more significant 
drain on retirement savings for these groups. These results indicate 
that policies to encourage participants to keep DC plan balances in tax-
preferred retirement accounts, perhaps by making rollover of plan 
assets a default action in plans, may have a broad positive impact on 
retirement savings. 

Changing Retirement Decisions or Contribution Limits Would Affect 
Savings Primarily for Higher-Income Workers: 

Other changes we make in our projections related to plan features or 
individual behavior affect average replacement rates overall, but with 
less impact on lower-income workers' replacement rates and on the 
number of workers with zero plan savings at retirement. These scenarios 
include assumed changes in annual contribution limits and retirement 
decisions (see table 5). 

Table 5: Projected Average Annuity Equivalents and Replacement Rates 
from DC Plan Balances at Retirement, by Income, Under Different Model 
Assumptions: 

Baseline results, individual-level: Annuity equivalent (per year, 2007 
dollars); 
Overall: 18,784; 
By income quartile: 1: 1,850; 
By income quartile: 1: 6,554; 
By income quartile: 3: 16,635; 
By income quartile: 4: 50,098. 

Baseline results, individual-level: Replacement rate (percent); 
Overall: 22.2; 
By income quartile: 1: 10.3; 
By income quartile: 1: 18.2; 
By income quartile: 3: 26.3; 
By income quartile: 4: 33.8. 

Baseline results, individual-level: Percent of workers with no DC 
savings; 
Overall: 36.8; 
By income quartile: 1: 63.0; 
By income quartile: 1: 39.8; 
By income quartile: 3: 27.9; 
By income quartile: 4: 16.4. 

Raise annual contribution limits: Annuity equivalent (per year, 2007 
dollars); 
Overall: 21,056; 
By income quartile: 1: 1,879; 
By income quartile: 1: 6,583; 
By income quartile: 3: 16,999; 
By income quartile: 4: 58,763. 

Raise annual contribution limits: Replacement rate (percent); 
Overall: 23.6; 
By income quartile: 1: 10.5; 
By income quartile: 1: 18.3; 
By income quartile: 3: 26.9; 
By income quartile: 4: 38.5. 

Raise annual contribution limits: Percent of workers with no DC 
savings; 
Overall: 36.7; 
By income quartile: 1: 63.0; 
By income quartile: 1: 39.9; 
By income quartile: 3: 27.9; 
By income quartile: 4: 16.2. 

Workers delay retirement 1 year: Annuity equivalent (per year, 2007 
dollars); 
Overall: 19,873; 
By income quartile: 1: 1,876; 
By income quartile: 1: 6,895; 
By income quartile: 3: 17,826; 
By income quartile: 4: 52,895. 

Workers delay retirement 1 year: Replacement rate (percent); 
Overall: 23.3; 
By income quartile: 1: 10.5; 
By income quartile: 1: 19.0; 
By income quartile: 3: 28.0; 
By income quartile: 4: 35.6. 

Workers delay retirement 1 year: Percent of workers with no DC savings; 
Overall: 36.9; 
By income quartile: 1: 63.5; 
By income quartile: 1: 40.3; 
By income quartile: 3: 27.2; 
By income quartile: 4: 16.3. 

Workers delay retirement 3 years: Annuity equivalent (per year, 2007 
dollars); 
Overall: 22,710; 
By income quartile: 1: 2,151; 
By income quartile: 1: 7,623; 
By income quartile: 3: 19,897; 
By income quartile: 4: 61,170. 

Workers delay retirement 3 years: Replacement rate (percent); 
Overall: 25.7; 
By income quartile: 1: 12.1; 
By income quartile: 1: 20.7; 
By income quartile: 3: 30.5; 
By income quartile: 4: 39.4. 

Workers delay retirement 3 years: Percent of workers with no DC 
savings; 
Overall: 36.8; 
By income quartile: 1: 63.1; 
By income quartile: 1: 39.9; 
By income quartile: 3: 28.5; 
By income quartile: 4: 15.7. 

Source: GAO calculations using projected savings from PENSIM model. 

Note: All results are individual level. Model assumptions include the 
following: 1) workers fully annuitize all accumulated DC plan balances 
at retirement, between age 62 and 70; 2) participants invest all plan 
assets in life cycle funds; 3) stocks earn an average annual 6.4 
percent real return, except where specified. Replacement rates equal 
annuitized income from lifetime DC plan savings divided by annualized 
career earnings. See app. I for more details. 

[End of table] 

We model projected retirement savings assuming that annual DC 
contribution limits for employees rise from $15,500 to $25,000, and the 
combined employer-employee maximum contribution level rises from 
$45,000 to $60,000, starting in 2007.[Footnote 43] Higher annual 
maximum contributions affect projected savings almost exclusively among 
the highest-income group, indicating that few workers earning less are 
likely to contribute at existing maximum levels. The highest income 
quartile replacement rises from 33.8 to 38.5 percent, while replacement 
rates hardly change in the lower income groups. Similarly, this 
scenario has almost no impact on the percentage of workers with DC plan 
savings at retirement. 

Finally, we model retirement savings in two scenarios in which workers 
delay retirement by 1 or 3 years. Encouraging workers to retire later 
has been suggested as a key element in improving retirement income 
security, by increasing earnings, allowing more time to save for 
retirement, and reducing the length of retirement. In our projections, 
delaying retirement not only provides more years to contribute to and 
earn returns on plan balances but also might raise annual retirement 
income because older retirees receive more annuity income for any given 
level of savings, holding all else equal. In our projections, working 
longer modestly raises retirement savings in our projections. Working 
one extra year changes projected annuity income by 5.8 percent, but has 
little effect on the percentage of people with no DC savings in our 
projections. Delaying retirement by 3 years raises annuity income from 
DC plans by 20.9 percent on average, with replacement rates rising from 
22.2 percent in the baseline to 25.7 percent overall.[Footnote 44] The 
3-year delay increases annuity levels somewhat evenly across income 
groups, with higher-income workers showing slightly higher increases. 
Overall, working an extra 3 years raises average replacement rates 
about as much as universal account rollover would, but with little 
reduction in workers with no retirement savings. Thus, while working 
longer would likely raise workers' incomes, and in most cases 
retirement benefits from other sources such as Social Security, our 
projections show that this change alone would have a modest impact on 
retirement income from DC plans, particularly regarding lower-income 
workers and those not already saving in DC plans in the baseline. 

Recent Changes and Proposals Could Have Positive Effects on DC Plan 
Coverage, Participation, and Savings: 

Recent regulatory and legislative changes and proposals could have 
positive effects on DC plan coverage, participation, and saving. The 
Pension Protection Act of 2006 (PPA) facilitated the adoption of 
automatic plan features by plan sponsors that may increase DC 
participation and savings within existing plans.[Footnote 45] Proposals 
to expand the saver's credit could similarly encourage greater 
contributions by low-wage workers who are already covered by a DC plan. 
Other options, like the so-called "State-K" proposal, in which states 
would design and partner with private financial institutions to offer 
low-cost DC plans employers could provide to employees, would seek to 
expand coverage among workers without current plans by encouraging 
employers to sponsor new plans. Other options would try to increase 
retirement account coverage by increasing the use of IRAs or creating 
new retirement savings vehicles outside of the voluntary employer- 
sponsored pension framework. Such proposals include automatic IRAs, in 
which employers would be required to allow employees through automatic 
enrollment to contribute to IRAs by direct payroll deposit, or 
universal accounts proposals, in which all workers would be given a 
retirement account regardless of whether they had any employment based 
pension coverage. 

Changing Default Plan Features and Reforming Saver's Credit Could Raise 
DC Plan Savings, but May Have Limited Impact on Coverage: 

Changing certain traditional DC plan defaults may have a significant 
impact on DC participation and savings. Research suggests that 
employees exhibit inertia regarding plan participation and 
contributions, which can reduce DC savings by failure to participate or 
increase savings over time.[Footnote 46] To reverse the effects of 
these tendencies, some experts have suggested changing default plan 
actions to automatically sign up employees for participation, escalate 
contributions, and set default investment options unless workers opt 
out.[Footnote 47] Some studies have shown that automatic enrollment may 
increase DC plan participation. For example, one study of a large firm, 
automatic enrollment increased participation from 57 percent for 
employees eligible to participate 1 year before the firm adopted 
automatic enrollment to 86 percent for those hired under automatic 
enrollment.[Footnote 48] Another study finds that, prior to automatic 
enrollment, 26 to 43 percent of employees at 6 months' tenure 
participated in the plan at three different companies; under automatic 
enrollment 86 to 96 percent of employees participated.[Footnote 49] 
Some also advocate automatically rolling over DC savings into an IRA 
when employees separate from their employers to further increase 
retirement savings. Our own simulations shows that universal account 
rollover to a tax-preferred account, such as a new plan or an IRA, 
would increase projected retirement savings by 11 percent on average, 
with the biggest percentage increases for lowest-income workers. 

Various regulatory and legislative changes have focused on default DC 
plan features. In 1998, the IRS first approved plan sponsor use of 
automatic enrollment--the ability for plans to automatically sign 
employees up for a 401(k) plan (from which the employee can opt out), 
and--subsequently issued several rulings that clarified the use of 
other automatic plan features and the permissibility of automatic 
features in 403(b) and 457 plans.[Footnote 50] Accordingly, the 
percentage of 401(k) plans using automatic plan features has increased 
in recent years. One annual study of plan sponsors found that in 2004, 
12.4 percent of 401(k) plans were automatically enrolling participants, 
and this number increased to 17.5 percent of plans in 2005.[Footnote 
51] The percentage of plans automatically increasing employee 
contributions also rose from 6.8 percent in 2004 to 13.6 percent in 
2005. Some experts have argued that initially, some plan sponsors may 
have been hesitant to use automatic plan features because of legal 
ambiguities between state and federal law. However, clarifications 
relating to automatic enrollment and default investment in the PPA have 
led some plan sponsors and experts to expect more plans to adopt 
automatic plan features.[Footnote 52] 

Automatic DC plan features, however, may create complications for 
sponsors and participants that may limit any effect on savings and 
participation. Auto enrollment may not help expand plan sponsorship; in 
fact, sponsors who offer a matching contribution may not want to offer 
automatic enrollment if they believe this will raise their pension 
costs.[Footnote 53] Also, if sponsors automatically invest 
contributions in a low-risk fund such as a money market fund, this 
could limit rates of return on balances. However, choosing a risky 
investment fund could subject automatic contributions to market 
losses.[Footnote 54] Some employees may not realize they have been 
signed up for a plan, and may be displeased to discover this, 
particularly if their automatically invested contributions have lost 
money.[Footnote 55] 

Proposals Seek to Increase Plan Coverage and Savings, Particularly for 
Lower-Income Workers: 

Other proposals would target plan formation or increase participation 
and retirement savings by expanding worker access to other account- 
based retirement savings vehicles like IRAs. Some of these alternative 
retirement savings proposals are voluntary in design, while others are 
more universal. 

State-K: The State-K proposal aims to make low-cost retirement plans 
available to employers who do not otherwise sponsor a plan.[Footnote 
56] Under a State-K program, state government entities would design and 
administer a 401(k) plan that an employer within the state could choose 
to offer its employees. By pooling resources and sharing costs across 
many different employers, states would seek to create a lower-cost plan 
that employers, particularly smaller ones, that do not sponsor any plan 
might find attractive enough to adopt and offer to their employees. In 
one proposed program from Maryland, expenses to implement, maintain, 
and administer the plan would be paid from contributions to, or income 
or assets of, the program.[Footnote 57] The 401(k) designs made 
available by a State-K program would use automatic enrollment and low- 
cost default investment, with limited and simple features to minimize 
administrative costs and simplify choices for employers and employees. 
Contributions accumulated by these State-K funds would be kept separate 
from state employee retirement plan funds, but State-K funds could be 
pooled in order to lower overall plan investment costs. By decreasing 
the costs and assisting small employers with compliance and plan 
administration, the State-K could be expected to increase incentives 
for employers to sponsor plans, thus increasing employee access to 
account-based retirement plans. However, it is unclear to what extent 
employers would adopt such plans. 

The Automatic IRA: The Automatic IRA proposal would make direct deposit 
or payroll deduction saving into an IRA available to all employees by 
requiring employers that do not sponsor any retirement plan to offer 
withholding to facilitate employee contributions. To maximize 
participation, employees would be automatically enrolled at 3 percent 
of pay, or could elect to opt out or to defer a different percentage of 
pay to an IRA, up to the maximum IRA annual contribution limit ($4,000 
for 2007; $5,000 for 2008). Employers would not be required choose 
investments or set up the IRAs, which would be provided mainly by the 
private-sector IRA trustees and custodians that currently provide them. 
Employers also would not be required or permitted to make matching 
contributions, and would not need to comply with the Employee 
Retirement Income Security Act of 1974 (ERISA) or any qualified plan 
standards such as non-discrimination requirements. Employers, however, 
would be required to provide notice to employees, including information 
on the maximum amount that can be contributed to the plan on an annual 
basis.[Footnote 58] One congressional proposal would require employers, 
other than small or new ones, to offer payroll deposit IRA arrangements 
to employees not eligible for pension plans and permit automatic 
enrollment in such IRAs in many circumstances. Participating IRAs would 
be required to offer a default investment consisting of life cycle 
funds similar to those offered by the Thrift Savings Plan, the DC plan 
for federal workers, or other investments specified by a new entity 
established for that purpose.[Footnote 59] 

Universal accounts: Similar to the automatic IRA, universal account 
(UA) proposals aim to establish retirement savings accounts for all 
workers, and vary slightly based on employment-based pension access. 
Additionally, some proposals would have employers contribute to the 
account, whereas other proposals would also have the federal government 
match contributions. One proposal suggests a 2 percent annual 
contribution from the federal government regardless of individual 
contributions, while another would provide for individual contributions 
only, capped at $7,500 per year.[Footnote 60] In 1999, the Clinton 
Administration proposed a UA to be established for each worker and 
spouse with earnings of at least $5,000 annually. Individuals would 
receive a tax credit of up to $300 annually. Additionally, workers 
could voluntarily contribute to the account up to specified amounts 
with a 50 to 100 percent match by the federal government.[Footnote 61] 
This match would come in the form of a tax credit, and total voluntary 
contributions, including government contributions, would be limited to 
$1,000. Both the credit and the match would phase out as income 
increases, providing a progressive benefit and targeting low-and middle-
income workers. Federal contributions would have revenue implications, 
while requiring employer contributions could increase employer 
compensation costs. 

Other proposals would expand the size and scope of the saver's credit 
to encourage greater contributions by those low-wage workers who are 
already covered by a DC plan that allows employee contributions. 
Currently, the saver's credit, originally proposed in 2000 as an 
outgrowth of the 1999 UA proposal as a government matching deposit on 
some voluntary contributions to IRAs and 401(k) plans, provides a 
nonrefundable tax credit to low-and middle-income savers of up to 50 
percent of their annual IRA or 401(k) contributions of up to $2,000. 
However, according to one analysis, because the credit is 
nonrefundable, only about 17 percent of those with incomes low enough 
to qualify for the credit would receive any benefit if they contributed 
to a plan. Some analysts think that expanding the saver's credit, or 
creating direct transfers such as tax rebates or deposits into 
retirement savings accounts, could increase plan contributions 
specifically for low-and middle-income workers.[Footnote 62] Making the 
saver's credit refundable to the participant could also provide a 
direct transfer to the tax filer in lieu of a retirement account match, 
but offers no assurance that funds would be saved or deposited into a 
retirement account.[Footnote 63] A refundable tax credit would also 
have revenue implications for the federal budget. 

Concluding Observations: 

The DC plan has clearly overtaken the DB plan as the principal 
retirement plan for the nation's private sector workforce, and its 
growing dominance suggests its increasingly crucial role in the 
retirement security of current and future generations of workers. The 
current DC-based system faces major challenges, like its DB-based 
predecessor, in terms of coverage, participation, and lifetime 
distributions. Achieving retirement security through DC plans carries 
particular challenges for workers, since accumulating benefits in an 
account-based plan requires more active commitment and management from 
individuals than it does for DB participants. Since workers must 
typically sign up and voluntarily reduce their take home pay to 
contribute to their DC plans, invest this money wisely over their 
working years, and resist withdrawing from balances prior to 
retirement, it is perhaps to be expected that even those who have the 
opportunity to participate save little. While our results on both 
current and projected plan balances suggest that while some workers 
save significant amounts toward their retirement in DC plans, a large 
proportion of workers will likely not save enough in DC plans for a 
secure retirement. 

Of particular concern are the retirement income challenges faced by 
lower earners. Many of these workers face competing income demands for 
basic necessities that may make contributions to their retirement plans 
difficult. Further, the tax preferences that may entice higher-income 
workers to contribute to their DC plans may not entice low-income 
workers who have plan coverage, since these workers face relatively low 
marginal tax rates. Our model results suggest that other measures, such 
as automatic enrollment and rollover of funds may make a difference for 
some lower income workers. Should pension policy, as embodied by the 
automatic provisions in PPA, continue to move in this direction, it 
should focus on those workers most in need of enhanced retirement 
income prospects. 

Agency Comments: 

We provided a draft of this report to the Department of Labor and the 
Department of the Treasury, as well as to five outside reviewers. 
Neither agency provided formal comments. We incorporated any technical 
comments we received throughout the report as appropriate. 

As agreed with your office, unless you publicly announce its contents 
earlier, we plan no further distribution until 30 days after the date 
of this letter. At that time, we will send copies of this report to the 
Secretary of Labor, the Secretary of the Treasury, 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 website at [hyperlink, 
http://www.gao.gov]. 

If you have any questions concerning this report, please contact 
Barbara Bovbjerg at (202) 512-7215. Contact points for our Office 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 III. 

Sincerely yours, 

Signed by: 

Barbara D. Bovbjerg: 

Director Education, Workforce, and Income Security Issues: 

[End of section] 

Appendix I: Scope and Methodology: 

To analyze saving in DC plans, we examined data from the Federal 
Reserve Board's Survey of Consumer Finances (SCF). This triennial 
survey asks extensive questions about household income and wealth 
components. We used the latest available survey, from 2004. The SCF is 
widely used by the research community, is continually vetted by the 
Federal Reserve and users, and is considered to be a reliable data 
source. The SCF is believed by many to be the best source of publicly 
available information on household finances. Further information about 
our use of the SCF, including sampling errors, as well as definitions 
and assumptions we made in our analysis are detailed below. We also 
reviewed published statistics in articles by public policy groups and 
in academic studies. 

To analyze how much Americans can expect to save in DC plans over their 
careers and the factors that affect these savings, we used the Policy 
Simulation Group's (PSG) microsimulation models to run various 
simulations of workers saving over a working career, changing various 
inputs to model different scenarios that affect savings at retirement. 
PENSIM is a pension policy simulation model that has been developed for 
the Department of Labor to analyze lifetime coverage and adequacy 
issues related to employer-sponsored pensions in the United States. We 
projected account balances at retirement for PENSIM-generated workers 
under different scenarios representing different pension features, 
individual behavioral decisions, and market assumptions. See below for 
further discussion of PENSIM and our assumptions and methodologies. 

To analyze those plan-or government-level policies that might best 
increase participation and savings in DC plans, we synthesized 
information gathered from interviews of plan practitioners, financial 
managers, and public policy experts, as well as from academic and 
policy studies on DC plan participation and savings. We also researched 
current government initiatives and policy proposals to broaden 
participation in account-based pension plans and increase retirement 
savings. 

We conducted our work from July 2006 to October 2007 in accordance with 
generally accepted government auditing standards. 

Methodology and Assumptions Using Survey of Consumer Finances Data: 

The 2004 SCF surveyed 4,522 households about their pensions, incomes, 
labor force participation, asset holdings and debts, use of financial 
services, and demographic information. The SCF is conducted using a 
dual-frame sample design. One part of the design is a standard, multi- 
stage area-probability design, while the second part is a special 
oversample of relatively wealthy households. This is done in order to 
accurately capture financial information about the population at large 
as well as characteristics specific to the relatively wealthy. The two 
parts of the sample are adjusted for sample nonresponse and combined 
using weights to provide a representation of households overall. In 
addition, the SCF excludes people included in the Forbes Magazine list 
of the 400 wealthiest people in the United States. Furthermore, the 
2004 SCF dropped three observations from the public data set that had 
net worth at least equal to the minimum level needed to qualify for the 
Forbes list. 

The SCF is a probability sample based on random selections, so the 2004 
SCF sample is only one of a large number of samples that might have 
been drawn. Since each sample could have provided different estimates, 
we express our confidence in the precision of our particular sample's 
results as a 95 percent confidence interval (e.g., plus or minus 4 
percentage points). This is the interval that would contain the actual 
population value for 95 percent of the samples we could have drawn. As 
a result, we are 95 percent confident that each of the confidence 
intervals in this report will include the true values in the study 
population. 

All percentage estimates based on GAO analysis of 2004 SCF data used in 
this report have 95 percent confidence intervals that are within plus- 
or-minus 4 percentage points, with the following exceptions described 
in table 6 below. 

Table 6: Sampling Errors Greater than 4 Percentage Points for 
Percentage Estimates at the 95 Percent Confidence Interval: 

Estimate: Percentage of workers aged 60-64 participating in current DC 
plans; 
Value: 45; 
Sampling error: +/-7.6. 

Estimate: Percentage of households who had previously received lump-sum 
distributions who cashed out all funds; 
Value: 47; 
Sampling error: +/- 4.7. 

Estimate: Percentage of households who had previously received lump-sum 
distributions who rolled over all funds; 
Value: 50; 
Sampling error: +/
-4.3. 

Estimate: Percentage of workers, in the lowest income quartile and 
offered a retirement plan, participating in current DC plans; 
Value: 60; 
Sampling error: +/-5.8. 

Source: GAO analysis of 2004 SCF data. 

[End of table] 

Other numerical estimates based on GAO analysis of 2004 SCF data used 
in this report have 95 percent confidence intervals that are within 25 
percent of the estimate itself, with exceptions described in table 7. 
[Footnote 64] 

Table 7: Sampling Errors Greater Than 25 Percent for Numerical 
Estimates at the 95 Percent Confidence Interval: 

Median total DC balances for all workers with either a current or 
former DC plan; 
Estimate (2003 dollars): 24,200; 
Sampling error (Plus or minus): +/-5,303. 

Median total DC balances for all households with either a current or 
former DC plan; 
Estimate (2003 dollars): 27,940; 
Sampling error (Plus or minus): +/-4,762. 

Median total DC balances for workers with either a current or former DC 
plan (aged 18-29); 
Estimate (2003 dollars): 6,160; 
Sampling error (Plus or minus): +/-3,791. 

Median total DC balances for workers with either a current or former DC 
plan (aged 30-39); 
Estimate (2003 dollars): 14,580; 
Sampling error (Plus or minus): +/-2,852. 

Median total DC balances for workers with either a current or former DC 
plan (aged 40-49); 
Estimate (2003 dollars): 31,580; 
Sampling error (Plus or minus): +/-7,460. 

Median total DC balances for workers with either a current or former DC 
plan (aged 50-59); 
Estimate (2003 dollars): 43,200; 
Sampling error (Plus or minus): +/-12,113. 

Median total DC balances for workers with either a current or former DC 
plan (aged 60-64); 
Estimate (2003 dollars): 60,600; 
Sampling error (Plus or minus): +/-23,435. 

Mean total DC balances for workers with either a current or former DC 
plan (aged 18-29); 
Estimate (2003 dollars): 11,865; 
Sampling error (Plus or minus): +/-2,507. 

Mean total DC balances for workers with either a current or former DC 
plan (aged 30-39); 
Estimate (2003 dollars): 34,851; 
Sampling error (Plus or minus): +/-6,507. 

Mean total DC balances for workers with either a current or former DC 
plan (aged 40-49); 
Estimate (2003 dollars): 64,124; 
Sampling error (Plus or minus): +/-8,271. 

Mean total DC balances for workers with either a current or former DC 
plan (aged 50-59); 
Estimate (2003 dollars): 126,301; 
Sampling error (Plus or minus): +/-25,156. 

Mean total DC balances for workers with either a current or former DC 
plan (aged 60-64); 
Estimate (2003 dollars): 135,299; 
Sampling error (Plus or minus): +/-34,933. 

Median total DC balances for workers with either a current or former DC 
plan (aged 55-64); 
Estimate (2003 dollars): 50,000; 
Sampling error (Plus or minus): +/-22,406. 

Median Total of Cashed Out Funds for households that cashed out all 
lump sum distributions; 
Estimate (2003 dollars): 6,800; 
Sampling error (Plus or minus): +/-1,429. 

Median Total of Rolled Over Funds for households that rolled over all 
lump sum distributions; 
Estimate (2003 dollars): 24,200; 
Sampling error (Plus or minus): +/-5,303. 

Median total DC balances for workers of median income or lower with 
either a current or former DC plan; 
Estimate (2003 dollars): 9,420; 
Sampling error (Plus or minus): +/-2,165. 

Median total DC balances for workers of median wealth or lower with 
either a current or former DC plan (aged 50-59); 
Estimate (2003 dollars): 13,800; 
Sampling error (Plus or minus): +/-5,244. 

Median total DC balances for workers of median wealth or lower with 
either a current or former DC plan (aged 60-64); 
Estimate (2003 dollars): 18,000; 
Sampling error (Plus or minus): +/-14,766. 

Median total DC balances for workers with either a current or former DC 
plan (with a DB plan); 
Estimate (2003 dollars): 20,820; 
Sampling error (Plus or minus): +/-2,652. 

Median total DC balances for workers with either a current or former DC 
plan (without a DB plan); 
Estimate (2003 dollars): 31,560; 
Sampling error (Plus or minus): +/-7,640. 

Source: GAO analysis of 2004 SCF data. 

Note: We calculated monthly and yearly incomes as annuity equivalents 
using the Thrift Savings Plan calculator [hyperlink, 
http://calc.tsp.gov], assuming an interest rate of 5.250 percent, 
single life benefits beginning at age 65, no joint survivor benefits, 
and level payments. 

[End of table] 

Because of the complexity of the SCF design and the need to suppress 
some detailed sample design information to maintain confidentiality of 
respondents, standard procedures for estimate of sampling errors could 
not be used. Further, the SCF uses multiple imputations to estimate 
responses to most survey questions to which respondents did not provide 
answers. Sampling error estimates for this report are based on a 
bootstrap technique using replicate weights to produce estimates of 
sampling error that account for both the variability due to sampling 
and due to imputation.[Footnote 65] 

The SCF collects detailed information about an economically dominant 
single individual or couple in a household (what the SCF calls a 
primary economic unit), where the individuals are at least 18 years 
old. We created an additional sample containing information on 7,471 
individuals by separating information about respondents and their 
spouses or partners and considering them separately. When we discuss 
individuals in this document, we are referring to this sample. When we 
refer to all workers, we are referring to the subpopulation of workers 
within this individual sample. In households where there are additional 
adult workers, beyond the respondent and the spouse or partner, who may 
also have earnings and a retirement plan, information about these 
additional workers is not captured by the SCF and is therefore not part 
of our analysis. It is also important to note that the SCF was designed 
to be used as a household survey, and some information could not be 
broken into individual-level information. Where that was the case, we 
presented only household-level information. 

We defined "worker" relatively broadly and opted to begin with the set 
of all those who reported that they were both working and some other 
activity, including for example, "worker plus disabled" and "worker 
plus retired." We then excluded those workers who reported that they 
were self-employed from our analysis. Our definition of DC plans 
includes the following plans: 401(k); 403(b); 457; thrift/savings plan; 
profit-sharing plan; portable cash option plan; deferred compensation 
plan, n.e.c; SEP/SIMPLE; money purchase plan; stock purchase plan; and 
employee stock ownership plan (ESOP). 

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

Our analysis of the 2004 SCF yielded slightly lower participation rates 
than other data sets that consider pensions. For example, 2004 Bureau 
of Labor Statistics (BLS) data indicate a somewhat higher rate of 
active participation in DC accounts, 42 percent, compared with our 
finding of 36 percent. One possible factor contributing to this 
difference is that BLS surveys establishments about their employees, 
while SCF surveys individuals who report on themselves and their 
households; it is possible that the SCF respondents may be failing to 
report all retirement accounts, while BLS is capturing a greater 
proportion of them. Also, the SCF considered both public and private 
sector workers, while the BLS statistic is only for private sector 
workers. Differences may also be explained by different definitions of 
workers and participation, question wording, or lines of questioning. 
The SCF appears to provide a lower bound on the estimation of pension 
coverage among 4 major data sets.[Footnote 66] 

Methodology and Assumptions Using PENSIM Microsimulation Model: 

To project lifetime savings in DC pensions, and related retirement 
plans with personal accounts, and to identify the effects of changes in 
policies, market assumptions, or individual behavior, we used the 
Policy Simulation Group's (PSG) Pension Simulator (PENSIM) 
microsimulation models.[Footnote 67] PENSIM is a dynamic 
microsimulation model that produces life histories for a sample of 
individuals born in the same year. The life history for a sample 
individual includes different life events, such as birth, schooling 
events, marriage and divorce, childbirth, immigration and emigration, 
disability onset and recovery, and death. In addition, a simulated life 
history includes a complete employment record for each individual, 
including each job's starting date, job characteristics, pension 
coverage and plan characteristics, and ending date. The model has been 
developed by PSG since 1997 with funding and input by the Office of 
Policy and Research at the Employee Benefits Security Administration 
(EBSA) of the U.S. Department of Labor with the recommendations of the 
National Research Council panel on retirement income modeling. 

PENSIM sets the timing for each life event by using data from various 
longitudinal data sets to estimate a waiting-time model (often called a 
hazard function model) using standard survival analysis methods. PENSIM 
incorporates many such estimated waiting-time models into a single 
dynamic simulation model. This model can be used to simulate a 
synthetic sample of complete life histories. PENSIM employs continuous- 
time, discrete-event simulation techniques, such that life events do 
not have to occur at discrete intervals, such as annually on a person's 
birthday. PENSIM also uses simulated data generated by another PSG 
simulation model, SSASIM, which produces simulated macro-demographic 
and macroeconomic variables. 

PENSIM imputes pension characteristics using a model estimated with 
1996 to 1998 establishment data from the BLS Employee Benefits Survey 
(now known as the National Compensation Survey). Pension offerings are 
calibrated to historical trends in pension offerings from 1975 to 2005, 
including plan mix, types of plans, and employer matching. Further, 
PENSIM incorporates data from the 1996-1998 Employee Benefits Survey 
(EBS) to impute access to and participation rates in DC plans in which 
the employer makes no contribution, which BLS does not report as 
pension plans in the NCS. The inclusion of these "zero-matching" plans 
enhances PENSIM's ability to accurately reflect the universe of pension 
plans offered by employers. PENSIM assumes that 2005 pension offerings, 
included the imputed zero-matching plans, are projected forward in 
time. 

PSG has conducted validation checks of PENSIM's simulated life 
histories against both historical life history statistics and other 
projections. Different life history statistics have been validated 
against data from the Survey of Income and Program Participation 
(SIPP), the Current Population Survey (CPS), Modeling Income in the 
Near Term (MINT3), the Panel Study of Income Dynamics (PSID), and the 
Social Security Administration's Trustees Report. PSG reports that 
PENSIM life histories have produced similar annual population, taxable 
earnings, and disability benefits for the years 2000 to 2080 as those 
produced by the Congressional Budget Office's long-term social security 
model (CBOLT) and as shown in the Social Security Administration's 2004 
Trustees Report. According to PSG, PENSIM generates simulated DC plan 
participation rates and account balances that are similar to those 
observed in a variety of data sets. For example, measures of central 
tendency in the simulated distribution of DC account balances among 
employed individuals is similar to those produced by an analysis of the 
Employee Benefit Research Institute (EBRI)-Investment Company Institute 
(ICI) 401(k) database and of the 2004 SCF. GAO performed no independent 
validation checks of PENSIM's life histories or pension 
characteristics. 

In 2006, EBSA submitted PENSIM to a peer review by three economists. 
The economists' overall reviews ranged from highly favorable to highly 
critical. While the economist who gave PENSIM a favorable review 
expressed a "high degree of confidence" in the model, the one who 
criticized it focused on PENSIM's reduced form modeling. This means 
that the model is grounded in previously observed statistical 
relationships among individuals' characteristics, circumstances, and 
behaviors, rather than on any underlying theory of the determinants of 
behaviors, such as the common economic theory that individuals make 
rational choices as their preferences dictate and thereby maximize 
their own welfare. The third reviewer raised questions about specific 
modeling assumptions and possible overlooked indirect effects. 

Assumptions Used in Projecting DC Plan Balances at Retirement: 

PENSIM allows the user to alter one or more inputs to represent changes 
in government policy, market assumptions, or personal behavioral 
choices and analyze the subsequent impact on pension benefits. Starting 
with a 2 percent sample of a 1990 cohort, totaling 104,435 people at 
birth. our baseline simulation includes some of the following key 
assumptions and features. For our report, we focus exclusively on 
accumulated balances in DC plans and ignore any benefits an individual 
might receive from DB plans or from Social Security. Our reported 
benefits and replacement rates therefore capture just one source of 
potential income available to a retiree. 

* Workers accumulate DC pension benefits from past jobs in one rollover 
account, which continue to receive investment returns, along with any 
benefits from a current job. At retirement, these are combined into one 
account. Because we focus on DC plan balances only, we assume all 
workers are ineligible to participate in DB plans and do not track 
Social Security benefits. 

* Plan participants invest all assets in their account in life cycle 
funds, which adjust the mix of assets between stocks and government 
bonds as the individual ages. Stocks return an annual nonstochastic 
real rate of return of 6.4 percent and government bonds have a real 
return of 2.9 percent per year. In one simulation, we use the 
government bond rate on all plan assets. Using different rates of 
return reflect assumptions used by OCACT in some of its analyses of 
trust fund investment. 

* Workers purchase a single, nominal life annuity, typically at 
retirement, which occurs between the ages of 62 and 70. Anyone who 
becomes permanently disabled at age 45 or older also purchases an 
immediate annuity at their disability age. We eliminate from the sample 
cohort members who: 1) die before they retire, at whatever age; 2) die 
prior to age 55; 3) immigrates into the cohort at an age older than 25; 
or 4) becomes permanently disabled prior to age 45. We assume that the 
annuity provider charges an administrative load on the annuity such 
that in all scenarios the provider's revenues balance the annuity costs 
(i.e., zero profit). 

* Replacement rates equal the annuity value of DC plan balances divided 
by a "steady earnings" index. This index reflects career earnings, 
calibrated to the Social Security Administration's age-65 average wage 
index (AWI). PENSIM computes steady earnings by first computing the 
present value of lifetime wages. Then, it calculates a scaling factor 
that, when multiplied by the present value of lifetime earnings for a 
1990 cohort member earning the AWI from ages 21 to 65, produces the 
individual's present value of lifetime earnings. This scaling factor is 
multiplied by AWI at age 65, then adjusted to 2007 dollars. Using this 
measure as opposed to average pay for an individual's final 3 or 5 
years of working, minimizes the problems presented by a worker who has 
irregular earnings near the end of his or her career, perhaps because 
of reduced hours.: 

* For household replacement rates, we use a combined annuity value of 
worker-spouse lifetime DC plan savings and a combined measure of steady 
family earnings. 

Starting from this baseline model, we vary key inputs and assumptions 
to see how these variations affect pension benefits and replacement 
rates at retirement. Scenarios we ran include: 

Universal rollover of DC plan balances. All workers with a DC balance 
roll it over into an Individual Retirement Account or another qualified 
plan upon job separation, as opposed to cashing out the balance, in 
which case the money is assumed lost for retirement purposes. 

Immediate eligibility and participation in a plan. A worker who would 
be offered a plan has no eligibility waiting period and immediately 
enrolls. This does not necessarily mean that the participant makes 
immediate or regular contributions; contribution levels are determined 
stochastically by PENSIM based on worker characteristics. 

Delayed retirement. Workers work beyond the retirement age determined 
by PENSIM in the baseline run. In one scenario, workers work up to one 
extra year; in another, they delay retirement for up to 3 years, 
although 70 remains the maximum retirement age. 

Raised contribution limits. We set annual contribution limits starting 
in 2007 to $25,000 per individual, up from $15,500 under current law, 
and $60,000 for combined employer-employee contributions, up from 
$45,000 under current law. These limits rise with cost of living 
changes in subsequent years, as is the case in our baseline model. 

PENSIM Cohort Summary and Cross-Sectional Statistics: 

Lifetime summary statistics of the simulated 1990 cohort's workforce 
and demographic variables give some insight into the model's projected 
DC savings at retirement that we report (see tables 8 and 9). The 
78,045 people in the sample who have some earnings, do not immigrate 
into the cohort after age 25, live to age 55, and retire (or become 
disabled at age 45 or older), work a median 29.4 years full-time and 
2.1 part-time, with median "steady" earnings of $46,122 (in 2007 
dollars). Those whose earnings fall in the lowest quartile work full- 
time for only a median 14.1 years, while working part-time for 9.1 
years, and 13.4 years for their longest-tenured job; this group's 
median annual steady earnings measure $16,820. In contrast, those in 
the highest-quartile of earnings work for a median 34.8 years, 
including 19.5 years for their longest job, and have median steady 
earnings of $126,380 per year. The results also show that pension 
coverage varies somewhat across income groups. About 83 percent of 
workers in the lowest income quartile have at least one job in which 
they are covered by a DC plan throughout their working careers, and are 
eligible for DC plan coverage for a median 9.4 years. In contrast, at 
least 90 percent of workers in the highest three income quartiles have 
some DC coverage during their careers. Those in the highest income 
quartile are eligible for DC participation for a median 25.2 years 
throughout their career. 

Table 8: Summary Statistics, PENSIM 1990 Cohort: 

Demographic variables, full sample: N, full sample; 
Full sample: 104,435; 
By income quartile: 1: [Empty]; 
By income quartile: 2: [Empty]; 
By income quartile: 3: [Empty]; 
By income quartile: 4: [Empty]. 

Demographic variables, full sample: N, for replacement rate 
calculations[A]; 
Full sample: 78,045; 
By income quartile: 1: 19,511; 
By income quartile: 2: 19,511; 
By income quartile: 3: 19,512; 
By income quartile: 4: 19,511. 

Demographic variables, full sample: Percent female; 
Full sample: 49.5; 
By income quartile: 1: 73.8; 
By income quartile: 2: 55.6; 
By income quartile: 3: 44.8; 
By income quartile: 4: 28.2. 

Demographic variables, full sample: Education (median); 
Full sample: Some college; 
By income quartile: 1: High school graduate; 
By income quartile: 2: High school graduate; 
By income quartile: 3: Some college; 
By income quartile: 4: College graduate. 

Demographic variables, full sample: Percent who work for at least one 
DC sponsor during career; 
Full sample: 90.4; 
By income quartile: 1: 83.2; 
By income quartile: 2: 90.8; 
By income quartile: 3: 92.7; 
By income quartile: 4: 95.1. 

Demographic variables, full sample: Percent whose longest-held job 
offered DC pension; 
Full sample: 73.3; 
By income quartile: 1: 56.3; 
By income quartile: 2: 71.7; 
By income quartile: 3: 79.2; 
By income quartile: 4: 86.2. 

Source: GAO calculations of PENSIM simulation of 1990 cohort. 

Note: Percentage female and education medians are for entire sample; 
all other statistics are for only those used in the replacement rate 
calculations. 

[A] Excludes cohort members who have no lifetime earnings; immigrate 
after age 25; die prior to retiring or becoming disabled; or become 
disabled prior to age 45. 

[End of table] 

Table 9: Sample Summary Statistics, PENSIM 1990 Cohort, Medians: 

Workforce variables: Years working full-time; 
Full sample: 29.4; 
By income quartile: 1: 14.1; 
By income quartile: 2: 27.9; 
By income quartile: 3: 31.8; 
By income quartile: 4: 34.8. 

Workforce variables: Years working part-time; 
Full sample: 2.1; 
By income quartile: 1: 9.1; 
By income quartile: 2: 2.2; 
By income quartile: 3: 1.1; 
By income quartile: 4: 0.5. 

Workforce variables: Steady earnings (annual, 2007 dollars); 
Full sample: 46,122; 
By income quartile: 1: 16,820; 
By income quartile: 2: 34,950; 
By income quartile: 3: 60,777; 
By income quartile: 4: 126,380. 

Workforce variables: Number of jobs over lifetime; 
Full sample: 5.0; 
By income quartile: 1: 5.0; 
By income quartile: 2: 5.0; 
By income quartile: 3: 5.0; 
By income quartile: 4: 5.0. 

Workforce variables: Duration of longest job, years; 
Full sample: 17.2; 
By income quartile: 1: 13.4; 
By income quartile: 2: 16.9; 
By income quartile: 3: 18.3; 
By income quartile: 4: 19.5. 

Workforce variables: Retirement age; 
Full sample: 63.0; 
By income quartile: 1: 62.0; 
By income quartile: 2: 63.0; 
By income quartile: 3: 63.0; 
By income quartile: 4: 64.0. 

Workforce variables: Years eligible for DC pension; 
Full sample: 18.5; 
By income quartile: 1: 9.4; 
By income quartile: 2: 17.6; 
By income quartile: 3: 21.2; 
By income quartile: 4: 25.2. 

Source: GAO calculations of PENSIM simulation of 1990 cohort. 

[End of table] 

Cross-sectional results of the sample cohort also provide some insights 
into the model's assumptions, as well as some further insights into the 
relatively low projected sample replacement rates (see table 10). These 
statistics describe the working characteristics for each employed 
individual at a randomly determined age sometime between 22 and 62 in 
order to provide a snapshot of a "current" job for most of the sample. 
Among those employed at the time of the survey, 61.8 percent had an 
employer who sponsors a DC plan. Of these workers with a plan offered, 
94 percent were eligible to participate, and among those eligible 67 
percent participated. Taking all of these percentages together, this 
means that at any one time only 38.9 percent of the working population 
actively participated in a DC plan in our projections. Even among these 
participants, only 56.9 percent reported making a contribution to the 
plan in the previous year, while 45.7 percent had an employer 
contribution. Median combined employer-employee contributions in the 
previous year were 6.2 percent of earnings in our simulation. 

Table 10: Cross-Sectional Pension Characteristics of Sample: 

Age at survey; 
Average: 42.1; 
Median: 42.1. 

Percent of sample employed; 
Average: 71.5; 
Median: [Empty]. 

Current job duration (years); 
Average: 8.0; 
Median: 5.9. 

Job offers DC plan; 
Average: 61.8; 
Median: [Empty]. 

Among offered, percent eligible to participate; 
Average: 93.9; 
Median: [Empty]. 

Among eligible, percent participating; 
Average: 67.0; 
Median: [Empty]. 

Past year's employee contribution (percent of earnings); 
Average: 4.3; 
Median: 3.9. 

Past year's employer contribution (percent of earnings); 
Average: 3.1; 
Median: 0. 

Total contribution (percent of earnings); 
Average: 7.4; 
Median: 6.2. 

Cumulative returns per year in plan (2007 dollars); 
Average: 1,303; 
Median: 383. 

Cumulative returns in current plan (2007 dollars); 
Average: 22,318; 
Median: 180. 

Among eligible, percent contributing in past year; 
Average: 56.9; 
Median: [Empty]. 

Among eligible, percent with employer contribution in past year; 
Average: 45.7; 
Median: [Empty]. 

Source: GAO calculations of PENSIM simulation of 1990 cohort. 

Note: Results reflect one time snapshot of each member of the sample at 
a randomly determined age between 22 and 62. 

[End of table] 

[End of section] 

Appendix II: Comparison of DC Plan Projections Based on PENSIM to Other 
Studies: 

Other studies have projected DC plan savings for workers saving over 
their entire working careers. These studies generally find higher 
projected replacement rates from DC plan savings than our simulations 
do. However, each study makes different key assumptions, particularly 
about plan coverage, participation, and contributions. 

A 2007 study by Patrick Purcell and Debra B. Whitman for the 
Congressional Research Service (CRS) simulates DC plan replacement 
rates based on earnings, contributions, and the rate of return on plan 
balances. CRS projects savings for households that begin saving at age 
25, 35, or 45. The study estimates 2004 earnings using the March 2005 
CPS as starting wages, and assumes that households experience an annual 
wage growth rate of 1.1 percent. Households are randomly assigned a 6 
percent, 8 percent, or 10 percent retirement plan contribution rate 
every year from their starting age until age 65. The study assumes 
households allocate 65 percent of their retirement account assets to 
Standard & Poor's 500 index of stocks from ages 25 to 34, 60 percent to 
stocks from ages 35 to 44, 55 percent to stocks from ages 45 to 54, and 
50 percent to stocks from age 55 and above, with the remaining 
portfolio assets invested in AAA-rated corporate bonds. A Monte Carlo 
simulation based on historical returns on stocks and bonds determines 
annual rates of return. Replacement rates represent annuitized DC plan 
balances at age 65 divided by final 5-year average pay. 

After running the simulations, CRS finds variation in replacement rates 
depending on rate of return, years of saving, and earning percentile. 
In the CRS "middle estimate," the unmarried householder that saves for 
30 years, has annual household earnings in the 50th percentile, 
contributes 8 percent each year until retirement, and earns returns on 
contributions in the 50th percentile would have a 50 percent 
replacement rate (see table 11). The projected replacement rate jumps 
to 98 percent at 40 years of saving, and 22 percent at just 20 years of 
saving. Assuming a 6 percent annual contribution reduces projected 
replacement rates by about 10 to 30 percent. For example, an unmarried 
householder at the 50th percentile of annual earnings and the 50th 
percentile of returns saving for 40 years is projected to have a 
replacement rate of 72 percent at a 6 percent annual contribution (see 
table 12). All CRS estimates, however, exceed those we report in 
projections in this report, in part because CRS assumes constant 
participation in, and contributions to, a DC plan. In addition, CRS 
calculated annuity equivalents of accumulated DC balances based on 
current annuity prices; for younger workers retiring several decades 
into the future, we would expect the price of a given level of annuity 
income to be higher than today's levels because of longer life 
expectancies. This would lower the replacement rates for any projected 
lump sum. 

Table 11: Retirement Savings and Income Replacement Rates for Unmarried 
Householders, Annual Total Contributions Equal to 8 Percent of 
Household Earnings and 50th Percentile of Returns (2004 Dollars): 

Baseline replacement rates (percent); 
Age household begins saving: 25; 
Annual household earnings percentile: 75[TH]: .91; 
Annual household earnings percentile: 50[TH]: .98; 
Annual household earnings percentile: 25[TH]: 1.05. 

Baseline replacement rates (percent); 
Age household begins saving: 35; 
Annual household earnings percentile: 75[TH]: .47; 
Annual household earnings percentile: 50[TH]: .50; 
Annual household earnings percentile: 25[TH]: .53. 

Baseline replacement rates (percent); 
Age household begins saving: 45; 
Annual household earnings percentile: 75[TH]: .21; 
Annual household earnings percentile: 50[TH]: .22; 
Annual household earnings percentile: 25[TH]: .24. 

Source: Patrick Purcell, and Debra B. Whitman. "Retirement Savings: How 
Much Will Workers Have When They Retire?" Congressional Research 
Service, January 29, 2007. 

Note: Replacement rates represent annuitized DC plan balances at age 65 
divided by final 5-year average pay. 

[End of table] 

Table 12: Retirement Savings and Income Replacement Rates for Unmarried 
Householders, Annual Total Contributions Equal to 6 Percent of 
Household Earnings and 50th Percentile of Returns (2004 dollars): 

Baseline replacement rates (percent); 
Age household begins saving: 25; 
Annual household earnings percentile: 75[TH]: .67; 
Annual household earnings percentile: 50[TH]: .72; 
Annual household earnings percentile: 25[TH]: .77. 

Baseline replacement rates (percent); 
Age household begins saving: 35; 
Annual household earnings percentile: 75[TH]: .35; 
Annual household earnings percentile: 50[TH]: .37; 
Annual household earnings percentile: 25[TH]: .40. 

Baseline replacement rates (percent); 
Age household begins saving: 45; 
Annual household earnings percentile: 75[TH]: .16; 
Annual household earnings percentile: 50[TH]: .17; 
Annual household earnings percentile: 25[TH]: .18. 

Source: Purcell and Whitman, 2007. 

Note: Replacement rates represent annuitized DC plan balances at age 65 
divided by final 5-year average pay. 

[End of table] 

A 2005 study, by Sarah Holden of ICI and Jack VanDerhei of EBRI, 
simulates, as a baseline scenario, retirement savings at age 65 for a 
group in their 20s and 30s in the year 2000. The baseline assumes 
workers are continuously covered by a DC plan throughout their career, 
and that workers will continuously participate. However, the authors 
also run the model assuming this group will have participation rates 
similar to current rates by allowing workers to not be covered by, 
participate in, or contribute to a DC plan. Their model also 
incorporates the possibility that a participant might cash out a DC 
plan balance upon leaving a job. Replacement rates are calculated by 
earnings quartile for participants retiring between 2035 and 2039 as 
the annuity value of age-65 plan balances divided by final 5-year 
average pay. 

The EBRI/ICI baseline projections, starting with a sample of plan 
participants, show a median replacement rate of 51 percent for the 
lowest earnings quartile and 67 percent for the highest. (See table 
13). The authors analyze the effect of other plan or behavioral 
assumptions. For example, replacement rates fall significantly when the 
projections relax the assumption of continuous ongoing eligibility for 
a 401(k) plan, although they remain higher than our projections, 
perhaps because the projections start with current participants and 
assume continuous employment. When the authors include nonparticipants 
and assume automatic enrollment with a 6 percent employee contribution 
and investment of assets in a life cycle fund, replacement rates rise 
significantly from projections without automatic enrollment. Although 
they project a larger effect on replacement rates resulting from 
automatic enrollment than our projections show, EBRI/ICI similarly 
shows a greater increase in savings for lower-income workers. 

Table 13: Median Replacement Rates from DC Plan Balances for Workers 
Turning 65 between 2030 and 2039, by Income Quartile: 

Replacement rates. 

Income quartile: 1[ST]; 
Baseline, assuming continuous eligibility, participants only: .51; 
Without continuous eligibility: .23; 
With continuous eligibility, including non-participants: .23; 
With continuous eligibility and automatic enrollment, including 
nonparticipants: .52. 

Income quartile: 2[ND]; 
Baseline, assuming continuous eligibility, participants only: .54; 
Without continuous eligibility: .23; 
With continuous eligibility, including non-participants: .33; 
With continuous eligibility and automatic enrollment, including 
nonparticipants: .54. 

Income quartile: 3[RD]; 
Baseline, assuming continuous eligibility, participants only: .59; 
Without continuous eligibility: .25; 
With continuous eligibility, including non-participants: .43; 
With continuous eligibility and automatic enrollment, including 
nonparticipants: .57. 

Income quartile: 4[TH]; 
Baseline, assuming continuous eligibility, participants only: .67; 
Without continuous eligibility: .28; 
With continuous eligibility, including non-participants: .56; 
With continuous eligibility and automatic enrollment, including 
nonparticipants: .63. 

Source: Sarah Holden and Jack VanDerhei. "The Influence of Automatic 
Enrollment, Catch-Up, and IRA Contributions on 401(k) Accumulations at 
Retirement." Investment Company Institute Perspective, vol. 11, No. 2, 
July 2005); co-published as EBRI Issue Brief No. 283 (Washington, D.C.: 
Jul. 2005). 

Note: Replacement rates equal income generated from DC savings divided 
by final 5-year average salary. 

[End of table] 

A forthcoming study by Poterba, Venti, and Wise uses the Survey of 
Income and Program Participation (SIPP) to project DC plan balances at 
age 65. In order to project participation, the authors assume that DC 
plan sponsorship will continue to grow, although more slowly than 
during recent decades. They calculate participation by earnings deciles 
within 5-year age intervals. The authors assume that 60 percent of plan 
contributions are allocated to large capitalization equities, and 40 
percent to corporate bonds, and assume an average nominal rate of 
return of 12 percent for equities and 6 percent for corporate bonds. In 
addition, the authors run a projection assuming the rate of return on 
equities is 300 basis points less than the historical rate. They 
determine a person's likelihood of DC plan participation based on age, 
cohort, and earnings, as well as the probability of cashing out an 
existing DC plan balance when someone leaves a job. The authors 
simulate earnings histories based on data from the Health and 
Retirement Study (HRS), and impute earnings for younger cohorts for 
which data are not available. They assume an annual combined employee- 
employer contribution rate of 10 percent for each year an individual 
participates, and do not account for increases in annual contributions 
or changes made to DC plans in the Pension Protection Act, such as a 
possible increase in participation by automatically enrolling 
employees. 

The authors find retirement savings for individuals retiring by decade 
between 2000 and 2040 by lifetime earning deciles. For workers in the 
fifth earnings decile retiring in 2030 at age 65, the authors project a 
mean DC plan balance of $272,135 in 2000 dollars, and $895,179 for the 
highest earning decile (see table 14). Earners in the lowest and second 
deciles, however, project average balances of $1,372, and $21,917. The 
projected average DC plan assets for 2030 retirees fall to $179,540 for 
the fifth decile of earnings, $614,789 for the highest decile, and $810 
for the lowest decile when the authors assume an annual rate of return 
300 basis points below the historic rate of return (see table 15). 

Table 14: Mean Projected DC Plan Assets for Cohorts Retiring in 2000, 
2010, 2020, 2030, and 2040, by Lifetime Earnings Deciles (in 2000 
dollars): 

Baseline retirement savings ($). 

Cohort: 2000; 
Lifetime earning decile: 1[ST]: 0; 
Lifetime earning decile: 5[TH]: 19,437; 
Lifetime earning decile: 10[TH]: 166,405. 

Cohort: 2010; 
Lifetime earning decile: 1[ST]: 158; 
Lifetime earning decile: 5[TH]: 28,367; 
Lifetime earning decile: 10[TH]: 343,137. 

Cohort: 2020; 
Lifetime earning decile: 1[ST]: 366; 
Lifetime earning decile: 5[TH]: 166,268; 
Lifetime earning decile: 10[TH]: 577,632. 

Cohort: 2030; 
Lifetime earning decile: 1[ST]: 1,372; 
Lifetime earning decile: 5[TH]: 272,135; 
Lifetime earning decile: 10[TH]: 895,179. 

Cohort: 2040; 
Lifetime earning decile: 1[ST]: 3,688; 
Lifetime earning decile: 5[TH]: 489,558; 
Lifetime earning decile: 10[TH]: 1,242,580. 

Source: James Poterba, Steven F. Venti and David A. Wise. "New 
Estimates of the Path of 401(k) Assets," forthcoming in J. Poterba, 
ed., Tax Policy and the Economy, vol. 22 (Chicago: University of 
Chicago Press, 2008). 

[End of table] 

Table 15: Mean Projected DC Plan Assets for Cohorts Retiring in 2000, 
2010, 2020, 2030, and 2040, Assuming Rate of Return on Equities is 300 
Basis Points Less than Historic Rate, by Lifetime Earnings Deciles (in 
2000 dollars): 

Retirement savings (2000 dollars). 

Cohort: 2000; 
Lifetime earning decile: 1[ST]: 0; 
Lifetime earning decile: 5[TH]: 19,437; 
Lifetime earning decile: 10[TH]: 166,405. 

Cohort: 2010; 
Lifetime earning decile: 1[ST]: 147; 
Lifetime earning decile: 5[TH]: 75,555; 
Lifetime earning decile: 10[TH]: 315,294. 

Cohort: 2020; 
Lifetime earning decile: 1[ST]: 335; 
Lifetime earning decile: 5[TH]: 128,920; 
Lifetime earning decile: 10[TH]: 454,171. 

Cohort: 2030; 
Lifetime earning decile: 1[ST]: 810; 
Lifetime earning decile: 5[TH]: 179,540; 
Lifetime earning decile: 10[TH]: 614,789. 

Cohort: 2040; 
Lifetime earning decile: 1[ST]: 2,072; 
Lifetime earning decile: 5[TH]: 292,902; 
Lifetime earning decile: 10[TH]: 785,150. 

Source: Poterba, Venti, and Wise, forthcoming 2008. 

[End of table] 

Finally, a 2007 study by William Even and David Macpherson estimates 
replacement rate for those continuously enrolled in a DC plan between 
36 and 65 years of age. The authors simulate a sample using the SCF, 
and generate an age earnings profile for their sample using data on 
pension-covered workers in the 1989 SCF. The authors also use the SCF 
to generate annual contributions to DC plans, which are estimated using 
a person's earnings, age, education, gender, race, ethnicity, martial 
status, union coverage, and firm size. The authors also create an 
artificial sample for workers who are predicted to be eligible for a DC 
plan, but choose not to participate. Finally, the authors assume three 
different rates of return on pension contributions: a 3 percent rate of 
return based on historical returns on government bonds; a historic 
returns portfolio based on an account mix of 75 percent in stocks split 
between large and small capital equities, and 25 percent split between 
long term corporate bonds, long-term government bonds, midterm 
government bonds, and Treasury bills; a 6.5 percent real rate of 
returns based on the average real rate of return on DC plans from 1985 
to 1994 for plans with over 100 participants. In calculating annuity 
rates, the authors rely on mortality tables for group annuitants as 
opposed to the population as a whole, and do not include the charge the 
company makes for marketing and administrative expenses. 

The authors find that replacement rates vary by income distribution. 
For example, low-income workers who are continuously enrolled in a DC 
plan at the median replacement rate distribution are estimated to have 
a 30 percent replacement rate. (see table 16) The average replacement 
rate for such workers is 44 percent. Middle-income and high-income 
workers have median replacement rates 31 percent and 35 percent 
respectively. The authors' estimates are likely higher than ours 
because the authors assume continuous enrollment. 

Table 16: Replacement Rates by Income: 

Income: Low; 
Replacement rate: .30. 

Income: Middle; 
Replacement rate: .31. 

Income: High; 
Replacement rate: .35. 

Source: William Even and David Macpherson. "Defined Contribution Plans 
and the Distribution of Pension Wealth." Industrial Relations, Vol. 46, 
Number 3. July, 2007: 

[End of table] 

[End of section] 

Appendix III: GAO Contact and Staff Acknowledgments: 

Contact: 

Barbara D. Bovbjerg (202) 512-7215: 

Staff Acknowledgments: 

In addition to the contact above, Charles A. Jeszeck, Mark M. Glickman, 
Katherine Freeman, Leo Chyi, Charles J. Ford, Charles Willson, Edward 
Nannenhorn, Mark Ramage, Joe Applebaum, and Craig Winslow made 
important contributions to this report. 

[End of section] 

Footnotes: 

[1] Further, the Social Security normal retirement age for receiving 
full benefits has begun rising from age 65 until reaching age 67 
starting in 2027, a fact that will reduce benefits, relative to current 
rules, for those retiring at a given age. 

[2] The SCF is conducted every 3 years to provide detailed information 
on the balance sheet, pension, income, and other demographic 
characteristics of U.S. families. The 2004 survey is the most recently 
published survey. Data from the SCF are widely used by branches of the 
U.S. government and major economic research centers. Further detail 
about the SCF and GAO's analysis can be found in app. I. 

[3] A typical "final average pay" plan might set annual benefits equal 
to 1.5 percent of the average of the employee's final 5 years of 
earnings multiplied by the employee's tenure at the firm in years. 

[4] Beginning in 2006, plans were permitted to allow employees to 
designate some contributions to Roth 401(k) plans, which are not 
excluded from current income but allow for tax-free withdrawals in 
retirement. Economic Growth and Tax Relief Reconciliation Act of 2001, 
Pub. L. No. 107-16, § 617, 115 Stat. 38, 103-06 (codified at 26 U.S.C. 
§ 402A). 

[5] 26 U.S.C. § 72(t). The Internal Revenue Code (IRC) sets limits on 
annual contributions to DC plans by both employees and employers. 26 
U.S.C. § 415(c). In 2007, an employee may make up to $15,500 in tax- 
deductible contributions into a DC plan, and employee and employer 
combined contributions cannot exceed $45,000. A worker age 50 or older 
may contribute an additional $5,000 in annual "catch-up" contributions. 
The IRC exempts distributions from DC plans from an additional 10 
percent tax if taken for certain purposes. 26 U.S.C. § 72(t)(2). For 
example, if the employee becomes disabled, needs funds for medical 
purposes, or if the distribution is taken upon separation of service at 
age 55, the additional tax does not apply. 

[6] This tax expenditure includes Keogh plans ($10 billion), 401(k) 
plans ($41 billion), employee stock ownership plans ($2 billion), and 
the saver's credit ($1 billion). Summing these figures does not take 
into account any interactions. In addition, the tax expenditure for DB 
plans measured $49 billion and for IRAs measured $4 billion. 

[7] The saver's credit is a credit against federal income tax available 
to low-and middle-income taxpayers based on their qualified 
contributions to 401(k) and other retirement savings plans and to IRAs. 
26 U.S.C. § 25B. The Pension Protection Act of 2006 made the saver's 
credit permanent and indexed qualifying taxable income levels for 
inflation. Pub. L. No. 109-280 §§ 812 and 833(a), 120 Stat. 997, 1003- 
04. 

[8] About 87 percent of all 401(k) plans generally allow participants 
to choose how much to invest, within federal limits, and to select from 
a menu of diversified investment options selected by the employer 
sponsoring the plan, such as an assortment of mutual funds that include 
a mix of stocks, bonds, and money market investments. 

[9] A DC plan sponsor may make an automatic distribution of a 
participant's account balance when the participant leaves a job if the 
balance does not exceed $5,000. However, if the balance exceeds $1,000, 
the sponsor must automatically roll this money over into a default IRA 
or keep the balances in the plan, unless the participant explicitly 
chooses otherwise. 26 U.S.C. § 401(a)(31). 

[10] The Pension Benefit Guaranty Corporation (PBGC) guarantees the 
payment of most private sector DB pension benefits in the event of 
sponsor termination of an underfunded plan, up to certain limits. 29 
U.S.C. §§ 1322 and 1322a. For plans terminating in 2007, PBGC pays a 
maximum of $4,125 per month for a single straight-life annuity to a 65- 
year-old retiree, with lower guarantees for younger retirees. 

[11] See Craig Copeland, "Individual Account Retirement Plans: An 
Analysis of the 2004 Survey of Consumer Finances," Employee Benefit 
Research Institute Issue Brief, No. 293 (Washington, D.C.: May 2006). 

[12] A typical employer match might, for example, equal 50 percent of 
employee contributions on the first 6 percent of deferred employee 
salary. 

[13] Some employers sponsor cash balance plans, which have some 
characteristics of both DB and DC plans. While cash balance plans 
express accrued benefits in terms of a lump-sum balance, they are DB 
plans in which benefits are determined by formula. For more on cash 
balance plans, see GAO, Private Pensions: Information on Cash Balance 
Pension Plans, GAO-06-42 (Washington, D.C.: Nov. 3, 2005); Cash Balance 
Plans: Implications for Retirement Income, HEHS-00-207 (Washington, 
D.C.: Sept. 29, 2000); and Private Pensions: Implications of 
Conversions to Cash Balance Plans, HEHS-00-185 (Washington, D.C.: Sept. 
29, 2000). 

[14] Many factors affect how much a person will need: age at 
retirement, life expectancy, living expenses, health expenses, 
investment returns, inflation, and personal tolerance for risk. For 
summaries of this research through 2002 and 2003, see GAO, Private 
Pensions: Improving Worker Coverage and Benefits, GAO-02-225 
(Washington, D.C.: April 9, 2002), pp. 41-44; and Congressional Budget 
Office, Baby Boomers' Retirement Prospects: An Overview, November 2003. 

[15] For an example of such calculations, see Alicia Munnell, Anthony 
Webb, and Luke Delorme; "Retirement at Risk: A New National Retirement 
Index," Center for Retirement Research, June 2006 available at 
[hyperlink, http://www.crr.bc.edu]. In this research, replacement rates 
are calculated for households with differing lifetime income levels. 
Lower-income households are found to need a higher replacement rate due 
to lower saving during working years while higher-income households 
need a lower replacement rate. Because higher-income households tend to 
save relatively more while working, once retired they will no longer 
need to save a relatively large fraction of income. 

[16] Most studies have found that at least 50 percent of households are 
likely to have adequate retirement income, while at least 20 percent of 
households are likely not to have adequate income. Research findings 
differ on the remaining 30 percent of households. 

[17] Data reported by the Social Security Administration (SSA) do not 
consider lump-sum withdrawals from retirement accounts, such as DC 
plans or IRAs, as income, and hence these statistics do not include non-
annuitized savings. 

[18] SSA defines a low earner as someone whose career average earnings 
are about 45 percent of the national average wage index (AWI), while a 
high earner has career average earnings of about 160 percent of AWI. 

[19] These are sample estimates based on the 2004 SCF and are subject 
to sampling error. Most SCF percentage estimates have 95 percent 
confidence intervals between plus-or-minus 4 percentage points of the 
percentage itself. The majority of SCF estimates of medians and means 
have 95 percent confidence intervals within plus-or-minus 25 percent of 
the estimate itself. Exceptions to these rules are presented in 
appendix I. The SCF is designed to produce estimates for a subset of 
the household, referred to as the "primary economic unit." There are 
limitations in extending this to an individual or worker level of 
analysis. Additional details can also be found in app. I. 

[20] The 95 percent confidence interval for this estimate is from 
$27,594 to $72,406. While the range is broad, this statistic still 
effectively illustrates the limited DC savings of those currently near 
retirement age. 

[21] Employees are to be eligible for pension plans once they reach one 
year of service or the age of 21. In order to fulfill the year of 
service requirement, employees must have worked for the employer for 12 
months and fulfilled 1,000 hours of service." 26 U.S.C. §§ 401(a)(3), 
410(a)(1)(A) and (3)(A). 

[22] This participation figure includes all retirement plans, including 
both DB and DC plans. Because employees covered by DB plans are by 
definition, almost universally participants, the participation rate for 
DC plans is below the 84 percent figure. 

[23] Retirement plans rolled over from a former employer could have 
originally been either DC or DB plans. Also, any retirement plans from 
a former employer that were converted into an annuity would not be 
captured in these "total balance" statistics. 

[24] We calculated this yearly income, as an annuity equivalent using 
the Thrift Savings Plan calculator [hyperlink, http://calc.tsp.gov], 
assuming an interest rate of 5.25 percent, single life benefits 
beginning at age 65, no joint survivor benefits, and level payments. 
The 95 percent confidence interval for this annuity estimate is from 
$2,431 to $6,377 per year for life. 

[25] 26 U.S.C. §§ 401(a) and 408(d). 

[26] 26 U.S.C. § 72(t). See footnote 5 above for distributions that do 
not trigger a 10 percent early-withdrawal additional tax. 

[27] Households included in this analysis of lump-sum distributions are 
restricted to those where the head of household is under age 59 in 
order to approximate those that would be subject to penalties for 
cashing out the retirement funds. Percentages do not add up to 100 
percent because of rounding. 

[28] These rollover and cash-out figures look at all cash settlements 
from past jobs. The SCF does not specify the original account type, so 
the analysis includes all retirement plans or pensions that were 
converted into a lump-sum distribution or settlement. 

[29] See "Lump-Sum Distributions," EBRI Notes, vol. 26, No. 12 
(Washington, D.C.: Dec. 2005). 

[30] Profit Sharing/401(k) Council of America. "49TH Annual Survey: 
PSCA's Annual Survey of Profit Sharing and 401(k) Plans Reflecting the 
2005 Plan Experience" (Chicago, Ill.: 2006). 

[31] Jack VanDerhei, Sarah Holden, Craig Copeland, and Luis Alonso, 
"401(k) Plan Asset Allocation, Account Balances, and Loan Activity in 
2006." EBRI, Issue Brief, No. 308 (Washington, D.C.: Aug. 2007). 

[32] See GAO, 401(k) Pension Plans: Loan Provisions Enhance 
Participation but May Affect Income Security for Some, GAO/HEHS-98-5 
(Washington, D.C; Oct. 1, 1997). 

[33] Since some older workers may have reduced their hours or are both 
retired and working, they may be earning less than they had been 
through most of their working life. Household wealth can more 
accurately reflect their financial situation than income can. 

[34] The 95 percent confidence interval for this estimate of total DC 
balances of workers aged 60-64 at or below the median level of wealth 
is from $3,234 to $32,766. The 95 percent confidence intervals for 
these annuity estimates would be from $24 to $240 per month or from 
$285 to $2,883 per year. 

[35] Replacement rates equal the annuity value of DC plan balances 
divided by a "steady earnings" index. This index reflects career 
earnings, calibrated to the Social Security Administration's age-65 
average wage index (AWI). See app. I for further details about these 
calculations. 

[36] We computed these results using the PENSIM simulation model, which 
creates a hypothetical birth cohort and models the cohort's lives from 
birth to death, including all life events such as marriages, births, 
education and job decisions, pension coverage and behavior, and 
retirement. Our simulations and results exclude any DB and Social 
Security benefits and calculate benefits from DC plans only. Annuity 
amounts assume full annuitization of DC balances at retirement for a 
single lifetime annuity not indexed to inflation. See app. I for 
detailed information about the projections and input assumptions used 
to produce the results presented in this section. 

[37] Other studies that do similar DC plan balance projections that 
focus primarily on workers with continuous plan coverage generally find 
higher savings levels and replacement rates than we report in this 
section. We discuss these studies in more detail in app. II. 

[38] All annuity equivalents for accumulated DC savings presented in 
this section are not indexed to inflation, and hence would lose 
purchasing power over time. 

[39] This 2.9 percent annual return on assets represents a projected 
rate of return on U.S. Treasury bonds. While stocks have had greater 
long-term historical average returns, annual stock returns also exhibit 
more variance than Treasury bonds, and therefore we model DC plan 
projections under different scenarios. For more discussion of the 
appropriate rate to use in projections, see app. I. 

[40] We recalculate income quartiles for the sub-samples, and thus the 
income cut-offs for each quartile differ from those in the full-sample 
baseline. 

[41] While this scenario eliminates waiting periods for eligibility and 
participation among workers of firms that sponsor plans, it does not 
necessarily imply that workers are making a contribution to a plan each 
period, nor does it affect the likelihood that a firm will offer a DC 
plan. PENSIM determines periodic contribution levels among participants 
based on plan features and worker characteristics. See app. I. 

[42] In our baseline scenario, workers cash out account balances 36 
percent of the time when leaving a job. 

[43] The baseline projections assume that annual contribution limits 
continue to rise in the future from 2007 limits of $15,500 for employee 
contributions and $45,000 for combined employer-employee contributions. 

[44] We would expect the effect on annuity income to exceed that on 
replacement rates because working longer may also raise our measure of 
career earnings that we use in the denominator of the replacement rate 
calculation. 

[45] Pub. L. No. 109-280, § 902, 120 Stat. 780, 1033-39. 

[46] Jodi DiCenzo, "Behavioral Finance and Retirement Plan 
Contributions: How Participants Behave, and Prescriptive Solutions," 
EBRI Issue Brief No. 301 (Washington, D.C.: Jan. 2007). 

[47] See William G. Gale, J. Mark Iwry, and Peter R. Orszag, "The 
Automatic 401(k): A Simple Way to Strengthen Retirement Savings," The 
Retirement Security Project, No. 2005-1 (Washington, D.C.: Mar. 2005); 
and William G. Gale, J. Mark Iwry, and Spencer Walters, "Retirement 
Saving for Middle-and Lower-Income Households: The Pension Protection 
Act of 2006 and the Unfinished Agenda," The Retirement Security 
Project, No. 2007-1 (Washington, D.C.: 2007). 

[48] Brigitte C. Madrian and Dennis F. Shea. "The Power of Suggestion: 
Inertia in 401(k) Participation and Savings Behavior." The Quarterly 
Journal of Economics, vol. 116, Issue 4 (Nov. 2001). The authors 
compare participation rates for those hired at a large firm in the 
health-care and insurance industry, before automatic enrollment (with 1 
year of service required to be eligible), and those hired after 
automatic enrollment (with immediate eligibility for all employees, and 
all newly hired employees auto-enrolled). 

[49] Our simulation projections show that immediate plan eligibility 
and participation would increase the average annuity equivalent of 
retirement savings for workers by almost 40 percent, with replacement 
rates for lowest-income quartile workers' savings growing from 10.3 
percent in our baseline model to 25.4 percent. This simulation 
undoubtedly overstates the effect of auto enrollment provisions under 
PPA; since employers are not compelled to offer auto enrollment, it may 
apply only to new employees, and it is difficult to predict the effect 
of auto enrollment on employer's willingness to sponsor a plan. 

[50] See Rev. Rul. 98-30 and 1998-1 C.B. 1273. In 2000, additional 
rulings provided that various automatic plan features were permissible, 
including automatic enrollment of both newly hired employees and 
nonenrolled incumbent employees, and automatic enrollment in 403(b) and 
457, as well as in 401(k), plans, and allowed automatic investment. See 
Rev. Rul. 2000-8, 2000-1 C.B. 617, Rev. Rul. 2000-35, 2000-2 C.B. 138; 
Rev. Rul. 2000-33, and 2000-1 C.B. 142. However, there was no certainty 
regarding the extent to which employers would face greater exposure to 
liability for losses to employee accounts due to default investment 
choices in automatic enrollment plans. The PPA addressed this 
uncertainty by directing the Department of Labor to provide a measure 
of additional protection to plan fiduciaries that use certain kinds of 
default investments, PPA § 624, 120 Stat. 980 and 72 Fed. Reg. 60,452 
(Oct. 24, 2007), and the Department recently issued final regulations 
providing such protection. 

[51] See Profit Sharing/401(k) Council of America. "48th Annual Survey 
of Profit Sharing and 401(k) Plans Reflecting 2004 Plan Experience" 
(Chicago, Ill.: 2005); and "49th Annual Survey: PSCA's Annual Survey of 
Profit Sharing and 401(k) Plans Reflecting the 2005 Plan Experience" 
(Chicago, Ill.: 2006, August 2007). 

[52] For instance, PPA specified that the Employee Retirement Income 
Security Act of 1974 (ERISA) pre-empted state payroll or anti- 
garnishment laws that might be interpreted to prohibit automatic 
deferral of employees pay. PPA § 902(f), 120 Stat. 1039. 

[53] One recent study suggests the absence of an employer match in an 
automatic-enrollment 401(k) scheme has little impact on participation. 
Thus, the authors suggest that a non-matching automatic-enrollment type 
scheme, such as an automatic IRA, could have a significant impact on 
participation and savings. This study finds that the absence of an 
employer match in an automatic 401(k) plan only modestly decreases 
participation, implying that an automatic payroll deposit plan such as 
an IRA could be successful in increasing participation even without an 
employer match. (John Beshears et al. "The Impact of Employer Matching 
on Savings Plan Participation Under Automatic Enrollment." National 
Bureau of Economic Research, Working Paper 13352, August 2007.) 

[54] The Department of Labor's final regulation regarding default 
contributions, which is to take effect December 24, 2007, specifies 
what constitutes a "qualified default investment alternative," which 
includes life cycle funds, balanced funds, and managed accounts, as 
well as grandfathered and short-term principal preservation funds. 72 
Fed. Reg. 60,470-80. 

[55] PPA specifies that each employee will be given ninety days to 
withdraw from the automatic enrollment plan and distributions taken 
within this time period will be penalty free. PPA § 902, 120 Stat. 780, 
1033-39. 

[56] Several states have proposed or have been exploring State-Ks, 
including Washington, Maryland, Michigan, and Vermont. For additional 
information on the State-K concept, see J. Mark Iwry, "State-K: A 
Strategy for Using State-Assisted Saving to Expand Private Pension 
Coverage." Supplemental Written Testimony before the Sub-Committee on 
Long-Term Growth and Debt Reduction of the Committee on Finance, US 
Senate. Jun. 29, 2006; and J. Mark Iwry, "Growing Private Pensions: A 
Supporting Role for the States," BNA Tax Management Compensation 
Planning Journal, vol. 34, no. 12 (Washington, D.C.: Dec. 1, 2006). 

[57] H.B. 1414, 423D Gen. Assem., Reg. Sess. (Md. 2006). 

[58] See J. Mark Iwry and David C. John, Testimony Before the 
Subcommittee on Health, Employment, Labor, and Pensions of the House 
Education and Labor Committee (Nov. 8, 2007) and "Pursuing Universal 
Retirement Security Through Automatic IRAs," Retirement Security 
Project, No. 2007-2 (Washington, D.C.: 2007). The automatic IRA 
proposal differs from existing employment-based IRAs, such as the SEP- 
IRA and SIMPLE IRAs, which require employer contributions. In the SEP, 
small business employees 21 and over who have worked at least 3 of the 
past 5 years for the employer, and who earn at least $450 qualify. 26 
U.S.C. § 408(k). Contributions cannot discriminate in favor of highly 
compensated employees. SIMPLE plans can be adopted by employers who 
have no more than 100 employees who earned at least $5,000 the previous 
year. 26 U.S.C. § 408(p). In 2007, employee contributions to a SIMPLE 
IRA were limited to $10,500, including elective deferrals and employer 
deferrals. Employers must either match employee contributions dollar 
for dollar up to 3 percent of pay, with the option of contributing as 
little as 1 percent of pay for 2 out of 5 years, or make automatic non- 
elective contributions to employee accounts of 2 percent of pay. In 
addition, employers can also allow employees to make direct payroll 
contributions to IRAs without making matching contributions to the IRA. 
26 U.S.C. § 408(q). 

[59] Automatic IRA Act of 2007, S. 1141, 110th Congress (as introduced 
April 18, 2007) and H.R. 2167, 110th Congress (as introduced May 3, 
2007). 

[60] See David M. Walker, "Fiscal and Retirement Challenges." 
Presentation to the UJA Foundation of New York, September 2007, and 
Burman, Leanord E, William G. Gale and Peter R. Orszag. "The 
Administration's Savings Proposals: Preliminary Analysis." Tax 
Analysts, 2003. 

[61] See GAO, Social Security: Evaluating Reform Proposals. GAO/AIMD/ 
HEHS-00-29 (Washington, D.C; Nov. 4, 1999). See also U.S. Department of 
the Treasury, General Explanations of the Administration's Fiscal Year 
2001 Revenue Proposals (Feb. 2000), pages 49-52. 

[62] See William G. Gale, J. Mark Iwry, and Peter R. Orszag, "The 
Saver's Credit: Issues and Options," (Washington, D.C.: May 3, 2004). 
One field experiment examined savings behavior for tax filers who were 
given the opportunity to open and contribute to an account, including 
an IRA, at the time of tax filing. Some of these filers were offered a 
match with a structure similar to the saver's credit that directly 
deposited funds into the account for the filers. Before the experiment, 
about 2 percent of all filers opened such an account, and average 
contributions for all taxpayers who opened an account were about $382. 
During the experiment, about 3 percent of those not offered a match 
opened an account and contributed an average of $765, about 8 percent 
of those offered a 20 percent match opened an account and contributed 
an average of $1,102, and about 14 percent of those offered the 50 
percent match opened an account and contributed and average of $1,108. 
See Esther Duflo et al. "Saving Incentives for Low-and Middle-Income 
Families: Evidence from a Field Experiment with H&R Block." National 
Bureau of Economic Research, Working Paper 11680. Sept. 2005. 

[63] See William G. Gale, J. Mark Iwry, and Peter Orszag. "Improving 
Tax Incentives for Low-Income Savers: The Saver's Credit." Urban 
Institute, Discussion Paper No. 22, page 13, and footnote 43 
(Washington, D.C.: Jun. 2005). 

[64] Other estimates would include estimated means, medians, or 
projected annuities based on an estimated median account balance. 

[65] Additional information and details about this technique are 
available from the Federal Reserve Board, [hyperlink, 
http://www.federalreserve.gov/pubs/oss/oss2/about.html]. 

[66] Comparison data sets are the Survey of Income and Program 
Participation, the Current Population Survey, and the Department of 
Labor Form 5500 series. See Geoffrey Sanzenbacher; "Estimating Pension 
Coverage Using Different Data Sets," Center for Retirement Research, 
August 2006, for additional discussion on this topic. 

[67] For more information on PSG microsimulation models, see 
[hyperlink, http://www.polsim.com]. For more details on PENSIM, see 
Martin Holmer, Asa Janney, and Bob Cohen, PENSIM Overview, available 
from [hyperlink, http://www.polsim.com/overview.pdf]. 

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

Congressional Relations: 

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

Public Affairs: 

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