Order Code RL30922
CRS Report for Congress
Received through the CRS Web
Retirement Savings and Household Wealth:
A Summary of Recent Data
Updated June 28, 2004
Patrick J. Purcell
Specialist in Social Legislation
Domestic Social Policy Division
Congressional Research Service ˜ The Library of Congress

---

Retirement Savings and Household Wealth: A
Summary of Recent Data
Summary
Pension analysts refer to Social Security, employer-sponsored retirement plans,
and personal savings as the “three-legged stool” of retirement income, but for some
workers at least one of the legs is missing. Many workers fail to save adequately for
retirement and many are not covered by an employer-sponsored retirement plan.
Data from the Federal Reserve Board’s Survey of Consumer Finances (SCF) indicate
that only 58% of households with an employed head or spouse between the ages of
21 and 64 included at least one worker who participated in an employer-sponsored
retirement plan in 2001. Most of them participated in savings and thrift plans, in
which the worker must decide whether to contribute to the plan, and how to invest
the funds. Only 25% of households included at least one worker who was covered
by a defined benefit pension plan that guarantees a fixed monthly payment for life.
The Federal Reserve Board collects data on household assets and liabilities
through its SCF. The most recent data from this survey were collected in 2001.
According to the SCF, 47.8 million households with at least one worker between the
ages of 21 and 64 — (63%) — owned one or more retirement accounts in 2001. An
estimated 27.8 million households — (37%) — did not own a retirement savings
account of any kind. Among the households who owned a retirement savings
account of any kind in 2001, the mean value of all such accounts was $95,943. The
median value of all accounts was $27,000. The median value of the retirement
accounts held by households headed by a worker between the ages of 55 and 64 was
$55,000 in 2001. For a 65-year-old retiring in June 2004, $55,000 would be
sufficient to purchase a level, single-life annuity that would pay $400 per month,
based on the federal Thrift Savings Plan’s current annuity interest rate of 4.125%.
A balance of $55,000 would be sufficient to purchase a joint-and-survivor annuity
of $385 per month at age 65 at an interest rate of 4.125%.
The Bureau of the Census collects data on household assets and liabilities
through its Survey of Income and Program Participation (SIPP). The SIPP most
recently collected data on household assets and liabilities in late 2001. According to
the SIPP, an estimated 39.7 million households with at least one worker between the
ages of 21 and 64 — (52%) — owned one or more retirement accounts, including
IRAs, Keogh accounts, and 401(k)-type accounts in 2001. An estimated 37.3 million
such households — (48%) — did not own a retirement savings account of any kind.
Among the 39.7 million households that owned a retirement savings account of any
kind in 2001, the mean value of all such accounts was $56,800. The median value
of all the households’ accounts was $22,400.
Both surveys show that rates of retirement plan ownership and average account
balances rise steadily with income and with level of education. Homeowners and
married couples are more likely to have a retirement account than are renters or
single persons. Both surveys show that while the rate of IRA ownership among
employees of small businesses differs only a little from that of workers at large
businesses, workers at firms with more than 100 employees are much more likely to
participate in a 401(k)-type plan than are employees of smaller businesses.

---

Contents
Background: America’s Aging Population . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Congress and Retirement Saving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
Employer-Sponsored Retirement Plans . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Trends in Retirement Plan Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Retirement Savings of American Households . . . . . . . . . . . . . . . . . . . . . . . . 5
The Survey of Consumer Finances . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Defining the Terms of the Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Summary of Retirement Plan Ownership . . . . . . . . . . . . . . . . . . . . . . . 6
Retirement Account Balances by Age of Household Head . . . . . . . . . . 7
Retirement Plan Ownership and Demographic Traits . . . . . . . . . . . . . . 8
Household Net Worth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
The Survey of Income and Program Participation . . . . . . . . . . . . . . . . 12
SIPP Data on Retirement Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Policy Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Appendix: Why Do the Survey Results Differ? . . . . . . . . . . . . . . . . . . . . . . . . . 20
Why Was the Survey Conducted? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
What Questions Were Asked? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Who Was Asked the Questions? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
When Were the Questions Asked? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
List of Tables
Table 1. Participation in Retirement Plans at Current Main Job in 2001 . . . . . . . 4
Table 2. Retirement Account Balances in 2001 . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Table 3. Retirement Account Balances by Age in 2001 . . . . . . . . . . . . . . . . . . . . 8
Table 4. Ownership of Individual Retirement Accounts/Keogh Accounts
in 2001 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Table 5. Ownership of 401(k)-type Plans from Current or Past Job in 2001 . . . . 11
Table 6. Median Household Net Worth in 2001, by Age of Household
Head . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
Table 7. Household Retirement Account Balances in 2001 . . . . . . . . . . . . . . . . 14
Table 8. Household Retirement Account Balances by Age in 2001 . . . . . . . . . . 15
Table 9. Ownership of Individual Retirement Accounts/Keogh Accounts
in 2001 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Table 10. Ownership of 401(k)-type Plans from Current or Past Job in 2001 . . . 17
Table 11. Median Household Net Worth in 2001, by Age of Household
Head . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

---

Retirement Savings and Household Wealth:
A Summary of Recent Data
Background: America’s Aging Population
The aging of the American population and the impending retirement of the
“baby boom” generation will place significant strains over the next several decades
on Social Security and on retirees’ own financial resources. The decline in birth rates
since the 1960s and increases in life expectancy will result in fewer workers relative
to the number of retirees. Consequently, Social Security benefits will have to be
financed by a working population that is shrinking relative to the number of retirees
and retirees will have to stretch their savings and other assets over longer periods of
retirement than their parents and grandparents experienced.
Americans Are Living Longer Then Ever Before. The average life
expectancy of Americans born in 1960 was 69.7 years. It has been estimated that
those who were born in 2000 will live for an average of 76.4 years.1 A man who
reached age 65 in 1960 could expect to live another 13 years, while a woman who
turned 65 in 1960 had a remaining life expectancy of 16 years. A man who reached
age 65 in 2000 could expect to live another 15.6 years, while a woman who turned
65 in 2000 had a remaining life expectancy of 19.4 years. As more people live into
old age, the age-profile of the population will shift. In 1960, 16.7 million people in
the United States — 9.2% of the population — were age 65 or older. In 2000, there
were 35.0 million Americans age 65 or older, representing 12.4% of the population.
By 2025, according to projections made by the Bureau of the Census, there will be
62 million people age 65 or older, comprising 18.5% of the U.S. population.
These demographic trends will strain the components of the traditional “three-
legged stool” of retirement income: Social Security, pensions, and personal saving.
The Social Security Board of Trustees has estimated that the Social Security trust
fund will be exhausted by 2044 unless actions are taken to preserve it.2 Pensions are
the second largest source of income among the elderly, after Social Security, but only
about half of all workers in the United States have pension coverage through their
jobs. Moreover, the traditional pension that provides a lifelong annuity is becoming
less common. Today, more workers participate in savings and thrift plans than in
traditional pension plans. A key characteristic of these savings plans is that the
worker must actively participate, deciding whether to contribute to the plan, how
much to contribute, and how to invest the funds. Workers who do not choose to
save, or who save too little, may face difficult financial circumstances in retirement.
1 U.S. National Center for Health Statistics, Vital Statistics of the United States.
2 Social Security and Medicare Boards of Trustees, Status of the Social Security and
Medicare Programs: A Summary of the 2004 Annual Reports, Mar. 2004.

---

CRS-2
Congress and Retirement Saving
Congress has acted several times over the years to encourage workers to save
for retirement, mainly by allowing income taxes to be deferred on amounts that
workers or their employers contribute to certain types of savings plans established to
prepare for retirement. For example:
! The Technical Amendments Act of 1958 (P.L. 85-866) added
Internal Revenue Code Section 403(b), authorizing deferral of taxes
on employer and employee contributions to retirement plans of
religious, charitable, educational, research, and cultural institutions.
! The Self-Employed Individuals Tax Retirement Act of 1962 (P.L.
87-792) authorized tax-deferred Keogh Plans (after Representative
Eugene J. Keogh of New York) for workers who are self-employed.
! The Employee Retirement Income Security Act of 1974 (P.L.
93-406) authorized Individual Retirement Accounts (IRAs) in which
eligible contributions and investment earnings are tax-deferred.
! The Revenue Act of 1978 (P. L. 95-600) added Internal Revenue
Code Section 401(k). Employers and employees can make pre-tax
contributions to these retirement plans. Earnings are tax-deferred.
! The Revenue Act of 1978 also added Internal Revenue Code
Section 457 to permit state and local government employees to defer
income taxes on a portion of salary that is deposited into a retirement
plan.
! The Taxpayer Relief Act of 1997 (P.L. 105-34) authorized the Roth
IRA, which accepts only after-tax contributions but provides for
tax-free distributions.
! The Economic Growth and Tax Relief Reconciliation Act of 2001
(P.L. 107-16) increased the maximum contribution to an IRA to
$3,000 in 2002, 2003, and 2004; to $4,000 in 2005, 2006, and 2007,
and to $5,000 in 2008. It then will be indexed to inflation in $500
increments. Individuals age 50 and older may contribute an
additional $500 to an IRA in 2002 through 2005 and $1,000 in each
year thereafter. The law increased the maximum employee salary
deferral under I.R.C. §401(k) to $11,000 in 2002 and by an
additional $1,000 each year through 2006, when it will reach
$15,000. After 2006, the maximum deferral will be indexed to
inflation. The law raised the maximum contribution to retirement
plans under §§403(b) and 457 to the same amount as under section
401(k). People age 50 or older can make additional contributions to
retirement plans authorized under I.R.C. §§401(k), 403(b), or 457.
The maximum permissible additional contribution is $1,000 in 2002,
$2,000 in 2003, $3,000 in 2004, $4,000 in 2005, and $5,000 in 2006.
This amount is indexed to inflation in years after 2006.

---

CRS-3
Employer-Sponsored Retirement Plans
Social Security, employer-sponsored retirement plans, and personal savings are
sometimes called the “three-legged stool” of retirement income, but for many
workers at least one of the legs is missing. Coverage under Social Security is nearly
universal, but access to employer-sponsored retirement plans is limited. Data from
the Federal Reserve Board’s Survey of Consumer Finances (SCF) indicate that in
2001, 58% of households in which the head or spouse was a worker between the ages
of 21 and 64 included at least one worker who participated in an employer-sponsored
retirement plan. (See Table 1.) Forty-seven percent of households with a worker
between the ages of 21 and 64 included someone who participated in a defined
contribution plan. Only 25% of households included at least one worker who was
covered by a defined benefit plan that guarantees a fixed monthly payment for life.
Trends in Retirement Plan Design. Over the past 25 years, there has been
a shift in the distribution of retirement plans and of plan participants from defined
benefit plans to defined contribution plans. In a defined benefit or “DB” plan, the
retirement benefit is usually paid as a lifelong annuity based on the employee’s length
of service and average salary in the years just before retirement. DB plans usually
are funded by employer contributions. A defined contribution or “DC” plan is much
like a savings account maintained by the employer on behalf of each participating
employee. The employer contributes a specific dollar amount or percentage of pay,
which is invested in stocks, bonds, or other assets. The employee usually contributes
to the plan, too. In a DC plan, it is the employee who bears the investment risk. At
retirement, the balance in the account is the sum of all contributions plus interest,
dividends, and capital gains — or losses. The account balance is usually distributed
as a single lump sum. Many large employers recently have converted their traditional
DB pensions to hybrid plans that have characteristics of both DB and DC plans, the
most popular of which has been the cash balance plan. In a cash balance plan, the
benefit is defined in terms of an account balance. The employer makes contributions
to the plan and pays interest on the accumulated balance. However, these account
balances are merely bookkeeping devices. They are not individual accounts owned
by the participants. Legally, therefore, a cash balance plan is a defined benefit plan.
The Growth of Defined Contribution Plans. After Section 401(k) was
added to the IRC in 1978, DC plans overtook traditional defined benefit pensions in
number of plans, participants, and total assets. Earlier defined contribution plans had
been funded exclusively by employer contributions. In a 401(k) plan, however, both
the employer and the employee can make contributions. These characteristics leave
most of the responsibility of preparing for retirement with the employee, who must
decide whether to participate, how much to contribute, and how to invest the assets.
In 1998 and 2000, the IRS issued rulings that permit employers to enroll employees
automatically in a 401(k) plan. IRS Revenue Ruling 98-30 allows employers to
provide for automatic enrollment into 401(k) plans for newly eligible employees.
Under a so-called “negative election” workers may be automatically enrolled in their
employer’s retirement savings plan. Employees who are enrolled automatically must
be given an option to drop out of the plan. IRS Revenue Ruling 2000-8 allows
automatic enrollment in 401(k) plans of current employees who had not elected to
participate.

---

CRS-4
Table 1. Participation in Retirement Plans at Current Main Job in 2001
Number of
Any Type
Defined
Defined
Both
Household Head Characteristics Householdsa
of Planb
Contributionb
Benefitb
Typesb
Age
Under 35
21,372
48.1%
41.0%
14.6%
7.5%
35 to 44
22,440
63.1
52.7
25.6
15.2
45 to 54
19,759
64.7
49.5
33.2
18.0
55 or older
12,043
54.0
40.3
25.7
12.0
Race
White, non-Hispanic
56,993
59.8
48.6
26.1
14.9
Black, Hispanic, or Asian
18,620
51.7
40.4
19.5
8.2
Sex and Marital Status
Couple
43,791
65.0
52.5
29.9
17.5
Single Male
16,183
50.3
41.0
18.1
8.8
Single Female
15,639
45.6
35.6
15.8
5.8
Education
Did not graduate High School
8,938
32.1
24.3
12.9
5.1
High School graduate
22,404
53.8
41.7
22.7
10.6
Some college
18,776
57.3
45.8
22.9
11.4
College graduate
25,495
70.8
59.2
31.3
19.7
Household income (annual)
Under $25,000
15,450
24.9
22.8
7.9
1.6
$25,000 to $49,999
22,391
51.9
38.8
18.6
5.4
$50,000 to $74,999
15,765
72.5
59.0
31.5
18.0
$75,000 or more
22,008
76.6
65.3
37.1
26.0
Own or rent home
Own
50,887
66.5
54.2
29.7
17.3
Rent
24,726
39.9
30.9
13.8
4.8
Full time or part time worker
Not currently working
3,276
33.0
26.2
14.3
7.5
Part time
6,915
29.5
23.3
10.8
4.7
Full time (35+ hours per week)
65,422
62.1
50.1
26.4
14.4
Establishment size
Not currently working
3,276
33.0
26.2
14.3
7.5
Under 20 employees
19,288
29.4
24.3
9.1
4.0
20 to 99 employees
11,135
53.4
45.0
17.9
9.6
100 to 499 employees
12,378
67.2
54.3
25.9
13.0
500 or more employees
29,536
76.9
60.7
37.5
21.4
Covered by Union contract?
Not currently working
3,276
33.0
26.2
14.3
7.5
Union
13,539
81.7
53.9
51.3
23.5
Non-union
58,798
53.7
46.0
18.9
11.2
Have an IRA or Keogh plan?
Yes
25,238
66.8
53.8
30.3
17.2
No
50,375
53.3
43.0
21.6
11.2
Total
75,613
57.8%
46.6%
24.5%
13.2%
Source: CRS analysis of the Federal Reserve Board’s 2001 Survey of Consumer Finances.
Notes: Includes single persons as well as families.
a. Households with an employed head or spouse age 21-64, in thousands.
b. Percentage of households in which head or spouse participated, by type of plan.

---

CRS-5
Retirement Savings of American Households
With the trend away from defined benefit plans to defined contribution plans,
workers now bear much of the responsibility of preparing for retirement. Workers
whose employers offer savings or “thrift” plans such as those authorized under
Sections 401(k), 403(b), and 457 of the Internal Revenue Code can accumulate assets
on a tax-deferred basis while they are working. In addition, most people with earned
income can contribute to an IRA. In both cases, taxes are paid when the funds are
withdrawn, and a penalty may apply if the withdrawals occur before retirement.3 For
many people, the marginal income tax rate that they will face in retirement will be
lower than the rate that was applied to their earnings prior to retirement.
The Survey of Consumer Finances. The SCF is sponsored by the Board
of Governors of the Federal Reserve System in cooperation with the Department of
the Treasury. It is conducted once every three years to collect information on the
assets and liabilities of households, their pension coverage, sources and amounts of
income, and demographic characteristics. Data from the SCF are widely used by
analysts at the Federal Reserve, in other branches of government, and by scholars in
private-sector research organizations and academic institutions. Since 1992, SCF data
have been collected by the National Organization for Research at the University of
Chicago (NORC). The most recent available SCF data were collected in 2001.4 For
the 2001 SCF, members of 4,449 households were interviewed.
Defining the Terms of the Analysis. In the SCF, each household is
divided into a “primary economic unit” (PEU) and everyone else in the household.
The PEU in the SCF is the economically dominant single individual or couple in the
household together with everyone else who is economically interdependent with
them. In the tables that follow, the primary economic unit is called a household. The
tables present information on the retirement savings of households in which the head
or spouse was an employed adult between 21 and 64 years old.5 According to the
SCF there were 75.6 million households in which the head or spouse was an
employed adult between 21 and 64 years old in 2001. We restricted the analysis to
households with a worker between the ages of 21 and 64 because younger workers
have low rates of participation in retirement plans and are generally more concerned
with establishing themselves in their careers than in accumulating assets for
retirement. Workers age 65 or older are more likely than those under 65 to have
retired from their career jobs and to be working part-time. The tables do not include
the portion of retirement wealth that is represented by the present value of benefits
accrued under Social Security and employer-sponsored defined-benefit pension plans.
They include only the balances accumulated in IRAs, Keogh plans for the self-
3 In a traditional IRA, pre-tax contributions can be made only if the worker is not covered
by an employer-sponsored retirement plan or has income below amounts specified in law.
All investment earnings accrue on a tax-deferred basis. Roth IRAs accept only after-tax
contributions; however, withdrawals from a Roth IRA during retirement are tax-free.
4 For more information, see [http://www.federalreserve.gov/pubs/oss/oss2/scfindex.html].
5 The “household head” is the person named by the family for purposes of the survey.

---

CRS-6
employed, and employer-sponsored defined contribution plans, such as those
authorized under Sections 401(k), 403(b), and 457 of the tax code.
The following tables show the retirement savings of American households in
which there was at least one worker between the ages of 21 and 64. The tables show
the number of households that owned at least one retirement account as well as the
average balances held in those accounts.

Summary of Retirement Plan Ownership. According to the SCF, there
were 75.6 million households with an employed head or spouse between the ages of
21 and 64 in 2001. An estimated 47.8 million of these households (63.2%) owned
one or more retirement accounts, including IRAs, Keogh accounts, 401(k) accounts
and other employer-sponsored savings or thrift plans. (See Table 2). According to
the SCF, an estimated 36.4 million households (48.1%) owned a 401(k)-type plan,
25.2 million (33.4%) owned an IRA or Keogh plan, and 13.8 million (18.3%) owned
both an IRA/Keogh and a 401(k) plan. An estimated 27.8 million households with
at least one worker between the ages of 21 and 64 (36.8%) did not own a retirement
savings account of any kind, according to the SCF.
The data displayed in Table 2 also summarize the average account balances
among households that owned at least one retirement account. According to the
SCF, among the 47.8 million households that owned a retirement savings account of
any kind in 2001, the mean value of all such accounts was $95,943. The median
value of all the households’ accounts was $27,000. (Half of the households owned
accounts totaling more than $27,000 and half owned accounts with a total value of
less than $27,000.) The mean balance of 401(k)-type plans was reported as $68,320,
and the median balance was $20,000. The mean balance of IRA/Keogh plans was
reported as $83,240 on the SCF, and the median account balance was reported as
$21,000.
Means and Medians
The average values of retirement accounts are shown in this report in
terms of both the mean and the median values. The mean is a simple arithmetic
average.6 It is calculated by adding up the reported values of all accounts and
then dividing this total by the number of account-holders. As a measure of
central tendency — what an “average” represents — the mean is flawed
because it can be influenced by a relatively small number of unusually high or
low values. The median is another kind of average that is more representative
of the population because it is not biased by unusually high or low values. The
median is calculated by ordering all of the observed values from highest to
lowest and finding the value that lies exactly at the midpoint of the distribution.
One half of all observed values are greater than the median and the other half
are less than the median.
6 A survey weight has been assigned to each household. The weights sum to the number of
households in the U.S. The means in the tables are the weighted means for each observation.

---

CRS-7
Table 2. Retirement Account Balances in 2001
Mean
Median
Number of
Percent of
Value of
Value of
Householdsa Households Accounts Accounts
All households
75,613
100%

Owned either an IRA/Keogh or a 401(k)-type plan
47,806
63.2
All retirement accounts, all types
$95,943
$27,000
Owned a 401(k) or other type of thrift planb
36,385
48.1
All 401(k)-type accounts
68,320
20,000
All retirement accounts, all types
97,298
29,000
Owned an IRA or Keogh planc
25,238
33.4
All IRA/Keogh accounts
83,240
21,000
All retirement accounts
144,880
49,000
Owned both a 401(k) and an IRA/Keogh plan
13,817
18.3
All 401(k)-type accounts
112,589
35,000
All IRA/Keogh accounts
76,307
24,000
All retirement accounts, all types
188,896
83,000
Owned neither an IRA/Keogh nor a 401(k)-type
27,807
36.8
Source: CRS analysis of the Federal Reserve Board’s Survey of Consumer Finances.
Notes: Includes single persons as well as families.
a. Households with an employed head or spouse age 21-64, in thousands.
b. May also have owned an IRA or Keogh plan.
c. May also have owned a 401(k)-type plan.
Retirement Account Balances by Age of Household Head. A
worker’s age is an important consideration when evaluating the adequacy of his or
her retirement wealth. The more time that a worker has until reaching retirement age,
the greater will be the opportunity for additional contributions and investment
earnings to build up the account balance. Table 3 shows average retirement account
balances, categorized by the age of the household head. The SCF indicates that 10.9
million households headed by a worker 21 to 34 years old (51.2% of households in
this age group) owned one or more retirement accounts. The SCF data show mean
and median account balances for these households of $19,123 and $7,000,
respectively. Even among workers 55 to 64 years old, median retirement account
balances were not very large. The median value of the retirement accounts held by
households headed by a worker between the ages of 55 and 64 was $55,000 in 2001,
according to the SCF. For a 65-year-old retiring in June 2004, $55,000 would be
sufficient to purchase a level, single-life annuity that would pay $400 per month,
based on the federal Thrift Savings Plan’s current annuity interest rate of 4.125%.7
A balance of $55,000 would be sufficient to purchase a joint-and-survivor annuity
of $385 per month at age 65 at an interest rate of 4.125%.
7 Example is based on 50% annuity for a spouse three years younger than the annuitant.

---

CRS-8
Of course, the median account values reflect only balances of households that
own a retirement account. The data collected by the SCF show a median account
balance of $55,000 among households headed by a worker 55 to 64 years old. This
means that half of all retirement account owners in this age category had total
account balances of more than $55,000 and half had account balances of less than
$55,000. However, when we take into account those who had no retirement accounts,
and thus had total retirement account balances of zero, 7.8 million households headed
by a worker 55 to 64 years old — 65.1% of households in this age group — had total
retirement account balances of $55,000 or less in 2001.
Table 3. Retirement Account Balances by Age in 2001
Households
Percent
Mean
Median
Number of
with
with
Value, All Value, All
Age of Household Head
Householdsa
Accounts
Accounts Accountsb Accountsb
21 to 34 years old
Value of all retirement accounts
21,372
10,944
51.2%
$19,123
$7,000
35 to 44
Value of all retirement accounts
22,439
14,815
66.0
65,583
29,000
45 to 54
Value of all retirement accounts
19,759
13,643
69.0
132,741
48,000
55 or older
Value of all retirement accounts
12,043
8,403
69.8
189,779
55,000
All households
Value of all retirement accounts
75,613
47,806
63.2
95,943
27,000
Source: CRS analysis of the Federal Reserve Board’s Survey of Consumer Finances.
Notes: Includes single persons as well as families. Includes defined contribution plan account balances from both
current and past employment.
a. Households with an employed head or spouse age 21-64, in thousands.
b. Means and medians reflect balances in all types of plans.
Retirement Plan Ownership and Demographic Traits. Table 4 and
Table 5 show the rate of ownership and average account balances for IRAs and
Keogh plans in 2001 as reported on the SCF. The rates of ownership and average
account balances are shown in these tables in relation to the demographic
characteristics of the household head. In summary:
! IRA ownership and average account balances rose steadily with
household income;
! Households headed by a white worker were twice as like as those in
which the head was a non-white worker to own an IRA;
! Married couples were almost twice as likely as unmarried
individuals to have owned an IRA, in part because these data

---

CRS-9
measure retirement plan ownership at the household level, and many
married couples include two workers;
! IRA ownership rose with education, and college graduates were
much more likely than those who had not graduated from college to
own an IRA;
! homeowners were about three times as likely as renters to own an
IRA;
! IRA ownership differed little between full-time workers and part-
time workers;
! IRA ownership among employees of small businesses differed only
a little from that of workers at large businesses;
! Union membership appears to have little relationship to IRA
ownership, and
! Households where one or more workers participated in a 401(k)
were more likely to own an IRA than households in which no one
participated in a 401(k) plan.
Many of the relationships between demographic characteristics and 401(k)
participation were similar to the relationships between demographic characteristics
and IRA ownership, but there were some differences. For example, while IRA
ownership increased in each age group, 401(k) ownership dropped in the 55-or-older
category. This can likely be attributed in part to the large number of people who roll
over 401(k) account balances into an IRA when they retire. While 401(k) ownership
was greater among households headed by a white worker than a non-white worker,
the difference was not as great as the difference in the rate of IRA ownership by race.
Likewise, while 401(k) ownership was greater among couples than singles, the
difference was not as great as the difference in the rate of IRA ownership by marital
status. Finally, while IRA ownership differed little among employees of small firms
and large firms, 401(k) ownership was higher among workers at large businesses than
among workers at small businesses by a ratio of about 2:1.

---

CRS-10
Table 4. Ownership of Individual Retirement Accounts/Keogh Accounts in 2001
Number of
Percent that Own Mean Balance in Median balance in
Households with
an IRA or Keogh
All IRA/Keogh
All IRA/Keogh
Household Head Characteristics
One or More
Planb
Plansb
Plansb
Workersa
Age
Under 35
21,372
20.3%
$ 15,081
$ 7,000
35 to 44
22,440
30.9
43,277
15,000
45 to 54
19,759
41.0
102,943
40,000
55 or older
12,043
48.6
153,961
45,000
Race
White, non-Hispanic
56,993
38.5
90,926
25,000
Black, Hispanic, or Asian
18,620
17.7
31,926
7,500
Sex and Marital Status
Married
43,791
42.1
98,521
28,000
Single Male
16,182
21.3
47,481
12,000
Single Female
15,639
21.5
36,172
7,500
Education
Did not graduate High School
8,938
12.3
24,558
10,000
High School graduate
22,404
22.2
53,718
15,000
Some college
18,776
27.0
50,145
14,000
College graduate
25,495
55.2
110,178
31,000
Household income (annual)
Under $25,000
15,449
10.1
16,563
5,000
$25,000 to $49,999
22,391
23.3
30,486
10,000
$50,000 to $74,999
15,765
33.4
60,479
16,000
$75,000 or more
22,008
60.0
121,055
40,000
Own or rent home
Own
50,887
42.2
93,538
27,000
Rent
24,726
15.3
24,640
7,000
Full time or part time worker

Not currently working
3,276
32.4
165,397
43,000
Full time (35+ hours per week)
65,422
33.5
76,235
20,000
Part-time
6,914
32.3
113,121
40,000
Establishment size
Not currently working
3,276
32.4
165,397
43,000
Under 20 employees
19,288
36.5
110,187
30,000
20 to 99 employees
11,135
31.9
57,069
16,500
100 to 499 employees
12,378
27.6
72,041
20,000
500 or more employees
29,536
34.4
68,915
16,000
Covered by Union contract?
Not currently working
3,276
32.4
165,397
43,000
Union
13,539
29.9
50,490
10,000
Non-union
58,797
34.2
85,492
24,000
Have a 401(k) type plan?
Yes
36,385
38.0
76,307
24,000
No
39,228
29.1
91,627
20,000
Total
75,613
33.4%
$ 83,240
$ 21,000
Source: CRS analysis of the Federal Reserve Board’s 2001 Survey of Consumer Finances.
Notes: Includes single persons as well as families.
a. Households with an employed head or spouse age 21-64, in thousands.
b. Percentage of households in which head or spouse participated in plan, by type of plan.

---

CRS-11
Table 5. Ownership of 401(k)-type Plans from Current or Past Job in 2001
Number of
Percent that Own
Households with
Mean Balance in Median Balance
One or More DC
Household Head Characteristics
One or More
All Such Plans
in All Such Plans
Plansb
Workersa
Age
Under 25
21,372
41.8%
$16,098
$ 6,000
35 to 44
22,440
54.5
54,920
24,000
45 to 54
19,759
51.4
96,271
30,000
55 or older
12,043
42.2
136,518
43,000
Race
White, non-Hispanic
56,993
50.1
77,863
24,000
Black, Hispanic, or Asian
18,620
42.1
33,547
9,000
Sex and Marital Status
Married
43,791
53.8
85,816
27,000
Single Male
16,182
43.0
40,537
10,000
Single Female
15,639
37.7
31,197
8,000
Education
Did not graduate High School
8,938
25.9
32,038
6,000
High School graduate
22,404
43.6
33,838
12,000
Some college
18,776
47.3
49,564
16,000
College graduate
25,495
60.5
106,383
31,000
Household income (annual)
Under $25,000
15,449
20.2
9,076
2,200
$25,000 to $49,999
22,391
40.3
21,795
7,800
$50,000 to $74,999
15,765
60.4
41,538
21,000
$75,000 or more
22,008
66.9
126,703
50,000
Own or rent home
Own
50,887
55.8
81,542
27,000
Rent
24,726
32.3
21,244
4,650
Full time or part time worker
Not currently working
3,276
31.5
64,670
25,000
Full time (35+ hours per week)
65,422
51.2
68,116
20,000
Part time
6,914
26.5
74,101
11,000
Establishment size
Not currently working
3,276
31.5
64,670
25,000
Under 20 employees
19,288
26.2
95,334
18,000
20 to 99 employees
11,135
46.3
52,653
12,000
100 to 499 employees
12,378
55.9
54,909
16,000
500 or more employees
29,536
61.7
70,581
23,000
Covered by Union contract?
Not currently working
3,276
31.5
64,670
25,000
Union
13,539
54.2
52,204
23,000
Non-union
58,797
47.7
72,673
18,000
Have an IRA or Keogh plan?
Yes
25,238
54.7
112,589
35,000
No
50,375
44.8
41,217
11,000
Total
75,613
48.1%
$68,320
$20,000
Source: CRS analysis of the Federal Reserve Board’s 2001 Survey of Consumer Finances.
Notes: Includes single persons as well as families.
a. Households with an employed head or spouse age 21-64, in thousands.
b. Percentage of households in which head or spouse participated in plan, by type of plan.

---

CRS-12
Household Net Worth. Most households have forms of wealth other than
retirement accounts on which they will be able to draw during retirement. More than
95% of workers in the United States are covered by Social Security, and about a
quarter of households included a worker who participated in a defined-benefit
pension plan in 2001.8 In addition, many workers have assets that could be used to
pay expenses during retirement. For example, the most valuable asset owned by
most people is their home, and some may find when they are older that they prefer
to live in a smaller house or apartment, or they may choose to move to an area where
property taxes and other living expenses are lower than where they lived during their
working years. In addition to equity in their homes, many individuals have financial
assets, equity in businesses, real estate, or other valuables that can either provide a
stream of income through interest, dividends, or rents, or that can be fully or partially
liquidated to finance their consumption needs during retirement. “Net worth” is the
difference between a household’s assets and liabilities. The median net worth of all
households in the United States in 2001, categorized by the age of the household
head, is shown in Table 6.
Table 6. Median Household Net Worth in 2001,
by Age of Household Head
(All households)
Age of Household Head
Amount
Under 35 years old
$11,600
35 to 44
77,600
45 to 54
132,000
55 to 64
181,500
65 to 74
176,300
75 or older
151,400
All households
86,100
Source: Federal Reserve Bulletin, January 2003, vol. 89, no. 1.
The Survey of Income and Program Participation. The Bureau of the
Census collects data on household assets and liabilities through its Survey of Income
and Program Participation (SIPP). The households selected to participate in the
SIPP are a nationally representative sample of the civilian, noninstitutionalized
population of the United States. The SIPP is a longitudinal survey, meaning that it
measures changes in the economic and demographic characteristics of individuals
and households over time. Households that participate in the survey are interviewed
once every four months over a period that ranges from 2½ years to four years. At
each interview, respondents are asked to provide information covering the four
months since the previous interview. This four-month span is called the “reference
period” for the interview. While it was designed as a longitudinal survey, the SIPP
also can be used to study characteristics of the population at a point in time
8 See Table 1.

---

CRS-13
(cross-sectional analysis) by looking at the data from a particular four-month
reference period.
The SIPP is an important source of information about the demographic and
economic status of United States residents. By collecting data on labor force
participation, sources of income, and participation in federal and state programs, the
SIPP provides a wealth of information about government programs and their effects
on the economic situations of families and individuals. For example, the SIPP can
be used to examine participation in means-tested programs (such as Temporary Aid
to Needy Families, Food Stamps, and Medicaid) and in programs that are not means-
tested (such as Social Security and Medicare). In addition to asking about amounts
and sources of income, the SIPP collects information on asset ownership to provide
a more complete picture of the total economic resources available to families and
individuals. The SIPP data on household wealth and asset ownership presented in
this report are the most recent available from the Bureau of the Census.
According to the Census Bureau, the main objective of SIPP is to provide
accurate and comprehensive information about the income and program participation
of individuals and families in the United States, and about the principal determinants
of income and program participation. The large sample size of the SIPP allows it to
be used to analyze the demographic and economic characteristics of various sub-
populations of interest to policy makers. The 2001 panel of the SIPP began with
35,000 households. The third wave of the 2001 panel of the SIPP — on which the
analysis in this report is based — included more than 27,000 households. The SIPP
collects detailed information on income from all sources, on taxes paid, and on
participation in government transfer programs. More importantly for purpose of this
report, the SIPP collects data on household assets and liabilities, including ownership
of IRAs and Keogh accounts for the self-employed, and participation in employer-
sponsored retirement savings plans, such as those authorized under Sections 401(k),
403(b), and 457 of the Internal Revenue Code.9
SIPP Data on Retirement Savings. The SIPP most recently collected data
on household assets and liabilities in late 2001. In comparing the data from the SIPP
with that collected on the SCF, it is important to note that the Census Bureau and the
Federal Reserve Board define families differently. The Census Bureau defines a
family as any two or more people related by birth, marriage, or adoption who live
together. Households may contain more than one family, but all members of the
household who are related by birth, marriage, or adoption are members of the same
family. In the SCF, each household is divided into a “primary economic unit” (PEU)
and everyone else in the household. The PEU is the economically dominant single
individual or couple in the household. It is roughly equivalent to a family in the
SIPP. Although the Federal Reserve Board sometimes refers to the PEU as a family,
Fed staff have noted that “the term is more comparable to the U.S. Bureau of the
Census definition of ‘households’ than to their use of ‘families.’”10 The SCF and the
SIPP match up well with each other in counting the number of U.S. households.
According to the SIPP, there were 77 million households with an employed head or
9 See [http://www.sipp.census.gov/sipp/intro.html].
10 Aizcorbe, Kennickell, and Moore, Federal Reserve Bulletin, vol. 89, no.1 (2003), p. 2.

---

CRS-14
spouse between the ages of 21 and 64 in 2001, while according to the SCF there were
75.6 million such households in 2001.
Summary of Account Ownership. According to the SIPP, an estimated
39.7 million households headed by a worker 21 to 64 years old (51.5% of all such
households) owned one or more retirement accounts, including IRAs, Keogh
accounts, 401(k) accounts and other employer-sponsored savings or thrift plans in
2001. An estimated 31.8 million households (41.2%) owned a 401(k)-type plan, 19.5
million (25.4%) owned an IRA or Keogh plan (mostly IRAs), and 11.6 million
(15.1%) owned both an IRA/Keogh and a 401(k)plan. An estimated 37.3 million
households with at least one worker between the ages of 21 and 64 (48.5%) did not
own a retirement savings account of any kind.11 Among the 39.7 million households
that owned a retirement savings account of any kind in 2001, the mean value of all
such accounts was $56,764. For households with more than one account, this is the
mean value of all accounts summed together. The median value of all the
households’ accounts was $22,425. (Half of the households owned accounts totaling
more than $22,425 and half owned accounts with a total value of less than $22,425.)
Mean retirement account balances reported on the SIPP were substantially lower than
those reported on the SCF, but the median account balances on the two surveys
differed by much less.
Table 7. Household Retirement Account Balances in 2001
Mean
Median
Number of
Percent of
Value of
Value of
Householdsa Households Accounts Accounts
All households
76,997
100%
Owned either an IRA/Keogh or a 401(k)-type plan
39,683
51.5
All retirement accounts, all types
$56,764
$22,425
Owned a 401(k) or other type of thrift planb
31,756
41.2
All 401(k)-type accounts
43,464
17,000
All retirement accounts, all types
60,149
25,000
Owned an IRA or Keogh planc
19,520
25.4
All IRA/Keogh accounts
44,700
18,000
All retirement accounts
80,194
38,000
Owned both a 401(k) and an IRA/Keogh plan
11,593
15.1
All 401(k)-type accounts
59,782
29,000
All IRA/Keogh accounts
45,705
19,000
All retirement accounts, all types
105,490
63,000
Owned neither an IRA/Keogh nor a 401(k)-type
37,314
48.5
Source: CRS analysis of the 2001panel of the Census Bureau’s Survey of Income and Program Participation.
Notes: Includes single persons as well as families.
a. Households with an employed head or spouse 21 to 64 years old, in thousands.
b. May also have owned an IRA or Keogh plan.
c. May also have owned a 401(k)-type plan.
11 These figures also indicate that 20.2 million households had only a 401(k)-type plan:
(31.8-11.6=20.2). An estimated 7.9 million households had only an IRA: (19.5-11.6=7.9).

---

CRS-15
Account Balances, by Age . The SIPP and the CPS show very similar
mean and median account balances among households headed by someone between
the ages of 21 and 34, although the SIPP reports fewer households owning retirement
accounts in this age group. According to the SIPP, 9.1 million households — 42.4%
of households headed by a person 21 to 34 years old — owned at least one retirement
account. The SIPP reports mean and median account balances of $18,000 and
$7,300, respectively. The data collected on the SIPP show a median account balance
of $45,000 among households headed by a worker 55 to 64 years old. However,
when we take into account those who had no retirement accounts — and thus had
retirement account balances of zero — 8.1 million households headed by a worker
55 to 64 years old — 71.2% of households in this age group — had total retirement
account balances of $45,000 or less, according to the SIPP.
Table 8. Household Retirement Account Balances by Age in 2001
Percent
Households
Mean
Median
Number of
with One
Age of Household Head
that Owned
Value, All Value, All
Householdsa
or More
Accounts
Accountsb Accountsb
Accounts
21 to 34 years old
Value of all retirement accounts
21,543
9,129
42.4% $17,978
$ 7,300
35 to 44
Value of all retirement accounts
23,339
12,310
52.7
49,838
25,000
45 to 54
Value of all retirement accounts
20,750
11,713
56.4
75,470
35,500
55 or older
Value of all retirement accounts
11,364
6,531
57.5
90,485
45,000
All households
Value of all retirement accounts
76,997
39,683
51.5
56,764
22,425
Source: CRS analysis of the 2001 panel of the Census Bureau’s Survey of Income and Program Participation.
Notes: Includes single persons as well as families. Includes defined contribution plan account balances from both
current and past employment.
a. Households with an employed head or spouse age 21-64, in thousands.
b. Means and medians reflect balances in all types of plans.
Table 9 and Table 10 show the rate of ownership and average account balances for
IRAs and Keogh plans as reported on the SIPP. The rates of ownership and average
account balances are shown in these tables in relation to the demographic
characteristics of the household head. As was noted earlier, among all households
that included a worker between the ages of 21 and 64, rates of IRA ownership and
average account balances were higher on the SCF than on the SIPP. Nevertheless,
the relationship between IRA ownership and demographic characteristics are broadly
similar in the two surveys.

---

CRS-16
Table 9. Ownership of Individual Retirement Accounts/Keogh Accounts
in 2001
Number of
Percent That
Mean Balance in Median Balance
Households with
Own an IRA or
All IRA/Keogh
in All IRA/Keogh
Household Head Characteristics
One or More
a
Keogh Planb
Plans
Plans
Workers
Age
Under 35
21,543
15.4%
$ 14,150
$ 6,000
35 to 44
23,339
24.3
33,150
15,000
45 to 54
20,750
29.7
54,900
24,000
55 or older
11,364
38.3
68,560
30,000
Race
White, non-Hispanic
64,250
27.9
45,500
18,450
Black, Hispanic, or Asian
12,747
12.4
35,500
13,000
Sex and Marital Status
Married
44,672
30.1
50,240
20,000
Single Male
14,219
19.2
32,800
14,000
Single Female
18,105
18.4
32,000
12,000
Education
Did not graduate High School
7,649
5.3
25,200
10,000
High School graduate
21,146
15.5
35,300
14,000
Some college
24,860
22.7
35,700
15,000
College graduate
23,341
43.7
53,500
20,000
Household income (annual)
Under $25,000
14,650
9.6
36,500
14,000
$25,000 to $49,999
24,546
17.8
30,100
10,000
$50,000 to $74,999
17,088
28.3
38,600
16,000
$75,000 or more
20,713
43.1
56,400
24,750
Own or rent home
Own
51,792
32.3
48,100
20,000
Rent
25,205
11.1
24,100
8,000
Full time or part time worker

Not currently working
6,258
23.5
56,500
21,000
Full time (35+ hours per week)
51,878
24.9
42,800
16,500
Part-time
18,861
27.2
46,100
18,000
Establishment size
Not reported
14,716
30.0
60,800
25,500
Under 25 employees
17,826
23.2
43,500
15,000
25 to 99 employees
11,924
22.5
40,500
15,000
100 or more employees
20,758
24.8
35,900
15,000
Public sector employees
11,772
26.7
41,640
16,500
Covered by Union contract?
Yes
9,860
23.5
37,960
15,000
No
67,137
25.6
45,600
18,000
Have a 401(k)-type plan?
Yes
31,756
36.5
45,700
19,000
No
45,241
17.5
43,200
16,000
Total
76,997
25.4%
$ 44,700
$ 18,000
Source:
CRS analysis of the 2001 panel of the Census Bureau’s Survey of Income and Program Participation.
Notes: Includes single persons as well as families.
a. Households with an employed head or spouse age 21-64, in thousands.
b. Percentage of households in which head or spouse participates in plan, by type of plan.

---

CRS-17
Table 10. Ownership of 401(k)-type Plans from Current or Past Job in 2001
Number of
Percent That
Households with
Own One or
Mean Balance
Median Balance
Household Head Characteristics
One Or More
More DC Plansb In All Such Plans in All Such Plans
a
Workers
Age
Under 25
21,543
36.1
$15,100
$ 6,000
35 to 44
23,339
43.8
41,600
20,000
45 to 54
20,750
45.0
58,400
27,000
55 or older
11,364
39.0
66,000
30,000
Race
White, non-Hispanic
64,250
43.0
45,300
18,000
Black, Hispanic, or Asian
12,747
32.5
31,400
10,800
Sex and Marital Status
Married
44,672
46.9
50,100
21,500
Single Male
14,219
35.0
36,500
14,000
Single Female
18,105
32.2
25,400
8,500
Education
Did not graduate High School
7,649
16.7
19,700
8,000
High School graduate
21,146
34.6
31,900
12,500
Some college
24,860
41.7
35,100
14,000
College graduate
23,341
54.9
59,200
25,000
Household income (annual)
Under $25,000
14,650
12.8
17,150
5,000
$25,000 to $49,999
24,546
33.7
23,200
8,000
$50,000 to $74,999
17,088
50.1
36,000
16,000
$75,000 or more
20,713
63.0
65,000
35,000
Own or rent home
Own
51,792
48.7
49,300
22,000
Rent
25,205
26.0
20,800
6,500
Full time or part time worker
Not currently working
6,258
32.6
52,600
21,500
Full time (35+ hours per week)
51,878
43.7
42,700
17,500
Part time
18,861
37.5
43,300
15,000
Establishment size
Not reported
14,716
27.7
50,600
20,000
Under 25 employees
17,826
32.5
39,800
15,000
25 to 99 employees
11,924
44.4
38,950
14,000
100 or more employees
20,758
55.3
46,400
19,000
Public sector employee
11,772
43.4
39,900
19,000
Covered by Union contract?
Yes
9,860
46.6
39,100
18,000
No
67,137
40.5
44,200
17,000
Have an IRA or Keogh plan?
Yes
20,115
57.6
59,800
29,000
No
65,008
31.0
34,100
13,000
Total
76,997
41.2%
$43,464
$17,000
Source: CRS analysis of the 2001 panel of the Census Bureau’s Survey of Income and Program Participation.
Notes: Includes single persons as well as families.
a. Households with an employed head or spouse age 21-64, in thousands.
b. Percentage of households in which head or spouse participates in plan, by type of plan.

---

CRS-18
Household Net Worth. The net worth of households is the difference
between their assets and liabilities. As with the average balances in retirement
accounts that are the principal focus of this report, household net worth as measured
by the SIPP is lower than the household net worth recorded on the SCF.12
Table 11. Median Household Net Worth in 2001,
by Age of Household Head
(All households)
Age of Household Head
Amount
Under 35 years old
$ 5,700
35 to 44
42,300
45 to 54
85,000
55 to 64
122,800
65 to 74
123,500
75 or older
115,800
All households
58,550
Source: CRS analysis of the 2001 panel of the Census Bureau’s
Survey of Income and Program Participation.
Policy Issues
Are Americans saving adequately for retirement? The median retirement
account balances reported by respondents to the Survey of Consumer Finances would
not by themselves provide an income in retirement that most people in the United
States would find to be adequate. The median retirement account balance reported
on the SCF among households headed by a worker 55 to 64 years old who owned one
or more retirement accounts was just $55,000 in 2001. This amount would provide
a monthly annuity of $400 per month to a person retiring at age 65 if converted to an
annuity. Moreover, as the data in preceding tables show, an estimated 28 million
households headed by a worker under age 65 had no retirement savings accounts in
2001.
Although most workers in the United States — about 96% — are covered by
Social Security, only about 25% of households include a worker who participates in
12 On the public use file of the SIPP, total household wealth is defined as the sum value for
all adults in the household of home equity, net equity in vehicles, business equity, interest-
earning assets held in banking institutions, interest earning assets held in other institutions,
equity in stocks and mutual fund shares, equity in real estate other than the home, equity in
other assets, and equity in IRA and Keogh accounts. For this table of net worth, the Census
Bureau has added the sum value for all adults in the household of all 401(k) plans and other
thrift plans. This is consistent with the method used by the Federal Reserve Board, which
includes the value of such accounts in the estimates of household wealth that it derives from
the Survey of Consumer Finances. Household debt is the sum of debts owed by all adults
in the household, including home mortgages, home equity loans, other real estate debts,
automobile loans, installment loans, credit card debt, and other household debt.

---

CRS-19
a defined-benefit pension plan. For workers who do not have coverage through a
defined-benefit pension, personal saving is an essential element of preparing for
retirement. Whether workers save by putting money aside in an account that is
earmarked specifically for retirement or by accumulating other assets on which they
can draw after they have retired is not necessarily important. The act of saving is of
greater consequence to retirement security than the manner in which it is
accomplished. Nevertheless, the fact that 28 million households headed by a worker
under age 65 have no retirement savings accounts indicates that, at the very least,
many people are not using the most tax-efficient means of setting aside funds for
retirement.
While the widespread adoption of tax-favored retirement savings plans over the
past 25 years can be viewed as a substantial public policy success, greater personal
saving will be needed for the current generation of workers to maintain their desired
standard of living in retirement. The uncertain future of Social Security and the
declining prevalence of traditional defined-benefit pensions that provide a guaranteed
lifelong annuity have put much of the responsibility for preparing for retirement on
the shoulders of the worker. The low rate of personal saving in the United States,
and the lack of any retirement savings accounts among millions of American
workers, indicate that there is a need for greater awareness among the public about
the importance of setting aside funds to prepare for life after they have stopped
working.
References
Aizcorbe, Ana M., Arthur B. Kennickell, and Kevin B. Moore. “Recent Changes in
U.S. Family Finances: Evidence from the 1998 and 2001 Survey of Consumer
Finances.” The Federal Reserve Bulletin, vol. 89, no. 1 (January 2003).
Orzechowski, Shawna and Peter Sepielli. Net Worth and Asset Ownership of
Households: 1998 and 2000. U.S. Census Bureau. Current Population Reports.
May 2003, pp. 70-88.

---

CRS-20
Appendix: Why Do the Survey Results Differ?
The tables in this report summarize the rate of retirement plan ownership and
the mean and median balances of retirement accounts as reported on the Census
Bureau’s Survey of Income and Program Participation (SIPP) and the Federal
Reserve Board’ Survey of Consumer Finances (SCF). The survey results differ
substantially with respect to two of these variables: the rate of retirement plan
ownership and the mean value of households’ retirement account balances. They
differ much less with respect to the median value of households’ retirement account
balances. The results of a survey depend mainly on four factors, the most important
of which is why the survey is being conducted. The reasons for conducting the
survey critically influence what questions are asked, who is asked, and to a lesser
extent, when they are asked. The SIPP and the SCF differ with respect to all four.
Why Was the Survey Conducted?
The Federal Reserve Board conducts the SCF “to provide detailed information
on the finances of U.S. families.”13 The SCF questionnaire has been developed over
the years to elicit responses that provide the most complete information on household
finances that can be obtained without subjecting the respondents to an interview of
such length that their willingness to participate in the survey is adversely affected.
The method of selecting households to participate in the survey has been developed
to take into account the skewed distribution of wealth in the United States. (A
relatively small percentage of households own a substantial percentage of total
wealth.) The sample of households is selected to ensure that a sufficient number of
high-income households are included so that even relatively infrequently held assets
are represented in the survey.
In contrast to the SCF, the SIPP is a multi-purpose survey that must satisfy the
needs of the many government agencies that co-sponsor it and that make use of the
data collected. According to the Census Bureau, the main objective of SIPP is “to
provide accurate and comprehensive information about the income and program
participation of individuals and families in the United States, and about the principal
determinants of income and program participation. SIPP data allow the government
to evaluate the effectiveness of federal, state, and local programs.”14 The portions of
the survey that focus on collecting information on household assets and liabilities are
shorter and the questions less detailed than on the SCF. The method of selecting
households to participate in the survey takes into account the higher rate of
participation in income-support programs among low-income households.
What Questions Were Asked?
The SIPP contains a “core” set of questions that focus on income and
employment. These are asked during each of the six to twelve interviews in which
a household participates over the 24- to 48-month life of each panel of the survey.
13 See [http://www.federalreserve.gov/pubs/oss/oss2/about.html].
14 See [http://www.sipp.census.gov/sipp/intro.html].

---

CRS-21
Among the core set of questions are two that ask whether anyone in the household
owned an IRA or Keogh plan during the previous four months and whether anyone
in the household owned a 401(k) plan or other thrift plan during the same period. In
addition to the core questions, the SIPP includes in each wave one or more “topical
modules” that ask detailed questions on topics of particular interest to policymakers
and analysts. A topical module on assets, liabilities, and eligibility for means-tested
government programs is typically administered three times over the course of each
panel of the SIPP. In the topical module, respondents who answered “yes” to either
of the questions about IRA/Keogh ownership or 401(k) ownership in the core are
asked up to six follow-up questions:
! As of the last day of the reference period did ... have any Individual
Retirement Accounts — any IRAs — in ...’s own name?
! As of the last day of the reference period, what was the total balance
or market value (including interest earned) of the IRA accounts in
...’s own name?
! As of the last day of the reference period, did ... have a Keogh
account in ...’s own name?
! As of the last day of the reference period, what was the total balance
or market value of assets in ...’s Keogh account(s)?
! As of the last day of the reference period, did ... have any 401K or
thrift plans in ...’s own name?
! As of the last day of the reference period, what was the total balance
or market value (including interest earned) of any 401K or thrift
plans held in ...’s own name?15
The SCF asks a more detailed series of questions about ownership of, and
account balances in, IRAs, Keogh plans, 401(k) plans, and other employer-sponsored
thrift plans. One advantage of this more extensive series of questions is that they
provide more opportunities for respondents to introduce information about plans that
might have been overlooked or forgotten during the first few questions. Furthermore,
because the SCF is designed primarily to collect information about household
finances, respondents are likely to be more focused on these aspects of their families
and more prepared with papers and documents that would help them provide
complete information. Interviewers also are able to focus their efforts at maximizing
participant cooperation on a single topic. In contrast, because the SIPP interview
must move from one subject area to another to meet its more diverse objectives,
respondents are more likely to overlook or omit some information about their assets
and liabilities when answering those questions.
15 The corresponding variables on the SIPP data file are EALR, TALRB, EALK, TALKB,
EALT, and TALTB. These are “person-level” variables, but they can easily be summed
over all members of a family or household.

---

CRS-22
Who Was Asked the Questions?
The SIPP and the SCF differ in sample size and, more importantly, in sampling
methodology. The third wave of the 2001 panel of the SIPP, during which the data
analyzed in this report were collected, was conducted among approximately 27,000
households. The analysis in this report was based on records of 18,629 respondents
representing an estimated 77 million households in which the household head or
spouse was employed and between the ages of 21 and 64. For the 2001 SCF,
members of 4,449 households were interviewed. The analysis in this report was
based on the records of 3,230 respondents representing an estimated 75.6 million
households in which the household head or spouse was employed and between the
ages of 21 and 64.
Both the SIPP and the SCF are conducted among samples of households that —
when used in conjunction with the appropriate sample weights — are intended to be
representative of the civilian, non-institutional resident population of the United
States.16 The survey results must be used with the appropriate sample weights
because neither survey employs simple random sampling of the population. A simple
random sample consists of a number (n) of individuals or households chosen from
the population in such a way that every set of n individuals or households has an
equal chance of being selected for the sample. Both the SIPP and SCF employ
complex sample designs that are intended to allow detailed analysis of sub-
populations that are of particular interest to policymakers. In the case of the SIPP,
the sub-populations of particular interest are low-income households and households
headed by African-American and Hispanic householders. In the SCF, the sub-
population of interest is households that hold specific forms of wealth — such as
individual corporate stocks, corporate bonds, and municipal bonds — that are owned
by a relatively small percentage of all U.S. households. Consequently, each survey
over-samples some households. The SIPP over-samples households in census tracts
with above-average rates of poverty and the SCF over-samples high-income
households using data derived from tax returns.
According to the documentation of the Survey of Consumer Finances,
The SCF combines two techniques for random sampling. First, a standard multi-
stage area probability sample (a geographically based random sample) is selected
to provide good coverage of characteristics, such as home ownership, that are
broadly distributed in the population. Second, a supplemental sample is selected
to disproportionately include wealthy families, who hold a relatively large share
of such thinly held assets as noncorporate businesses and tax-exempt bonds.
Called the list sample, this group is drawn from a list of statistical records
derived from tax returns. Of the 4,449 interviews completed for the 2001 SCF,
2,917 were from the area probability sample, and 1,532 were from the list
sample. The 2001 survey represents 106.5 million families.17
16 Thus, both surveys exclude uniformed military personnel (unless living off-base in
civilian housing), residents of nursing homes and prisons, and Americans living abroad.
17 Federal Reserve Bulletin, vol. 89, no. 1 (Jan. 2003), p. 30.

---

CRS-23
The Census Bureau employs a two-stage sample design to select the SIPP
sample. First, primary sampling units (PSUs) are selected from a list of counties and
cities. Counties are either grouped with adjacent counties to form PSUs or a single
county may constitute a PSU by itself. Second, specific addresses are selected within
the PSUs for inclusion in the survey sample.18 In the SIPP, the Census Bureau over-
samples the low-income population, based on decennial census information. As
described by the Census Bureau:
Housing units within each PSU were split into high- and low-poverty strata. If
the housing unit received the Census long form that included income questions,
the unit’s poverty status was determined directly; for other housing units, poverty
status was assumed on the basis of responses to Census short-form items
predictive of poverty rates. The Census Bureau then sampled the low-income
stratum at 1.66 times the rate of the high-income stratum in each PSU. Compared
with the number of cases produced without over-sampling, this over-sampling
produced an 18 percent increase in the number of cases in and near poverty at
Wave 1. Even greater gains occurred in some subgroups, such as blacks and
Hispanics in poverty, with a gain in the number of sample cases as high as 24
percent. The increases in effective sample sizes were somewhat smaller after
allowance was made for the increased variance associated with differential
weighting. Also, the sample sizes for the higher income and higher age groups
were reduced.19
When Were the Questions Asked?
The reference periods for the SIPP and SCF data presented in this report differ
slightly. The SIPP questions on retirement assets were asked between October 2001
and January 2002, and they reflect account balances at the end of September 2001
through the end of December 2001, depending on when each particular household
was interviewed. The SCF data were collected between May and December 2001.
It’s unlikely that this small difference in the dates of data collection between the two
surveys contributed significantly to the difference in rates of ownership and average
account balances between the two surveys.
In summary, the SIPP and the SCF differ in their main purpose. The SIPP is
intended to collect information on a much wider range of topics than the SCF.
Consequently, the surveys differ with respect to the number of questions asked about
retirement assets, with the questions on the SCF being both more numerous and more
detailed than those on the SIPP. Because the main focus of the SCF interview is
household assets and liabilities, and because the questions it asks are more numerous
and detailed, the data on assets and liabilities from the SCF are likely to be more
representative of the full population than the asset and liability data from the SIPP.
The SIPP and the SCF also differ with respect to sample size and the sub-
populations that they over-sample. The greater sample size of the SIPP allows more
detailed analysis of certain sub-populations with above-average rates of poverty, such
as female-headed households and or households headed by African-Americans. In
18 For more information, see [http://www.sipp.census.gov/sipp/selecting.html].
19 For more information, see [http://www.sipp.census.gov/sipp/oversample.html].

---

CRS-24
general, the standard error of an estimate falls as the sample size increases. (Think,
for example, of the “margin of error” that is reported along with the results of most
polls. The larger the number of people in the poll, the smaller is the margin of error.)
The difference in sample size, however, does not adequately explain the lower rates
of ownership of retirement plans and the lower mean balances of those accounts as
reported on the SIPP. If, for example, exactly the same survey were conducted
among two random samples of the population that differed only in that one sample
was significantly larger than the other, the survey results should be centered around
the same mean, with the difference attributable mainly to the “sampling error”
inherent in selecting any sample from a population.
While the difference in sample size between the SIPP and the SCF is not likely
to have contributed significantly to the different results obtained in the two surveys,
the specific sub-populations selected for over-sampling may have contributed to
these differences. Survey designers take account of the effect of over-sampling by
assigning an appropriate survey weight to each observation. The weighted number
of observations is matched to a set of control totals — such as the number of people
with certain characteristics that were counted in a recent census of the population.
It is possible, however, that even if the weighted number of observations matches the
control totals for a population along a range of characteristics such as age, race, and
sex, that the weights may not fully capture differences between the sample and the
population with respect to some other traits — such as ownership of retirement
assets, for example. We merely note here that the SCF — which over-samples high
income households — shows significantly higher rates of account ownership and
higher mean balances than does the SIPP — which over-samples low-income
households.20
Researchers who have compared the asset values reported on the SIPP with
aggregate data from the National Income and Product Accounts have found that the
values of financial assets are under-reported on the SIPP. In part because it more
closely matches these national aggregate totals, the SCF is considered by many
economists to be more accurate than the SIPP in representing the assets owned by
American households. The asset data collected on the SIPP, however, are extremely
useful in public policy analysis, both because the large sample size allows analysis
of sub-populations at high risk of needing assistance through government income
support programs, and because the SIPP is a rich source of information on economic,
demographic, and social characteristics that are not recorded on the SCF or other
surveys.
20 The SIPP data indicate that 19% of households with a working head or spouse between
the ages of 21 and 64 had annual incomes under $25,000 in the year preceding the survey,
and 26.9% had incomes of $75,000 or more. The SCF data indicate that 20.4% of
households had had incomes under $25,000 in the year preceding the survey while 29.1%
of households had incomes of $75,000 or more.

---