Over the past 25 years, an important change has occurred in the structure of employer-sponsored

retirement plans in the private sector. Although the percentage of the workforce who participate in

employer-sponsored retirement plans has remained relatively stable at approximately half of all

workers, the type of plan by which most workers are covered has changed from defined benefit

(DB) pensions to defined contribution (DC) plans. The responsibilities of managing a DB plan—

making contributions, investing the assets, and paying the benefits to retired workers and their

survivors—lie mainly with the employer. In a typical DC plan, the worker must decide whether to

participate in the plan, how much to contribute, how to invest the contributions, and what to do

with the money in the plan when he or she changes jobs or retires. As a result of the shift from

DB plans to DC plans, workers today bear more responsibility for preparing for their financial

security in retirement.

According to data collected by the Federal Reserve Board, 45% of households in which the

householder or spouse was employed contributed to employer-sponsored retirement plans in

2004, and 58% owned a retirement account of any kind. Among married-couple households in

which the householder was under age 35, the median balance in all retirement accounts owned by

the household was $19,000 in 2004. Among unmarried householders, the median retirement

account balance in 2004 was just $7,000. Among married-couple households headed by

individuals between 45 and 54 years old, median retirement assets in 2004 were $103,200.

Unmarried householders aged 45 to 54 had a median balance of $32,000. Most households that

participated in defined contribution plans in 2004 contributed between 3% and 10% of pay to the

plan. Younger households with median earnings contributed about 5% of pay, while median-

earnings households 45 and older contributed about 6% of pay.

The report also presents the results of an analysis of the amount of retirement savings that

households might be able to accumulate by age 65 under a number of different scenarios. The

analysis shows how varying the age at which households begin to save for retirement, the

percentage of their earnings that they save, and the rate of return on investment can affect the

amount of retirement savings the household will accumulate. Using Monte Carlo methods that

simulate the variability of investment rates of return, we found that a married-couple household

that contributed 8% of pay annually for 30 years beginning at age 35 to a retirement plan invested

in a mix of stocks and bonds could expect have accumulated $468,000 (in 2004 dollars) by age 65

if rates of return were at the median over the 30-year period. Nevertheless, given the variability of

rates of return, there is a 5% chance that the couple would have $961,000 or more and a 5%

chance that the couple would have $214,000 or less. Higher contribution rates and longer

investment periods lead to higher account balances, but also increase the impact of the variability

of investment rates of return. At a 10% contribution rate over 30 years, the household could

expect to accumulate $594,000, with a 90% probability that account would total between

$301,000 and $1.2 million. Saving 8% of pay over 40 years, the household could expect to

accumulate $844,000, with a 90% probability that the account would total between $370,000 and

$2 million. This report will not be updated.

Trends in Retirement Plan Design...................................................................................................1

Growing Prevalence of Defined Contribution Plans.................................................................2

Retirement Savings of American Households.................................................................................3

Retirement Account Balances in 2004......................................................................................4

Amount of Contributions..........................................................................................................7

How Much Might Workers Accumulate by Age 65?......................................................................11

Methods ................................................................................................................................... 12

What is “Monte Carlo” Analysis?.................................................................................................13

Simulation Results: Retirement Account Balances at Age 65.......................................................13

Variability of Investment Rates of Return...............................................................................14

Length of Investment Period...................................................................................................15

Contribution Rates..................................................................................................................16

Household Earnings................................................................................................................17

Simulation Results: Measuring Retirement Income Adequacy.....................................................18

Annuities: Insurance Against Longevity Risk...............................................................................19

Measuring Earnings Replacement Rates.................................................................................20

Married Couples Versus Singles.............................................................................................22

Detailed Simulation Results....................................................................................................24

Policy Considerations....................................................................................................................27

Conclusion ..................................................................................................................................... 28

Figure 1. Effect of Variability in Investment Rates of Return on Retirement Savings at

Age 65........................................................................................................................................14

Figure 2. Effect of Age at Which Savings Begins on Retirement Savings at Age 65...................16

Figure 3. Effect of Annual Contribution Rate on Retirement Savings at Age 65..........................17

Figure 4. Effect of Household Earnings on Retirement Savings at Age 65...................................18

Figure 5. Estimated Range of Earnings Replacement Rates at Age 65 for a Married-

Couple Household with Median Earnings..................................................................................22

Figure 6. Estimated Retirement Savings at Age 65 or Married Couple and Unmarried

Househol ders .............................................................................................................................. 23

Figure 7. Earnings Replacement Rates at Age 65 of Married Couples and Unmarried

Househol ders .............................................................................................................................. 24

Table 1. Number of Plans and Active Participants, by Type of Plan, 1980-2003............................1

Table 2. Household Participation in Defined Contribution Plans at Current Employer in

2004 .............................................................................................................................................. 5

Table 3. Household Retirement Account Balances in 2004............................................................6

Table 4. Contributions to Employer-Sponsored Plans in 2004........................................................8

Table 5. Contributions to Employer-Sponsored Plans in 2004......................................................10

Table 6. Retirement Savings and Income Replacement Rates, Based on Annual Total

Contributions Equal to 8% of Household Earnings...................................................................25

Table A-1. Household Earnings in 2004, by Age and Marital Status of Householder and

Percentile Rank of Earnings.......................................................................................................30

Table A-2. Annual Total Return on Stocks and Bonds and Annual Rate of Change in the

Consumer Price Index, 1926-2005.............................................................................................31

Table A-3. Retirement Savings and Income Replacement Rates, Based on Annual Total

Contributions Equal to 6% of Household Earnings...................................................................32

Table A-4. Retirement Savings and Income Replacement Rates, Based on Annual Total

Contributions Equal to 10% of Household Earnings.................................................................34

Appendix. Historical Data and Estimates of Savings....................................................................30

Author Contact Information..........................................................................................................36

Over the past 25 years, an important change has occurred in the structure of employer-sponsored

retirement plans in the private sector. Although the percentage of the workforce who participate in

employer-sponsored retirement plans has remained relatively stable at approximately half of all

workers, the type of plan by which most workers are covered has changed. In 1980, the majority

of workers participated in defined benefit (DB) pensions. (See Table 1.) Generally, workers in DB

plans do not have to elect to participate. All covered workers earn benefits under the plan, and the

benefits typically are based on the number of years of service by the employee and some measure

of the worker’s average salary. At retirement, benefits typically are paid as an annuity that

provides the retiree with a monthly income for life. The Employee Retirement Income Security

Act of 1974 (ERISA, P.L. 101-508) requires an employer that sponsors a defined benefit (DB)

plan to establish a trust fund that holds assets sufficient to pay the retirement benefits earned by

the workers who participate in the plan. The responsibilities of managing a DB plan—making

contributions, investing the assets, and paying the benefits to retired workers and their

survivors—lie mainly with the employer.

Table 1. Number of Plans and Active Participants, by Type of Plan, 1980-2003

Today, a majority of workers participate in 401(k)-type plans rather than in traditional defined

benefit pensions. These are called defined contribution (DC) plans. Defined contribution plans are

much like savings accounts maintained by employers on behalf of each participating employee. In

a typical DC plan, the worker must decide whether to participate in the plan, how much to

contribute, how to invest the contributions, and what to do with the money in the plan when he or

she changes jobs or retires. Thus, in a DC plan, it is the employee who bears the investment risk

and who is ultimately responsible for prefunding his or her retirement income. As a result of the

shift from DB plans to DC plans, workers today bear more responsibility for preparing for their

financial security in retirement. Decisions that workers make—or fail to make—from the time

that they first enter the workforce can have a substantial impact on their wealth and income many

decades in the future. Understanding how workers have responded to these challenges and

opportunities may help Congress develop policies that will assist workers in making the best

possible decisions to provide for their financial security in retirement.

This CRS report presents information on trends in retirement plan design and then summarizes

data collected by the Federal Reserve Board on the retirement savings accumulated by workers

and the rates at which they are saving for retirement. The report also presents the results of an

analysis conducted by CRS on the amount of retirement savings that workers might be able to

accumulate by age 65 under a number of different scenarios. The analysis shows how varying

each of several factors, including the age at which households begin to save for retirement, the

percentage of their earnings that they save, and the rate of return on investment can affect the

amount of retirement savings the household will accumulate. The accumulated savings is then

converted into an annuity to illustrate the share of pre-retirement earnings that the accumulated

retirement savings could replace.

The rapid growth of defined contribution plans began in the 1980s. In 1978, Congress added

section 401(k) to the Internal Revenue Code, which allowed employees to contribute part of their

current pay into a retirement plan on a pre-tax basis. In 1981, the Internal Revenue Service (IRS)

published regulations for “cash or deferred arrangements” established under §401(k), into which

employees can make pre-tax contributions, and in which interest, dividends, and capital gains

accrue on a tax-deferred basis until the money is withdrawn. Since that time, DC plans have

overtaken defined benefit pensions in the number of plans, the number of participants, and total

assets. In 2006, only 20% of all workers in the private sector were included in defined benefit

pension plans, while 43% participated in defined contribution plans. About 12% of workers

participated in both types of plan.

DB and DC plans also differ with respect to participation. In general, all workers who meet the

requirements for coverage under a DB plan automatically earn benefits under the plan. The

employer prefunds the benefits that will be paid to all eligible employees when they reach

retirement age. In contrast, in most DC plans, the employee must elect to participate. The

employee also must decide how much to contribute to the plan, and how to invest the

contributions. According to the U.S. Department of Labor, 20% of workers whose employers

sponsored DC plans did not participate in these plans in 2006.

One way to boost enrollment in DC plans would be to enroll all eligible employees automatically.

More firms, particularly among large employers, have adopted automatic enrollment in recent

years. According to the Profit Sharing/401(k) Council of America, 17% all 401(k) plans had

automatic enrollment in 2005, up from 11% in 2004. Automatic enrollment had been adopted by

34% of plans with 5,000 or more participants by 2005, compared to just 4% of plans with fewer

than 50 participants. The Pension Protection Act of 2006 (P.L. 109-280) contains provisions that

are intended to encourage employers to adopt automatic enrollment in defined contribution plans.

Plans with this feature will be exempted from certain tests for discrimination in favor of highly-

compensated employees, a practice that is prohibited by law.

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 §401(k), §403(b), and §457 of the Internal Revenue

Code can accumulate assets on a tax-deferred basis while they are working. Most people with

earned income also can contribute to an individual retirement account (IRA). In 2007, IRA

contributions of up to $4,000 (or $5,000 for people 50 and older) are tax-deductible for workers

who are not covered by a retirement plan at work. In these plans, taxes are paid when the funds

are withdrawn, and a penalty may apply if the withdrawal occurs before retirement. Another

option is to save for retirement in a Roth IRA. Roth IRAs accept only after-tax contributions;

however, withdrawals from a Roth IRA during retirement are tax-free.

The Survey of Consumer Finances

According to the Survey of Consumer Finances, there were 84.7 million households with one or

more workers in 2004 and in 44.5% of these households either the householder, the householder’s

spouse, or both participated in a defined contribution retirement plan. (See Table 2.) Some

workers do not participate because their employer does not offer a plan; however, data from the

Department of Labor indicate that among workers whose employer offers a DC plan, 20% do not

participate.

Participation in employer-sponsored defined contribution plans varied with the age and marital

status of the householder. Participation was lowest among households in which the householder

was under age 35 (37%) and highest among households in which the householder was between

the ages of 45 and 54 (52%). Participation was higher among married-couple households (51%)

than among unmarried householders (36%), partly because married-couple households had more

workers. However, married-couple households had higher participation rates at all ages than

households headed by unmarried persons.

Table 2 also shows the percentage of participating households in which either the household, the

employer, or both contribute to the plan. Ninety percent of participating households reported that

they contributed to the plan in which they participated, while 83% reported that the employer

contributed to the plan. Three-fourths of all participating households reported that both the

household and the employer contributed to the plan.

Age and marital status are both important considerations when evaluating the adequacy of a

household’s retirement savings. Couples obviously need more income to support themselves than

single persons (although they do not necessarily need twice as much income.) Younger workers

have more time to save than older workers, and can reap the benefits of compound interest over a

longer period. As the data presented later in this report will demonstrate, workers who wait until

middle age to start saving for retirement face an uphill struggle in accumulating adequate

retirement assets.

Table 3 shows the retirement account balances of households that owned one or more retirement

accounts in 2004, categorized by the age and marital status of the household head. The first

column shows the balances in all of the DC plans at the current main jobs of the householder and

his or her spouse. The second column shows the balances in all retirement accounts owned by the

household, including accounts at their current jobs, balances held in accounts at former

employers, and balances in individual retirement accounts (IRAs). The third column of Table 3

shows the ratio of household retirement saving to annual household earnings. For example, the

second row of Table 3 shows that among married-couple households in which the householder

was under age 35, the median balance in all retirement accounts owned by the household was

$19,000. This amount was equal to 26.7% of the median annual earnings of those households.

Similar ratios are used later in this report to illustrate a measure of the adequacy of retirement

savings.

Table 3 also shows the 75

the 75 percentile, married couple households headed by persons under age 35 had total

retirement assets of $44,000. In other words, three-fourths of married-couple households headed

by persons under age 35 had total retirement assets of $44,000 or less in 2004, while one-fourth

of all such households had total retirement assets of more than $44,000. At the 25 percentile,

married couple households headed by persons under age 35 had total retirement assets of $5,400

in 2004.

Among married-couple households headed by individuals between 45 and 54 years old, median

retirement assets in 2004 were $103,200. Households headed by unmarried individuals had

retirement assets that were lower at every age than those of married couples, both in absolute

terms and as a ratio of their current earnings. Among households headed by single persons

between the ages of 45 and 54, for example, median retirement assets in 2004 were $32,000, or

less than a third of the median retirement assets of married-couple households in this age group.

Likewise, at the 75 percentile, households headed by unmarried individuals between the ages of

45 and 54 had total retirement assets of $80,000, compared to assets of $275,000 among married ^{th}

couple households in this age group. At the 25 percentile, households headed by unmarried

individuals between the ages of 45 and 54 had total retirement assets of $11,400, compared with

assets of $30,000 among married couple households.

Eventually, most households will have to begin spending their retirement assets. Most choose to

do so through periodic withdrawals, while others choose to convert some or all of their retirement

assets into a guaranteed stream of income by purchasing an annuity. An individual retiring at age

65 in January 2007 with $119,500—the median retirement account balance among married-

couple households head by persons age 55 and older—could purchase a level, single-life annuity

that would pay $826 per month ($9,912 per year) or a joint and 100% survivor annuity paying

$662 per month ($7,944 per year), based on the current annuity interest rate of 5.25%. These

amounts would replace just 19% and 15%, respectively, of the median household earnings of

$52,000 among all married-couple households headed by individuals who were 60 to 64 years old

in 2004.

Table 2. Household Participation in Defined Contribution Plans at Current

Employer in 2004

Table 3. Household Retirement Account Balances in 2004

The amount that a household accumulates in a DC plan depends on the amount that the employer

and employee have contributed to the plan and the investment gains or losses on those

contributions. The maximum permissible annual contributions by workers and employers are

limited by federal law, but few workers contribute the legal maximum.

permissible employee contribution to defined contribution plans was the lesser of 100% of

earnings or $13,000 per worker. Workers age 50 and older were permitted to contribute an

additional $3,000. The maximum total contribution, including both employee and employer

contributions, was $41,000 per worker in 2004.

Table 4 shows the annualized dollar amount of contributions to defined contribution plans per

household in 2004. Table 5 shows household contributions, employer contributions, and total

contributions as a percentage of household earnings. In both tables, the first column of data shows

the amount of household contributions, the second shows the amount of employer contributions,

and the third column shows the total contribution to the plan. The employer and employee

contributions do not sum to the total contribution because in some cases only the household

contributed to the plan, and in other cases only the employer contributed.

At each age, married-couple households contributed more to DC plans than households headed by

unmarried persons. Among both married-couple households and single households and across all

age groups, employee salary deferrals into defined contribution plans were larger than employer

contributions. (See Table 4.) Among married-couple households headed by persons under 35, the

median household contribution in 2004 was $3,680, and the median employer contribution was

$2,520. The median total contribution was $5,520. Among households headed by unmarried

persons under 35, the median household contribution in 2004 was $2,080, and the median

employer contribution was $1,400. The median total contribution was $3,120.

As a percentage of pay, the contributions of married-couple households and households headed

by unmarried individuals differed less than the dollar amounts of their contributions. (See Table

5.) The median contribution among households headed by individuals under age 45 was about 5%

for both single and married households. Both married-couple households and singles ages 45 to

54 typically contributed about 6% of earnings. Overall, household contributions ranged from ^{th}

about 3% of household earnings at the 25 percentile of contributions to about 10% of household

earnings at the 75 percentile of contributions.

Table 4. Contributions to Employer-Sponsored Plans in 2004

Table 5. Contributions to Employer-Sponsored Plans in 2004

The previous section described the amounts that households had accumulated in retirement

savings accounts and how much they were contributing to their retirement plans in 2004, as

reported in the Federal Reserve Board’s Survey of Consumer Finances. This section uses income

data from the Census Bureau’s Current Population Survey and statistical software that simulates

the variability of investment rates of return to estimate future retirement account balances and to

demonstrate how several variables can affect the amount of retirement savings that households

could accumulate by age 65.

As was shown by the data displayed in Table 2, only 45% of working households participated in

employer-sponsored defined contribution plans in 2004. Some households that did not participate

in employer-sponsored plans saved for retirement in individual retirement accounts (IRAs), but

data from the SCF indicate that most households that did not participate in an employer-

sponsored plan also did not own an IRA. Households that do not save for retirement may be

reducing their future incomes significantly, but by how much? If a household starts to save, what

variables might affect the amount that they have accumulated by the time the householder reaches

age 65? These questions are addressed in this section of the report.

As noted in the introduction, workers must decide not only whether to save for retirement, but

also how much to save, how to invest their savings, and what to do with their accumulated

savings each time they change jobs and when they reach retirement. A number of variables can

affect the amount that households have accumulated in their retirement accounts by the time they

reach retirement age, including the following:

• household earnings;

• the amount that the household saves;

• the age at which the householder begins to save, and thus the number of years over

which contributions and investment earnings accumulate; and

• the average annual rate of return earned by the household’s retirement account.

To estimate household savings, household earnings also had to be estimated. For the base year of

our analysis, earnings were estimated in 2004 at every age from 25 through 64 for married-couple

and unmarried households at the 75, 50, and 25 percentiles of earnings from the March 2005

CPS. This produced an age-earnings-marital status matrix with 240 cells. (See Appendix Table

A-1.) For each later year in the simulation, earnings were increased by 1.1%, which is the

estimated long-run growth rate of real wages as projected by the Office of the Actuary of the

Social Security Administration. For example, the median earnings in 2004 of a married-couple

household headed by a 25 year-old was $41,000. To estimate the same household’s earnings one

year later when the householder would be age 26, the 2004 median earnings of married-couple

households headed by a 26-year-old ($45,600) were multiplied by 1.011, which resulted in

estimated household earnings of $46,102. The following year, when the householder would be

age 27, we estimated household earnings as $51,106, which is the 2004 earnings of a married-

couple household headed by a 27 year-old, ($50,000) multiplied by 1.011. The process for each

household through age 64 was repeated. For simplicity, we assumed that married-couple

households would remain married-couple households throughout the period of the analysis and

that unmarried households would remain unmarried.

Having estimated household earnings each year, we next had to choose the percentage of earnings

that each household would contribute to its retirement account annually. The data on contribution

rates from the 2004 Survey of Consumer Finances indicated that most households contribute

between 3% and 10% of earnings to their employer-sponsored retirement plan. Based on these

data, we estimated the retirement savings that would accumulate by age 65, assuming that

households contributed either 6%, 8%, or 10% of household earnings to the account every year,

starting at age 25, age 35, or age 45. Assuming the householder retires at age 65, these starting

dates would result in periods of saving for retirement lasting 40 years, 30 years, and 20 years,

respectively. Households that do not save every year would accumulate less than was estimated

for those that contribute consistently.

To estimate the amount that households would have accumulated in their retirement account by

age 65, the annual rate of return on the funds invested in those accounts had to be estimated.

Rather than assume that the rate of return in each year would be the long-term average rate of

return on a mixed portfolio of stocks and bonds, we used a Monte Carlo simulation process in

which the rate of return in each year was randomly selected from the range of likely rates of

return implied by the historical returns on stocks and bonds. Many financial advisors recommend

that investors shift their portfolios away from stocks and into bonds as they approach retirement.

Therefore, in our simulations, households allocated 65% of assets to the Standard & Poor’s 500

index of stocks from the ages of 25 to 34, 60% to stocks from 35 to 44, 55% to stocks from 45 to

54, and 50% to stocks after age 55. In each case, the remainder of the portfolio was assumed to be

invested in AAA-rated corporate bonds. The accounts were re-balanced after each year of the

simulation so that the portfolio would start the next year at the chosen allocation between stocks

and bonds. The model also takes into account the correlation between stock and bond returns.

Figures 1 through 4 illustrates the likely range of retirement savings that would be accumulated

by married-couple households and unmarried householders with high, medium, and low earnings,

based on several different contribution rates, lengths of investment period, and investment rates

of return. Referred to are households with earnings at the 75 percentile for their age and marital

status as “high earners,” and to those with earnings at the 50 and 25 earnings percentiles as

“median earners” and “low earners,” respectively.

Figure 1 illustrates how the variability of investment rates of return can affect the amount of

retirement assets that households accumulate. In this example, we estimated the value of a

retirement account balance at age 65 for a married-couple household with median earnings that,

beginning at age 35, contributed 8% of earnings each year for 30 years to an account that was

invested in a mix of stocks and bonds. Each of the 1,000 simulations of a 30-year investment

period produced a unique mean rate of return for the 30-year period. Of these, the median real

rate of return over the 30-year period was 5.5% At this rate of return, the household’s retirement

account balance (in 2004 dollars) would be $468,000 when the householder reached age 65.

Figure 1. Effect of Variability in Investment Rates of Return on Retirement Savings

at Age 65

Figure 1 also shows what this household would accumulate in its retirement account if the

average rate of return over the 30-year investment period was significantly higher than average or

lower than average. Based on the history of stock and bond returns from 1926 through 2005, and

given the allocation of investment assets that we chose, there is a 5% chance that the household’s

retirement account would earn an average annual real rate of return of 1.7% or less over the 30-

year investment period. The historical record of returns suggests that such a low rate of return has

approximately a one-in-twenty chance of occurring in any given 30-year period. In the event that

investment returns were at the 5 percentile, the household would have $214,000 (in 2004

dollars) in its retirement account when the householder reached age 65. This is less than half of

the amount that it would have accumulated if the average rate of return over the 30-year period

were equal to the median real rate of return of 5.5%.

On the other hand, stock and bond markets might perform better-than-average over the period

when the household is saving for retirement, in which case it will accumulate more assets than it

would have in a period of average investment returns. Based on historical returns, in any 30-year

investment period, there is a 95% probability that the average real rate of return on the mix of

assets in the household’s retirement account would be 9.3% or less. However, there is a 5%

chance that average real rate of return would be higher than 9.3%. If the household were to attain

a 9.3% average real rate of return on its investments over 30 years, it would have a retirement

account balance of $961,000 (in 2004 dollars) at age 65. The amounts displayed in Figure 1

illustrate that variability in rates of return will inevitably lead to some uncertainty in retirement

planning. Households can decide when to begin saving for retirement and how much to save, but

variability in rates of return is beyond their control and yet has a great impact on their assets at

retirement.

The age at which a worker starts saving for retirement can dramatically affect the amount that he

or she has accumulated at retirement age. Beginning to save at a younger age results in larger

total contributions and allows investment gains to compound over a longer time. Figure 2 shows

the retirement account balance at age 65 of a married-couple with median household earnings that

contributes 8% of pay each year to a retirement account invested in a mix of stocks and bonds

over periods of 20, 30, and 40 years. The household that begins saving at age 25 saves for 40

years, while the households that begin saving at ages 35 and 45 save for 30 years and 20 years,

respectively. For each of the three investment periods, we show the account balance at age 65 if

the average rate of return for the investment period were equal to the median rate of return, and if

the average rate of return were significantly below the median (at the 5 percentile of likely rates

of return) or significantly above the median (at the 95 percentile of likely rates of return).

If the couple were to begin saving at age 25 and earned the median rate of return over 40 years,

their account balance at age 65 would be $844,000 (in 2004 dollars). Delaying the start of

retirement saving until age 35 would result in an account balance of $468,000, or just 55% of the

amount they would have accumulated had they started saving at age 25. Waiting until age 45 to

begin saving for retirement would result in an account balance of $213,000 at 65, or one-fourth of

the amount they would have had at 65 if they had contributed 8% a year starting at age 25.

Of course, a couple that delays the start of saving until age 45 might get lucky and invest during a

period of above-average rates of return. For example, if rates of return during the twenty years

when they were saving were in the 95 percentile of the likely rates of return, the couple would

accumulate $372,000 (in 2004 dollars) by age 65, or about 75% more than if rates of return

during that period were at the 50 percentile. On the other hand, even a couple that begins to save

at age 25 may have the misfortune to invest during a period of below-average returns. A couple

with median household earnings that contributes an amount equal to 8% of earnings annually for

40 years to a retirement account that is invested in a mix of stocks and bonds could expect to

accumulate $844,000 (in 2004 dollars) by age 65 if investment returns over that period were in

the 50 percentile of likely returns, but they would have just $370,000 if investment returns were

at the 5 percentile of likely rates of return. Workers cannot control the variability of investment

rates of return, but they can choose to begin saving while they are young. As the data presented in

Figure 2 demonstrate, this usually will lead to much greater wealth at retirement than they could

achieve if they were to wait until 35 or 45 to begin saving.

Figure 2. Effect of Age at Which Savings Begins on Retirement Savings at Age 65

Figure 3 illustrates the effect of contribution rates on retirement savings. A married-couple

household that contributed 6% of household earnings each year to a retirement account invested

in a mix of stocks and bonds starting at age 35 would accumulate $353,000 (in 2004 dollars) by

age 65 if the real rate of return over that period were at the 50 percentile of likely rates of return.

By contributing 8% of earnings, the household would have $468,000 by age 65, or 33% more

than if they had contributed 6% of pay to their account. Had the couple contributed 10% per year

for 30 years, their account balance at age 65 would be $594,000, or 68% higher than the amount

resulting from a 6% annual contribution.

Variation in rates of return can affect retirement accumulations significantly, even for those who

contribute a greater amount to their retirement account. For example, the results displayed in

Figure 3 show that if the average real rate of return over 30 years were in the 5 percentile of

likely rates of return, a couple that contributed 6% of pay annually to a retirement account

invested in a mix of stocks and bonds could expect to have accumulated just $170,000 (in 2004

dollars) by age 65. This is less than a quarter of the amount that they would have accumulated if

rates of return during that period had been at the 95 percentile. Similarly, with returns at the 5

percentile, a couple contributing 10% of earnings annually will have accumulated $301,000 by

age 65, or about half of what they would have accumulated at the median rate of return, and only

a quarter of the $1.2 million that would have accumulated if rates of return were at the 95

percentile of likely rates of return.

Figure 3. Effect of Annual Contribution Rate on Retirement Savings at Age 65

Figure 4 shows account balances at age 65 (in 2004 dollars) for married couples at three different

earnings levels, assuming that they contributed 8% of earnings each year beginning at age 35 to a

retirement account invested in a mix of stocks and bonds. A married-couple household with

median earnings would have accumulated $468,000 (in 2004 dollars) by age 65 if stock and bond

market returns were at the median during the 30-year investment period. A high-earning

household with earnings at the 75 percentile would have accumulated $706,000 dollars, while a

low-earning household with earnings at the 25 percentile would have accumulated just

$289,000.

Figure 4. Effect of Household Earnings on Retirement Savings at Age 65

For most people, the main purpose of saving for retirement is to have money available to replace

the earnings that they will lose when they stop working. The proverbial “three-legged stool” of

retirement income—private pensions, Social Security, and personal savings—is missing a leg for

many households because the number of defined benefit pensions has declined substantially over

the past 25 years. Most workers in the U.S.—about 96%—are covered by Social Security, but

Social Security is not designed to replace all of a worker’s earnings. The Social Security

Administration estimates that for a career-long average-wage earner retiring at the full retirement

age, Social Security will replace about 41% of their career-average earnings. For a career-long

low-wage earner, Social Security will replace an estimated 55% of average earnings. For a career-

long high-wage earner, Social Security will replace just 27% of their average earnings.

The percentage of pre-retirement earnings that is replaced by Social Security, pensions, or other

income is called the earnings replacement rate. The estimated value of retirement accounts at age

65 in constant 2004 dollars is an absolute measure of retirement assets. The earnings replacement

rate is a relative measure of retirement assets. To calculate earnings replacement rates, we first

needed to decide on an appropriate measure of earnings. For most households, a relevant measure

would be their average earnings in the years just before they reach retirement age. As our measure

of pre-retirement earnings, we used average household earnings of married-couple and single-

person households headed by persons 60 to 64 years old. We used average earnings in 2004 as

reported on the CPS for households headed by persons 60 to 64 years old as the base measure of

earnings, and projected these amounts forward at a growth rate of 1.1% per year.

To calculate the earnings replacement rate from households’ retirement savings, we converted the

account balance into a stream of income. This was done by calculating the annuity value of the

retirement account. An annuity is a contract between an individual and a financial institution—

usually an insurance company—in which, in exchange for a premium paid by the annuity

purchaser, the insurer promises to pay the individual an income for life or for a fixed period of

years. The premium for an “immediate income” annuity is usually a single payment, which the

insurer then invests. A life annuity insures the purchaser (and his or her surviving spouse in the

case of a joint and survivor annuity) against longevity risk, which is the risk that the individual

will outlive his or her retirement assets. We calculated the percentage of pre-retirement earnings

that would be replaced if the household’s entire retirement account balance were converted to a

life annuity. For example, if a married-couple household with median earnings contributed an

amount equal to 8% of earnings to a retirement account each year for 30 years and earned the

median rate of return over that period of time, the couple would have accumulated $468,000 (in

2004 dollars) when the householder reached age 65. In January 2007, this amount would purchase

a joint and 100% survivor annuity that would replace 42% of the household’s average annual

earnings in the five years before the householder reached age 65. (On the CPS, income is

reported before taxes. The replacement rate here is pre-tax annuity income divided by pre-tax

earnings.)

Figures 1 through 4 illustrate how variation in rates of return, length of investment period,

contribution rates, and household income can affect the amount of retirement savings that

households have accumulated by age 65. Figure 5 combines in a single graph the effect that

variation in investment rate of return, length of investment period, and contribution rates can have

on earnings replacement rates for a married-couple household with median household earnings.

In this graph, the value of the household’s retirement account at age 65 (measured in 2004

dollars) has been converted to a joint and 100% survivor annuity, and the income from the

annuity is compared with average household earnings over the five years before retirement.

On the vertical axis, the graph shows the annuity value of the household’s retirement account,

measured as the percentage of average pre-retirement earnings the annuity would replace. On the

horizontal axis, we show three investment periods of 20, 30, and 40 years, which correspond in

our analysis to saving for retirement starting at ages 45, 35, and 25, respectively. The vertical bars

show the range of replacement rates that could be achieved from converting the household’s

retirement account to a joint and 100% survivor annuity at age 65, depending on the percentage

of pay that the household contributed to its account each year and the investment rate of return.

On each bar, the replacement rates corresponding to low, average, and high rates of return are

represented by the square, circle, and diamond, respectively. For example, the left-most (and

shortest) vertical bar in Figure 5 shows that if a married couple with median earnings invested

6% of household earnings for 20 years, their retirement account balance at age 65 could purchase

an annuity that would pay an amount equal to just 8% of their final average pay, assuming that

investment returns over the 20-year period were in the 5 percentile of likely returns. The annuity

value of their account would replace 14% of their final average earnings if investment returns

were at the median, and the annuity would replace 27% of their final average pay if investment

returns over the 20-year period were in the 95 percentile of likely returns.

Moving from left to right across Figure 5, the vertical bars representing earnings replacement

rates both begin and end at higher replacement rates, representing the effects on retirement

savings of higher contribution rates and longer investment periods. The greater length of the bars

as contribution rates rise and investment periods grow longer illustrates the impact that variability

of rates of return can have on retirement account accumulations, and in turn on the annuity

income that one could purchase with those accounts. Looking at the left-most panel, we see that if

a household were to begin saving 8% of earnings at age 45, the annuity value of its retirement

account could vary by more than 20 percentage points, depending on whether the real rate of

return in the period is significantly above average (at the 95 percentile of returns) or

significantly below average (at the 5

investment returns, a median-earning couple saving 8% of earnings per year would have a

retirement account balance that, if converted to an annuity, could replace 33% of its average pre-

retirement earnings. If investment returns during that period are significantly below average,

however, the annuity value of their retirement account would replace just 11% of the couple’s

pre-retirement average earnings.

The longer the investment period, the greater the difference in retirement accumulations that

results from the variability of returns. After a 40-year period of high investment returns (at the

95 percentile of likely returns), a median-earning couple saving 8% of earnings per year would

have a retirement account balance that, if converted to an annuity, could replace 180% of its

average pre-retirement household earnings. In other words, if they converted their entire

retirement account to an annuity, the income would be almost twice their average household

earnings in the five years preceding retirement. If investment returns during that period were

significantly below average (at the 5 percentile of likely returns), the annuity value of their

retirement account would replace only about one-third of the couple’s pre-retirement average

earnings.

The account balance that a household would accumulate over 40 years of investing during a

period of below-average returns would be much less than the amount that the couple would have

accumulated if they’d had the good fortune to have invested during a period of above-average

rates of return. Nevertheless, in our simulations the annuity value of an account accumulated by

contributing 8% of earnings over 40 years of very low investment returns was almost the same as

the annuity value of an account accumulated by contributing 8% of earnings over a 20-year

period of well above-average investment returns. Workers who begin to save at a young age can

accumulate substantial retirement assets even in periods of low returns, and they will be far better

off at retirement than those who delay saving in the event that investment returns are at or above

the historical average.

Figure 5. Estimated Range of Earnings Replacement Rates at Age 65 for a Married-

Couple Household with Median Earnings

The data presented in Figures 1 through 5 illustrate account balances for married-couple

households at age 65 under specific assumptions about the age at which savings begin, annual

contribution rates, investment rates of return, and household earnings. Figure 6 compares the

accumulated retirement savings at age 65 (in 2004 dollars) of married-couple households and

unmarried householders with median earnings who contribute 8% of earnings beginning at age 35

to a retirement account invested in a mix of stocks and bonds. Account balances are higher for

married couples because their higher earnings produce larger contributions, but as the data

presented in Figure 7 demonstrate, if the couple purchases a joint and 100% survivor annuity,

their earnings replacement rate would be lower than that of an unmarried householder purchasing

a single-life annuity.

If, over a 30-year investment period, total rates of return on stocks and bonds were at their

historical median, a married-couple with median earnings that contributed 8% of pay annually to

a retirement plan invested in a mix of stocks and bonds could accumulate an estimated $468,000

(in 2004 dollars) by age 65. This is almost twice as much as the estimated $247,000 that would be

accumulated by an unmarried householder with median earnings contributing 8% of pay to a

retirement account over the same period. Nevertheless, the annuity value of the retirement

account, expressed as a percentage of pre-retirement earnings, would be higher for the unmarried

householder than for the married-couple household. (See Figure 7.) The higher earnings

replacement rate for the unmarried householders is due largely to the fact that in these examples,

the annuity estimates are based on a joint and 100% survivor annuity for the married-couple

household and a single-life annuity for the unmarried householder. If the married-couple

household elected to purchase an annuity without a survivor benefit (i.e., a single-life annuity

based on the age of the householder), their annuity income would be higher during the life of the

annuity purchasers, but the surviving spouse would no longer receive annuity income after the

death of the annuity purchaser.

Figure 6. Estimated Retirement Savings at Age 65 or Married Couple and

Unmarried Householders

Figure 7. Earnings Replacement Rates at Age 65 of Married Couples and Unmarried

Householders

Table 6 displays the estimated retirement account balances (in 2004 dollars) at age 65 for

households that contribute an amount equal to 8% of household earnings to a retirement account

for periods of 40 years (beginning at age 25), 30 years (beginning at age 35), and 20 years

(beginning at age 45). Retirement account accumulations are shown both for married-couple

households and unmarried householders with high, median, and low career-average earnings.

These earnings levels are represented by households with earnings at the 75 percentile, 50

percentile, and 25 percentile among households of the same age and marital status. Similar

tables showing retirement account balances resulting from contributions equal to 6% of earnings

and 10% of earnings are shown in the Appendix to this report. (See Table A-3 and Table A-4.)

For each of six types of household, as defined by household earnings and the marital status of the

householder, Table 6 shows the estimated retirement account balance at age 65 resulting from

annual contributions equal to 8% of pay over periods of 20, 30, and 40 years, depending upon

whether the average real rate of return on investment during that period was at the 95, 50, or 5

percentile of likely returns.

For example, a median-earning married-couple household that began contributing 8% of pay

annually at age 35 to a retirement account invested in a mix of stocks and bonds could expect, on

average, to have accumulated $468,000 (in 2004 dollars) in its retirement account by age 65. This

is the amount that would result if investment returns over those 40 years fell in the middle—the

50 percentile—of the likely range of possible returns, based on the distribution of real rates of

total return on stocks and bonds that occurred between 1926 and 2005. If rates of return over the

investment period were well above average—at the 95 percentile of likely returns—the

household’s estimated retirement account balance at age 65 would be $961,000 (in 2004 dollars).

On the other hand, if the average rate of return earned over the investment period was well below

average—at the 5 percentile of likely rates of return—the household would have accumulated

just $214,000 (in 2004 dollars) by age 65.

Table 6 also displays two relative measures of retirement savings: the ratio of the account balance

to the household’s average earnings in the five years before the householder reached age 65 and

the annuity value of the retirement account balance at age 65. For example, the estimated

retirement account balance at age 65 of $468,000 (in 2004 dollars) for a median-earning married-

couple household that contributed 8% of pay annually to a retirement account over 40 years

beginning at age 25 would be equal to 7.0 times the household’s average earnings (in 2004

dollars) during the five years when the householder was ages 60 to 64. Based on current interest

rates, if this amount were converted to a level, joint and 100% survivor annuity, it would replace

an estimated 42% of the household’s average earnings in the five years that the householder was

ages 60 to 64. In our simulations, given an annual retirement plan contribution equal to 8% of

earnings, the ratio of account balances at age 65 to household earnings from ages 60 to 64 ranged

from a low of 1.4 times earnings for low-earning married-couple and unmarried households that

begin to save at age 45 and invest during a period of low investment returns to a high of 31times

earnings for a high-earning unmarried householder who begins saving at age 25 and invests

during a period of high investment returns. Replacement rates ranged from a low of 10% for low-

earning married-couple households and 12% for low-earning unmarried households who invest in

a period of below-average returns to 202% for high-earning married-couple households and 257%

for high-earning unmarried households who invest in a period of above-average returns.

Table 6. Retirement Savings and Income Replacement Rates, Based on Annual Total

Contributions Equal to 8% of Household Earnings

The uncertain future of Social Security and the declining prevalence of defined-benefit pensions

that provide a guaranteed lifelong income have put much of the responsibility for preparing for

retirement directly on workers. Saving for retirement will be especially important for workers

who are not included in a defined-benefit pension where they are employed, which includes about

80% of all workers in the private sector. Even among those who are saving for retirement,

retirement account balances are generally low when compared to household earnings. As the data

displayed in Table 3 showed, the median account balance in 2004 among married-couple

households headed by persons 55 to 64 years old was equal to just 1.6 times the median earnings

of those households. Among unmarried householders between the ages of 55 and 64, median

retirement savings were equal to just 1.4 times median earnings.

The low rate of personal saving in the United States and the lack of any retirement savings

accounts among millions of American households 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. Most workers in the United States will need to begin saving more of

their income if they wish to maintain a standard of living in retirement comparable to that which

they enjoyed while working. The alternatives would be to work longer or to greatly reduce their

standard of living in retirement.

Employers in the United States are not required to offer pensions to their employees and workers

are not required to save for retirement. Because both of these activities are voluntary, most policy

proposals for boosting workers’ retirement saving are intended to make workers more aware of

the importance of saving and to make the act of saving easier for both workers and employers.

Although most defined contribution plans continue to require employees to elect to participate, a

growing number of plans have adopted automatic enrollment of eligible workers, and the Pension

Protection Act of 2006 (P.L. 109-280) included provisions intended to encourage more employers

to adopt automatic enrollment in their retirement plans.

saving would be to promote greater use of payroll deduction for individual retirement accounts

(IRAs). Because IRAs are not employer-sponsored plans, there would be little administrative

burden for employers, and payroll deduction would be an easy way for workers to send money

directly to a retirement account.

The Pension Protection Act also made permanent the Retirement Savings Tax Credit, originally

enacted in 2001, and provided for indexing the income thresholds over which the credit is phased

out. Some policy analysts have suggested that if this credit were made refundable, it would

encourage more lower-income workers to save for retirement. Another proposal would disregard

amounts in retirement savings plans for purposes of determining eligibility for certain means-

tested federal aid programs. On the employer side, it has been proposed to give employers a tax

credit for the cost of maintaining retirement savings plans to encourage more employers to offer

such plans.

With respect to promoting secure lifetime income for retirees, policy proposals have focused on

providing incentives for people to purchase life annuities. For example, under one proposal,

individuals would not pay federal taxes on one-half of the income generated by annuities that

promise lifetime payments. There would be an annual limit of $20,000 on the amount of annuity

income that an individual could exclude from federal taxes each year. Another policy option

would be to require employers that offer defined contribution plans to offer retiring workers the

opportunity to purchase an annuity through the employer, as is currently required of defined

benefit plans.

Are Americans saving adequately for retirement? There is no simple answer to this question

because circumstances vary so greatly from one household to another. Data from the Federal

Reserve Board’s Survey of Consumer Finances indicate that fewer than half of all working

households participated in an employer-sponsored retirement savings plan in 2004, and fewer

than a third of all working households owned an individual retirement account. For this report, we

estimated the amounts that married-couple households and unmarried householders with high,

medium, and low earnings could accumulate in their retirement accounts by age 65, depending on

the percentage of earnings that they save, the age at which they begin saving, and the total real

rate of return in stock and bond markets during the period that they are investing. Two of these

three variables—the contribution rate and the age at which they begin to save—are more or less

under the control of the worker.

As the results displayed in Figure 5 illustrate, starting to save while young and doing so

consistently every year is perhaps the single most effective way to assure that one reaches

retirement with adequate savings. For a household with median annual earnings, even a relatively

low annual contribution equal to 6% of earnings will, at the median likely rate of return, grow

over 40 years to an amount that, if converted to an annuity, would replace more than half of the

household’s average pre-retirement earnings. At a 10% contribution rate, the annuity value of the

account would replace more than 90% of the household’s pre-retirement earnings, assuming rates

of return are at the median. It is also important to note that our estimates are based on the

assumption that the household contributes to a retirement plan every year for a period of 20, 30,

or 40 years. Because of interruptions in employment, unexpected expenses, and other reasons,

many households do not contribute to a retirement plan every year.

Unfortunately, we cannot safely assume that rates of return over the next 20, 30, or 40 years will

be “average.” In our analysis, we simulated the variability in rates of return through a Monte

Carlo process. We found that, although the average annual rate of total return over 30 years on the

mixed portfolio of stocks and bonds that we chose for our analysis would be 5.5%, there was a

5% chance that the real rate of return would be 1.7% or lower and a 5% chance that it would be

9.3% or higher. This variability in rates of return is something over which workers have no

control, and which will inevitably lead to some uncertainty in retirement planning. While it may

be easier for workers to focus on what they are likely to accumulate in their retirement accounts

“on average,” ignoring the variability of investment rates of return could lead to poor decisions

that might be avoided if workers were better informed about the way that variability in

investment rates of return can affect their retirement savings over time. A worker who is told that

the most likely real rate of return on his or her investments is 5.5% might save more or less than

if he or she were told that the most likely real rate of return will be between 1.7% and 9.3%. Both

statements are true, but the second more clearly conveys the uncertainty that characterizes any

estimate of likely future rates of return on investment.

Table A-1. Household Earnings in 2004, by Age and Marital Status of Householder

and Percentile Rank of Earnings

Table A-2. Annual Total Return on Stocks and Bonds and Annual Rate of Change in

the Consumer Price Index, 1926-2005

Table A-3. Retirement Savings and Income Replacement Rates, Based on Annual

Total Contributions Equal to 6% of Household Earnings

Table A-4. Retirement Savings and Income Replacement Rates, Based on Annual

Total Contributions Equal to 10% of Household Earnings