To a worker contemplating retirement, there is perhaps no more important question than “How

long will my money last?” Congress has a strong interest in the income security of older

Americans because much of their income is either provided directly from public programs like

Social Security, or in the case of pensions and retirement accounts, is subsidized through tax

deductions and deferrals.

Many retirees must decide how to convert retirement account balances into income and how to

preserve the accounts in the face of several kinds of risk.

• Longevity risk is the risk that the individual will exhaust his or her account before

death and experience a substantial decline in income.

• Investment risk is the risk that the assets in which the individual has invested his

or her retirement account will decline in value.

• Inflation risk is the risk that price increases will cause the individual’s retirement

income to decline in purchasing power.

• Unexpected events such as divorce, the death of a spouse, the cost of medical

care, or a need for long-term care services are also risks.

There are strategies for dealing with each of these risks, but no single strategy can deal effectively

with all of them. For example, purchasing a life annuity insures against longevity risk and it shifts

the investment risk to the insurer. However, purchasing an annuity depletes the purchaser’s

available assets by the amount of the premium. These assets are no longer available to the retiree

in the event of a catastrophic illness or other unexpected major expense. To date, the demand for

annuities has been low. There are many reasons for the low demand for annuities, but one of the

most important has been that many potential annuity purchasers do not value the longevity

insurance provided by annuities at its market price.

Retirees who choose not to purchase life annuities must decide how much to withdraw from their

retirement accounts each year. Because they face uncertainty with respect to both life expectancy

and the rate of return on investment, this decision carries its own risks. If withdrawals are too

large, retirees risk spending down their savings too quickly, possibly leaving them impoverished.

If withdrawals are too small, they might spend too little and leave substantial assets unspent when

they die.

An analysis conducted by CRS indicates that under specific conditions there is a 95% or greater

probability that a man who retires at age 65 will not exhaust his retirement account before the

earlier of death or age 95 if his initial withdrawal does not exceed 5% of the account balance and

later withdrawals are the same in inflation-adjusted dollars. Under the same conditions, there is a

95% or greater probability that a woman who retires at age 65 will not exhaust her retirement

account before the earlier of death or age 95 if her initial withdrawal does not exceed 4.5% of the

account balance and later withdrawals are the same in inflation-adjusted dollars.

Introduc tion ..................................................................................................................................... 1

The Nature of Risk in Retirement...................................................................................................2

Longevity Risk..........................................................................................................................2

Inflation Risk.............................................................................................................................3

Investment Risk.........................................................................................................................4

Unexpected Events....................................................................................................................4

Annuities as a Source of Retirement Income..................................................................................5

The Current Market for Annuities.............................................................................................6

Group Annuities..................................................................................................................6

Individual Annuities............................................................................................................7

Tax Treatment of Annuities.....................................................................................................10

Taxation of Income from a Fixed Annuity........................................................................10

Taxation of Income from a Variable Annuity.....................................................................11

Tax Exclusion for Long-Term Care Insurance...................................................................11

Consumer Protections and the Regulatory Environment........................................................12

State Regulation of Annuities...........................................................................................12

Federal Regulation of Annuities.......................................................................................13

Why Is Demand for Individual Annuities So Low?................................................................13

Complexity and Lack of Transparency in Annuity Expenses...........................................14

Lack of Flexibility in Dealing with Unexpected Expenses...............................................15

Retirement Account Withdrawal Strategies...................................................................................16

How Long Will a Retirement Account Last with Fixed Annual Withdrawals?.............................16

Initial Rate of Withdrawal.......................................................................................................17

Investment Portfolio................................................................................................................18

Probability of Assets Lasting for at Least a Specific Number of Years............................18

Portfolio of 65% Stocks and 35% Bonds..........................................................................18

Portfolio of 35% Stocks and 65% Bonds..........................................................................19

Estimates Incorporating Life Expectancy...............................................................................20

Estimates of Variable Annual Withdrawals.............................................................................24

Summary of Withdrawal Strategies: Balancing Risks...................................................................29

Figure 1. Growth in Annuity Sales Relative to Growth in the S&P 500 Index...............................9

Table 1. Estimated Percentage of Individuals Age 65 in 2004 Surviving to Selected Ages...........3

Table 2. Annual Average Rate of Return Over 10- Year Period......................................................4

Table 3. Monthly Income By Gender and Joint/Survivor Option for Fixed Annuity With

a $100,000 Premium, 2008..........................................................................................................8

Table 4. Tax Treatment of Annuities...............................................................................................11

Table 5. Estimated Probability a Retirement Account Will Last for at Least a Specific

Number of Years.........................................................................................................................19

Table 6. Probability of Retirement Account Lasting to at Least a Given Age, Including

Mortality Risk, Retirement at Age 60........................................................................................22

Table 7. Probability of Retirement Account Lasting to at Least a Given Age, Including

Mortality Risk, Retirement at Age 65........................................................................................23

Table 8. Probability of Retirement Account Lasting to at Least a Given Age, Including

Mortality Risk, Retirement at Age 70........................................................................................24

Table 9. Variable Annual Withdrawals Based on Life Expectancy, Retirement at Age 60...........27

Table 10. Variable Annual Withdrawals Based on Life Expectancy, Retirement at Age 65.........28

Table 11. Variable Annual Withdrawals Based on Life Expectancy, Retirement at Age 70.........29

Table B-1. Life Expectancy at Each Age, in Years........................................................................32

Appendix A. What is “Monte Carlo” Analysis?............................................................................31

Appendix B. United States Life Tables, 2004...............................................................................32

Author Contact Information..........................................................................................................33

The 78 million members of the “baby boom” generation are beginning to retire.

years of accumulating assets to spend in retirement, they now must decide how to convert these

assets into a steady stream of income. Because of the trend away from defined benefit (DB)

pensions to defined contribution (DC) plans—such as 401(k) plans—future retirees will be less

likely to have a guaranteed stream of income from defined benefit pensions. Furthermore, while

Social Security will provide a guaranteed income to most retirees, it will replace only a relatively

small proportion of their pre-retirement income. As a result of these trends, many future retirees

will rely greatly on their savings to finance their consumption during retirement.

A retiree who is deciding how to convert wealth into retirement income will have to balance

many risks. Increases in average life expectancy will mean that future retirees will have to ensure

that their wealth will last through a retirement that could span 30 or 40 years. Increased volatility

in equity markets, the effects of inflation on purchasing power, and the possibility of substantial

expenses for medical treatment and long-term care will further complicate this decision.

There are a number of ways to convert retirement wealth into income. One option is to purchase a

life annuity from an insurance company. In exchange for payment of an initial premium, a life

annuity pays a guaranteed income throughout retirement, regardless of how long the purchaser

lives. Life annuities represent only about 5% of individual annuities sold in the United States.

Most annuities sold in the U.S. are deferred annuities, which are tax-deferred retirement savings

accounts. At retirement, a deferred annuity can be converted to a life annuity; yet, relatively few

deferred annuities are converted to life annuities. Likewise, most individuals who have

accumulated retirement savings in 401(k) plans or individual retirement accounts (IRAs) choose

to access these funds through lump-sum distributions or periodic withdrawals rather than by

purchasing a life annuity.

There are many reasons why relatively few retirees choose to purchase a life annuity as a source

of retirement income. Among these reasons are that

• most retirees receive annuity income from Social Security;

• about one-third of retirees receive annuity income from defined benefit pensions;

• the fees charged by annuity providers can be high and the fee structure is not

transparent;

• purchasing an annuity reduces the assets available to the retiree to meet

unexpected expenses, and breaking the contract is costly; and

• there have been instances of deceptive sales practices by some agents, many of

whom receive large commissions for each sale.

Another option for converting retirement savings into income is to take periodic withdrawals

from a retirement account. Some retirees attempt to “self-annuitize” by basing the amount of each

periodic withdrawal from the account on their remaining life expectancies. Retirees who self-

annuitize take on the responsibility of managing their investments and also the risk that they will

live longer than average.

This report has two sections. The first section describes four kinds of risk that retirees face in

retirement: longevity risk, investment risk, inflation risk, and the risk of large, unexpected

expenses for medical care or long-term care. It then describes the basic features of life annuities

and examines some of the reasons that the market for these annuities remains small, in spite of the

protection that they provide against outliving one’s retirement assets. The second section of the

report describes various strategies for self-annuitizing and presents the results of an analysis that

CRS conducted to estimate the probability that an individual who elects to self-annuitize would

exhaust his or her retirement assets before death.

Decisions about how to draw down assets in retirement must take into account many risks. These

include longevity risk (i.e., the risk that one will live beyond the average life expectancy),

inflation risk, investment risk, and the risk associated with unexpected financial shocks from

widowhood, divorce, medical care, or the need for long-term care services.

The U.S. population is living longer than previous generations. A man who reached age 65 in

1960 could expect to live another 13.0 years, while a woman who turned 65 in 1960 had a

remaining life expectancy of 15.8 years. A man who reached age 65 in 2004 could expect to live

another 17.1 years, while a woman who turned 65 in 2004 had a remaining life expectancy of

20.0 years. Looking beyond the averages, more than one-fourth of women who reach age 65 are

likely to live to age 90, and 12.4% are likely to live to age 95. One out of six men who attain age

65 will live to age 90, and an estimated 6% will live to age 95. (See Table 1). Individuals who

underestimate the likelihood of living into very old age might spend their assets too quickly,

depleting their savings while they still have many years to live. This could lead to a decline in

their standard of living, and possibly to increased reliance on public assistance programs.

Table 1. Estimated Percentage of Individuals Age 65 in 2004

Surviving to Selected Ages

As overall prices in the economy rise over time, the purchasing power of income declines unless

income increases at the same rate that prices increase. Few pensions in the private sector provide

regular cost-of-living increases. Data collected by the Department of Labor indicate that fewer

than 5% of pensions in the private sector provide regular cost-of-living adjustments to retirees.

Retirees must account for the potential impact of inflation on their investment portfolios as they

decide how to draw down their retirement wealth.

Social Security is one of the few sources of retirement income that is fully inflation-indexed each

year. Yet, even Social Security’s annual cost-of-living adjustments may not fully protect

retirement income from the effects of inflation. The annual cost-of-living adjustment for Social

Security is based on the increase in the overall level of prices. Prices for some categories of

expenditures that the elderly use at higher rates than younger consumers – such as health care –

have been growing faster than overall prices. Over the period from 1980 to 2007, the average

annual rate of inflation for goods and services averaged 3.5%, but prices for medical care rose at

an average annual rate of 5.9%. For those who have above-average out-of-pocket health care

expenditures, these price increases can significantly reduce the amount of income available for

other expenditures. In addition, Medicare Part B premium increases are tied to program cost

increases which have historically exceeded growth in general inflation. By 2025, these premiums

are expected to equal over 50% of the average Social Security benefit, as compared to 27% in

2006.

Investment risk is the possibility that an individual’s retirement assets might decline in value

because specific stocks, bonds, or other assets depreciate. Diversification can reduce investment

risk, because declines in the value of some assets are likely to be fully or partially offset by gains

in the value of other assets. Stock and bond mutual funds, for example, help protect individuals

from investment risk by purchasing securities from many companies in a variety of industries. In

a mutual fund, investment losses from companies that are performing poorly may be offset by

investment gains from companies that are performing well.

Investment risk includes market risk, which is the possibility that an individual’s retirement assets

will decline in value because of an overall decline in asset prices, as when the stock market falls.

Even a well-diversified portfolio of stocks will not protect the value of an individual’s retirement

from depreciating if stock values fall across the board, as they have in 2008. Although a

diversified portfolio can moderate investment risk, there have been extended periods of time

when declines in the stock market and low interest rates resulted in negative returns on

investment. For example, between 1929 to 2006 there were 14 ten-year periods of negative

annual rates of return after adjusting for inflation. Table 2 shows the five worst of those 14 ten-

year periods. There have also been instances when negative rates of return spanned 15 years.

Between 1966 and 1981, real rates of return on stocks averaged -0.4%, and the average annual

real rate of return on short-term government securities was -0.2%. Those who retire at the

beginning of a period of negative returns could face significant reductions in retirement wealth.

Table 2. Annual Average Rate of Return Over 10- Year Period

Unexpected events can adversely affect retirement income. These include losing a spouse through

death or divorce, high medical expenses, and the need for long-term care services. According to a

recent study by the Urban Institute, more than two-thirds of adults age 70 and older experienced

at least one such financial shock over a nine-year period. Widowhood occurred among nearly

one-third of married adults over age 70. The study found that widowhood was more likely to

reduce women’s wealth than men’s wealth.

Unexpected health care costs can also reduce retirement wealth. Although the majority of retirees

are covered by Medicare, deductibles and co-payments can be significant for those who are

seriously ill. Medicare provides only limited coverage for long-term hospital stays and nursing

home care. A recent analysis of the Health and Retirement Survey found that 6% of households

aged 75-84 paid more than 50% of their income for out-of-pocket medical expenses. These

costs can be especially high at the end of life when a surviving spouse may face large medical

bills and an accompanying reduction in retirement income. A study in 2002 found that out-of-

pocket costs for end-of-life medical expenses could average about $23,000 (adjusted to 2007

dollars). While the introduction of Medicare Part D may offset some of these costs, even under

this program an individual who is not eligible for a low-income subsidy could be responsible for

more than $6,000 in copayments for prescription medicines in 2009.

Another potential economic shock is the cost of long-term care services.

that over two-thirds of individuals aged 65 and older will require long-term care services at some

point in their lives. In 2008, the annual average cost of a nursing home stay is $68,255 for a semi-

private room and $76,285 for a private room. These costs greatly exceed the 2007 median

income of $29,730 among households in which either the householder or householder’s spouse

was 65 or older. Medicare does not cover extended stays in nursing homes, and Medicaid

coverage generally is available only to individuals who are poor or become poor by spending

down their assets on long-term care services. For those who are not Medicaid-eligible and who

have not purchased private long-term care insurance, long-term care costs must be paid out-of-

pocket.

A life annuity – also called an immediate annuity – is an insurance contract that provides income

payments on specified dates in return for an initial premium. Life annuities can help protect

retirees against some of the financial risks of retirement, especially longevity risk and investment

risk. Life annuities pay income to the purchaser for as long as he or she lives, and in the case of

joint-and survivor annuities, for as long as the surviving spouse lives. In addition, some annuities

offer limited protection against inflation through annual increases. However, the annual increases

must be paid for by accepting a lower initial monthly annuity income. Other annuities allow the

purchaser to share in the investment gains from growth in equity markets as a way to offset the

effects of inflation; however, such annuities also require the purchaser to share in the investment

losses if markets fall.

Despite the potential advantages of annuities in reducing longevity risk and investment risk, life

annuities continue to represent a small proportion of all annuities sold in the U.S. In recent years,

deferred annuities – used as tax-deferred savings vehicles – have outsold life annuities by a ratio

of almost 20 to 1. As noted earlier, the irrevocable nature of annuity contracts appears to be an

important factor in dissuading many retirees from purchasing life annuities. Although insurers

have devised options to assure that the payments will continue after the purchaser’s death – joint-

and-survivor annuities, and term-certain annuities, for example – these options add to the cost of

the annuity. Because of the projected increase in the number of older Americans over the next 20

years, and the concurrent increase in the potential market for life annuities, insurers are likely to

continue to add features to their annuity products in an attempt to broaden their appeal to retirees.

Annuities are either provided through employer-sponsored pension plans (the group market) or

are purchased directly by individuals (the individual market).

Annuities from defined-benefit pensions can protect retirees from investment risk and longevity

risk. The benefit formula under most defined benefit pensions is based on the worker’s years of

service and final average salary. The individual’s benefit does not vary with investment returns,

nor does it decrease as a result of increases in average life-expectancy. The employer that

sponsors the plan bears both the investment risk and the longevity risk. If investment returns fall

below expectations, or if the plan’s actuaries project increases in life expectancy, the plan sponsor

must provide additional funding to the plan to meet these costs. Since the mid-1980s, the

proportion of workers who retire with a defined benefit pension has declined substantially. The

number of workers participating in defined benefit pension plans fell from 26.9 million in 1985 to

20.6 million in 2006, a decline of 23%.

Another reason for the decline in the number of workers retiring with an employer-sponsored

annuity has been an increase in the number of defined benefit plans that offer the option of taking

a lump-sum rather than an annuity. Of all those who are covered under a defined benefit plan,

more than half are offered the choice between a lump sum and an annuity. When offered this

choice, many individuals choose the lump-sum option. According to a study by the Vanguard

Group, only 40% of defined benefit plan participants who are offered a lump-sum choose to

receive an annuity.

the form of a defined benefit annuity is likely to decline in the future.

Annuities purchased directly from insurance companies or from insurance agents or brokers by

individuals are called individual annuities. Two features that vary across individual annuities are

the timing of payments and the rates of return.

Annuities can either be deferred, in that premiums are paid and assets are accumulated while the

individual is working and payment of income is deferred until the worker retires, or immediate, in

which a single premium is paid in exchange for lifelong stream of income that begins

immediately.

A deferred annuity is similar to a savings account in which individuals can accumulate money

over time. Investment earnings accrue on a tax-deferred basis. Deferred annuities represent nearly

95 percent of annuity sales. While individuals holding a deferred annuity can convert the funds

into a guaranteed stream of income in retirement, most take a lump-sum payment or a series of

periodic withdrawals.

An immediate annuity provides a guaranteed monthly income for a specified period of time in

exchange for a one-time premium payment. Income from an annuity can be either fixed or

variable. Income from an annuity may be a received for specific number of years, as with a term

certain annuity, for the life of the annuitant, as in a single-life annuity, or for the lives of both the

annuitant and his or her spouse, as in a joint and survivor annuity. Under a joint and survivor

annuity, the surviving spouse is eligible to receive income until he or she dies. The survivor

benefit is typically 50%, 75%, or 100% of the income received while the annuity purchaser was

living. As noted earlier, while immediate annuities can protect retirees from longevity risk and

transfer the investment risk to the insurer, only 5% of individual annuities currently sold are

immediate annuities.

Table 3 shows some examples of monthly income from an immediate annuity. Most annuities

purchased in the private sector provide different incomes for men and women. Men can purchase

an annuity with a higher monthly income than women for the same premium because the average

life expectancy of a man is lower than for a woman of the same age. The average lifetime values

of the annuities for a man and woman of the same age would be equal. Employer-sponsored

pension plans must offer unisex annuities to retirees. Under a unisex annuity, a man would

receive a lower monthly annuity income than he would get from a gender-based annuity,

reflecting the higher average life expectancy of a group that includes both men and women.

Likewise a woman would receive a higher monthly annuity income from a unisex annuity than

she would receive would from a gender-adjusted annuity.

Table 3. Monthly Income By Gender and Joint/Survivor Option

for Fixed Annuity With a $100,000 Premium, 2008

Annuities are similar to other investment vehicles in that purchasers earn a rate of return on their

premium investment. The rate of return on the annuity can be fixed, indexed, variable, or some

combination of the three. The risk, regulation and fee structure varies across these different types

of annuities.

A fixed annuity pays a fixed monthly payment for the term of the annuity. The amount of the

monthly payment is determined at the time the annuity is purchased. The income that any given

premium amount will purchase depends mainly on the age of the purchaser (and the age of the

purchaser’s spouse in the case of a joint and survivor annuity) and prevailing market interest rates

for medium-term bonds at the time the annuity is purchased. In 2007, fixed annuities represented

22% of annuity sales.

Figure 1. Growth in Annuity Sales Relative to Growth in the S&P 500 Index

1998 1999 2000 2001 2002 2003 2004 2005 2006

A variable annuity offers annuity purchasers a choice of a wide range of investment options that

can vary in value over time. The investment options can include stocks, bonds and money market

portfolios. The monthly income provided by a variable annuity will fluctuate according to the

investment performance of the funds in which the annuity premium is invested. Income from a

variable annuity can decline if the investment underlying the annuity loses value. Some variable

annuity policies offer limited protection against declines in value through a guaranteed minimum

income. In 2007, the majority of annuity sales (67%) comprised variable annuities.

An equity-index annuity earns a rate of return based on the performance of an equity index fund.

Examples of equity indexes used include Dow Jones Industrial Average, Lehman Brothers

Aggregate U.S. Index, and Standard and Poor’s (S&P) 500 Composite Stock Price Index. If the

underlying index declines, the income provided by the annuity will also decline. To reduce this

risk, the insurer in some cases may provide a guaranteed minimum payment to protect the

consumer against market fluctuations. This minimum, however, costs more to annuity purchasers

by reducing their initial annuity income. In 2007, 11% of annuity sales were indexed annuities.

The demand for each type of annuity is influenced by rates of return in equity markets. A strong

equity market reduces demand for fixed annuities while a weak stock market increases demand

for fixed annuities. During the strong equity market in late 1990s, growth in variable annuities

sales exceeded growth in fixed annuity sales. By early 2000, when equity markets were earning

negative rates of return, the demand for variable annuities fell. After 2000, growth in variable

annuity sales dominated the market. (See Figure 1.)

Inflation-indexed annuities preserve the purchasing power of annuity income by providing a

lifetime stream of income that increases with inflation. Treasury Inflation Protected Securities

(TIPS) could be used as an investment vehicle by insurers that would like to offer inflation-

indexed annuities, but few U.S. insurers offer such annuities. Potential purchasers have proven

unwilling to accept a substantially lower initial payment in exchange for protection against

inflation. Some insurers offer graded annuities that provide annual increases in payments, which

are typically capped at 3%, but graded annuities do not fully protect against inflation that exceeds

the annual cap.

The taxation of an annuity differs between the accumulation phase and the payout phase.

Deferred annuities receive favorable tax treatment in the accumulation phase. Returns on

investment are not taxed in the year they are earned. During the payout phase, taxation of

annuity income differs between payments taken as withdrawals and payments taken as a

life annuity. In either case, annuity income is taxed at ordinary income tax rates rather

than at capital gains tax rates.

income is received as

• a single lump sum,

• a series of withdrawals that are not annuitized, or

• a life annuity (either variable or fixed).

When funds are withdrawn as a lump sum, the amount of the distribution that exceeds the

amount the annuity owner invested is subject to taxation at the owner’s ordinary income

tax rate.

is considered to consist of investment earnings until all investment gains have been

withdrawn.

Each payment from a fixed annuity is treated as consisting partly of the investment gains, which

are taxable, and partly as a return of principal, which is not taxable. The proportion of each

payment that is excluded from taxable income is determined by dividing the amount that the

individual paid into the annuity by the total amount that would be paid out over an average life

expectancy, according to life tables published by the Internal Revenue Service. Each payment is

multiplied by this ratio to determine the fraction of the annuity payment that is not subject to

income taxes. For example, consider an individual who has paid a $100,000 premium for an

immediate annuity at age 65. In November 2008, a 65 year-old male would receive monthly

income of about $675 for a premium of $100,000. IRS life tables indicate that, on average, the

purchaser will receive annuity payments for 20 years. The total expected lifetime income from

the annuity therefore would be $162,000. The proportion of each payment that would be

excluded from taxable income would be 100,000/162,000, or 61.7%. The other 38.3% percent of

each payment would be subject to income taxes.

With a variable annuity, income varies with the performance of the underlying investments. The

amount of income from a variable annuity that is excluded from taxable income is computed by

dividing the premiums paid for the annuity by the number of years that payments are expected to

be made to the annuitant. For a life annuity, this would be the annuitant’s life expectancy as

determined using IRS tables. In the case of a 65 year-old who had paid a premium of $100,000,

the amount of each monthly annuity payment that would be excluded from taxable income would

be $416.

An exception to the taxation of annuity income was included in the Pension Protection Act of

2006 (P.L. 109-280). Beginning in 2010, withdrawals from annuity contracts that are used to pay

for qualified long-term care insurance premiums are not subject to income tax. However, the

investment in the annuity contract is reduced by the amount of the long-term care insurance

premiums. This means that a larger percentage of the future income received from the annuity

will represent investment gains and will be subject to income taxes.

Table 4. Tax Treatment of Annuities

Consumer protections are intended to ensure that information used to sell an annuity is truthful,

and that individuals who purchase an annuity fully understand the future consequences with

respect to retirement income. The key challenge in the development and enforcement of

consumer protections is that some annuity products are regulated exclusively at the state level

while others are also regulated by federal agencies. Because state governments have primary

jurisdiction over regulation of fixed annuities, there is wide variation in regulation across states.

Further, while variable annuities are regulated by the federal Securities and Exchange

Commission (SEC), equity-index annuities, which are designed to track the performance of a

common stock index, such as the S&P 500 or the Russell 2000, are regulated primarily by the

states.

Like other insurance products, most annuities are regulated by the states. State laws govern the

organization and licensing of insurance companies and their agents and state insurance

departments oversee insurance company operations. These state laws and regulations also govern

marketing and sales practices as well as insurer requirements.

To help guide states in their oversight efforts, the National Association of Insurance

Commissioners (NAIC) has developed language for “model laws and regulations” to provide

guidelines for legislators to modify and adopt in their respective states. The NAIC Model Act has

not been uniformly adopted across states, thus leaving potential gaps in consumer protections. As

of February 2008, 17 states have not adopted the NAIC model regulations on suitability. The

NAIC Model Suitability language requires insurance companies to give objective financial

information to potential purchasers, and it requires agents to use a standardized form to determine

whether an annuity would be suitable for the potential purchaser. Some state laws ban the use of

professional designations or titles – such as Senior Financial Advisor – that might mislead senior

consumers into thinking the advisor has special financial expertise related to the needs of older

consumers.

A similar problem exists with respect to disclosure requirements in annuity contracts. The NAIC

Annuity Disclosure Model Regulation requires certain information to be disclosed, including

information about premiums and how they are charged, a summary of the options and restrictions

for accessing money, and an outline of fees. According to the NAIC, 35 states have adopted the

NAIC disclosure regulation.

The states also play a role in protecting annuity owners against the insolvency of annuity insurers.

To do this, each state has a state guaranty association to provide a financial safety net for each

line of insurance and to ensure that coverage continues if an insurer becomes insolvent. State laws

require insurers to become members of the guaranty associations in every state in which they are

licensed to do business. The actual coverage for annuity contracts varies from state to state, but

cash values and annuity benefits are usually protected up to at least $100,000. However, coverage

is not provided for variable annuity contracts. Variable annuity contracts are held in separate

accounts by insurers and they are protected from the general creditors of the insurance company

in the event of insolvency. Although the Securities Investor Protection Corporation (SIPC)

protects against fraud in variable annuity sales, it does not provide any relief to investors whose

variable annuities decline due to falling prices in equity markets.

Because the assets underlying variable annuities are invested in equities, they are regulated by the

federal government through the Securities and Exchange Commission (SEC). Equity-index

annuities, which are based on market indexes, are not yet regulated by the SEC (see discussion

below). Federal securities laws require certain disclosure documents, including a prospectus, to

be given to investors. Certain disclosure documents must also be filed with the SEC. In addition,

written marketing materials, such as advertisements, are subject to federal regulation.

Federal law prohibits agents who sell variable annuities from making untrue statements of

material fact or failing to state a material fact that is necessary to prevent the statement from

being misleading. Annuity agents also have a fiduciary duty to provide full and fair disclosure

of all material facts to their clients and their prospective clients, including all statements in

advertising materials. Currently, equity-index annuities are regulated by the states, rather then the

SEC. To address this inconsistency in regulation between variable and equity-index annuities, the

Securities and Exchange Commission has issued a proposed regulation that would extend federal

securities laws with respect to full and fair disclosure and sales practice protections to certain

equity-index annuities.

Despite some of the potential advantages of individual life annuities, immediate annuities remain

a small part of retirement assets held in the U.S. Of those who purchased immediate annuities in

2003, the median age of purchase was 70. The Internal Revenue Code requires owners of

individual retirement accounts to begin taking withdrawals from their accounts and including the

withdrawals in their taxable income no later than April of the year in which they reach age 70½.

Some owners of IRAs use immediate annuities to meet the requirement to take these “required

minimum distributions” from their retirement accounts. Without this requirement, demand for

annuities might even be lower.

There are several reasons why the demand for annuities is low despite the aging of the

population. Some potential purchasers may already feel they have a sufficient amount of

annuitized income from Social Security, and about a third of people 65 and older also have

annuity income from defined benefit pensions. Another reason may be the amount and non-

transparency of fees and expenses charged by insurance companies. Further, annuity contracts are

not easily canceled, and many individuals fear that after purchasing an annuity they may later

need a large sum of money to pay for unexpected expenses, such as long-term care or health

expenses. Even among people who understand that it is important to insure against longevity risk,

some fear that they will die before reaching their normal life expectancy, and will end up “losing

the bet” with the insurance company that sold the annuity. Finally, recent adverse publicity about

deceptive sales practices in the annuity market has added to concerns among potential buyers of

immediate annuities.

Annuity providers impose a number of fees and expenses that are complex and are not transparent

to the annuity purchaser. Even a rather “simple” prospectus identifying the various fees can be

more than 50 pages long.

Fees and expenses fall into three main categories: surrender charges, investment fees, and

insurance charges. Fees and expenses vary depending on the type of annuity (fixed or variable).

Both fixed and variable annuities have surrender charges, which are fees for cancelling the

contract before a specified number of years have passed. A typical surrender charge starts at 7%

of the premium in the first year of the contract and declines by 1% a year until it reaches zero.

However, a few companies have surrender charges of up to 15% to 25% of the annuity

premium.

Variable annuities have two additional fees associated with managing assets. These are

investment management fees and insurance charges. Investment management fees cover the cost

of managing the different funds across investment accounts. Investment fees vary depending on

the type of investment portfolio chosen. Insurance charges include administration, sales

commissions, and mortality and expense charges. Mortality and expense charges average 1.15%

of the average value of investment and cover three components of the insurance guarantee:

• Mortality premium or guarantee of income over one’s lifetime;

• Death benefit to protect beneficiaries (also called survivor benefit); and,

• The cost of the minimum income guarantee.

These fees vary by product design. Typically, a fixed annuity with a joint and survivor option is

subject to all three components. However, a variable annuity without a joint and survivor option

and no minimum guarantee would only be subject to the mortality premium.

It is difficult for consumers to identify and understand each fee charged for an annuity. Each

insurer has a different format for disclosing information. The insurance industry has recognized

this problem and has begun to standardize fee disclosure. A group of insurance organizations is

working to develop a simple, standardized disclosure document that presents information about

fees in a consumer-friendly manner.

Another factor affecting the cost of annuities is that people who buy annuities tend to be those

who expect to live longer than average. A person who chooses to purchase an annuity may have

information about his or her health, habits, or family history that the insurance company does not

have regarding their risk of living longer than average. This phenomenon, called “adverse

selection,” leads to higher annuity premiums than insurers would otherwise have to charge if

longevity risk were spread over the entire population. Estimates of the cost of adverse selection

vary. Some studies have found that adverse selection reduces income to annuity purchasers by 4

cents to 10 cents per dollar of premiums paid. A more recent study defined “potential

annuitants” more narrowly to only include those with sufficient wealth to purchase an annuity.

When re-defined in this manner, potential annuitants tend to live considerably longer than

average and thus would receive a better deal from an annuity than the average person. According

to this analysis, the impact of adverse selection on annuity prices is only about half as great as

previously estimated, or about 2 cents to 5 cents per premium dollar. If participation in

individual annuities were broader, the effect of adverse selection would be reduced and

annuitants’ income would be higher.

Once an individual purchases an immediate annuity, the decision is not easily reversible. Most

states require a 10-day look back period during which a buyer can change his or her mind, but

after this, canceling an annuity contract will result in substantial surrender charges. Part of the

lack of flexibility in annuity contracts was recently addressed in the Pension Protection Act of

2006, which allows funds used to purchase an annuity to be withdrawn tax-free to buy long-term

care insurance. It is important to note that long-term care insurance must be purchased well in

advance of actually needing long-term care services. Annuity owners would have to make the

decision to purchase long-term care insurance in the early years of the payout phase.

Dying Before Getting Full Value. Annuity buyers pay for the insurance component of the annuity,

which guarantees a monthly income no matter how long the annuitant lives. Some people are

reluctant to purchase an annuity out of fear that they will die before they get back the premiums

that they paid into it. Joint and survivor annuities and term-certain annuities can assure that

annuity payments will continue even if the purchaser dies earlier than he or she expected, but

these options reduce the monthly payments that the annuitant receives while he or she is living.

Adverse Publicity and Lack of Knowledge About Annuities. Another factor affecting current

demand for annuities may be adverse publicity surrounding false advertising and deceptive sales

practices employed by insurance agents selling equity-index annuities. These practices may have

led some consumers to avoid all annuity products. Equity-index annuities, however, currently

account for only 11% of annuity sales. Further, recent proposals by the SEC to strengthen

regulations for sales of equity-index annuities may help to alleviate consumers’ concerns in the

future.

Although annuities offer protection from longevity risk and investment risk, relatively few people

use their retirement savings to purchase an annuity. Most people choose instead to take periodic

withdrawals from their retirement accounts. Individuals who purchase life annuities transfer the

responsibility for managing assets and the risk of outliving their assets to an insurance company.

In contrast, retirees who “self-annuitize” take on the responsibility of managing their investments

and also the risk of living longer than average. Annuity purchasers, however, give up control over

the assets that they use to purchase their annuities, while those who take periodic withdrawals

have the money in their retirement accounts available to meet large, unexpected expenses that

may arise during retirement.

For those who choose to take periodic withdrawals, there are two basic approaches to taking

money out of their retirement accounts. The first approach attempts to “smooth” consumption

over the period of retirement through equal (inflation-adjusted) withdrawals each year. This

method provides a steady income from year to year, but as the examples presented in this report

will illustrate, it can be difficult to choose a rate of withdrawal that can be sustained in the face of

uncertain life expectancy and variable rates of return on investment. Another strategy for drawing

down retirement assets is to take withdrawals that are based on the individual’s remaining life

expectancy in the year that each withdrawal is taken. This method of withdrawing money from a

retirement account is prescribed by law for the required minimum distributions that all owners of

traditional IRAs and retired owners of 401(k) accounts must begin taking after they reach age

70½ . Under this approach, it is unlikely that the individual will exhaust his or her savings, but

withdrawals can vary substantially from year to year. This can make planning and budgeting

difficult.

An individual who chooses to take periodic withdrawals might want to have some idea how likely

his or her chosen strategy is to succeed. The first step in such an assessment is to define success.

Financial planners often advise clients that they should adopt a method of withdrawing retirement

funds that will result in a high probability that their savings will last 30 to 40 years. Assuming

that 40 years is a reasonable upper bound for the number of years that a retirement account might

need to last, the next task for the retiree is to determine the minimum probability of success that

he or she is willing to accept. There is no fixed standard for the minimum probability of success

that a retiree should be willing to accept for the annual rate of withdrawal that he or she chooses.

For purposes of illustration, we have highlighted in the tables that follow the combinations of

initial withdrawal rates and investment allocations that resulted in an account balance lasting for a

given number of years (or to a given age) in 95.0% or more of our simulations.

One way to evaluate the likely success of a withdrawal strategy is to determine the probability

that the retiree’s assets will last for at least a specific number of years, assuming that rates of

return on investment will vary from year to year. To estimate this probability, CRS developed a

model that estimates how long a sum of money will last, assuming a particular initial rate of

withdrawal and the probable distribution of annual rates of return on investment. The model

estimates the annual rate of return on investment through a Monte Carlo simulation process in

which the rate of return in each year is based on the distribution of annual total returns on stocks

and bonds over the 82 years from 1926 through 2007. (See Appendix A for a description of

Monte Carlo simulation processes.)

It is important to note that an annual rate of withdrawal that minimizes the risk of exhausting a

retirement account – whether by delaying the initial withdrawal or by taking “small” annual

withdrawals – will increase the likelihood that the account will have substantial assets remaining

at the time of the owner’s death. Delaying the initial withdrawal and taking relatively small

withdrawals both help to preserve assets in the event that the individual outlives his or her normal

life expectancy or experiences below-average rates of investment return. However, for the

individual who lives to his or her normal life expectancy and experiences average rates of return

on investment, the end result of successfully reducing the risk of exhausting his or her account

may be a substantial unexpended account balance at the time of death.

CRS estimated the probability that assets would last for five periods of time ranging from 20

years to 40 years at five initial withdrawal rates ranging from 4.0% to 6.0% of the value of the

account when the first withdrawal is taken. We chose this range of withdrawal rates because “a

large body of research on ‘safe’ withdrawal rates for individuals has determined that a real

withdrawal rate in the neighborhood of 4 percent of the initial retirement portfolio has a ‘low’

chance of running out of money,” and because recent research has demonstrated that initial rates

of withdrawal equal to 7.0% or more are very likely to exhaust the assets too rapidly. In the

estimates presented in the following tables, the withdrawal rate is stated as the percentage of the

initial withdrawal from the account. Subsequent withdrawals are equal to the first withdrawal in

constant (inflation-adjusted) dollars, but they may represent a different percentage of the

remaining account balance.

Many financial advisors recommend that retirees should keep 50% or more of their retirement

savings invested in a diversified portfolio of stocks because stocks have historically achieved a

higher long-run average rate of return than bonds. The higher long-run average rate of return on

common stocks compared to bonds acts as a form of longevity insurance for the retiree.

Generally, advisors recommend that the remainder of assets should be invested in bonds and

money market securities that are less susceptible than stocks to large capital losses.

In our analysis, we simulated withdrawals from two portfolios. In one set of simulations, retirees

allocated 65% of assets to the Standard & Poor’s 500 index of stocks and invested the remainder

in AAA-rated corporate bonds. In the second set of simulations, retirees allocated 35% of assets

to the Standard & Poor’s 500 index of stocks and invested the remainder in AAA-rated corporate

bonds. In all of the simulations, withdrawals were taken at the beginning of each year. Accounts

were re-balanced annually so that the portfolio would start each year at the chosen allocation

between stocks and bonds. The model also took into account the correlation between annual

returns on stocks and bonds. The effects of account fees and income taxes were ignored for

purposes of this analysis.

The data presented in Table 5 illustrate the likelihood that savings will last for at least a given

number of years under several initial rates of withdrawal and under the two investment portfolios

described above. Panel 1 of Table 5 shows the probability that a retirement account will last for at

least a given number of years, assuming that 65% of the assets in the account are invested in

stocks represented by the Standard & Poor’s 500 index and 35% of the assets are invested in

AAA-rated corporate bonds. Panel 2 of Table 5 shows the probability that savings will last for at

least a given length of time, assuming that 35% of the assets are invested in stocks and 65% of the

assets are invested in bonds.

The results presented in Table 5 indicate that at an initial withdrawal rate of 4.0%, there is a

98.5% probability that an account invested in this portfolio will last for at least 20 years. The

longer the period of time over which withdrawals are taken, the lower the likelihood that a given

rate of withdrawal will continue to be successful. At a 4.0% initial rate of withdrawal, there is a

92.5% chance that the account will last for 30 years or more, and an 86.8% chance that it will last

for at least 40 years.

For any given number of years, the likelihood of an account lasting for at least that length of time

is lower for higher initial rates of withdrawal. While there is a 98.5% chance that an account will

last for at least 20 years at a 4.0% initial rate of withdrawal, this probability drops to 94.1% for a

5.0% initial rate of withdrawal and to 84.2% for a 6.0% initial rate of withdrawal.

Panel 2 of Table 5 shows the probability that a retirement account in which 35% of assets are

invested in stocks and 65% of assets are invested in bonds would last for at least 20, 30, or 40

years. The results of the simulations indicate that for periods of time of 20 years or more, the

likelihood of exhausting a retirement account is higher for an account with a 65% allocation to

bonds compared to an account with a 65% allocation to stocks.

The likelihood that an account would last for at least 20 years is slightly higher for an account

with a 65% allocation to bonds compared to an account with a 65% allocation to stocks. This

result occurs because a portfolio that is more heavily invested in stocks has a greater chance than

a portfolio mainly invested in bonds of experiencing a large capital loss. If this happens in the

early years of retirement, the account may be depleted rapidly. Over longer periods of time,

however, the probability of running out of money is substantially higher with a portfolio in which

65% of assets are invested in bonds compared to a portfolio in which 65% of assets are invested

in stocks because of the lower expected average annual rate of return on bonds.

Table 5. Estimated Probability a Retirement Account Will Last

for at Least a Specific Number of Years

The estimates presented in Table 5 illustrate the probability that a retirement account will last for

at least a given number of years, assuming a particular initial rate of withdrawal, a specific

allocation of investments between stocks and bonds, and the estimated annual rates of return on

investment. These results are likely to underestimate the likelihood that a particular strategy will

succeed because they do not account for individual mortality. Because not all account owners will

live until the end of a fixed interval of time, the probability of an account having money in it at

the earlier of either the account owner’s death or the end of the interval is greater than the

probability of the account having money in it at the end of the interval. To ignore the possibility

of the account owner dying in any year will result in underestimating the probability that a

particular rate of withdrawal will sustain the account throughout the owner’s lifetime. As other

researchers have noted, the relevant consideration in planning retirement withdrawals is “the

probability of running out of money in the retirement life span, whether that span is shorter or

longer than a predetermined number of years.”

Estimates of the likelihood that a retirement account will last for a particular number of years

should include the probability that the account owner will survive for that number of years. This

can be done by including in the model of retirement withdrawals the probability of the individual

surviving from one year to the next. To incorporate the effect of mortality on the probability that a

retiree will exhaust his or her retirement account before the earlier of either the retiree’s death or

the attainment of a particular age, CRS added two variables to its model. One variable accounts

for the individual’s age at retirement and in each succeeding year, and the other makes each

annual withdrawal conditional on the individual having survived from one year to the next. The

probability of survival from year to year was based on male and female life expectancies taken

from cohort life tables. An individual’s withdrawal strategy was considered to have been

successful if there was a 95.0% or higher probability that he or she had a positive account balance

in the year of death or, if still living, at the ages of 80, 85, 90, 95, and 100.

Table 6, Table 7, and Table 8 show the estimated probabilities that the retirement accounts of

men and women who retire at ages 60, 65, or 70 will last until the ages of 80, 85, 90, 95, and 100,

taking into account the probability of the individual dying in each year. The initial withdrawal

rates range from 4.0% to 6.0% and subsequent withdrawals are assumed to be equal in real value

(i.e., adjusted for inflation) to the initial withdrawal.

of success for two portfolios. In one portfolio, 65% of assets are invested in stocks and 35% in

bonds, and in the other portfolio, 65% of assets are invested in bonds and 35% in stocks.

The data presented in Table 6 show the estimated probabilities of a retirement account lasting

until ages 80, 85, 90, 95, and 100 for men and women retiring at age 60. Table 5 showed that at a

4.0% initial rate of withdrawal from an account invested 65% in stocks and 35% in bonds, there

is a 98.5% chance that a retirement account will last for at least 20 years. The results presented in

Table 6show that under the same set of assumptions, but incorporating the effects of mortality, a

man retiring at 60 has a 99.1% chance that his retirement account will last until at least age 80,

while for a woman retiring at age 60 there is a 98.9% chance that her account will last until at

least age 80. In this case, the probabilities are almost equally high in both Table 5 and Table 6,

and there is little difference in the probabilities of success between and women. Both of these

results change for the probability of success at later ages.

Table 5 showed that at a 4.0% withdrawal rate, the probability of a retirement account lasting for

at least 30 years was 92.5%. The data in Table 6 show that once the effects of mortality are taken

into consideration, there is a 97.3% chance that a man who retires at 60 and takes an initial

withdrawal of 4.0%, will still have some money in the account at age 90, while a woman retiring

at age 60 and taking an initial withdrawal of 4.0% has a 96.1% chance of still having money in

her account on her 90 birthday. The probability that a woman will still have money in her

account at any given age is lower than the probability for a man because a woman has a higher

probability of having survived to that age. (See Appendix B.)

In simulations representing retirement at age 60, a withdrawal rate of 4.0% was successful in

95.0% or more of simulations for both men and women under both investment portfolios at all

ages up to 100. (See Table 6.) A withdrawal rate of 6.0% failed to achieve a 95.0% success rate

for either men or women under either investment portfolio even just to age 80. For both men and

women retiring at age 60, a withdrawal rate of 5.0% or higher carries a high risk that their

retirement accounts will be exhausted before they have attained their normal life expectancies.

Table 7 and Table 8 show that, compared to retiring at age 60, delaying retirement until 65 or 70

can substantially increase the likelihood that an individual will not exhaust his or her retirement

account before he or she dies. In simulations representing retirement at age 65, withdrawal rates

of 4.0% and 4.5% were successful in 95.0% or more of simulations for both men and women

under both investment portfolios at all ages up to 100. (See Table 7.) A withdrawal rate of 6.0%

achieved a 95.0% success rate under either investment portfolio for both men and women only up

to age 80. For men who retire at age 65, a withdrawal rate of 5.5% or higher carries a substantial

risk that their retirement accounts will be exhausted if they live to age 90 or older. For women

who retire at age 65, a withdrawal rate of 5.0% or higher carries a substantial risk that their

retirement accounts will be exhausted if they live to age 90 or older.

In simulations representing retirement at age 70, withdrawal rates of 4.0%, 4.5%, and 5.0% were

successful in 95.0% or more of simulations for both men and women under both investment

portfolios at all ages up to 100. (See Table 8.) A withdrawal rate of 5.5% was successful up to age

100 in 94.7% or more of cases for men under either investment portfolio, but succeeded in 95.0%

or more of cases only to age 90 for women. For men who retire at age 70, a withdrawal rate of

6.0% or higher carries a substantial risk that their retirement accounts will be exhausted if they

live to age 95 or older. For women who retire at age 70, a withdrawal rate of 5.5% or higher

carries a substantial risk that their retirement accounts will be exhausted if they live to age 95 or

older.

Table 6. Probability of Retirement Account Lasting to at Least

a Given Age, Including Mortality Risk, Retirement at Age 60

Table 7. Probability of Retirement Account Lasting to at Least a Given Age,

Including Mortality Risk, Retirement at Age 65

Table 8. Probability of Retirement Account Lasting to at Least a Given Age,

Including Mortality Risk, Retirement at Age 70

The estimates shown in Table 5, Table 6, Table 7, and Table 8 are based on simulations of a

withdrawal strategy that produces annual income of a constant real (inflation-adjusted) amount.

The individual takes an initial annual withdrawal equal to a particular percentage of the account

balance, and all subsequent withdrawals are equal to the real value of the initial withdrawal. In

this respect, the withdrawal strategy mimics an inflation-indexed annuity by providing a steady

annual income. However, an individual who self-annuitizes must, in effect, insure against his or

her own longevity by holding substantial “reserves” in the account to protect against the

possibility of outliving his or her savings. Because insurers diversify the risk of longevity over all

annuity purchasers, they are able to pay a higher annual income than an individual who self-

annuitizes could safely withdraw from his or her account. Taking withdrawals from a retirement

account that are equal to the amount that would be paid by an annuity purchased from an

insurance company exposes the retiree to the risk of outliving his assets.

If a retiree is willing to allow the amount withdrawn from the account to vary from year to year

based on investment returns and remaining life expectancy, he or she can reduce the likelihood of

fully depleting the account before dying. The trade-off for reducing this risk is that the

individual’s annual income will be less predictable. This can make planning and budgeting more

difficult. One such strategy bases each annual withdrawal on the current account balance and the

individual’s remaining life expectancy. This withdrawal rule is mandated under the Internal

Revenue Code for owners of traditional IRAs and retired owners of 401(k) plans after they attain

age 70½ “to ensure that retirees consume their tax-qualified retirement pension accounts instead

of leaving them as bequests for their heirs.”

Table 9, Table 10, and Table 11 show how annual withdrawals that are based on the individual’s

remaining life expectancy in the year that the withdrawal is taken can vary over the course of the

person’s retirement. Table 9 illustrates the variability of withdrawals for a man and a woman each

of whom retires at age 60 with an account balance of $100,000. Table 10 and Table 11 show

withdrawals for men and women who retire at ages 65 and 70, respectively, also with initial

account balances of $100,000. In 2004, a 60 year-old man had a remaining life expectancy of

20.8 years and a woman had a remaining life expectancy of 24.0 years. Under the 1/E(t)

withdrawal rule, the 60 year-old man would withdraw 1/20.8 (4.8%) of his account balance and

the 60 year-old woman would withdraw 1/24.0 (4.2%) of her account. One year later, the 61 year-

old man would have a remaining life expectancy of 20.0 years and would withdraw 1/20.0 (5.0%)

of his account balance. The 61 year-old woman would have a remaining life expectancy of 23.2

years and would withdraw 1/23.2 (4.3%) of her account balance. As the retiree ages, his or her

remaining life expectancy falls, and the percentage of the account that he or she withdraws rises.

At age 75, for example, a man has a remaining life expectancy of 10.7 years and so he would

withdraw 1/10.7 (9.3%) of his remaining account balance. A 75 year-old woman has a remaining

life expectancy of 12.8 years and would withdraw 1/12.8 (7.8%) of her remaining account

balance.

Under this method, annual withdrawals are a continually rising fraction of the remaining account

balance, but the real dollar of value of the withdrawals may rise or fall from year to year,

depending on the investment performance of the retirement account. In a “worst-case scenario”

the aging retiree’s withdrawals would be a rising fraction of a shrinking account balance.

Although the retiree would not fully deplete the account—because the withdrawal is never equal

to 100% of the remaining account balance—if the account shrinks in value due to a decline in

asset values, the withdrawals could grow smaller from year to year.

Withdrawals based on remaining life expectancy will vary in size (measured here in constant

dollars) from year to year. For example, the top panel of Table 9 shows the range of withdrawals

taken between the ages of 60 and 100 by a man who retires at age 60 with an account balance of

$100,000. Across 10,000 simulations, the typical male retiree could expect his annual

withdrawals to range from $4,465 to $9,160 with an average withdrawal of $6,394. In 5% of the

simulations, however, the smallest annual withdrawal was $1,409 or less, and in 5% of the

simulations, the largest withdrawal was $22,803 or more. A woman with the same portfolio

retiring at age 60 could expect her annual withdrawals to range from $3,940 to $9,729 with an

average withdrawal of $6,342. (See Panel 3 of Table 9. In 5% of the simulations, the smallest

annual withdrawal was $1,530 or less, and in 5% of the simulations, the largest withdrawal was

$25,968 or more.

Panels 2 and 4 of Table 9 show that when the simulations were based on a portfolio in which

65% of assets were invested in bonds and 35% invested in stocks, the average withdrawal was

smaller than in the case of the more stock-heavy portfolio. The typical male retiring at 60 with a

$100,000 portfolio invested 65% in bonds and 35% in stocks could expect his annual withdrawals

to range from $4,491 to $7,370 with an average withdrawal of $5,734. (Again, all amounts are in

constant dollars.) In 5% of all the simulations, however, the smallest annual withdrawal was

$1,100 or less, and in 5% of the simulations, the largest withdrawal was $13,184 or more. A

woman with the same portfolio who retires at age 60 could expect her annual withdrawals to

range from $3,985 to $7,465, with an average withdrawal of $5,516. (See Panel 4 of Table 9.) In

5% of all the simulations, the smallest annual withdrawal was $1,349 or less, and in 5% of the

simulations, the largest withdrawal was $14,460 or more.

For individuals who retire at age 65 or at age 70, average annual withdrawals will be higher than

for those who retire at 60 because they will be based on shorter remaining life expectancies.

Table 10 shows the estimated range of withdrawals taken between the ages of 65 and 100 by men

and women who retire at age 65 with initial account balances of $100,000. Table 11 shows the

estimated range of withdrawals taken between the ages of 70 and 100 by men and women who

retire at age 70 with initial account balances of $100,000.

For a man retiring at age 65 with an initial account balance of $100,000 of which 65% is invested

in stocks and 35% is invested in bonds, annual withdrawals could be expected to range from

$5,284 to $9,103 with an average withdrawal of $6,861. (See Panel 1 of Table 10.) In 5% of the

simulations, however, the smallest annual withdrawal was $1,460 or less and in 5% of the

simulations the largest withdrawal was $19,513 or more. A woman with the same portfolio

retiring at age 65 could expect her annual withdrawals to range from $4,653 to $9,293 with an

average withdrawal of $6,582. In 5% of the simulations, the smallest annual withdrawal was

$1,554 or less, and in 5% of the simulations, the largest withdrawal was $22,126 or more.

For a man retiring at age 70 with an initial account balance of $100,000 of which 65% is invested

in stocks and 35% is invested in bonds, annual withdrawals could be expected to range from

$6,241 to $9,552 with an average withdrawal of $7,606. (See Panel 1 of Table 11.) In 5% of the

simulations, however, the smallest annual withdrawal was $1,412 or less, and in 5% of the

simulations, the largest withdrawal was $17,822 or more. A woman with the same portfolio

retiring at age 70, could expect her annual withdrawals to range from $5,504 to $9,282 with an

average withdrawal of $7,083. In 5% of the simulations, the smallest annual withdrawal was

$1,519 or less, and in 5% of the simulations, the largest withdrawal was $19,314 or more.

Table 9. Variable Annual Withdrawals Based on Life Expectancy,

Retirement at Age 60

Table 10. Variable Annual Withdrawals Based on Life Expectancy,

Retirement at Age 65

Table 11. Variable Annual Withdrawals Based on Life Expectancy,

Retirement at Age 70

The uncertainties that retirees face with respect to both life expectancy and annual rates of return

on investment make choosing a withdrawal strategy for their retirement accounts one of the most

complicated financial decisions of their lives. The decision is even more complicated if retirement

assets must be managed over the joint life expectancies of a couple. In light of these complex

considerations, some analysts have suggested that “the withdrawal phase of retirement planning

may well require more professional guidance and expertise than the accumulation phase.”

A retiree who wishes to achieve a predictable annual income can take annual withdrawals that are

equal in inflation-adjusted dollars. An individual who chooses a rate of withdrawal that is too

high risks spending down the account too quickly, possibly leaving the person impoverished. An

individual who chooses a rate of withdrawal that is too low risks spending down the account too

slowly, unnecessarily reducing his or her consumption and leaving substantial assets unspent at

death. On the other hand, the retiree can choose to take withdrawals that vary from year to year

based on the current balance in the account and the retiree’s remaining life expectancy. This

strategy can result in highly variable annual income.

The results of the analysis that CRS conducted indicate that under certain conditions there is a

95.0% or greater probability a man who retires at age 65 will not fully deplete his retirement

account before the earlier of his death or age 90 if his initial withdrawal does not exceed 5.0% of

the account balance and if later withdrawals are equal to the first in inflation-adjusted dollars.

Under the same conditions, there is a 95.0% or greater probability that a woman who retires at

age 65 will not fully deplete her retirement account before the earlier of her death or age 90 if her

initial withdrawal does not exceed 4.5% of the account balance and if later withdrawals are equal

to the first in inflation-adjusted dollars. The results hold for both a portfolio invested 65% in

stocks and 35% in bonds and for one invested 35% in stocks and 65% in bonds.

The weight that individuals assign to each of the risks they face in retirement will vary from

person to person. No one withdrawal strategy will be optimal for everyone. Other researchers

have noted that “overall ... there is no clearly dominant strategy, because all involve trade-offs

between risk, benefit, and bequest measures, and individual preferences may vary.” One way to

balance these risks would be to segregate one’s retirement funds into two or more accounts and

adopt different withdrawal strategies for each. Likewise, one might use some retirement assets to

purchase an annuity while taking withdrawals from one or more accounts using one or more

withdrawal strategies. Many retirees, however, will not have accumulated enough retirement

savings to make these options practical.

Monte Carlo analysis is a method of estimating the probable outcome of an event in which one or

more of the variables affecting the outcome are random. The term was coined by mathematicians

in the 1940s who likened probability analysis to studying the games of chance played in the

casinos of Monte Carlo. One common use of Monte Carlo simulations is to illustrate how the

variability of investment rates of return can affect the balances in a retirement account. The

essence of a Monte Carlo estimation process is to simulate an event many times, allowing the

random variable to vary according to its mathematical mean and variance. Each outcome is then

ranked according to the likelihood of its occurrence. Using Monte Carlo methods, analysts can

estimate not just the result that will occur “on average,” but also the likelihood of results that are

significantly above or below the average. In other words, Monte Carlo methods of estimation

allow us to incorporate into our estimates the element of risk.

Monte Carlo estimation methods utilize not just the average value of a random variable, but also

the distribution of values around the average. For example, rates of return in the stock market

vary from year to year. The nominal rate of return on the Standard & Poor’s 500 index of stocks

averaged 10.3% between 1926 and 2007, but annual rates of return varied widely around this

average, producing a standard deviation of 20.0%. Likewise, while the nominal annual return on

AAA-rated corporate bonds averaged 6.3% between 1926 and 2007, the standard deviation

around this average was 7.0%.

To estimate the likely rate of return that an investment would achieve over a 40-year period, for

example, Monte Carlo simulation software generates a rate of return for each year based on the

distribution of probable rates of return, as derived from historical data. The program then

simulates the 40-year period a second time, again generating a rate of return for each year from

the probability distribution of rates of return. The process is repeated until the simulation is

completed, and thousands of 40-year investment periods have been simulated. The results of the

simulation—in this case, investment rates of return—are then ranked by percentiles. The model

CRS used also accounted for the correlation between the rates of return on stocks and bonds and

the effects of inflation on real annual returns.

In our simulation of a 40-year period in which 100% percent of assets were invested in common

stocks, the mean real rate of return in 10,000 iterations (simulating a 40-year period 10,000 times)

was 7.0%, which is the same as the actual mean real rate of return on common stocks in the

period from 1926 through 2007. (1.103/1.0305 = 1.70) However, in 5% of those 10,000 iterations,

the mean real rate of return over the 40-year period was 1.6% or less, while at the other extreme,

in 5% of the 10,000 iterations, the mean real rate of return over the 40-year period was 12.4% or

more. In terms of evaluating risk, these results imply an expected annual average real rate of

return on common stocks over any given 40-year period of 7.0%, and a 90% probability that the

average annual real rate of return over that period will be between 1.6% and 12.4%.

Table B-1. Life Expectancy at Each Age, in Years