How does the amount that current workers can expect to receive in future Social Security retirement benefits vary by state? This Heritage Foundation analysis shows that future retirement benefits will vary greatly from state to state. It also shows that typical workers in every state can expect to receive returns on their lifetime Social Security payroll taxes that are much lower than those they could expect to have earned from investing those same retirement taxes in even an ultra-conservative portfolio of U.S. Treasury Bonds.

Using federal government data on life expectancy and earnings in each state, Heritage analysts found that:

• Average young double-earner families in every state are likely to experience lifetime losses of potential income that run into hundreds of thousands of dollars. If workers in every state had been allowed to invest the retirement portion of their Social Security taxes in 401(k)-type investment plans, they could expect to accumulate by retirement significantly more income than they can expect to receive under Social Security. Based on a comparison with a portfolio of 50 percent equities and 50 percent U.S. Treasury Bonds, pre-tax losses under Social Security in 1997 dollars for an average two-earner couple born in 1967 with two children range from more than $1.09 million in the District of Columbia to $335,000 in South Dakota. Even when compared with a conservative portfolio composed entirely of U.S. Treasury Bonds, losses in 1997 dollars range from $592,133 in the District of Columbia to $98,517 in North Dakota. The 15 states with the largest dollar losses for young married couples under Social Security are shown in Table 1.

• Social Security pays a very low rate of return to almost all families born since 1945. Average families in every state can expect inflation-adjusted (or real) rates of return from Social Security that fall well below the 7 percent long-term real rate that the Social Security Administration's 1994-1996 Advisory Council found to exist on equities. Real rates of return from Social Security range from 3.83 percent for single-income couples in South Dakota born in 1945 with two children down to a negative 0.77 percent for an average double-earner couple in the District of Columbia born in 1975.

• Single workers fare particularly badly under Social Security. Single workers, who do not benefit from Social Security survivors or spousal benefits, have particularly low rates of return. Returns for single males range from 2.14 percent for a single male in Hawaii born in 1945 down to a negative 2.95 percent for single males in the District of Columbia born in 1975. Because of longer life expectancies, single females fare slightly better under Social Security with returns ranging from 2.78 percent for South Dakota females born in 1945 down to 0.27 percent for single females in the District of Columbia born in 1975. The 15 states with the lowest returns for single males born in 1975 are shown in Table 2

It is important to note that low life expectancies depress Social Security's rate of return by reducing the period during which retirees collect benefits. States with higher earnings levels also tend to have lower rates of return because Social Security is designed to transfer money from high-income to low-income retirees.

Why Do Rates of Return from Social Security Matter?

Social Security's rate of return measures the ultimate effect of the program on the lives of American workers and families. If the rate were nearly to equal what one could achieve from stocks and bonds, then it might make sense to devote current savings to other things besides retirement. But retirement rates of return from Social Security that are significantly poorer than returns from bonds or stocks, even after adjusting for inflation and risk, mean that more savings need to be allocated to future retirement needs.

The ability of individuals to make this important decision depends on seeing clearly their retirement rate of return from Social Security. Knowledge about Social Security's rate of return is especially important for low-income workers, who generally are less able to save additional dollars for retirement than higher-income workers.

Key Assumptions and Methodology
(for details, see Appendix)

• In order to focus just on the individual retirement issues surrounding Social Security, the estimated insurance cost of pre-retirement survivors benefits is subtracted from Old-Age and Survivors Insurance payroll taxes. Thus, only retirement income taxes and benefits are compared. Likewise, the Heritage Social Security model assumes no change in disability insurance. Holding disability insurance payments constant means that the rates of return in this paper reflect only the retirement portion of Social Security's many programs. Likewise, the Heritage Social Security model holds constant the pre-retirement survivors benefits and taxes that support this program.

• Future increases in life expectancy and wages are taken into account and, unless otherwise stated, are consistent with the intermediate assumptions of the Board of Trustees of the Social Security trust funds.

• "Rate of Return" is a statistic widely used to measure the income performance of an investment. It represents the annual rate of increase in the value of an investment and usually is expressed in percentage terms.

• All calculations are adjusted for inflation. Both the employee's and employer's share of payroll taxes are included in the calculations.

• Unless otherwise indicated, the "private" investment alternatives described in this study are based on tax-deferred IRA-type accounts, but with initial contributions not tax-deductible. These accounts may be subject to post-retirement income taxes.

The Heritage estimates of the rate of return help low-income workers to determine whether Social Security will provide them the retirement income they expect by referring only to that portion of Social Security that provides retirement income. The non-retirement components of Social Security are removed from consideration by subtracting pre-retirement survivors benefits and the taxes that support this separate insurance program from the retirement rate of return calculations. Similarly, the Heritage analysis does not include disability insurance taxes or benefits in its retirement rate of return estimates. The study assumes that both the survivors and disability programs will continue unaffected by privatization of the retirement portion of payroll taxes. (See Appendix, Basic Assumptions and Methodology, for a more complete explanation of the methodology of this calculation.)

Until now, debate on the future of Social Security has focused mainly on the future financial solvency of the system. However, to focus only on the future balances of the trust funds ignores the key problem faced by the Social Security program: In its present form, Social Security acts to reduce the potential lifetime wealth of the great majority of current participants. In theory, it would be possible to ensure the program's financial viability by preserving its current form while cutting benefits or raising payroll taxes. However, such solutions, while balancing the trust funds, would reduce Social Security's rate of return even further below its current level.

Defenders of the current Social Security system argue that its rates of return are not a pressing concern because the program was intended to provide a basic retirement income and stopgap benefits for the spouses of deceased workers. Such an argument might be persuasive if Social Security taxes were a minor burden, but Social Security taxes are not low. The Social Security program began in 1937 with a 2 percent payroll tax rate. By 1972, workers were being taxed for the Social Security Old-Age and Survivors Insurance (OASI) program alone at 8.1 percent on the first $21,500 (in 1997 dollars) of earnings. In 1997, workers paid 10.7 percent on the first $65,400 of employment income.¹ According to U.S. Department of Labor data, the average American family now spends a higher proportion of income on Social Security taxes than on housing.²

These high payroll taxes mean that many American workers, especially those at low income levels, have few dollars left over for private, supplemental retirement savings. Many low- and moderate-income families are forced to rely on Social Security as their major, if not sole, form of "wealth." Thus, the key criterion by which the Social Security system should be judged is this: Do workers receive an adequate return from the large amount of taxes they are forced to pay into the system?

Since an understanding of the returns from the current system is necessary to ensure a productive national dialogue on Social Security, economists from The Heritage Foundation's Center for Data Analysis examined the pattern of returns for various groups of Americans. Heritage analysts calculated the rate of return from Social Security for the population as a whole and for workers by income level, family structure, race and ethnicity, age, and gender.³ The publication of Social Security rates of return by geographic area offers still additional information on Social Security to the American public and will better enable them to decide on the best reform.

What Do These Figures Mean?

Table 3 illustrates the foregone income effects of Social Security using the example of a married couple, both of whom were born in 1967, who now have two children, and who both earn the average wage prevailing in their state. The table compares Social Security's retirement benefits for this couple with amounts that they could expect to accumulate by retirement if they were permitted to invest the retirement portion of their payroll taxes in either 30-year U.S. Treasury Bonds or a conservative portfolio composed of 50 percent U.S. Treasury Bonds and a broad equity index that follows the Standard and Poor's 500. The dollar differences between Social Security's benefits and the amounts that could be saved through the private investment of payroll taxes are shown under the heading Loss Under Current Social Security Compared with Private Investment in columns 5 and 6.

The differences in columns 5 and 6 measure the absolute dollar amount that such a family loses under Social Security in each state and the District of Columbia compared with what they could expect to receive had their taxes been placed in private investment accounts. All amounts in Table 3 are expressed in terms of 1997 dollars, which means that taxes paid, projected Social Security benefits, and investment alternatives have been adjusted for inflation.

States with the largest dollar losses under the current Social Security system tend to be highly populated and heavily urbanized, with relatively high earnings and substantial minority populations. The large dollar differentials also reflect the comparatively high taxes that young workers now pay to obtain their future Social Security retirement benefits. The states with the lowest dollar losses tend to be comparatively rural and to have low average earnings combined with relatively high life expectancies.

As can be seen in Map 1, there is a geographic concentration of such states with relatively low absolute-dollar losses from Social Security in the upper Midwest.⁴ There is a concentration of states with large absolute-dollar losses in highly urbanized Eastern, Midwestern, and Western states.

Map 2 illustrates differences across states in Social Security's rate of return for single males.⁵ As can be seen from Map 2, single male workers in Southern states who were born in 1975 have generally low rates of return from Social Security. Single males in a number of large urban states also have low rates of return from Social Security. Higher rates of return for males are found in the upper Midwest and in certain Western states.

Map 3 shows Social Security's rate of return for single females born in 1975.⁶ As is the case with male workers, single female workers in rural Western and upper Midwest states have comparatively higher rates of return than the national average. However, in contrast to the case of single men, single women in heavily urbanized Eastern and Midwestern states tend to fare worse than single women in the South.

Tables 4, 5, 6, and 7 show the real rate of return from Social Security for all family types born between 1935 and 1975. In all states, single workers and dual-income married couples fare worse than single-earner married couples who pay tax on only one earner's wages while collecting benefits for both the worker and the spouse. Single males have the lowest rates of return because of shorter life expectancies. In all cases, the real rate of return on Social Security declines sharply for workers born between 1935 and 1975, with young workers facing the lowest returns. In five states and the District of Columbia, typical single male workers born in 1975 have negative rates of return, which means they can expect to get back less in inflation-adjusted dollars than they pay into the Social Security system.

Conclusion

Social Security's original aim was to help low- and moderate-income workers provide for themselves during their retirement years. However, as this and other Heritage studies have shown, the current Social Security system actually works to decrease the lifetime income of most participants by driving retirement income below what could be achieved through private investments.

What is true at the national level about Social Security's rates of return can be seen in each state, but the degree of permanent income loss varies significantly between states. Social Security imposes particularly heavy burdens on workers in states with low life expectancies or above-average incomes. In general, Southern and heavily urbanized states have the lowest rates of return on the retirement portion of the current Social Security program.

This analysis of the Social Security system probably underestimates the total costs of the current system. It makes no attempt, for example, to include the economic benefits to states that most likely would flow from privatizing all or a portion of Social Security's retirement. Substantially increasing private savings improves the chances for faster economic growth, higher wages, and increased employment.

The superior returns available in bond and stock funds also raise the probability that retirees will be able to leave cash estates to their children or other heirs. Such estates constitute direct infusions of cash into local communities that can be used to open a grandchild's own retirement savings account, expand a local business, or pay for someone's educational or medical expenses.

Although debate on Social Security at times has focused entirely on obscure technical terms (such as "replacement ratios" and "long-range actuarial balance") that mean little to people busy with making a living and raising a family, there is little doubt that the outcome of the debate will be profoundly important to families in every one of the 50 states and the District of Columbia. For example, whether the current system will continue to exist in its present form is a matter of great concern to a young Ohio couple struggling to accumulate enough wealth to give themselves and their children a better life. It is equally crucial to the single Mississippi male who must watch as his biggest source of retirement wealth yields him a negative return.

For almost every type of worker and family in every state--from Alaska to Florida and from Connecticut to New Mexico--retiring under the current Social Security system means having less retirement income and passing on fewer dollars to the next generation than would be the case if that family were allowed to place their Social Security taxes in private retirement accounts.

William W. Beach is John M. Olin Senior Fellow in Economics and Director of the Center for Data Analysis at The Heritage Foundation.

Why Do Social Security Rates of Return
Differ Between States?

The two main factors that drive differences in the return from Social Security are:

1. Earnings. Social Security retirement benefits are based on a worker's taxable earnings. During retirement a low-income worker will be paid a monthly Social Security benefit that is a larger share of his taxable earnings than the benefit paid to a similar high-income worker.

2. Life Expectancy. If life expectancy in a state is low, workers will tend to die sooner and collect fewer Social Security Old Age benefits. In states with longer life expectancies, beneficiaries will live longer and collect more benefits during their lifetimes.

APPENDIX:
BASIC ASSUMPTIONS AND METHODOLOGY

The authors used The Heritage Foundation's Social Security Rate of Return Microsimulation Model to compare the benefits that different types of families can expect to receive from Old-Age and Survivors Insurance (OASI) with the Social Security taxes they pay during their working lives.

The Heritage model treats taxes paid over a worker's lifetime as a series of investments. Social Security's rate of return is the rate of return on these payroll taxes that would buy an annuity equal in value to the Social Security benefits payments. This yield is the difference between OASI benefits payments (after subtracting any applicable income taxes) and the amounts paid to the Old-Age and Survivors Insurance trust fund through payroll taxes. In the model and this paper, all amounts are adjusted for inflation and expressed in terms of 1997 purchasing power.

The Heritage Foundation model includes both portions of OASI taxes: the share paid by employers and the share paid directly by the employee. However, in calculating the return, an amount is removed from taxes paid that is equal to the premium on a term life insurance policy which has the same value as benefits that are paid to children of workers (and the spouse caring for their children) who die before retirement. This means the calculations do not unfairly include the cost of the spousal benefit when figuring the rate of return in terms of retirement income.

Heritage analysts also assume that, from 2015, tax rates will increase by the amount that the Board of Trustees of the Social Security Trust Funds considers to be necessary to finance the OASI benefits contained in current law.

The earnings to which OASI tax rates are applied are based on a proportion of the Social Security Administration's Average Wage Index. Past values of this wage are taken from historical data contained in the 1997 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds, and future wage growth is based on the Trustees' best guess of what the rate of increase in the average wage will be.

All workers are assumed to begin work on their 21st birthday and to continue to work right up to the age at which they become entitled to Social Security's full Old-Age and Survivors benefit. For those retiring in 1997, this is age 65; but under current law, this retirement age is scheduled to increase gradually until reaching 67 for those born in 1960 and later. Earnings are adjusted to reflect the average wage earned in each state as a proportion of the national wage.

The model calculates post-retirement Old-Age and Survivors benefits to individuals according to formulas stipulated in current law and the "best guess" economic assumptions contained in the 1997 Trustees' Report, up to the date on which their life expectancy expires. Neither Disability Insurance taxes nor benefits are included in the model.

The model uses life expectancies drawn from the National Center for Health Statistics' 1992 Life Tables for the United States.⁷ The 1979-1981 Decennial Life Tables produced by the National Center for Health Statistics are used to calibrate life expectancy on a state-by-state basis.⁸

Throughout this study, comparisons are made between what families could accumulate during their working lives if they were able to invest their Social Security Old-Age and Survivors taxes (less the life insurance premium equal to the value of pre-retirement Survivors Insurance benefits) and what they can expect to receive, on average, in Old-Age and Survivors benefits. Different assumptions are entertained regarding the composition of the worker's portfolio of private investments.

For years prior to 1997, the historical inflation-adjusted rates of return on long-term U.S. Treasury Bonds⁹ and U.S. equities¹⁰ are used to determine, respectively, the rate of return on bonds and the rate of return on equities. For the period 1997 onward, Heritage analysts used forecasts of the real rates of return on 30-year long-term U.S. Treasury Bonds to estimate returns on bond investments.

These forecasts were made by WEFA, Inc., an economics consulting firm, and published in its Long-Term Macroeconomic Forecast for October 1997.¹¹ The eventual long-run average of these forecasts is a 2.8 percent real rate of return. The annualized real rate of return on equities is assumed to be 5.7 percent, which lies at the lower boundary of professional estimates of the long-run returns to equities.¹²

The Heritage Foundation Social Security Rate of Return Microsimulation Model

The Heritage Foundation Social Security Rate of Return Microsimulation Model computes the annualized rate of return from Social Security on the basis of the taxes that individuals or couples are projected to pay and the benefits they can expect to receive during their lifetime. The focus of the model is not to provide estimates of the "average" rates of return to existing populations, but rather to use data to construct representative individual and family types and to estimate the rates of return that those representative types will enjoy.

Internal Rate of Return

The internal rate of return is defined as the rate which will set the discounted value of the stream of Social Security Old-Age and Survivors Insurance tax payments (i.e., taxes [T_i]) equal to the discounted stream of income from the system (i.e., benefits [B_i]).

Discount Rate:

r is the discount rate such that:

Taxes

The taxes paid by an individual are calculated by multiplying the individual's taxable earnings and self-employment income in a given year by the OASI tax rate in that year. Each individual is assumed to begin work on his or her 21st birthday and to cease working on the date on which he or she is entitled by law to collect the full Social Security Old-Age benefit.¹³ The OASI tax rate remains at the current law level until the year 2015, after which tax rates are adjusted annually so that income and expenditures of the Old-Age and Survivors Insurance program are equal.¹⁴

The tax revenue in a given year is calculated by means of multiplying the earnings for that person by the OASI tax rate

T_i = x_i*W_i - L_i

where x is the OASI tax rate for year i; W_i is the total taxable wage, salary, and self-employment income for year i; and L_i is an amount equivalent to the value of a life insurance premium equal to the actuarial value of pre-retirement Survivors Insurance coverage.

Earnings

Earnings for workers in each state are assumed to be a proportion of Social Security's Average Wage Index¹⁵ for employed and self-employed workers. This proportion is calculated using the state's average wage as a percentage of average U.S. wages for 1996 as measured and reported by the U.S. Bureau of Labor Statistics.¹⁶

For periods after 1996, the average wage index is assumed to grow at the rate assumed under the "intermediate" projections made by the Social Security Trustees in their 1997 Annual Report.¹⁷ In the case of the Single-Earner Married Couple scenario, it is assumed that one spouse pays no OASI taxes during his or her lifetime. In the case of the Double-Earner Married Couple scenario, each earner is assumed to pay OASI taxes.

Post-Retirement Old-Age and Survivors Benefits

OASI benefits are calculated on the basis of the "bend point" formulas--the earnings levels from which benefit amounts are calculated--as specified under current law. For example, in order to calculate the monthly benefit amount for an individual who first becomes eligible for full Social Security Old-Age Benefits in 1995, the individual's Average Indexed Monthly Earnings (AIME) is calculated according to the formulas contained in current law. Individuals receiving benefits for the first time in 1997 are paid 90 percent of their AIME up to the $437 bend point, 32 percent of any earnings between the $437 and $2,635 bend points, and 15 percent of any amount in excess of $2,635 (up to the maximum amount of earnings which are taxable).

For years after 1997, these bend points are indexed at rates in the "intermediate" range projections made in the 1997 Trustees' Report. Benefits are paid up to the point of the individual's life expectancy. These tables are adjusted to incorporate fully the effect of changes in life expectancy that are estimated by the Trustees of the Social Security Trust Funds to occur over the period 1993-2070. These Social Security benefits may be subject to income taxes.

Survivors Insurance

For married couples, the value of pre-retirement Survivors Insurance--paid to children of deceased covered workers and the spouse taking care of them--is approximated by subtracting from taxes (T_i) the premium required to buy an equivalent term life insurance policy. Covered individuals are assumed to carry two 10-year term life insurance policies over 20 years between the ages of 35 and 55.

For each covered worker turning 35 in 1997 who has two children and earns the national average wage, the Survivors Insurance policy is estimated to be equivalent to a 10-year term life insurance policy worth $295,000. For each average-wage covered worker with two children who turns 45 in 1997, the Survivors Insurance policy is assumed to be equivalent to a 10-year term life insurance policy worth $194,700.

The market insurance annual premiums required to buy every $250,000 worth of insurance (in 1997) were estimated, respectively, to be $167 and $345 for a male and $150 and $230 for a female.¹⁸ The estimates of the life insurance component are indexed to changes in the earner's Primary Insurance Amount,¹⁹ which is used to calculate the worker's retirement benefit.

In the case of the single-earner married couple, each spouse is assumed to be the same age. After retirement, the couple is paid 150 percent of the benefit amount payable to a single beneficiary during the lifetime of the husband. During the period between the death of the husband and the death of the wife, the wife is paid 100 percent of the benefit amount payable to a single recipient.²⁰

Life Expectancy

Life expectancy by worker's age is estimated based on life-expectancy data contained in the National Center for Health Statistics' 1979-1981 State Life Tables.²¹ However, these estimates reflect only the demographic conditions that prevailed in 1979-1980 and do not reflect the long-term upward trend in life expectancy that factors such as improved health care and better nutritional standards will cause.

The Board of Trustees of the Social Security Trust Fund, for example, estimates that between 1997 and 2070, life expectancy at birth will increase by 5.8 years for males and 4.6 years for females and that life expectancy at age 65 will increase by 3 years for females and 2.9 years for males.²² In order to create life expectancy projections that embody these projected trends, it is necessary to adjust the 1979-1980 Life Tables. For workers in each state, these data were adjusted to conform with life expectancy estimates contained in the 1997 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors and Disability Insurance Trust Funds.

First, Heritage analysts made a slight adjustment in the 1979-1981 Life Tables by applying to them an age-weighted index that adjusts for the estimated national increase in life expectancy for each gender over 1979-1981 to 1997:

Q = E + J, and

J = ((O/65)*S + ((65 - O)/65)*X)

where

Q = 1997 "adjusted" static life expectancy;

J = age-weighted increase in life expectancy age between 1992 and 1997;

E = life expectancy based on 1979-1981 "static life tables";

O = age in 1979 (ranges from 5 to 49); and

S and X = respectively, the increase in life expectancy at birth and age 65 over 1979-1981 and 1997.

Second, Heritage analysts recognized that the gains in life expectancy in the post-1997 period will not be uniform across the age distribution. The Social Security Administration estimates that life expectancy at birth will increase much faster than life expectancy at age 65. In order to calculate the gain in life expectancy for individuals between these two points (birth and 65), an age-weighted index is used:

G = ((A/73)*B + ((65 - A)/73)*x')

where

G = overall gain in life expectancy for a particular age group over 1979-1981 to 2070;

A = age in 1997 (ages in the model range from 21 to 65);

B = gain in life expectancy at birth between 1997 and 2070; and

x' = gain in life expectancy at age 65 between 1997 and 2070.

G can be used to construct a projected life table for the single year 2070, where L is life expectancy for each age group in 1997 and G is the gain in life expectancy expected to occur for that particular age between 1997 and 2070:

L = Q + G.

However, this projection must also take into account the fact that life expectancy gains will be distributed over time as well as across the age distribution. The gains in life expectancy projected to occur will be spread across a period between now and 2070. The later a cohort is born, the greater the proportion of this increased longevity will be from where the cohorts can be assumed to benefit. In order to estimate the degree to which a given cohort will benefit from this increase in life expectancy, the following linear weighting equations were used:

"Dynamic" Life Expectancy = Y + R*(G)

where

Y = Q, or life expectancy in 1997;

R = ((2070 - V)/73); and

V = year in which the individual's life expectancy expires.

Endnotes

1. Social Security Administration, Social Security Bulletin, Annual Statistical Supplement for 1997, December 1997.

2. U.S. Department of Labor, Bureau of Labor Statistics, Consumer Expenditure Survey for 1995 (Washington, DC: U.S. Government Printing Office, June 1997), Table A.

3. William W. Beach and Gareth G. Davis, "Social Security's Rate of Return," Heritage Foundation Center for Data Analysis Report No. CDA98-01, January 15, 1998.

4. Map 1 displays data contained in column 6 of Table 3: "Loss Under Current Social Security Compared with Private Investment: 50% Equities/50% U.S. Treasury Bonds."

5. Map 2 displays data contained in column 9 of Table 4: "Real Rate of Return to Social Security by State for Single Males: 1975."

6. Map 3 displays data contained in column 9 of Table 5: "Real Rate of Return to Social Security by State for single Females: 1975."

7. National Center for Health Statistics, Vital Statistics of the United States, 1992 Life Tables, Vol. II, Section 6, 1997.

8. National Center for Health Statistics, U.S. Decennial Life Tables for 1979-81, Volume II, State Life Tables, No. 1-51, 1985.

9. Based on the real rate of return for long-term U.S. Treasury Bonds. The Federal Reserve Board's 10-year to 15-year Treasury Bond Index is used from 1950 to 1975; the 20-year Treasury Bond is used in 1976. From 1977 on, the 30-year bond is used.

10. Based on the real rate of return for the Standard and Poor's 500 Equity Index.

11. WEFA, Inc., formerly known as Wharton Econometric Forecasting Associates, is an internationally recognized economics consulting firm. Fortune 500 companies and prominent government agencies regularly use WEFA's forecasts and consulting products.

12. The 1994-1996 Social Security Advisory Council, for example, found that a long-run real rate of return on equities of 7 percent existed. Report of the 1994-1996 Advisory Council on Social Security, Vol. I: Findings and Recommendations, p. 35.

13. Currently, retirement age is 65. This age is scheduled by law to increase to 67 for workers born after 1960.

14. These tax rates are calculated using the intermediate assumptions in the 1997 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds.

15. As defined in the 1997 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds, p. 208.

16. U.S. Department of Labor, Bureau of Labor Statistics, "Average Annual Pay by State and Industry," news release dated September 11, 1997, Table 1, "State Average Annual Pay for 1995 and 1996 and Percent Change in Pay for All Covered Workers.

17. 1997 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds, pp. 52-62.

18. Based on lowest quotes available for contract from Budgetlife's World Wide Web page: http://www.budgetlife.com, on September 24, 1997.

19. As defined in the 1997 Annual Report of the Board of Trustees of the Federal Old-Age and Survivors Insurance and Disability Insurance Trust Funds, p. 216.

20. All life expectancy data used in this paper show that women have longer life expectancies than men.

21. National Center for Health Statistics, U.S. Decennial Life Tables, 1979-1981, Volume II, Nos. 1-51.

22. Ibid.