SOCIAL SECURITY

In the report, titled Strengthening Social Security and Creating Personal Wealth for All Americans and initially released on December 21, 2001, the President's Commission to Strengthen Social Security (PCSSS) presented three models for modifying the current Social Security program. Each of these models would include provisions for voluntary personal accounts and associated offsets to Social Security retirement benefits based on the earnings of workers who elect to have personal accounts.

This memorandum provides a description of the three models, as we understand them, and estimates of the expected effects of these models on selected aggregate and individual financial measures. The aggregate measures include the financial operations of the combined Trust Funds of the Old-Age and Survivors Insurance (OASI) and the Disability Insurance (DI) programs, aggregate flows and accumulations for personal accounts, effects on annual Federal unified budget balances, and cash flows from the General Fund of the Treasury to the OASDI Trust Funds. The individual measures include expected future total personal account accumulations and expected total benefit levels at retirement, under a range of assumptions. The terms, personal accounts and individual accounts, are used interchangeably in this memorandum.

All estimates are based on the intermediate assumptions of the 2001 OASDI Trustees Report, with additional assumptions related to returns on private securities, individual account and annuity administrative expenses, and individual account participation rates. These assumptions are described later in the memorandum. Estimates shown in this memorandum reflect the efforts of many individuals in the Office of the Chief Actuary, but particularly those of Jason Schultz, Michael Clingman, Michael Miller, Chris Chaplain, and Seung An.

I. Model 1 Specifications: 2-Percent Personal Account with Benefit Offset

a. Basic Provisions--Modification of OASDI Benefits

Under Model 1, OASDI benefit provisions would be unchanged from the specifications of current law. Thus, benefit levels specified in law for those who do not participate in the personal account option would be the same as under current law. However, based on the intermediate assumptions of the 2001 OASDI Trustees Report, OASDI Trust Funds and cash revenue would be insufficient to pay specified benefits through the next 75 years. Thus, under Model 1, as for current law, future modifications of revenue sources and/or benefit provisions would be needed to bring the program into long-range solvency.

b. Individual Accounts and Benefit Offset

Under this model, a voluntary option is provided starting in 2004 for workers covered under the OASDI program to have an amount equal to 2 percent of their OASDI taxable earnings deposited annually in a personal account. This option would be limited to workers who have not yet attained age 55 at the beginning of 2002.

Account contributions would be collected using the existing structure for collecting OASDI payroll tax contributions. In addition, account contributions would be managed by a central authority in a manner similar to that of the Federal Employee Thrift Savings Plan. Initially, available investment choices would be limited to a first tier of options that would include several broad index funds (equity, government bonds, and corporate and other bonds) plus several balanced funds. After several years, the board of the central authority would expand the options to include a second tier for individuals who had accumulated some threshold amount in their account. The second tier, still managed centrally, would offer a range of funds provided by approved private investment firms. The worker would select an investment firm and the funds offered by that firm. For both tiers, the central authority would maintain individual account records and would combine account transactions in aggregate amounts when dealing with the private investment firms.

For workers who participate in the individual account option, retirement and aged survivor benefits payable based on their earnings will be reduced according to a hypothetical account accumulation and annuity computation using a specified "offset yield rate". The offset yield rate for this plan is intended to be (or to average) 3.5 percent over price inflation. In practice, the offset yield rate could be computed as either (a) 3.5 percent above the realized or expected CPI inflation rate or (b) 0.5 percent above the realized or expected market yield on long-term Treasury bonds for each year.

The hypothetical account accumulation at retirement would be equal to the worker's personal account contributions accumulated using the specified offset yield rate for each past year. The retirement (and aged survivor) benefit offset would be equal to the computed amount of a CPI-indexed life annuity purchased with this hypothetical accumulation, and based on the expected future mortality, inflation, and real interest rates used for the intermediate assumptions of the most recent OASDI Trustees Report. Offset annuities would be based on expected unisex mortality for workers who are not married at retirement. Joint and 2/3 survivor life annuities would be computed for workers who are married at retirement, reflecting the actual ages of each spouse.

c. Financing of Individual Account Contributions

Model 1 is described as a flexible framework in which the personal account contributions might be financed entirely as a "redirect" of OASI payroll tax revenue, entirely from the General Fund of the Treasury, or with some combination of the two. Any portion of the contributions based on wages that is financed as a redirect from payroll tax revenue is assumed to be divided equally between employee and employer payroll taxes. Three variations on Model 1 are provided in the financial estimates in this memorandum all of which have a 2-percent total personal account contribution. These are "Model 1 (2+0)" with financing of account contributions entirely from OASI payroll tax revenue, "Model 1 (1+1)" with half (1-percentage-point) of the financing from payroll taxes and the rest from general revenue, and "Model 1 (0+2)" with financing entirely from general revenue.

d. Account Distributions and Taxation

Estimates provided in this memorandum assume that individuals would not have access to personal account accumulations prior to retirement. Allowing such access would diminish the account balance at retirement and thus the available retirement income thereafter. For death before retirement, account balances would be transferred to the account of the surviving spouse, if any, and otherwise to the worker's estate.

Upon entitlement to OASI benefits as a retired worker, aged spouse, or aged surviving spouse, the worker would have access to the account accumulation. Disabled workers would have access to their accounts when they convert to become retired worker beneficiaries. The benefit estimates in this memorandum assume that all account balances would be used to purchase life annuities at retirement. It is assumed that married workers would purchase joint and 2/3 survivor annuities. To the extent that lump-sum distributions are allowed under the model, monthly retirement annuity income would be diminished.

Personal account and annuity distributions would be treated like OASDI benefits for personal income tax purposes.

II. Model 2 Specifications: CPI Indexed OASDI Benefits and 4% (up to $1,000) Personal Account with Benefit Offset

Model 2 includes three basic provisions, an optional personal account with benefit offset, and a provision for additional transfers from the General Fund of the Treasury to the Trust Funds as needed.

a. Basic Provisions--Modification of OASDI Benefits

1) CPI-Indexed Benefits: Modify the primary insurance amount (PIA) formula factors (90, 32, and 15) starting in 2009, reducing them successively by the measured real wage growth in the second prior year. Modified PIA factors would be applicable for OASDI beneficiaries becoming eligible for benefits in 2009 and later. This provision would result in increasing benefit levels for individuals with equivalent lifetime earnings across generations (relative to the average wage level) at the rate of price growth (increase in the CPI), rather than at the rate of growth in the average wage level as in current law. Calculation of the average indexed monthly earnings (AIME) used in computing the PIA would be unaffected by this provision. This provision alone would increase the size of the long-range OASDI actuarial balance (reduce the actuarial deficit) by an estimated 2.07 percent of taxable payroll.

2) Enhanced Benefit Level for Low Earners: This provision would gradually raise the PIA starting 2009 with an ultimate increase for 2018 and later of 40.4 percent (relative to the level provided under provision 1 above) for a 30-year minimum wage worker.1 The combined effect of provisions 1 and 2 for such workers is expected to be a PIA equal to 120 percent of the aged poverty level for 2018. Thereafter the PIA would be indexed by the CPI as specified in provision 1, which is the same rate of growth specified for the poverty level.

The provision would provide the same 40.4 percent increase for 30-year workers with average earnings below that of the 30-year minimum wage worker. This 40.4 percent increase would be reduced for workers with higher career-average earnings levels (AIME), reaching 0 for those with AIMEs at twice the level of a 35-year minimum wage worker. For workers with more than 30 years of work, the percentage increase is maintained at the same level as specified for workers with the same AIME level and only 30 years of work. However, the percentage increase is reduced for workers with fewer than 30 years of work, reaching 0 for those with 20 or fewer years of work. Thus, no enhancement is provided by this provision for retirees with 20 or fewer years of employment. The year-of-work requirements would be "scaled" to the length of the elapsed period from age 22 to benefit eligibility for workers who become disabled or die before reaching age 62.2 The incremental effect of this provision after provision 1 would be to reduce the size of the long-range OASDI actuarial balance by an estimated 0.13 percent of taxable payroll.

The table below illustrates the effect of the benefit enhancement for workers with low earnings.

The benefit enhancement under this provision would be computed according to the following formula:

For all workers whose AIME is less than twice the AIME for a 35-year minimum wage worker, the PIA is multiplied by

1 + applicable percentage x AIME factor x coverage factor.

In the above formula,

"Applicable percentage" is equal to 4.04 percent for beneficiaries initially eligible in 2009, 8.08 percent for those initially eligible in 2010, ..., and 40.4 percent for those initially eligible in 2018 and later;
"AIME factor" is equal to

Here,

A = Twice the AIME of a 35-year minimum wage worker and

M = AIME for a 30-year minimum wage worker.

"Coverage factor" is equal to

In the above formula for the coverage factor, "QCs" represents the number of quarters of coverage earned by the worker prior to benefit eligibility. "Elapsed years" represents the number of years starting with the year the worker attains age 22 through the year prior to benefit eligibility, excluding periods of disabled worker entitlement.

3) Increased Benefits for Widow(er)s: Starting 2009, pay all aged surviving spouses (aged 62 or older) 75 percent of the benefit that would be received by the couple if both were still alive (including all applicable actuarial reductions and delayed retirement credits), if this is higher than their current benefit. The benefit provided by this option would be limited to what the survivor would receive as a retired worker beneficiary with a PIA equal to the average PIA of all retired worker beneficiaries for December of the year prior to becoming eligible for this option. Actuarial reduction for this limitation would be computed as if the survivor had begun receiving a retired worker benefit on the earliest of the actual ages upon which benefits began as an aged spouse, an aged surviving spouse, or a retired worker beneficiary, but not before 62. The incremental effect of this provision after provisions 1 and 2 would be to reduce the size of the long-range OASDI actuarial balance by an estimated 0.08 percent of taxable payroll.

The total combined effect of the basic provisions 1-3 would be to increase the size of the long-range OASDI actuarial balance (reduce the actuarial deficit) by an estimated 1.87 percent of taxable payroll.

b. Individual Accounts and Benefit Offset

Under this model, a voluntary option is provided starting in 2004 for workers covered under the OASDI program to have an amount equal to 4 percent of their OASDI taxable earnings, up to $1,000 (value for 2002, and wage indexed thereafter) deposited annually in a personal account. This option would be limited to workers who have not yet attained age 55 at the beginning of 2002.

Account contributions would be collected using the existing structure for collecting OASDI payroll tax contributions. In addition, account contributions would be managed by a central authority in a manner similar to that of the Federal Employee Thrift Savings Plan. Initially, available investment choices would be limited to a first tier of options that would include several broad index funds (equity, government bonds, and corporate and other bonds) plus several balanced funds. After several years, the board of the central authority would expand the options to include a second tier for individuals who had accumulated some threshold amount in their account. The second tier, still managed centrally, would offer a range of funds provided by approved private investment firms. The worker would select an investment firm and the funds offered by that firm. For both tiers, the central authority would maintain individual account records and would combine account transactions in aggregate amounts when dealing with the private investment firms.

For workers who participate in the individual account option, retirement and aged survivor benefits payable based on their earnings will be reduced according to a hypothetical account accumulation and annuity computation using a specified "offset yield rate". The offset yield rate for this plan is intended to be (or to average) 2 percent over price inflation. In practice, the offset yield rate could be computed as either (a) 2 percent above the realized or expected CPI inflation rate or (b) 1 percent below the realized or expected market yield on long-term Treasury bonds for each year.

The hypothetical account accumulation at retirement would be equal to the worker's personal account contributions accumulated using the specified offset yield rate for each year. The retirement (and aged survivor) benefit offset would be equal to the computed amount of a CPI-indexed life annuity purchased with this hypothetical accumulation, and based on the expected future mortality, inflation, and real interest rates used for the intermediate assumptions of the most recent OASDI Trustees Report. Offset annuities would be based on expected unisex mortality for workers who are not married at retirement. Joint and 2/3 survivor life annuities would be computed for workers who are married at retirement, reflecting the actual ages of each spouse.

c. Financing of Individual Account Contributions

Model 2 is a framework in which the personal account contributions would be financed entirely as a "redirect" of OASI payroll tax revenue. Contributions based on wages are assumed to be divided equally between employee and employer payroll taxes.

d. Account Distributions and Taxation

Estimates provided in this memorandum assume that individuals would not have access to personal account accumulations prior to retirement. Allowing such access would diminish the account balance at retirement and thus the available retirement income thereafter. For death before retirement, account balances would be transferred to the account of the surviving spouse, if any, and otherwise to the worker's estate.

Upon entitlement to OASI benefits as a retired worker, aged spouse, or aged surviving spouse, the worker would have access to the account accumulation. Disabled workers would have access to their accounts when they convert to retired worker beneficiaries. The benefit estimates in this memorandum assume that all account balances would be used to purchase life annuities at retirement. It is assumed that married workers would purchase joint and 2/3 survivor annuities. To the extent that lump-sum distributions are allowed under the model, monthly retirement annuity income would be diminished.

Personal account and annuity distributions would be treated like OASDI benefits for personal income tax purposes.

e. Provision for Additional Transfers from the General Fund of the Treasury

For any year in which the combined OASDI Trust Funds would fall below 100 percent of annual program cost, transfers would be made from the General Fund of the Treasury to maintain the Trust Funds at a level equal to annual outgo. This provision is intended to assure adequate financing during the "transition" associated with the individual account provision described above. To the extent to which workers choose to participate in the personal account, payroll tax revenue will be redirected from the Trust Funds beginning 2009, but benefit offsets associated with this option will not rise to substantial levels for many years. This provision would maintain OASDI solvency during the period for which individual accounts would reduce the net cash flow to the Trust Funds. This provision would have the additional effect of assuring that the OASDI Trust Funds would never become exhausted and thus the program would always remain solvent in the future.

III. Model 3 Specifications: Longevity Indexed OASDI Benefits and 2.5% (up to $1,000) Personal Account with Benefit Offset

Model 3 includes six basic provisions, an optional personal account with benefit offset, and a provision for additional transfers from the General Fund of the Treasury to the Trust Funds as needed.

a. Basic Provisions--Modification of OASDI Benefits and Dedicated Revenue

1) Longevity-Indexed Benefits: This provision would slow the growth across generations in the primary insurance amount (PIA) for all OASDI beneficiaries by an amount that would roughly offset the effects of increasing longevity on the average duration of benefit receipt for aged beneficiaries. Initially, PIA factors (90, 32, and 15) would be scheduled to be adjusted by a successive multiplier of 0.995 annually beginning 2009. This is about one-half the expected effect of "CPI-Indexing". This adjustment reduces monthly benefit levels by an amount equivalent to increasing the normal retirement age (NRA) for retired workers by enough to maintain a constant life expectancy at NRA, for any fixed age of benefit entitlement. Calculations of this adjustment use the mortality assumptions for the intermediate estimates of the 2001 OASDI Trustees Report and the actuarial reduction factors in current law. Under this provision, the 0.995 multiplier would be updated every 10 years (starting after 2010) to reflect actual historical increases in longevity as determined by the Social Security Administration for the most recent decade (as 2000 to 2010 for the first adjustment) and actuarial reduction factors in current law (without regard to provisions 2, 3, or 4 of this model). Note that this provision would apply in addition to the NRA increase already scheduled in current law. This provision alone would increase the size of the long-range OASDI actuarial balance (reduce the actuarial deficit) by an estimated 1.17 percent of taxable payroll.

2) Reduce Benefits for High Earners: Gradually reduce the third PIA factor, from 15 to 10, by 0.25 per year from 2009 through 2028. This reduction would be applied each year to the original 15 factor, prior to applying the cumulative effect of provision 1. This provision alone would increase the size of the long-range OASDI actuarial balance (reduce the actuarial deficit) by an estimated 0.16 percent of taxable payroll. The incremental effect of this provision after provision 1 would be to increase the size of the long-range OASDI actuarial balance by an estimated 0.14 percent of taxable payroll.

3) Enhanced Benefit Level for Low Earners: This provision would gradually raise the PIA starting 2009 with an ultimate increase for 2018 and later of 12 percent (relative to the level provided under provisions 1 and 2 above) for 30-year minimum wage worker.³ The combined effect of provisions 1, 2, and 3 for such workers is expected to be a PIA equal to 100 percent of the aged poverty level for 2018. Thereafter, the PIA would increase from one generation to the next at a rate that is expected to be about 0.5 percent per year faster than the growth in the CPI and the poverty level. Thus, PIA levels for such workers would be expected to rise to levels above 100 percent of the aged poverty level after 2018.

The provision would provide the same 12 percent increase for 30-year workers with average earnings below that of the 30-year minimum wage worker. This 12 percent increase would be reduced for workers with higher career-average earnings levels (AIME), reaching 0 for those with AIMEs equal to one-twelfth the average wage indexing series (AWI) for the second year prior to benefit eligibility. For workers with the same AIME levels, the percentage increase is raised for those with more than 30 years of work, reaching about 1.5 times as much (up to 18 percent) for those with 40 years of work or more. However, the percentage increase is reduced for workers with fewer than 30 years of work, reaching 0 for those with 20 or fewer years of work. Thus, no enhancement is provided by this provision for retirees with 20 or fewer years of employment. The year-of-work requirements would be "scaled" to the length of the elapsed period from age 22 to benefit eligibility for workers who become disabled or die before reaching age 62.⁴ The incremental effect of this provision after provisions 1 and 2 would be to reduce the size of the long-range OASDI actuarial balance by an estimated 0.13 percent of taxable payroll.

The following table illustrates the effect of the benefit enhancement for workers with low earnings.

The benefit enhancement under this provision would be computed according to the following formula:

For all workers with AIME less than one-twelfth the AWI for 2 years prior to eligibility, the PIA is multiplied by

1 + applicable percentage x AIME factor x coverage factor.

In the above formula,

"Applicable percentage" is equal to 1.2 percent for beneficiaries initially eligible in 2009, 2.4 percent for those initially eligible in 2010, ... , and 12 percent for those initially eligible in 2018 and later;
"AIME factor" is equal to

Here,

A = AWI for second year before eligibility, divided by 12 and

M = AIME for a 30-year minimum wage worker.

Note that A as defined for Model 3 is different than A as defined for Model 2.

"Coverage factor" is equal to the greater of zero and

1 + B x (QCs - 3 x elapsed years) / elapsed years.

with

In the above formula for the coverage factor, "QCs" represents the number of quarters of coverage earned by the worker prior to benefit eligibility. "Elapsed years" represents the number of years starting with the year the worker attains age 22 through the year prior to benefit eligibility, excluding periods of disabled worker entitlement.

4) Modify Actuarial Reduction and Increment Factors: The early retirement reduction factors and delayed retirement credits would be changed in an attempt to reflect the fact that the marginal increase in the full benefit level (i.e., the PIA) for earnings after reaching retirement eligibility age is, generally, relatively small. (Reduction and increment factors provided under current law are intended to provide actuarially equivalent lifetime benefits for a fixed earnings history regardless of the age at which retirement benefits start.) This relatively small marginal increase results from both the AIME formula, which uses 35 years of earnings, and the weighted PIA benefit formula. Together, these provide a larger marginal amount of benefit per dollar of additional earnings for low earners and for earnings earned early in a worker's career.

This provision is intended to provide a greater marginal incentive to work past the retirement earliest eligibility age (EEA). Because the degree of this marginal effect depends upon the extent and level of earnings a worker has had in earlier years, no absolute adjustment can be provided that would be appropriate for all workers. Rough estimates of adjustments to the reduction and increment factors have thus been developed.

The chart below displays the proposed monthly early retirement reductions that would be applicable for retired worker beneficiaries for the first 36 months for which benefits are received prior to NRA under both current law and the provision. (Different factors apply to aged spouse beneficiaries and aged widow beneficiaries.)

Similar increases for aged spouse beneficiaries would be applied, increasing the monthly reduction for the first 36 months of entitlement before NRA from 25/36 percent under present law to 30/36 percent under the provision.

The reductions that are proposed for the fourth and fifth year of benefit entitlement before NRA are 12/24% per month (current law reductions are 10/24% per month) for both retired worker and aged spouse beneficiaries. The reductions for the fourth and fifth year of entitlement before NRA are applicable to all new eligibles who reach age 62 after 2008.

The ultimate percentages of PIA payable for retired workers by age at initial benefit entitlement are shown in the table below.

The percentage of PIA payable for non-disabled aged widow beneficiaries newly eligible at age 60 would remain at 71.5 percent. The percentages payable for those newly eligible at ages between 60 and the NRA would scale linearly between 71.5 and 100 percent, as under present law.

The delayed retirement credit (DRC) under present law is scheduled to increase to 8% per year for workers attaining age 65 after 2007. Under this provision, the DRC would continue to increase at the rate of 0.5 percentage point every two years, with the first new increase applied to those attaining age 65 in 2010. An ultimate factor of 10 percentage points per year would be reached for workers reaching 65 after 2015. The delayed retirement credit applies for those months between NRA and age 70 in which no retired worker benefit is received.

Provision 4 alone would increase the size of the long-range OASDI actuarial balance (reduce the actuarial deficit) by an estimated 0.28 percent of taxable payroll.

5) Dedicated Transfers: Provide for dedicated transfers from the General Fund of the Treasury to the Trust Funds that would be specified in the law as percentages of OASDI effective taxable payroll on a year-by-year basis for years 2005 and later. The specified transfers are equal in size to the estimated net revenue that would be expected under two provisions (neither of which is specifically included in the model) based on estimates under the intermediate assumptions of the 2001 Trustees Report. One of these provisions is an increase in the OASDI taxable maximum that would raise the percentage of covered earnings taxable gradually to 86 percent between 2005 and 2009, and increase the level to maintain 86 percent thereafter. The other provision redirects the portion of the revenue from the taxation of OASDI benefits that is currently scheduled for the Medicare HI Trust Fund to the OASDI Trust Funds, phased in 10 percent in 2010, 20 percent in 2011, ..., and 100 percent in 2019 and later. The Commission did not endorse these two provisions upon which the amount of the transfer is based. In fact, the Commission recommends that the Congress consider a number of possible proposals that might provide the revenue specified under this provision. This provision alone would increase the size of the long-range OASDI actuarial balance (reduce the actuarial deficit) by an estimated 0.63 percent of taxable payroll.

6) Increased Benefits for Widow(er)s: Starting 2009, pay all aged surviving spouses (aged 62 or older) 75 percent of the benefit that would be received by the couple if both were still alive (including all applicable actuarial reductions and delayed retirement credits), if this is higher than their current benefit. The benefit provided by this option would be limited to what the survivor would receive as a retired worker beneficiary with a PIA equal to the average PIA of all retired worker beneficiaries for December of the year prior to becoming eligible for this option. Actuarial reduction for this limitation would be computed as if the survivor had begun receiving a retired worker benefit on the earliest of the actual ages upon which benefits began as an aged spouse, an aged surviving spouse, or a retired worker beneficiary, but not before 62. This provision alone would reduce the size of the long-range OASDI actuarial balance by an estimated 0.08 percent of taxable payroll.

The total combined effect of the basic provisions 1-6 would be to increase the size of the long-range OASDI actuarial balance (reduce the actuarial deficit) by an estimated 1.94 percent of taxable payroll.

b. Individual Accounts and Benefit Offset

Under this model, a voluntary option is provided starting in 2004 for workers covered under the OASDI program to have an amount equal to 2.5 percent of their OASDI taxable earnings, up to $1,000 (value for 2002, and wage indexed thereafter) deposited annually in a personal account. This option would be limited to workers who have not yet attained age 55 at the beginning of 2002. Participation in this option would require that the worker contribute an additional 1 percent of OASDI taxable earnings to the personal account each year. The 1-percent additional contribution would be subsidized in a progressive manner with a refundable tax credit that would be expected to have a cost (to the General Fund of the Treasury) of about 0.15 percent of OASDI taxable earnings if all workers participated.

Account contributions would be collected using the existing structure for collecting OASDI payroll tax contributions. In addition, account contributions would be managed by a central authority in a manner similar to that of the Federal Employee Thrift Savings Plan. Initially, available investment choices would be limited to a first tier of options that would include several broad index funds (equity, government bonds, and corporate and other bonds) plus several balanced funds. After several years, the board of the central authority would expand the options to include a second tier for individuals who had accumulated some threshold amount in their account. The second tier, still managed centrally, would offer a range of funds provided by approved private investment firms. The worker would select an investment firm and the funds offered by that firm. For both tiers, the central authority would maintain individual account records and would combine account transactions in aggregate amounts when dealing with the private investment firms.

For workers who participate in the individual account option, retirement and aged survivor benefits payable based on their earnings will be reduced according to a hypothetical account accumulation and annuity computation using a specified "offset yield rate". This hypothetical account and annuity computation would reflect only the personal account contributions provided as a redirect of payroll taxes (i.e., the 2.5 percent up to $1,000). The offset yield rate for this plan is intended to be (or to average) 2.5 percent over price inflation. In practice, the offset yield rate could be computed as either (a) 2.5 percent above the realized or expected CPI inflation rate or (b) 0.5 percent below the realized or expected market yield on long-term Treasury bonds for each year.

The hypothetical account accumulation at retirement would be equal to the worker's personal account contributions (excluding the additional 1-percent) accumulated using the specified offset yield rate for each year. The retirement (and aged survivor) benefit offset would be equal to the computed amount of a CPI-indexed life annuity purchased with this hypothetical accumulation, and based on the expected future mortality, inflation, and real interest rates used for the intermediate assumptions of the most recent OASDI Trustees Report. Offset annuities would be based on expected unisex mortality for workers who are not married at retirement. Joint and 2/3 survivor life annuities would be computed for workers who are married at retirement, reflecting the actual ages of each spouse.

c. Financing of Individual Account Contributions

Model 3 is a framework in which the voluntary 1-percent additional personal account contributions would be provided by the worker, with a progressive subsidy from the General Fund of the Treasury, as described above. For those who participate in the 1-percent additional contribution, the 2.5-percent (up to $1,000) personal account contribution would be financed entirely as a "redirect" of OASI payroll tax revenue. Contributions redirected from payroll tax revenue based on wages are assumed to be divided equally between employee and employer payroll taxes.

d. Account Distributions and Taxation

Estimates provided in this memorandum assume that individuals would not have access to personal account accumulations prior to retirement. Allowing such access would diminish the account balance at retirement and thus the available retirement income thereafter. For death before retirement, account balances would be transferred to the account of the surviving spouse, if any, and otherwise to the worker's estate.

Upon entitlement to OASI benefits as a retired worker, aged spouse, or aged surviving spouse, the worker would have access to the account accumulation. Disabled workers would have access to their accounts when they convert to retired worker beneficiaries. The benefit estimates in this memorandum assume that all account balances would be used to purchase life annuities at retirement. It is assumed that married workers would purchase joint and 2/3 survivor annuities. To the extent that lump-sum distributions are allowed under the model, monthly retirement annuity income would be diminished.

Personal account and annuity distributions would be treated like OASDI benefits for personal income tax purposes.

e. Provision for Additional Transfers from the General Fund of the Treasury

For any year in which the combined OASDI Trust Funds would fall below 100 percent of annual program cost, transfers would be made from the General Fund of the Treasury to maintain the Trust Funds at a level equal to annual outgo. This provision is provided to address the "transition costs" associated with the individual account provision described above. To the extent to which workers choose to participate in the personal account, payroll tax revenue will be redirected from the Trust Funds beginning 2009, but benefit offsets associated with this option will not rise to substantial levels for many years. This provision is intended to maintain OASDI solvency during the period for which individual accounts would reduce the net cash flow to the Trust Funds. This provision would have the additional effect of assuring that the OASDI Trust Funds would never become exhausted and thus the program would always remain solvent in the future.

IV. Assumptions Used for Financial Estimates

All estimates provided to the Commission have been based on the intermediate assumptions of the 2001 OASDI Trustees Report. This includes the ultimate assumption of a 3-percent ultimate real annual yield on long-term U.S. Treasury bonds (based on the effective market yield of all marketable Treasury bonds with a remaining duration of more than 4 years). A number of additional assumptions have been made for these estimates, as indicated below.

a. Personal Account Participation

Participation in the personal accounts would be optional in each of the three models developed by the Commission. The proportion of workers who would voluntarily participate cannot be determined with any degree of certainty. For this reason, estimates of the aggregate financial status of the Trust Funds, the effect on the Federal Unified Budget balance, and the effect on individual benefit levels are presented in this memorandum for three different levels of participation, 0 percent, 67 percent, and 100 percent.

Estimates for the basic provisions of each model represent the aggregate financial effects assuming no voluntary participation in personal accounts. Estimates presented for 67-percent participation are based on the assumption that two thirds of all potential personal account contributions are made. This condition could exist if two thirds of workers at every level of earnings participated. This condition could also be met, for example, if more than two thirds of high earners participated and less than two thirds of the remaining earners participated. Due to the size of the personal account contributions and the nature of the benefit offset provisions, aggregate financial estimates for these models are not very sensitive to the precise distribution of participation rates by earnings level, assuming that two thirds of all potential personal account contributions are made.

However, due to the nature of the three models, their likely levels of participation would differ. For Model 1, participation would be expected to be well below 100 percent because the benefit offset for participants would be expected to exceed the annuity distribution from the accumulation in a conservatively invested personal account (for example an account invested solely in long-term U.S. Treasury bonds). However, individuals who are interested in investing a substantial portion of their account in equities could expect to gain from participating. Thus, the assumption for 67-percent participation is likely to be the most appropriate of the three assumptions for Model 1.

For Model 2, participation would be expected to be higher. If the benefit offset yield rate is computed as 2 percent above the realized or expected inflation rate, actual net yields on personal accounts would generally, but not always, exceed the benefit offset yield rate. Due to this uncertainty, the 67-percent participation assumption is likely to be the most appropriate of the three assumptions in this case. However, if the benefit offset yield rate were computed as 1 percent below the realized or expected market yield on long-term Treasury bonds, 100 percent participation is the most appropriate assumption. Near universal participation is assumed in this case because Model 2 would provide for a benefit offset such that participants would gain by having an account as long as their individual account real yields (net of administrative expenses) are not 1 percent or more below what would have been achieved by investing solely in long-term Treasury bonds. Thus, even the most conservative investor could invest solely in Treasury bonds and be assured of coming out ahead as a result, as long as administrative expenses are less than 100 basis points (this is assumed to be true for the specified accounts).

For Model 3, less than 100 percent participation would be expected, and the 67-percent assumption is likely to be the most appropriate of the assumptions considered. Participation under Model 3 would be lower than under Model 2 for two reasons. First, in order participate, workers would need to make an additional contribution "out of pocket" of 1 percent of OASDI taxable earnings. Even with a subsidy of up to one half from the General Fund of the Treasury, this additional contribution would result in many low earners not participating. Second, for the personal account contribution that is financed by redirecting a portion of the worker's payroll taxes, the benefit offset that will later be applied is greater than under Model 2. This would reduce somewhat the likelihood of a net gain from opting for the personal account (assuming the benefit offset yield rate is computed as 2.5 percent over realized or expected inflation) or reduce the size of the assured net gain for the conservative investor (assuming the benefit offset yield rate is computed as 0.5 percent below the realized or expected market yield on Treasury bonds.).

b. Personal Account Accumulation

Workers are assumed to maintain personal-account portfolios that would have an average distribution of 50 percent in equity, 30 percent in corporate bonds, and 20 percent in U.S. Treasury long-term bonds. Equities are assumed to have an ultimate real annual yield of 6.5 percent, and corporate bonds are assumed to have an ultimate real annual yield of 3.5 percent, or one half of one percentage point higher than assumed for long-term U.S. Treasury bonds. An ultimate assumption of an annual administrative expense of 30 basis points is assumed for the accounts in all three models, consistent with the specifications of the account management.

These assumptions are critical for estimates of the expected effect of possible portfolio choices and yields on benefit levels. Thus, estimates of expected benefit levels for individuals under the models cover a range of possible yields, in order to provide a sensitivity analysis. On the other hand, aggregate financial estimates for the Trust Funds and the Federal Unified Budget are much less affected by variation in the yield achieved on personal accounts (because the benefit offsets are based on Treasury bond yield rates and thus are not affected by variations in the real yield on either equities or corporate bonds). A relatively small effect on aggregate financial status is realized from variation in personal account yields, however, because variations in account accumulations and distributions would also affect the level of trust-fund revenue derived from the taxation of benefits and account distributions. Because this affect is small, no sensitivity analysis to account yield assumptions is provided for aggregate financial estimates.

As mentioned above, the long-term ultimate average real yield on stock investments made in the future is assumed to be 6.5 percent, somewhat less than the 7-percent real yield that was assumed for the 1994-96 Advisory Council. This reduction in expected average yield is consistent with both (1) a growing consensus among economists that the market may value equities at somewhat higher average price-to-earnings ratios in the future based on broader access and a reduction in the perceived level of risk, and (2) the Trustees' increase in the assumed real yield on treasury bonds from the level assumed in 1995.

The expected ultimate average real portfolio yield for personal accounts would thus be 4.6 percent, net of administrative expense and is calculated as follows:

0.5 · 6.5% + 0.3 · 3.5% + 0.2 · 3.0% - 0.3% = 4.6%.

Due to the large degree of uncertainty associated with both the average portfolio distribution and future returns on equity (and corporate bonds), expected benefit levels are provided for two variations on the expected account yield. The first, referred to a "Low Yield" reflects an account yield equal to the assumed real return on long-term Treasury bonds, or 3 percent, less the administrative expense factor. This illustration is consistent with assuming that individuals will:

Invest more conservatively (100 percent in Treasury bonds),
Realize lower-than-expected returns on account assets (by 1.9 percentage points), or
View accounts on a "risk-adjusted" basis where all assets are assumed to have an ultimate real risk-adjusted return of 3 percent.

The second variation of the yield assumption is referred to as "High Yield" and is consistent with assuming that individuals will:

Invest more heavily in equity (60 percent rather than 50 percent) and less in bonds (24 percent in corporate bonds rather than 30 percent, and 16 percent in Treasury bonds rather than 20 percent), or
Realize higher-than-expected returns on account assets (0.32 percentage point higher on all assets or 0.64 percentage point higher on equities for an ultimate real equity yield of over 7.1 percent).

It should be noted that the difference between the central and high yield assumptions is smaller than the difference between the central and low yield assumptions. This is not intended to suggest that achieving the low yield over a lifetime is as likely as the achieving the high yield for an individual who invests 50% in equity, as assumed for the central assumption. For this investment portfolio the high yield is assumed to be more likely to occur than the low yield.

A range of administrative expense factors was assumed for individual accounts proposed by the 1994-96 Advisory Council on Social Security. For the Individual Account (IA) plan, individual contributions were assumed to be collected and recorded by a central institution, invested in large blocks with financial institutions, and invested in a limited number of indexed funds. Based on experience of the Teachers Insurance and Annuity Association College Retirement Equities Fund (TIAA-CREF) and the Federal Employee Thrift Savings Plan (TSP), it was assumed that the IA plan could be administered with an expense of 10.5 basis points per year. For the Personal Security Accounts (PSAs), individual accounts were assumed to be invested on an individual basis, resulting in an annual administrative expense of 100 basis points. Because the Commission's specifications for personal accounts are closer to the individual accounts for the IA plan than to the individual accounts for the PSA plan, an average ultimate administrative expense charge of 30 basis points appears to be reasonable. Some additional expense over the accounts of the IA plan seems reasonable because investment alternatives are intended to be much broader, including, at a minimum, more than one balance fund and potentially some actively managed funds.

c. Personal Account Distributions

Under these models, workers would not have access to account balances before retirement, defined as entitlement to Social Security retired worker, aged spouse, or aged surviving spouse benefits. In the event of a worker's death prior to such entitlement, the account balance would be transferred to the account of the surviving spouse, if any. In the absence of a current spouse, the account assets would pass to the worker's estate.

Upon the divorce of a worker who has not become entitled to benefits (as described above), the worker's personal account assets that accumulated during the marriage (including contributions during the marriage and returns on all assets during the marriage) are divided equally between the worker's and former spouse's accounts. If the worker has already become entitled to benefits (as described above) before the divorce, the annuity purchased with account assets will remain in force.

Any additional assets that accrue to a worker's account after annuitization, whether due to additional work, divorce or inheritance, are assumed to be immediately annuitized based on the worker's then current age and marital status. While full annuitization is assumed for the purpose of estimates presented in this memorandum, some degree of lump-sum distributions would be allowed under the Commission models. To the extent that a lump-sum distribution is selected, the available annuity would be diminished. However, the value to the retiree of the partial lump sum distribution would presumably be at least as great as the amount of annuity income that is foregone.

Estimates of benefit payments to individuals are computed for two different forms of life annuities. These are a CPI-indexed life annuity, and a variable life annuity. For the CPI-indexed life annuity, a net real yield equal to the assumed real yield on long-term Treasury bonds is assumed. This would require that annuity assets actually be invested with an expectation of a higher yield than for Treasury bonds in order to offset the administrative expense incurred by the annuity provider. For the purpose of these estimates, the administrative expense is assumed to be 30 basis points. This low expense factor for a CPI-indexed life annuity would likely only be provided by the Federal government, or by private financial institutions with special investment arrangements with the Federal government.

For the variable annuity, the "expected" level of monthly retirement income is greater because the Commission specified that the variable annuity would be invested in the same manner after retirement as before retirement, generally 50 percent in equities. Such investment in a variable annuity would lead to substantial variation in annual increases in annuity amounts. Increases in annual payments for an annuity at the rate of the increase in the CPI could not be assured. In fact, in years when the variable annuity portfolio substantially underperformed the expected return, benefit payments from the annuity could even be lower than in the prior year. Because of this uncertainty, we believe that variable life annuities would be selected by relatively few individuals. Thus, we put primary emphasis on estimates reflecting distributions with CPI-indexed life annuities.

V. Financial Estimates: Aggregate Measures of Effects on OASDI Financing, Individual Accounts, and the Federal Unified Budget

The attached tables reflect effects on the financial status of the OASDI program, including the benefit offsets based on contributions to personal accounts. For each model, the value of these benefit offsets is determined by accumulating the prior account contributions at the model-specific benefit offset yield rate (see descriptions of individual models above).

It is important to note that the two methods considered for computing the benefit offset yield rate would have the same "expected" effects on net benefit levels and on the financial status of the OASDI program. However, these two methods would have different effects on the sensitivity of benefit levels and OASDI financial status to variation in actual Treasury bond yields. If the benefit offset yield rate is computed as a fixed-percentage difference from realized or expected Treasury bond yields, then the net benefit level for the conservative investor (who invests solely in Treasury bonds) will be insensitive to (unaffected by) variation in actual bond yields. In addition, the sensitivity of OASDI financial status will ultimately be about the same as if no one opted for the personal accounts because variation in actual bond yields affects the present value of both payroll tax revenue redirected for PA contributions and benefit offsets to the same degree. But if the benefit offset yield rate is computed as a fixed percentage difference from the realized or expected inflation rate, then the sensitivity of net benefit levels to variation in actual Treasury bond yields will be much greater and the sensitivity of OASDI financial status will be considerably lower. This is true because, for example, a lower-than-expected Treasury bond yield will directly reduce the net benefit, dollar for dollar (because the offset is unaffected). While on the other hand, the OASDI Trust Funds will be partially insulated from the effects of the lower-than-expected bond yield because the benefit offset is unaffected.

a. Financial Operations of the Combined OASDI Trust Funds

Attached are eleven tables that provide a standard analysis of the financial effects of the three models, or plans, on the financial status of the Social Security OASDI program. These tables provide annual and 75-year-summarized cost rates, income rates, and balances for the OASDI program under the plans with the different participation rates described above. The first of these tables provides the estimated financial status of the OASDI program under present law. This table also reflects the financial status of the OASDI program under Model 1 if 0-percent participation in the personal account option were assumed (Model 1 specifies no basic changes to the OASDI program).

For Plan (Model) 1, four tables are provided. The first two are based on Plan 1 assuming that the 2-percent personal account contribution is financed completely with a redirect of OASDI payroll tax revenue. These are Plan 1--67p, which reflects a 2/3 individual account participation rate and Plan 1--100p, which reflects a 100 percent individual account participation rate. The third table, Plan 1(1+1) 67p, assumes that the 2-percent personal account contribution is financed one half with a redirect of OASDI payroll tax revenue and one half with General Fund revenue, and that the individual account participation rate is 2/3. The fourth table, Plan 1(0+2) 67p, assumes that the 2-percent personal account contribution is financed entirely with General Fund revenue, and that the individual account participation rate is 2/3. As indicated earlier, the nature of Model 1 suggests that 2/3 participation is the most reasonable assumption of the three discussed. No table is included for zero participation, because in this case, Plan 1 would be the same as current law.

For Plan 2, three tables are provided. The first, Basic Plan 2, includes the basic provisions of the Plan that affect OASDI benefit levels, but excludes both the individual account option and the provision for additional transfers to the Trust Funds from the General Fund of the Treasury as needed for OASDI solvency. The second, Plan 2T 67p, includes all provisions of the Plan and assumes 2/3 participation in the individual account option. The third, Plan 2T 100p, includes all provisions of the Plan and assumes 100 percent participation in the individual account option.

For Plan 3, three tables are provided. The first, Basic Plan 3, includes (1) the basic provisions of the Plan that affect OASDI benefit levels and (2) the specified, or dedicated transfers, from the General Fund of the Treasury starting 2005. However, Basic Plan 3 excludes (1) the individual account option and (2) the provision for additional transfers to the Trust Funds from the General Fund of the Treasury as needed for OASDI solvency. The second, Plan 3T 67p, includes all provisions of the Plan and assumes 2/3 participation in the individual account option. The third, Plan 3T 100p, includes all provisions of the Plan and assumes 100 percent participation in the individual account option. As indicated earlier, the nature of Model 3 suggests that 2/3 participation is the most reasonable assumption of the three discussed.

The table below summarizes the effects of the three models on the financial status of the OASDI Trust Funds under the 67 and 100 percent participation assumptions. More detailed analysis is provided in the attached tables.

For each year 2001 through 2076, the tables also provide:

The trust fund ratio (TFR) which is defined as OASDI Trust Fund assets as of the beginning of the year, expressed as a percentage of the outgo from the OASDI Trust Fund during the year;
The marginal change in the OASDI contribution rate (contribution rate directed to the Trust Funds), which reflects the change, from the prior year, in the OASDI contribution rate ⁵;
The net OASDI contribution rate; and
The change in the net OASDI contribution rate from that specified in current law broken out by (1) the change from payroll taxes redirected from the Trust Funds to individual account and (2) the change due to transfers from the General Fund to the Trust Funds.

b. Additional Aggregate Values for Trust Funds and Personal Accounts

A second set of ten tables for these models is attached (see "List of Tables") with a letter "a" following the table name. Each of these tables provides three additional sets of values. All values are expressed on a present value basis, i.e., current dollar values discounted to January 1, 2001 using the projected OASDI Trust Fund yield rates. These values are given for each year 2001 through 2076 and include:

Trust Fund levels under present law (PL) and the Plan as of the end of the year,
Net current accrual for future benefit offset under the Plan as of the end of the year,
Annual cash flows of the personal accounts, and
Personal account accumulations as of the end of the year.

The Trust Fund levels reflect the projected assets accumulated in the OASDI Trust Funds at the end of each year. Because the OASDI program does not have legal authority to borrow, these assets cannot become negative. Negative values in these tables are hypothetical, assuming the Trust Funds were able to borrow when necessary to fully pay scheduled benefits, with borrowing at the same interest rate specified for special issues to the Trust Funds. A negative value for a specific year represents the unfunded obligation for the period 2001 through the specific year.

Net current accrual for future benefit offset under each Plan is the currently accrued hypothetical amount of prior personal account contributions based on redirected payroll taxes that are expected to be applicable as a benefit offset in the future. This amount reflects deductions for accruals that have already been applied as benefit offsets and for accruals that were not applied (or are not expected to be applicable in the future) as offsets because of death by a worker before reaching retirement. It should be noted that these accruals are expressed in present value as of January 1, 2001, discounted at the OASDI Trust Fund yield rates, but that these amounts will actually "grow" through time at the benefit offset yield rate specified in each Plan. Thus, values of accruals at a particular date are not strictly comparable across Plans. It is also important to note that these accruals for future benefit offset are not equivalent to Trust Fund assets, as they are not available for payment of current benefits if needed.

Annual dollar flows and accumulations of the personal accounts are presented in the last three columns of these tables. These estimates are based on very specific assumptions that all personal account assets are converted to CPI-indexed life annuities at retirement (see description in the section on assumptions above). In practice, many individuals would likely annuitize only part of their personal account accumulation so estimated annuity assets are overstated to some degree. However this overstatement might be partially offset to the extent that some individuals would choose to purchase a variable life annuity, as described above, instead of the CPI-indexed life annuity. Total personal account and annuity assets (referred to as IA/Annuity assets in the tables) include both the assets of personal accounts held prior to retirement, and the assets held by the annuity provider after retirement. If the personal accounts are considered as a part of "Social Security", it is reasonable to combine the amounts of Trust Fund assets and personal accounts for a representation of total system assets.

The table below summarizes the effects of the three models on system assets and the net current accrual for future benefit offset under each Plan. More detailed analysis is provided in the attached tables.

c. Effects on Annual Federal Unified Budget Balances

A third set of ten tables for these models is attached (on pages 53 - 62) with a letter "b" following the table name. Each of these tables provides a rough estimate of the effects of the Plan on the annual Federal unified budget balance for calendar years 2004 through 2076. All values in these tables are presented in constant 2001 dollars (i.e., dollar amounts that are indexed back to 2001 based on the Consumer Price Index, CPI).

These estimates are based completely on the intermediate assumptions of the 2001 Trustees Report, including the trust-fund interest assumption (plus additional assumptions discussed above), and thus are not consistent with projections made by CBO and OMB (which use different assumptions). However, differences in payroll and benefit estimates are not large during the first 10 projection years so these values can be viewed as very rough approximations of the magnitude of effects on the unified budget balances through this period.

The first column in these tables provides the estimated contributions to personal accounts financed by redirecting payroll taxes plus, in the case of Plan 1 (1+1) and Plan 1 (0+2), the portion of the contributions financed from the General Fund of the Treasury. These contributions by the Federal government count as expenditures for the Federal unified budget.

A second column provides the amount of dedicated General Fund transfers to the Trust Funds (beginning 2005) specified for Plan 3, and is blank for other Plans. While these values are included in this table, it should be noted that such transfers do not affect the unified budget balance.

The third column provides the estimated amount of OASDI benefit offset based on earlier contributions to personal accounts. These benefit offsets reduce the amount paid to beneficiaries by the Trust Funds and thus reduce expenditures for the unified budget.

The fourth column provides the amount of other changes in OASDI cash flow under the Plan. These include specified modifications to OASDI benefit levels and changes in revenue to the Trust Funds based on taxation of benefits and disbursements from personal accounts. Additional transfers from the General Fund to the Trust Funds to achieve OASDI solvency are not included in this amount because they do not affect the unified budget balance.

A fifth column provides the estimated amount of income tax credit provided by the General Fund as a subsidy for the 1-percent out-of-pocket personal account contributions under Plan 3, and is blank for other Plans. This amount is an expenditure for the unified budget balance.

The sixth column provides the estimated "Change in Annual Unified Budget Cash Flow" for each Plan. This value reflects the amounts in the first 5 columns, and thus excludes the effects on interest obligations of the Federal government on publicly held debt.

The seventh column provides the estimated cumulative effect of the Plan through the end of the year on the amount of Federal debt held by the public, including interest in these changes. Note that these estimates assume that no other changes in Federal spending or income will occur other than those directly related to the Plan.

The eighth and final column provides the estimated "Change in Annual Unified Budget Balance", which includes changes in interest obligations to the public.

d. Annual Cash Flows from the General Fund of the Treasury to the OASDI Trust Funds

A fourth set of ten tables for these models is attached (on pages 63 - 72) with a letter "c" following the table name. Each of these tables provides the estimated annual net cash flow from the General Fund of the Treasury to the OASDI Trust Funds. All values in these tables are presented in constant 2001 dollars (i.e., dollar amounts that are indexed back to 2001 based on the CPI).

For comparison purposes, cash flow estimates are provided in each table for three different cases: * The Plan, assuming borrowing from the General Fund if needed to pay benefits (borrowing is expected to occur for Plan 1 only)
Present Law OASDI modified to allow borrowing from the General Fund to pay scheduled benefits, and
Present Law OASDI where only benefits payable with current financing provisions are being paid.

For each of these cases three columns are provided. The first column shows either estimates of the amount of borrowing needed from the General Fund to pay benefits or estimates of the amount of transfers from the General Fund as appropriate to the Plan. The second column is the estimated total net cash flow from the General Fund to the Trust Funds under the Plan, including transfers and borrowing. The third column is the total net cash flow for years starting with 2001 through the end of the given year, including accumulated interest cash flows for the period.

e. Aggregate Measures of OASDI Cash Flow for the 75-Year Period

Five aggregate measures of OASDI program cash flow are discussed in this section. The first two, actuarial balance and trust fund assets, are initially introduced earlier in this section. Aggregate gross cash-flow requirements from General Revenue (measure 4) and transition investment (measure 5) are presented in the Commission Report. The aggregate net cash-flow requirements from General Revenue, measure 5, is closely related to measure 4. All values presented below in the discussion of measures 3, 4, and 5 are based on 2/3 participation and, in the case of Model 1, assume all contributions are redirected from payroll taxes (Model 1 (2+0)).

1) Actuarial Balance: The traditional summary measure of cash flow for the OASDI Trust Funds over the 75-year long range valuation period is the actuarial balance. The actuarial balance expresses the net cash flow to and from the Trust Funds during the valuation period as a percentage of the effective taxable payroll (i.e., the tax base) for the period. Also included in the actuarial balance is the level of assets held in the Trust Funds at the beginning of the valuation period, and the cost of having a "contingency reserve" in the Trust Funds at the end of the period equal to the annual cost of the program. All values included in the actuarial balance are calculated on a present value basis. Thus, the actuarial balance provides a measure of whether the OASDI program will have sufficient net cash flow during the period, combined with starting assets in the Trust Funds, to allow for payment of scheduled benefits while leaving a reasonable contingency reserve at the end of the period. The estimated OASDI actuarial balance for present law and for each of the Commission's Plans is presented in section V.a., above.

2) Trust Fund Assets: The dollar level of assets held in the OASDI Trust Funds (also referred to as Trust Fund balance) at the end of the 75-year valuation period provides an aggregate measure of the net cash flow of the program over the valuation that is closely related to the actuarial balance. The Trust Fund balance at the end of the period, in present value terms, is equal to the net cash flow during the period plus the Trust Fund balance at the start of the period. In practice, the Trust Fund balance is not permitted to become negative because the OASDI program has no statutory authority to borrow. However, a theoretical projection of the Trust Fund balance as if borrowing were permitted is useful because it allows for a negative value which represents the accumulated additional revenue needed to fully pay scheduled benefits throughout the valuation period. This negative value, $3.2 trillion in present value dollars (discounted to 1-1-2001) under present law using the intermediate assumptions of the 2001 Trustees Report, is referred to as the "unfunded obligation" for the program. These values are presented for present law and each of the Commission's proposals in section V.b. above.

3) Aggregate Net Cash-Flow Requirements from General Revenue: Aggregate net cash-flow requirements from general revenue are more closely related to unified budget analysis than to the analysis of the specific financial needs of the Trust Funds. Aggregate net cash-flow requirements are computed consistent with the budget convention that assumes all scheduled benefits will be paid and that general revenue will finance any shortfall in OASDI financing. Moreover, this measure assesses the total cash flow from general revenues, including amounts that may be redeemed from Trust Fund assets. As a result, the total OASDI net cash-flow requirement from general revenue is $4.2 trillion in present value dollars (discounted to 1-1-2001) under present law for the 75-year period. This is $1 trillion higher than the unfunded obligation for the program, the difference being precisely the amount of Trust Fund assets held at the beginning of the period. Assuming 2/3 participation in the individual account option in each case, the aggregate net cash-flow requirement from general revenue would be $4.8 trillion, $2.3 trillion, and $2.9 trillion for Models 1, 2, and 3, respectively, in present value dollars as of 1-1-2001. Thus, net OASDI cash flow requirements from general revenue are increased by $0.6 trillion for Model 1 and reduced by $1.9 trillion and $1.3 trillion for Models 2 and 3, respectively.

4) Aggregate Gross Cash-Flow Requirements from General Revenue: Aggregate gross cash-flow requirements from general revenue are greater than net cash-flow requirements because they consider only years in which the OASDI program has a negative cash flow, and ignore years in which cash flow is positive. This approach is consistent with a view that years of negative OASDI cash-flow place a burden on general revenue sources that cannot be compensated for with positive OASDI cash flow in other years. Under current law and the intermediate assumptions of the 2001 Trustees Report, the gross OASDI cash-flow requirement from general revenue is $5.1 trillion in present value dollars (reflecting only years of negative cash flow starting in 2016). Assuming 2/3 participation in the individual account option in each case, the aggregate gross cash-flow requirement from general revenue would be $5.3 trillion, $2.8 trillion, and $3.4 trillion for Models 1, 2, and 3, respectively, in present value dollars as of 1-1-2001. Thus, gross OASDI cash flow requirements from general revenue are increased by $0.2 trillion for Model 1 and reduced by $2.3 trillion and $1.7 trillion for Models 2 and 3, respectively. These values are shown as item 3 in the summary table on page 18 of the Commission's Report.

5) "Transition Investment": There is no generally-accepted definition of what has been loosely referred to as the "transition cost" of changing the OASDI program. The concept of "transition investment", included as item 6 in the summary table on page 18 of the Commission's Report, provides one measure related to this concept. The designation as "transition investment" rather than transition cost is reasonable when additional costs are generated by a process designed to increase the extent of advance funding for the program.

The concept of "transition investment" adopted by the Commission is related to the estimated effects of the proposal on the net annual OASDI program cash-flow balance relative to all other entities, assuming borrowing by the Trust Funds were permitted when needed to pay benefits specified in the law. This cash flow is referred to as the OASDI annual balance (i.e., the difference between annual program cost with the payment of benefits specified in the law and annual income, excluding bond redemptions and borrowing from the General Fund of the Treasury).

Transition investment in any year is defined as the extent to which the OASDI net cash-flow balance (excluding any borrowing or bond purchase/redemptions from the General Fund of the Treasury) is lower under the proposal than under current law. Thus, a year for which the OASDI net cash-flow balance is higher under the proposal than under current law is deemed to be a year with no transition investment, even though a substantial contribution toward advance funding may be occurring.

This concept of "transition investment" may be evaluated in two different ways. The first counts any reduction in the annual net OASDI cash flow balance relative to current law (with borrowing authority). This would count a reduction from a present-law positive net cash-flow balance to a smaller positive net cash-flow balance under the proposal as transition investment. This way is consistent with the view that any positive current-law annual net OASDI cash-flow balance would be "saved" in the Trust Funds. Thus, the transition investment amount for a year would be the full difference between the net OASDI cash flow balances for current law and the proposal.

The second way of interpreting "transition investment" counts only the extent to which the net OASDI cash-flow balance is made negative or more negative than under current law (with borrowing authority). This would NOT count a reduction in present-law positive net OASDI cash-flow balance, except to the extent that the balance is made negative by the proposal. This way is consistent with the view that any positive current-law net annual OASDI cash-flow balance for a year would be spent on non-Social-Security Federal government obligations. Thus, this transition investment amount for the year excludes any current-law surplus.

The table above provides estimated values for these two ways of considering the concept of "Transition Investment" for the three models developed by the Commission (Model 1 is with all individual account contributions financed by redirecting payroll tax revenues). These values are shown in the summary table on page 23 of the Commission Report.

The years having a transition investment under the first way (i.e., where any reduction in OASDI net cash-flow balance is estimated) are 2004 through 2042, 2004 through 2025, and 2004 through 2020, for Models 1, 2, and 3, respectively. The years having a transition investment under the second way (i.e., where the estimated net OASDI cash-flow balance is made negative or more negative) are 2012 through 2042 for Model 1, 2010 through 2025 for Model 2, and 2014 through 2020 for Model 3. The dollar values given in the above table are present-value totals over these periods. Dividing these totals by the present-value total of GDP for the corresponding years yields the values expressed as a percent of GDP.

VI. Financial Estimates: Individual Measures of Effects on Retirement Benefits

For the purpose of this analysis, selected hypothetical individuals are assumed to participate fully in the available personal account option and to fully annuitize their account upon retirement (benefit entitlement) at age 65. Illustrations are provided for hypothetical workers retiring at 65 in 2012, 2022, 2032, 2042, 2052, and 2075.

For these hypothetical cases, earnings and personal account contributions are assumed to begin at age 21 (22 for steady maximum workers), or in the year 2004 if later. Annuities for married couples are assumed to be joint, with the survivor receiving two thirds of the monthly payment that is provided while both spouses are alive and entitled for benefits.

Four illustrative earnings levels are included. The "scaled" low, medium, and high earners have earnings patterns that reflect the relative probability of work and relative level of earnings by age during the period 1988-97. The absolute level of earnings in each case was set so that the Average Indexed Monthly Earnings (AIME) would be equal to that for a "steady" earner with low, average, and high earnings, respectively. For the steady average earner, earnings are at the SSA average wage index (AWI) for each year. For the steady low earner, earnings are at 45 percent of the AWI. The steady high earner has earnings at 160 percent of the AWI. The steady maximum worker is assumed to earn at or above the SSA taxable maximum each year prior to retirement. While these cases are hypothetical, the PIA for the medium (or steady average) earner is close to the median PIA for newly retired worker beneficiaries. See Social Security Administration Actuarial Note Number 144 for a full description of these hypothetical cases.

a. Expected Future Total Personal Account Accumulations at Retirement

The table titled "Wealth Estimated Accumulation of Personal Account Assets at Retirement at Age 65 for Plans 1, 2, and 3" provides estimated accumulated IA assets at age 65, just prior to annuitization, for the cases described above. As described in the section on assumptions, values are provided for the expected average personal account investment portfolio (50 percent in equity, 30 percent in corporate bonds, and 20 percent in Treasury bonds), as well as for a "Low Yield" and a "High Yield" sensitivity analysis. Estimates are provided in constant 2001 dollars.

b. Expected Total Benefit Levels at Retirement

Illustrations of benefit levels under these Plans are provided in 12 attached tables (see "List of Tables"). The first set of 6 tables is based on an assumption of full annuitization of personal account assets at retirement with a CPI-indexed life annuity on a joint and 2/3 survivor basis. This is believed to be the most likely choice for retirees as it would assure payments that would increase with the cost of living, and that would match the indexation of both OASDI benefit levels and benefit offsets under the Plan. The second set of 6 tables is based on the assumption of full annuitization of personal account assets with a variable annuity invested as before retirement. As discussed in the assumptions section, the variable annuity would provide a higher expected payment but could not assure increases from one year to the next that would keep up with the cost of living. In addition, the Models would permit a partial lump-sum distribution of an individuals account balance at retirement. Individuals who take partial lump-sum distributions would have lower monthly annuity payments based on the remaining personal account balance.

For each type of annuitization (CPI-indexed or variable), two tables are presented for each Plan 1, 2, and 3. One table illustrates the benefit levels of a married worker with a spouse who has earnings equal to those of the worker (2-earner couple). The other table illustrates the benefit levels of a worker with a non-earner spouse (1-earner couple). Monthly benefit estimates are presented in constant 2001 dollars as scheduled under present law, and as estimated under the Plan. Benefits are the amount payable based on a worker's earnings, and thus reflect one half of the couple's benefit in the 2-earner case, and the total couple benefit in the 1-earner case. Both spouses are assumed to reach 65 in the same year. Plan (proposal) benefits reflect:

% Basic Change for AllóThe percentage change in the benefit scheduled under present law based on the basic provisions of the Plan (note that this change applies to disability and survivor benefit cases, as well as to retirement cases),
% for PRA annuityóThe estimated monthly amount of the life annuity available based on full annuitization of the personal account accumulation assuming both spouses retire at age 65, expressed as a percentage of the present law scheduled benefit, and
% for Benefit OffsetóThe estimated amount of the benefit offset based on personal account contributions under the Plan, expressed as a percentage of the present law scheduled benefit.

The proposal benefit, reflecting the three factors above is presented in the tables first in constant 2001 dollars, but also in relative terms as a :

Percent of the present law scheduled benefit,
Percent of the present law payable benefit (reflecting reductions that would be needed starting in 2038), and
Percent of 2001 Real BenefitóThis is the ratio of the benefit payable under the Plan in constant 2001 dollars, to the amount payable to a worker with a comparable relative earnings history who retired at 65 in 2001.

For 2-earner married couples with unequal earnings, results would be between those shown for the 2-earner couples with equal earnings and for 1-earner couples. Single-life-annuity payment for an individual who is not married at retirement would be somewhat larger than for a married person with the same personal account.

Finally, it should be noted that estimates of personal account annuities and benefit offset amounts may tend to be somewhat overstated. Mortality for the individual account annuities calculated here is assumed to be the average for the total U.S. population, for all income levels. In fact, the expected mortality experience of annuitants, weighted by amount of assets to be annuitized, would be better (lower death rates) than for the general population. Individuals with lower accumulated assets due to lower lifetime earnings, or disability prior to retirement, tend to have higher mortality, all else being equal. Thus, the use of general-population mortality in these illustrations tends to understate the weighted life expectancy of annuitants, and overstate the size of the monthly annuity from individual account accumulations.

Stephen C. Goss

Alice H. Wade

Attachments: 54

List of Attached Tables

List of memos