March 10, 2005
Refer To:   
Senator Chuck Hagel
Stephen C. Goss, Chief Actuary
Alice H. Wade, Deputy Chief Actuary
Estimated Financial Effects of "The Saving Social Security Act of 2005"—INFORMATION

This memorandum provides estimates of the financial effects of the legislation you have developed and introduced on March 7, 2005. This bill, named "The Saving Social Security Act of 2005," would provide for voluntary personal accounts and for modifications of the benefit provisions of the Social Security program that are expected to result in solvency for the program throughout the 75-year long-range period. In addition, these provisions are projected to result in trust fund levels for the Social Security program that are stable or rising at the end of the long range period, thus meeting the criteria for sustainable solvency. That is, the Social Security (OASDI) program would be expected to be solvent for the foreseeable future under the intermediate assumptions of the 2004 Trustees Report and other assumptions described within this memorandum.

A detailed description of the provisions of the proposal is provided below based on our understanding of your intent from discussions with you, with Lou Ann Linehan, your Chief of Staff, and with Joe Cwiklinski of your staff. Following this description of the proposal is a report of the assumptions we have used and the projected financial effects of the proposal based on these assumptions.

Provisions of the Proposal

Workers who are age 45 or older on January 1, 2006 (those born in 1960 and earlier) will continue to participate in the OASDI program unchanged from current law. For workers under age 45 on January 1, 2006, personal accounts will be available on a voluntary basis and modifications in the OASI benefit provisions will apply.

1. Personal Accounts

Individuals who are under age 45 on January 1, 2006 (those born in 1961 and later) will have the option to participate in a personal account plan for earnings in calendar year 2006 and later. For those participating, contributions equal to 4 percent of their annual OASDI taxable earnings will be redirected from the OASI Trust Fund to their personal accounts.

Individual account contributions for a worker based on a year's earnings are not determinable until earnings are reported to and tabulated by the Social Security Administration. Because this reporting is made by employers on an annual basis, after the end of the calendar year, amounts for individual workers are not determinable for somewhat over a year, on average, after the date on which earnings are paid. Under the proposal, IA contributions during this initial period would be determined based on employer reports of aggregate OASDI taxable earnings of their employees, and would be invested in aggregate on behalf of the workers in the average portfolio. The accumulated IA contributions would be allocated to the individual accounts of specific workers as soon as current earnings reporting permits. Allocations would be made assuming that the pattern of earnings throughout the year was the same for all workers.

Personal accounts will be administered by a Central Administrative Authority (CAA) that would receive contributions and individual records of participants' earnings from the Social Security Administration. Individuals would be able to choose how their account assets would be invested from the selection of options currently being offered by the government employee Thrift Savings Plan (TSP), with the possible addition of a life cycle fund. The central administrative authority would maintain records and issue periodic statements to account holders. The IA management would be based on the design of the government employee TSP, with limited reporting requirements. Aggregated assets would be invested by privately managed investment firms. Through this approach it is assumed that IA administrative costs can be expected to be modest, ultimately around 0.30 percent of IA assets on average, with this being the charge for each account holder. This might require some Federal subsidy in early years for the IA, when account balances are low and start-up costs are incurred.

There would be no access to account assets prior to retirement except in the case of death. If a married worker dies before retiring, then the account assets would be transferred to the account of the surviving spouse. If a worker dies before retiring and has no current spouse, then the assets would pass to the workers estate. Workers who divorce before retirement would split evenly the amounts they each accumulated during the marriage based on individual account contributions during the marriage.

Upon retirement and entitlement to retired worker benefits (or at age 62 and entitlement to an aged spouse or aged surviving spouse benefit) personal account assets would be available to the individual and a CPI-indexed life annuity would be made available (joint and survivor annuity available for married couples) by the CAA. The annuity available from the CAA would be administered by the CAA with assets invested in the same manner as for individual accounts with the private investment firms.

Distributions from personal accounts will be treated like OASDI benefits for the computation of federal personal income tax liability, with the proceeds transferred to the OASI, DI, and HI Trust Funds, as in current law.

2. Benefit Offsets for Personal Account Participants

Upon entitlement to retired worker or aged (surviving) spouse benefits an offset would be computed based on a hypothetical account accumulation and would be applied to OASI benefits received. The hypothetical account, or "shadow account", would be the accumulation of all individual account contributions applied to the individual's account (through their own work, through divorce distributions, or from amounts transferred at the death of a spouse). These account contributions would be accumulated at 3 percent over the rate of inflation that had been realized in the CPI-W in years between the year of earnings upon which a contribution was based and the year of benefit entitlement. Reductions would be computed on the basis of a CPI-indexed life annuity calculation using unisex mortality rates, expected future inflation rates, and a valuation interest rate equal to 3 percent over the expected future rates of inflation, all reflecting the intermediate assumptions of the most recently published Trustees report at the time of benefit entitlement. The annuity calculation would be based on a joint and two-thirds survivor annuity for married couples1.

3. Raise the Normal Retirement Age to Age 68

For all individuals attaining age 62 in 2023 and later, the normal retirement age (NRA) will be 68. This is a 1-year increase in the NRA from the level for those attaining age 62 in 2022 (the last age cohort for which the individual account option would not be available). With eligibility for retired worker benefits still available at age 62, up to 6 years of early retirement would be possible. The percentage of the basic benefit level, or primary insurance amount (PIA), payable to workers retiring at 62 would be 65.5 percent, or 4.5 percentage points less than for retirement with 5 years of early retirement reduction. See provision 5 for further modification of early retirement reduction factors. The benefit reduction of 28.5 percent for entitlement to aged surviving spouse benefits at age 60 would be retained and scaled linearly by age between age 60 and the new NRA of 68.

4. Longevity Indexing of Retirement Benefits

For individuals attaining age 62 in 2024 and later, retired worker and aged (surviving) spouse benefits would be based on a modified PIA reflecting adjustments for increasing life expectancy. This modification would take the form of reductions in the PIA factors, 90, 32, and 15 that would gradually increase over time. The calculation would be based on the measured unisex life expectancy at age 68 based on the period life table produced by the Office of the Chief Actuary of the Social Security Administration using Social Security and Medicare data for the year 2020. This would be referred to as the "base life expectancy." In 2023, the Office of the Chief Actuary will compute a similar period life table based on data for the year 2021, and compute the NRA that would, based on this table, produce a life expectancy equal to the base life expectancy (for age 68 in the 2020 period life table). If the computed NRA based on the 2021 life table is greater than 68, then the actuarial reduction for retirement at age 68 under this increased NRA would be computed and would be applied as the reduction to the PIA factors (90, 32, and 15) for benefit eligibility in 2024.

For benefit eligibility in years after 2024, the reductions applied to the 90, 32 and 15 PIA factors would be computed as above by producing the period life table for the third prior year, computing the effective NRA that would have the same life expectancy as the base life expectancy, and compute the actuarial reduction for retirement at age 68 with this effective NRA. Based on the intermediate assumptions of the 2004 Trustees Report, the "effective NRA" is projected to increase by slightly less than 1 month per year after 2020, and the effective reduction in the PIA factors is projected to be about 0.5 percent (multiply by 0.995) each year.

The reduced PIA described above would not apply for the computation of benefits payable to surviving child beneficiaries, or to surviving spouse benefits with a child in care. Nor would the reduced PIA apply for benefits payable from the Disability Insurance Trust Fund. In these cases benefits would be based on the current PIA formula factors (90, 32, and 15). However, upon conversion to retirement status at NRA, disabled worker beneficiaries would have monthly benefits based on a weighted average of the PIA continued from disabled worker status, and the modified PIA for retired workers of the same age (reflecting years of disability freeze in the benefit computation). The weight on the continuation of the disabled worker PIA (WD) would equal the proportion of elapsed years2 after calendar year 2005 in which the individual was receiving disability benefits. The weight on the modified PIA for retired workers (WR) would equal 1 minus WD.

5. Modify Early Retirement and Delayed Retirement Factors

Early retirement and delayed retirement factors in current law are designed to produce approximately equivalent expected present values of future lifetime benefits for individuals who at age 62 choose among the possibilities of beginning benefit entitlement at ages 62 through 70. This provision increases the size of both early retirement reduction and delayed retirement increase factors, providing a greater incentive for workers to delay benefit entitlement and work to a higher age.

Early retirement factors would be increased gradually over 5 years, for those becoming eligible at age 62 in years 2024 through 2028. Over this period, the reduction for the first 3 years of early retirement would be gradually increased from 20 to 25 percent (from 6 2/3 percentage points per year to 8 1/3 percentage points per year.) The reduction for the next additional 2 years of early retirement would be increased from 10 percentage points (5 per year) to 12 percentage points (6 per year). And the reduction for the sixth year of early benefit entitlement would be increased from 4 1/2 percentage points per year to 5 1/2 percentage points per year.

Under current law, the delayed retirement credit (DRC) is scheduled to increase to 8 percentage points per year for those attaining age 65 in 2008. Under this provision, the DRC will be increased by an additional 0.5 percentage point every other year starting for those attaining age 65 in 2027, and reaching an ultimate level of 10 percentage points per year for those attaining age 65 in 2033.

6. General Revenues Transfers to the Trust Funds

In any year for which the Trustees intermediate projections show that the combined OASDI trust fund ratio is expected to fall below 100 percent of the annual cost of the program, transfers will be made from the General Fund of the Treasury sufficient to maintain the OASDI Trust Funds at a level no lower than 100 percent of annual program cost. This provision would guarantee solvency and sustainable solvency for the trust funds in any circumstance.


As indicated above, estimates provided in this memorandum are based on the intermediate assumptions of the 2004 Trustees Report.

Given the nature of the individual accounts and the offsets, individual account participation is expected to be substantial, but not universal. Because IA participation requires no additional contribution by workers, and the account assets would not be subject to potential future modification (as are basic monthly OASDI benefits), most individuals are expected to participate. The uncertainty of achieving a net real yield in excess of 3 percent on IA assets will cause some to pass on the option. On balance, we assume that about two thirds of the potential IA contributions for eligible workers will be realized through worker elections.

The ultimate average real yield on corporate bonds is assumed to be 3.5 percent, or 0.5 percentage point above the Trustees intermediate assumption for long-tern Treasury bonds. In addition, the long-term ultimate average annual real yield assumed for equities is 6.5 percent. This is somewhat lower than the historical real equity yield over the last several decades. On average, account holders are assumed to invest their assets about 50 percent in equities, 30 percent in corporate bonds and 20 percent in Treasury bonds. With an assumed ultimate administrative expense of about 0.30 percent (30 basis points) of assets per year, the average net annual yield is assumed to be about 4.6 percent over inflation. Variation in realized yield will, however, be substantial both across generations and among individuals within each generation. In order to understand the implications of this variability, a second estimate is provided reflecting a low yield, or risk-adjusted yield, for individual account assets equal to the assumed long-term Treasury bond yield of 3 percent over inflation.

A consensus appears to have formed among economists that equity pricing, as indicated by price-to-earnings ratios, may average somewhat higher in the long-term future than in the long-term past. This is consistent with broader access to equity markets and the belief that equities may be viewed as somewhat less "risky" in the future than in the past. Equity pricing will vary in the future as in the past. Price-to-earnings ratios were very high through 1999, and are now lower. The lower-than-historical average ultimate real yield assumed for equities purchased in future years is consistent with an average ultimate level of equity pricing somewhat above the average level of the past.

The assumption for an ultimate real equity yield of 7 percent that was used by the Office of the Chief Actuary until 2001 was developed in 1995 with the 1994-96 Advisory Council. At that time, the Trustees assumption for the ultimate average real yield on long-term Treasury bonds was 2.3 percent. Real yields on corporate bonds are believed to bear a close relationship to Treasury bond yields of similar duration. The 2004 Trustees Report includes the assumption that the ultimate real yield on long-term Treasury bonds will average 3 percent, or 0.7 percentage point higher than assumed in 1995. This increase in the assumed bond yield is consistent with a reduction in the perceived risk associated with equity investments.

It should be noted that the precise effects of implementing a plan that would result in a significant demand for equities and corporate bonds on the yields of these securities is not clear. This demand would likely be at least partially offset by reductions in demand for other investment mechanisms. For the purpose of these estimates, it is assumed that there will be no net dynamic feedback effects on the economy or on the financial markets.

Financial Effects of the Plan

Trust Fund Operations

Table 1 indicates that under the intermediate assumptions of the 2004 Trustees Report and the assumed average yields for equities and corporate bonds described above, the OASDI program is projected to be solvent throughout the 75-year projection period and to satisfy the criteria for sustainable solvency (stable or rising trust fund ratio, TFR, at the end of 75 years). The annual cost rate (cost of the OASDI program as a percent of payroll) is expected to decline steadily after 2033, reflecting the increasing extent of benefit reductions due to the basic benefit modifications and offsets associated with personal account participation. The annual balance (net cash-flow balance as a percent of payroll) is projected to be a small deficit of 0.22 percent of payroll for 2079 and declining. Continued decline would be expected after 2079 due to the continuation of the longevity indexing provision. The TFR is projected to be stabilized at 100 percent of annual cost between 2025 and 2077 through General Fund transfers. Thereafter, the TFR is projected to remain steady at 100 percent of annual program cost through 2079 and would be expected to later begin rising slowly.

General Fund transfers, expressed as a percent of taxable payroll, are projected to rise to a peak of 4.80 percent for 2030 through 2033 and decline thereafter, reaching zero for 2078 and later. The "effective" OASDI contribution rate is projected to be reduced from the nominal level of 12.4 percent by the amount redirected to personal accounts but increased temporarily by General Revenue transfers. Aggregate IA contributions are projected to reach 2.67 percent of taxable payroll for 2035 and later (two thirds of the 4-percent individual account contribution rate).

The actuarial deficit for the OASDI program over the 75-year projection period would be improved by an estimated 1.89 percent of taxable payroll, from an actuarial deficit of 1.89 percent of payroll projected under current law to an actuarial balance of zero under the plan.

Program Transfers and Assets

Table 1a provides an analysis of General Fund transfers under the plan and of OASDI Trust Fund and individual account assets. Columns 1 through 3 provide the estimated amounts of annual transfers to maintain a 100 percent TFR under the mechanism described above. Column 4 provides the cumulative total amount of these transfers from the General Fund of the Treasury through the end of each year. Under the plan, transfers from the General Fund would not be expected to be required until 2024; and the total expected transfers needed between 2024 and 2077 are $3.6 trillion in present value.  As seen in table 1d, column 7, this contributes to achieving a trust fund level of $0.2 trillion in present value at the end of 2078.

Total projected OASDI Trust Fund assets are shown in column 5 of table 1a. For purpose of comparison, the net OASDI Trust Fund assets are also shown for a theoretical Social Security program where borrowing authority is assumed for the trust funds. The theoretical Social Security program with borrowing authority is presented both without and with the General Fund transfers expected under this plan, in columns 8 and 9, respectively. Column 1 of table 1d shows that transfers or borrowing from the General Fund of the Treasury required to finance currently scheduled benefits through 2078 would total $3.7 trillion in present value, and this would leave the cash holdings of the trust fund at zero at the end of 2078.

If the individual accounts are considered as a part of a "total system", along with the OASDI program, then it is reasonable to consider "total system assets". These would be the sum of OASDI Trust Fund assets and IA assets (columns 5 and 6 of table 1a). Under the intermediate assumptions and assuming full annuitization of IA assets, total system assets are expected to be large and growing in real terms at the end of the 75-year projection period.

Effect on the Federal Unified Budget

Table 1b (present value dollars) and table 1b.c (constant dollars) provide estimates of the effect on federal unified budget cash flows and balances under this plan and these assumptions. All values in these tables represent the amount of the change that would be expected as a result of implementing the proposal, from the level that would be projected under current law. The effect of the plan on unified budget cash flow (column 4) would be expected to be negative initially, but positive starting 2040. This total cash flow change is the combination of the specific plan effects shown in columns 1-3. It is important to note that these estimates are based on the intermediate assumptions of the 2004 Trustees Report and thus are not consistent with estimates made by the OMB or the CBO based of their assumptions.

The fifth column provides the projected effect of implementing the plan on the Federal debt held by the public. For 2075 and later, the accumulated effect on debt held by the public is expected to be less than under current law scheduled benefits, assuming the law was changed to permit borrowing authority by the trust funds. The final column provides the projected effect on the annual unified budget balances, including both the cash flow effect in column 4 and the additional interest on the accumulated debt.

Cash Flow to the General Fund of the Treasury

Table 1c provides estimates of the net cash flow from the OASDI Trust Funds to the General Fund of the Treasury. Revenue paid by the Treasury to the trust funds for the redemption of the special-issue Treasury obligations held by the trust funds is included here as a negative cash flow to the General Fund. To the extent that General Fund transfers are invested in special-issue Treasury obligations and are subsequently redeemed, they will, upon redemption, affect cash flow between OASDI Trust Funds and the General Fund of the Treasury.

Values in Table 1c are shown as a percent of taxable payroll, in current dollars, in present value dollars as of 1/1/2004, and in constant 2004 dollars (discounted to 2004 with the projected growth in the CPI). For comparison purposes, net cash flow is also shown for a theoretical Social Security program where transfers from the General Fund of the Treasury to the OASDI Trust Funds are assumed to occur as needed to assure full payment of scheduled benefits in 2042 and later.

Change in Long-Range Trust Fund Assets/Unfunded Obligation

Table 1d provides estimates of the amount of assets in the combined OASI and DI Trust Funds at the end of each year, in present discounted value. Negative values do not indicate levels of trust fund assets as the program does not have borrowing authority. Instead, negative values reflect the magnitude of the unfunded obligation of the program through the end of the year. The first column presents these estimates under present law, where the unfunded obligation is $3.7 trillion through 2078, the end of the 75-year long-range period.

Columns 2 through 5 show the annual effects of the components of the proposal that move the OASDI program to elimination of the unfunded obligation. These include the basic OASI changes (the change in the NRA, the benefit formula, and the early retirement and delayed retirement factors, plus the effects on taxation of OASDI benefits and on IA distributions), IA contributions redirected from the trust funds to the IAs, offsets against OASI benefits due to IA participation, and the General Fund transfers needed to maintain solvency in 2024 through 2077.

The combination of the annual effects in columns 2 through 5 is accumulated in column 6, showing the effect on projected trust fund assets, or on the unfunded obligation, through the end of each year. Column 7 shows the resulting trust fund asset levels projected under the proposal. The overall effect of the proposal is to transform the projected $3.7 trillion long-range unfunded obligation for the program under current law into an expected positive trust fund balance of $0.2 trillion at the end of the period.

Estimates with Low or Risk-Adjusted Yields on Assets

Tables 2, 2a, 2b, 2b.c, 2c, and 2d provide an analysis of the implications of realizing actual real yields on individual account assets that are equal to the assumed average real yield on long-term Treasury bonds, or 3 percent. This may be viewed as either illustrating the case where the average real yield on equities and corporate bonds is no higher than on government bonds, or illustrating the effect of assuming "risk-adjusted" returns on equities and corporate bonds. In either case, the "expected" yield on annuitized assets is assumed to match the actual yield, on average.

The historically higher returns on equities and corporate bonds than on Treasury bonds is associated with the relatively higher degree of variability in the returns on these assets. One way of accounting for the combination of this increased variability and the associated higher expected return that is demanded in the markets is to portray the returns of the more variable assets as being the same as the relatively low-yield asset, Treasury bonds. Tables 2 serve this purpose. It should also be noted that while average real yields for equities have been at or below average bond yields for periods of a decade or so, the likelihood of having such a low average yield for a period of several decades seems extremely low.

Differences between Tables 1 and 2 are small because the only effect on the solvency of the trust funds due to variation in IA investment yields comes from the relatively small changes in income taxes paid on the IA distributions.

Stephen C. Goss

Alice H. Wade


List of Attached Tables

Table 1a
Table 1b
Table 1b.c
Table 1c
Table 1d
Table 2a
Table 2b
Table 2b.c
Table 2c
Table 2d

1Annuity payments are reduced by 1/3 when only one spouse is alive.

2Elapsed years are those years from ages 22 through 61.

List of memos