Revised February 4, 2005

By Jason Furman and Robert Greenstein

PDF of full report

View Related Reports

If you cannot access the files through the links, right-click on the underlined text, click "Save Link As," download to your directory, and open the document in Adobe Acrobat Reader.

Senator Lindsey Graham (R-SC) introduced a Social Security plan in the 108th Congress, reportedly developed with the help of White House staff, that is based on the main plan designed by President Bush’s Social Security Commission.  This analysis examines that plan.

The Graham plan is broadly consistent with the new details that the White House released on February 2, although the Graham plan (unlike the White House as of yet) includes specific proposals to reduce Social Security benefits.  (The box on page 4 compares the Graham plan to the President’s emerging plan.)  This analysis also considers the effects of Senator Graham’s subsequent proposal to raise the ceiling on the amount of wages subject to the Social Security payroll tax, which is not part of the plan that Graham introduced.

Executive Summary

Elements of the Plan

The Graham plan has four main parts:  (1) it would reduce Social Security retirement and survivors benefits by using “price indexing” rather than “wage indexing” to set initial benefits.  (2) it would create a system of voluntary private accounts into which workers could divert a portion of their payroll taxes amounting to four percent of their wages, up to a maximum of $1,300 per year.  For workers who chose this option, there would be a second reduction in Social Security benefits, with that reduction designed to partially repay the Social Security Trust Fund for the diversion of these workers’ payroll taxes from the Trust Fund to the private accounts.  (3) it would allow workers to contribute additional funds (beyond their payroll tax contributions) to their private accounts and would provide tax subsidies for these contributions.  (4) it would transfer several trillion dollars in general revenues to Social Security.

Effects on Beneficiaries

Effects on Social Security and the Rest of the Budget


Elements of the Graham Plan

Senator Graham introduced S. 1878, “The Social Security Solvency and Modernization Act of 2003,” in the 108th Congress.  The plan, which is a modified version of the “Model 2” plan developed by the President’s Commission to Strengthen Social Security, includes four principal components:

  1. Reductions in Social Security benefits.  The plan would reduce Social Security retirement and survivors benefits by using “price indexing” rather than “wage indexing” to set initial benefits.  These benefit reductions would apply regardless of whether a worker opted for a private account (see #2 below).  The reductions would be moderated somewhat for some low-income workers and widows and widowers.

    The shift to price-indexing would reduce future Social Security benefits substantially, as compared to the current benefit structure.  Under current law, initial Social Security benefits are designed to grow with wages, ensuring that the benefit replaces a constant fraction of lifetime earnings.  Under the price indexing proposal, initial benefits would instead grow with prices, which are projected to grow at least one percent more slowly than wages on an annual basis.  As a result, the benefit reductions under this proposal would grow steadily larger over time.
    [3]  The Social Security actuaries project that, compared to benefits under the current benefit structure, switching to price indexing would result in a benefit reduction of 26 percent (or $5,000 a year in 2004 dollars) for average wage-earners who retire in 2042, and 46 percent (or $12,000 in 2004 dollars) for average wage-earners who retire in 2075.

    Indeed, the benefit reductions resulting from the shift to price indexing are so large that they alone would more than restore long-term Social Security solvency, according to the Social Security actuaries.


How Does the Graham Plan Compare to the President’s New Details?

The White House released additional details of its Social Security plan in conjunction with the President’s State of the Union address on February 2.  The White House provided details about the private accounts and the associated reduction in benefits for individuals who choose the accounts.  The White House did not announce any measures to restore Social Security’s solvency.  (It acknowledged that the diversion of funds to private accounts, and the associated reduction of Social Security benefits in retirement for those who choose the accounts, would together have a neutral effect on Social Security solvency and do nothing to improve Social Security’s finances.)  Here is a thumbnail comparison of the Graham plan to the President’s new details:

  1. Reductions in Social Security benefits.  The President mentioned several possible benefit reductions in his State of the Union address but did not commit himself to any specific policy.  One of the proposals the President mentioned was shifting to “price indexing.”  This benefit reduction is included in the Graham plan and also in the principal plan developed by the President’s Commission to Strengthen Social Security.
  1. Private accounts.  The structure for private accounts in the Graham plan is similar to that laid out by the White House.  For high-earners, the Graham plan has smaller private accounts than the White House plan does.   (Graham proposes to cap contributions to the accounts, while the President would start with caps but phase them out.)  The Graham plan also offsets slightly less of the private accounts with subsequent Social Security benefit reductions.
  1. Additional contributions to private accounts.  The President has not specified whether individuals would be allowed to make additional contributions to their private accounts.  He has proposed to subsidize high-income savers outside of Social Security, a similar proposal to what Graham is proposing inside Social Security.
  1. General revenue transfers.  The President has not stated whether his plan will include general revenue transfers, although the plans that his commission developed to restore solvency all relied on general revenue transfer accounting mechanisms.


  1. Creation of individual accounts.  The plan includes private accounts into which workers who are 55 or younger in 2006 could direct a portion of their Social Security payroll taxes.  Workers would direct to these accounts an amount equal to four percent of their wages and salaries, up to a maximum of $1,300 per worker in 2006.  (The $1,300 ceiling would be adjusted upward each year to reflect growth in average wages.)

    Workers who chose to open individual accounts would experience a second benefit cut in order to repay the Social Security Trust Fund for the diversion of part of their payroll tax revenues away from the Trust Fund.  To determine the size of this second benefit cut, the Social Security Administration would calculate the amount that would be in a worker’s account if the payroll taxes diverted to the account had earned interest at a rate 0.3 percent below the rate paid on Treasury bonds.
    [4]  (In contrast, the President’s plan envisions a reduction in Social Security benefits based on what the diverted payroll taxes would have earned at the full Treasury bond rate, a slightly deeper Social Security in guaranteed benefits than the Graham plan envisions.)  This amount would then be converted to a monthly annuity value (i.e., to the amount that could be paid from the account on a monthly basis for the rest of the worker’s life, based on the assumption that the worker would have a normal mortality rate for someone who had reached the normal retirement age).  This monthly annuity value would be deducted from the worker’s Social Security check every month.


  1. Additional contributions to private accounts.  The plan would allow workers to make voluntary contributions to their private accounts, on top of the portion of their payroll taxes that was diverted to the accounts.  Low-income workers would receive a matching contribution from the government for additional contributions they made to the accounts.

    Workers would be allowed to make $5,000 in additional contributions each year.  These contributions would be treated for tax purposes like contributions to Roth IRAs:  they would not be tax-deductible, but interest would accrue on a tax-free basis and withdrawals in retirement would be tax free.  In a marked departure from the rules governing Roth IRAs, however, there would be no income limit on who could make these contributions.

    For ordinary workers, the option to contribute up to $5,000 a year to the accounts would provide no new tax incentive for retirement saving, since these workers already can make such contributions to IRAs.  (The current ceiling on annual contributions to an IRA is $4,000, but it will rise to $5,000 in 2008.)  Only 4 percent of families contribute the maximum amount to an IRA, and among middle-class families the portion is even lower.

    For high-income individuals, however, this would represent a major new tax break.  Under current law, high-income individuals with access to an employer-based retirement plan are not permitted to make tax-deductible contributions to a traditional IRA, and households cannot make contributions to a Roth IRA if their income exceeds $160,000.  The effect of this feature of the Graham plan would be the same as repealing the income limit on Roth IRAs:  it would enable high-income households with substantial assets to shift $5,000 every year from other taxable investment accounts to tax-sheltered private accounts.

    Indeed, research indicates that high-income households would tend to shift existing assets to these new accounts rather than to use these accounts for new saving.
    [5]  Thus, this feature of the plan would likely generate little additional overall saving.  The revenue losses would be substantial, totaling about one-sixth of the projected 75-year deficit in Social Security, and would further swell budget deficits.


  1. General revenue transfers.  The plan would require substantial general revenue transfers to Social Security.  According to estimates from the Social Security actuaries, the transfers would equal $3.2 trillion over 75 years in present value.[6]  Over an infinite horizon (i.e., measured into eternity), the transfers would total more than $4.4 trillion in present value.

    One final feature of the Graham plan bears noting.  Instead of participating in the private accounts, or not participating in the accounts but still being subject to the benefit reduction that resulted from price indexing, individuals would have a third option — of paying significantly higher payroll taxes and continuing to receive benefits under the current Social Security benefit structure.  Individuals electing this last option would be subject to a payroll tax rate of 14.4 percent (an increase of two percentage points over the current rate); they also would be subject to additional future increases in the payroll tax rate, depending on the program’s projected solvency.  The Social Security actuaries estimate that “very few” individuals would choose this option.

How Would the Private Accounts Work?

As noted, workers who chose to participate in the private accounts would have four percent of their average wages diverted to these accounts, up to a maximum of $1,300 in 2006.[8]  For example, a worker who earns $25,000 annually would have $1,000 directed to the account each year.  A worker who earns $50,000 would have the maximum of $1,300 directed to the account, which would equal 2.6 percent of the worker’s wages.  In aggregate, the Social Security actuaries project that 2.7 percent of payroll would be diverted into the private accounts.

At retirement, workers with private accounts would experience a reduction in their Social Security benefits equivalent to the amount that would be in their account if the payroll taxes diverted to the account had earned interest at a rate 0.3 percent below the rate paid on Treasury bonds.  To see how this would work in practice, suppose that Ms. Smith is an average wage-earner born in 1990, who starts to contribute 4 percent of her wages to the accounts in 2011, up to the maximum contribution allowed.  Assume also that Treasury bonds earn 3 percent above the inflation rate.  (This is Social Security Trustees’ projection of the rate that Treasury bonds will earn over time.)  When she retires, Ms. Smith’s benefits will be reduced by an amount equivalent to the size that her account would have reached if it had earned interest at 2.7 percent above the inflation rate (since 3 percent minus 0.3 percent equals 2.7 percent).

One reason why Ms. Smith’s account might earn too little to offset the additional reduction in Social Security benefits is that the administrative fees for her account might exceed the Social Security actuaries’ estimate of 0.3 percent per year.  Administrative costs have proven significantly higher than that for the private accounts established in the United Kingdom.  Administrative charges also are higher in most private pension plans in the United States.  A recent Center for Retirement Research analysis explains why administrative costs for private accounts may exceed 0.3 percent per year.[9]

Another reason why Ms. Smith’s account might earn too little to offset the benefit cut is that it might lose money in the financial markets.  Ms. Smith would be at particular risk of such an outcome if she retired in a year that the stock market was down and converted her account to an annuity.  Table 1 summarizes these three hypothetical outcomes for Ms. Smith’s investments.

Table 1

The Value of the Individual Accounts Under Different Return Scenarios


Low Return Scenario

Baseline Return (Risk-adjusted)

High Return

(Not risk-adjusted)

Assumed Return

1.5 percent

2.7 percent

4.6 percent





Gross account at retirement





Annuity Value








Amount repaid to Social Security at retirement





Annual Benefit Reduction




% Benefit Reduction








Net value





Annuity Value




% of Current Law Benefit








Effective tax rate on assets in the accounts

(This represents the Social Security benefit reduction to repay the Trust Fund as a percentage of the assets in the individual account.)




Note:  All numbers are in inflation-adjusted 2004 dollars and use the Social Security Trusteees’ assumptions.  Rates of return are inflation-adjusted.  The example assumes that a scaled-medium earner contributes 4 percent of wages up to the maximum amount allowed from age 21 in 2011 to age 64 in 2054 and retires at age 65.  The analysis assumes an actuarially fair annuity rate of 7.2 percent — that is, each $100 of assets is converted into an initial annual payment of $7.20, which grows with inflation thereafter and is paid each year until the beneficiary’s death.


It is important to note that regardless of whether a worker’s private account would fully offset the additional Social Security benefit cut imposed on accountholders, the worker also would face the benefit cut resulting from the adoption of price indexing.  Many workers who earned enough in their private accounts to come out ahead compared to workers who did not opt for private accounts would still receive significantly less (in Social Security benefits and private account earnings combined) than they would receive under the current benefit structure.


Effects on Beneficiaries

Table 2 outlines the plan’s expected effect on benefits.  The table shows CBO’s estimates of the Social Security benefits that would be paid to median wage-earners under the current benefit structure.  (These are referred to as “Scheduled Benefits.”)  It also shows CBO’s estimate of the benefits that would be paid if no action were taken to shore up Social Security, the program became “insolvent,” and benefits were reduced to match the available Social Security revenue.  (These are referred to as “Payable Benefits.”)

Finally, the table shows the benefits that would be paid to median wage-earners under the Graham plan — that is, the sum of their Social Security benefits and the monthly payments from their private accounts.  These figures are calculated using CBO’s assumptions and methodology, as set forth in CBO’s analysis of the President’s Social Security Commission plan.  (CBO applies “risk adjustment” in estimating income from individual accounts; see the box on page 11.)

As the table indicates:

Table 2

First-Year Annual Benefits for the Median Worker in Middle of the Income Scale

10-Year Birth Cohort Starting in Year

Current Law Benefits

Graham Plan

Scheduled  Benefits





Benefits (Social Security Plus Private Accounts)


Percentage Reduction, Compared to Scheduled Benefits

Percentage Reduction, Compared to Payable Benefits











































Source:  The columns for scheduled and payable benefits are taken directly from CBO, “Long-term Analysis of Plan 2 of the President’s Commission to Strengthen Social Security,” 2004.  The columns on the Graham plan are calculated using CBO assumptions and methodology, assuming the change to price indexing goes into effect in 2011, two years later than the effective date specified in S. 1878.  (S. 1878 was introduced in 2003.)  The benefits are for a worker who initially claims benefits at age 65.

It may seem surprising that the benefit reductions would be larger under the Graham plan than if no action were taken to shore up Social Security’s finances.  This is the case for a basic reason.  As explained below, the diversion of funds from Social Security to private accounts would enlarge Social Security’s deficit by about 50 percent over the next 75 years, necessitating much deeper benefit reductions to bring the system into balance.  These deeper benefit reductions would not be fully offset by income from the private accounts.

For Individuals with Disabilities, Benefits Would Drop Sharply Upon Retirement

For disabled beneficiaries, the Graham plan represents an improvement over the plan developed by the President’s Social Security Commission.  However, it still includes a troubling reduction in benefits when disabled beneficiaries reach the normal retirement age.

The plan would maintain current-law disability benefits for workers who are below the normal retirement age, which is currently 66 and is scheduled to rise to 67 over the next two decades.  But when a disabled beneficiary reached the normal retirement age, his or her benefits could be cut significantly under the plan.

Under Social Security, disabled beneficiaries are “converted” into retired beneficiaries when they reach the normal retirement age.  Under current law, this has no effect on their benefits.  Under the Graham plan, however, when people with disabilities reached the normal retirement age and began to be treated as retirees, their Social Security benefits would be recalculated using a mix of price indexing and wage indexing.  (The younger a person was when he or she became disabled, the greater would be the reliance on wage indexing.  The older such a person was when he or she became disabled, the heavier would be the reliance on price indexing.)  The result would be an abrupt benefit cut in old age.

For example, consider a hypothetical individual born in 2000 who becomes disabled at 42 and initially receives a disability benefit of $24,000, the same as under current law.  Upon reaching 67, the individual’s benefit would suddenly be reduced to $18,306 per year — a reduction of $5,700 or 24 percent.[10]  (These figures are in 2004 dollars.)

 New Tax Breaks for High-Income Families Costlier than
New Incentives for Lower-Income Families

Under the Graham plan, lower-income workers would receive a $100 match for the first $1 they contributed to their private account on top of their payroll tax contributions.  Any contributions they made beyond this first $1 would be matched at a 50 percent rate, up to a total match of $500 per year.  This option would be available to workers who earn up to $27,500.  The contributions would accumulate tax free and could be withdrawn tax free in retirement.

No Free Lunch:  Risk Adjusted Rates of Return

The Congressional Budget Office, the Bush administration’s Office of Management and Budget, and leading economists incorporate the effects of risk when they calculate the expected returns of a portfolio that includes stocks.  The basic concept they apply is that there is no such thing as a “free lunch” and that investments made in the stock market produce higher average returns over time but also carry a higher risk and may perform poorly and lose money.

The difference between the average return on stocks (often referred to as “equities”) and the average return on risk-free Treasury bonds is known as the “equity premium.”  According to standard economic theory, this premium represents compensation for the additional risk associated with investing in stocks.  In other words, since investments in stocks are riskier and can lose money, average returns on stocks have to be higher over time, or else no one would invest in them.  Most policy analysts subtract this risk premium when projecting the returns on a portfolio.  This produces what is known as a “risk-adjusted” rate of return.

This risk adjustment reflects the fact that, even over long periods, equities can carry substantial risk.  For example, over the past 130 years, while average stock returns have exceeded average bond returns, there have been eighteen 15-year periods in which the cumulative return on stocks was lower than the cumulative return on relatively safe bonds.  In six 15-year periods (the periods ending in 1920, 1921, 1979, 1980, 1981, and 1982), the inflation-adjusted rate of return on stocks actually was negative.1

CBO discussed this issue in a report it issued, in which it examined the effect of having the U.S. Government shift some of its own assets from Treasury bonds to stocks.  CBO explained: “Government investment in private securities does not offer a free lunch: although it would increase the expected value of budgetary resources, it would do so at the cost of exposing the government, future taxpayers, and beneficiaries of federal programs to greater risk.  If that risk was taken into account, the returns on private securities would be no greater than the returns on government securities.”2  In other words, CBO concluded that the rate of return that should be used in assessing the returns that stock-market investments would produce is the “risk-adjusted” rate of return.  The risk-adjusted rate of return is simply the expected rate of return on Treasury bonds, which are essentially risk free if held to maturity.

The Bush administration’s OMB applies the same principle in projecting the rate of return on stock investments made by the Railroad Retirement Board, the equivalent of the Social Security system for railroad workers.  OMB has explained that “Equities and private bonds earn a higher rate of return on average than the Treasury rate, but that return is subject to greater uncertainty.  Sound budgeting principles require that estimates of future trust fund balances reflect both the average return and the cost of risk associated with the uncertainty of that return….Economic theory suggests, however, that the difference between the expected return of a risky liquid asset and the Treasury rate is equal to the cost of the asset’s additional risk as priced by the market.  Following through on this insight, the best way to project the rate of return on the Fund’s balances is to use a Treasury rate.”3  In other words, in projecting the returns on the stock investments of the Railroad Retirement Board, OMB adjusts the returns for risk by assuming that the rate of return is equal to the rate paid on Treasury bonds.

Analysis of historic returns is from data compiled by Yale Professor Robert Shiller available at
CBO, 2002, Evaluating and Accounting for Federal Investment in Corporate Stocks and Other Private Securities, emphasis added.
Office of Management and Budget, 2002, Analytical Perspectives of the FY 2003 Budget, pp. 439-440, emphasis added.

Tax subsidies designed to encourage low-income workers to save have the potential to increase national saving.  Currently, most low-income workers receive neither tax benefits nor institutional support for retirement saving.  The plan’s new tax incentive for low-income workers could induce such workers to save more.

These subsidies should be paid for, however, not deficit-financed.  The Social Security actuaries estimate that this low-income subsidy would cost the non-Social Security budget 0.21 percent of payroll over 75 years.  That is equivalent to $136 billion in nominal dollars over 10 years or $419 billion in net present value over 75 years.

The plan’s other new tax break, which would allow people at all income levels to make $5,000 a year in tax-advantaged contributions to their private accounts, would be more costly – a preliminary estimate is that this tax break would cost about 0.28 percent of payroll over 75 years or about $600 billion in net present value, over 75 years.[11]  Despite its substantial cost, however, this tax break likely would not add to national saving.  In fact, it probably would reduce national saving.

As noted, this tax break would not appreciably improve the incentives that ordinary middle-class workers have to contribute to retirement savings accounts.  Only a tiny fraction of workers now make the maximum contribution allowed to an IRA or an employer-based retirement plan; few can afford to put the maximum amount aside.  Some 96 percent of taxpayers who are eligible for IRAs do not contribute the maximum amount permitted.  Since these individuals are not contributing the full amount they are allowed to contribute under current law, allowing them to make even larger contributions should have little effect.

For wealthy individuals, however, the story would be very different.  The plan’s new tax incentive would give them a windfall, enabling them to shift $5,000 every year from taxable savings and investment accounts to these new tax-sheltered accounts.  The result would be a costly deficit-financed tax break for high-income individuals.  (This windfall would be even larger than that under the Bush administration’s controversial Retirement Savings Accounts proposal, because the new $5,000 savings account would be in addition to existing traditional IRAs, which can be used by those high-income individuals who do not have access to an employer-based retirement plan.  Under the Administration’s RSA proposal, the RSAs would replace IRAs.)

Because this new deficit-financed tax break would likely cost more than the amount of new private saving it generated, it would be more likely to reduce national saving than to increase it.  (National saving equals the amount of private saving minus government budget deficits, or plus government budget surpluses, if surpluses exist.)

Together, the two new savings incentives described here would increase the unified budget deficits and debt over the next 75 years, adding to the nation’s long-term fiscal problems.


Effects on Social Security and the Rest of the Budget

By themselves, the individual accounts in the Graham plan would significantly worsen Social Security’s finances, because the Trust Fund would lose more from the diversion of payroll taxes to private accounts than it would gain from the reduction in benefits for the holders of those accounts.  This is the case for two reasons.  First, if the payroll taxes were not diverted to private accounts, they would be invested in Treasury bonds and grow at the rate those bonds offer.  The size of the subsequent reduction in Social Security benefits for those who elected private accounts, however, would be based on an interest rate that is 0.3 percent below the Treasury bond rate.  The second reason that the Trust Fund would end up a net loser from the creation of private accounts is that is would incur transition costs that would result from the delay between the diversion of funds to the private accounts and the reduction in workers’ Social Security benefits, which would occur decades later.

The Social Security actuaries project that the private accounts in the plan would increase Social Security’s actuarial deficit by $2 trillion in present value over 75 years, or 0.98 percent of payroll.[12]  Since the 75-year shortfall in Social Security currently equals $3.7 trillion in present value (or 1.89 percent of payroll), this means that the individual accounts would, by themselves, enlarge Social Security’s deficit by about 50 percent.  As a result, deeper benefit cuts would be needed to make Social Security solvent.

Plan Would Increase the Deficit at Least Through 2050

According to the Social Security actuaries, the Graham plan would increase the unified budget deficit by $2.4 trillion (in nominal dollars) from 2006 to 2015 and by an additional $4.9 trillion from 2016 to 2025; these sums would be raised by federal borrowing.  The plan would increase the deficit every year through 2050.  These deficits would result in a large increase in the national debt (i.e., the “debt held by the public”); the debt would remain at higher levels (than those at which it would be in the absence of the plan) through 2064.

By 2030, for example, the debt would be $10 trillion higher than it would be in the absence of the plan, according to the Social Security actuaries.  If the Administration’s tax cuts are extended and relief from the Alternative Minimum Tax is continued so the AMT does not affect tens of millions of families, this additional $10 trillion would bring the federal debt up to about 100 percent of the Gross Domestic Product by 2030.

It should be noted that these figures, generated by the actuaries, significantly understate the extent to which the plan would increase deficits and the debt, because these estimates do not include the full cost of the plan’s new tax cuts.  Estimating the cost of tax cuts unrelated to the payroll tax or to Social Security financing is beyond the purview of the Social Security actuaries.  As a result, the actuaries did not include an estimate of the cost of allowing $5,000 in annual, tax-preferred contributions.  Taking that tax cut into account would show the plan to increase deficits and the debt for even longer periods.

Plan Would Do Little to Increase National Saving

One of the principal goals of Social Security reform is supposed to be to increase national saving.  National saving is the sum of private saving and public saving or dissaving.  Public saving refers to government budget surpluses, which add to national saving.  Public dissaving refers to government budget deficits, which reduce national saving by soaking up some of the private saving that otherwise would be available for investment in the private sector.

When national saving increases, more funds are available for investments that can make the economy more productive; over time, the economy and national income grow more as a result.  A larger economy, in turn, would better enable the nation to afford the costs of Social Security, Medicare, and other commitments, as well as produce a more prosperous society.

The Graham plan, however, would do little to increase national saving for the foreseeable future.  The increased saving caused by the plan’s private accounts would be offset by the increased government deficits caused by the borrowing undertaken to establish those accounts.

In fact, the plan could end up reducing national saving.  This would occur if individuals reduced the amount they saved in IRAs and 401(k)s in light of the balances in their new individual accounts, as CBO and many economists assume people would do to some degree.  If this took place, the plan could generate more new public borrowing than private saving.  If that occurred and national saving declined, the country would be even less prepared for the coming retirement of the baby boomers.

Plan Would Not Materially Improve the Long-Run Fiscal Outlook

The Social Security actuaries have estimated that the Graham plan would improve Social Security’s actuarial balance by 2.55 percent of payroll over 75 years.  But the proposal would have the opposite effect on the non-Social Security portion of the budget, enlarging the deficit in that part of the budget by 2.1 percent of payroll.

Thus, the plan would result in only a modest overall improvement in the federal government’s long-run fiscal problems, assuming all of the scheduled benefit cuts actually materialized.  Securing such an improvement is supposed to be one of the principal goals of Social Security reform.  (See Table 3.)

The plan would have a negative impact on the non-Social Security budget for two reasons.  First, it includes large transfers from the non-Social Security budget to Social Security.  Senator Graham has indicated that these transfers would be financed in part by reductions in “inequitable federal subsidies,” sometimes referred to as corporate welfare.  His plan calls for a commission to identify such subsidies and envisions $800 billion in savings in this area over the first decade and even larger savings in subsequent decades.

Table 3 

The Impact of the Graham Plan on the Long-Run Fiscal Balance

Percent of payroll over 75 years


Impact on Social Security Balance

Impact on Non-Social Security Balance

Total Impact on Unified Budget Balance

Shift from wage to price indexing and other benefit changes




Establish individual accounts




General revenue transfers to Social Security




Shift revenues from HI to OASDI




Savings incentives for low-income workers




Expanded tax shelter for upper-income savers




Total Impact




*Includes interaction.
Source:  Impact on Social Security balance is from the Social Security actuaries’ memo, Table A and preliminary estimates of the cost of expanded tax shelters of upper-income households.


Such savings, however, are unlikely to materialize in anything close to the amounts that Senator Graham assumes.  Moreover, the plan would require that the transfer of funds from the rest of the budget to Social Security take place regardless of whether these ambitious savings are achieved.

The plan also would shift certain revenues from the Medicare Hospital Insurance Trust Fund to the Social Security Trust Fund.  This would improve the solvency of Social Security but worsen the solvency of Medicare by an equal amount.

The other reason why the Graham proposal would have a negative effect on the non-Social Security budget is that it includes substantial new tax cuts (described above), which would reduce non-Social Security revenues and increase the unified budget deficit.

In summary, the Graham plan would restore long-term Social Security solvency but do little to address the nation’s long-run fiscal imbalance.  By seeming to reform Social Security without taking steps that would improve the nation’s grim fiscal outlook, the plan would effectively put the United States in a worse long-term fiscal situation, because an opportunity to lessen the nation’s long-term fiscal problems would have been lost.


Would Raising the Payroll Tax Cap Address the Plan’s Shortcomings?

Senator Graham has recently suggested raising the ceiling on the amount of earnings subject to the Social Security payroll tax as a way to lessen the amount of deficit financing that would be needed to establish private accounts.  Raising the payroll tax cap may be a sound element of a Social Security plan.  It would fall far short, however, of producing the revenue needed to finance the plan’s “transition costs” fully and avoid the need for extensive borrowing.

For example, if the ceiling on taxable earnings, now $90,000, were immediately raised to about $145,000, it would raise about $600 billion over ten years.  At this level, 90 percent of wages would be under the cap and thus subject to taxation, the percentage of wages that was covered immediately after the 1983 reforms.  As noted, the private accounts element of the Graham plan would require $2.4 trillion in borrowing during the initial decade.  So if this $600 billion in savings were factored in, the plan would still require $1.8 trillion in borrowing over the first ten years (and additional borrowing in decades after that).  Deficits and the debt still would increase for a long period of time.

Moreover, an immediate increase to $145,000 in the ceiling on taxable earnings would represent a more aggressive increase in the ceiling than would be made by most other leading proposals that would raise the ceiling.  Former Social Security Commissioner Robert Ball has proposed raising the payroll tax ceiling gradually so that by 2036, some 90 percent of wages would be under the cap.  Based on estimates by the Social Security actuaries, this change would raise $284 billion from 2006 to 2015 (including the associated savings in interest payments on the debt).  That would leave a requirement for $2.1 trillion in borrowing over the first ten years.  Proposals in the Kolbe-Stenholm plan and the Diamond-Orszag plan to raise the ceiling on taxable earnings would raise less revenue.[13]

Over 75 years, the increase in the ceiling on taxable wages that Robert Ball proposed would reduce Social Security’s actuarial deficit by 0.61 percent of payroll.  That would not be enough to cover the costs of the plan’s private accounts, which would worsen Social Security finances by 0.98 percent of payroll.

Senator Graham has discussed raising the ceiling to $200,000 and potentially lowering the payroll tax rate.  In the absence of additional details, it is impossible to estimate how much revenue that would raise.  But even eliminating the cap entirely, and doing so immediately, would fall short of the amount needed to fund the private accounts without borrowing.[14]

Because these proposals to raise the payroll tax ceiling would not fully offset the transition costs involved in creating private accounts, they would reduce but not eliminate the need for borrowing to finance private accounts.



The Graham plan has serious shortcomings.  It would result in deep cuts in Social Security benefits, yet do little or nothing to improve the government’s long-term fiscal imbalance.  The plan rests on large-scale borrowing that would increase deficits and the national debt for decades.  In addition, embedded in the plan is a major new tax break that would establish a new tax-shelter opportunity for high-income households.

Raising the cap on the amount of wages subject to the Social Security payroll tax, as Senator Graham has recently suggested, could ease some of these problems, but only to a modest degree.  Most of the serious shortcomings would remain.

End Notes:

[1] Households with incomes over $160,000 are not eligible for Roth IRAs.  The income limits on IRAs reflect the expectation that for people above this income level, contributions to IRAs generally would not represent new saving, but merely a shift of assets made to take advantage of the tax break.

[2] The Graham plan does not use any timing gimmicks to hold down its ten-year cost.  As a result, it appears significantly more expensive than the new White House proposal.  In reality, once the White House plan was fully phased in, it would have larger private accounts than the Graham plan.  However, the White House reduced the estimate of the cost of its plan over the first ten years by presenting a ten-year cost estimate for its proposal that includes only five full years of implementation.  (Also of note, the Social Security actuaries assume a two-thirds participation rate in the White House plan but assume a 100 percent participation rate in the Graham plan.  The difference reflects the fact that the private accounts would be a modestly better deal under the Graham plan than under the White House plan.)

[3] For more detail, see Robert Greenstein, 1/28/2005, “So-Called ‘Price-Indexing’ Proposal Would Result in Deep Reductions Over Time in Social Security Benefits,” Center on Budget and Policy Priorities.

[4] This 0.3 percent differential is equivalent to the portion of a worker’s private account that the Social Security actuaries assume will be consumed by management fees.  The differential was built into the benefit-reduction formula to ensure that workers whose accounts earn interest at the Treasury bond rate or better will receive a total retirement benefit (from their accounts and their Social Security benefits combined) that at least equals the Social Security benefits they would have received if they had not chosen private accounts.  (See the example on page 6).

[5] See, for example, Eric Engen and William Gale, 2000, “The Effects of 401(k) Plans on Household Wealth:  Differences Across Earnings Groups,” NBER Working Paper.

[6] The transfers expressed as a percentage of payroll, are from the Social Security Office of the Actuary, “Estimated OASDI Financial Effects of ‘Social Security Solvency and Modernization Act of 2003’ Introduced by Senator Lindsey Graham,” 11/18/2003.  “Present value” is the amount today that, with interest, would exactly cover these future costs.

[7] The reason that this option would be so unattractive is that individuals today would have to pay the full cost of ensuring solvency for the next 75 years.  If a worker chose this option, his or her payroll tax increase would be set at a level not just to ensure that his or her own benefits were not reduced but also to make up for larger future deficits in Social Security.  Furthermore, once an individual elected this option, he or she would have to pay the full cost of adjusting to any subsequent adverse changes in Social Security’s financial outlook, which could result in further payroll tax increases.  In contrast, most other Social Security solvency plans phase in their adjustments over time.

[8] The maximum is indexed to wage inflation after 2006.

[9] Melissa Favreault et al., 2004, “Reform Model Two of the President’s Commission to Strengthen Social Security:  Distributional Outcomes Under Different Economic and Behavioral Assumptions,” Center for Retirement Research Working Paper 2004-19.

[10] The proportion of wage indexing and price indexing used would be determined by the number of years from ages 22 to 61 that the worker did not receive disability benefits.  In this example, the benefit after age 67 is halfway between the current law benefit and the benefit under price indexing.

[11] This is based on an extrapolation of a preliminary 10-year estimate by the Tax Policy Center.

[12] Without any repayment to Social Security from the private accounts, the accounts would impose a long-run cost for the Social Security system equal to 2.68 percent of payroll over 75 years.  The repayments recoup more than half of this cost, leaving the net cost to Social Security at 0.98 percent of payroll.

[13] Under these plans, the ceiling on taxable wages would be set to cover 87 percent of wages, less than the 90 percent proposed in the Ball plan.  The Diamond-Orszag plan would phase in this change over 60 years.

[14] Eliminating the ceiling entirely would raise at most 2.47 percent of payroll in 2013, under the extreme assumption that the change did not result in any behavioral changes or income shifting and that the increased contributions were not counted toward higher Social Security benefits.  The private accounts in the Graham plan cost 2.58 percent of payroll in 2013.