Skip Navigation

August 4, 1998
 

Honorable Bill Archer
Chairman
Committee on Ways and Means
U.S. House of Representatives
Washington, D.C. 20515

Dear Mr. Chairman:

In response to your letter concerning Professor Martin Feldstein's proposal to set up private savings accounts financed by tax credits, the Congressional Budget Office (CBO) has prepared the following analysis.

Feldstein originally sketched his proposal in two editorials in the Wall Street Journal last winter. CBO's analysis relies on two more recent articles by Feldstein that clarify some issues and also modify his earlier proposal. In its most recent version, that proposal would introduce a refundable tax credit of 2 percent of taxable payroll income for investment in personal retirement accounts (PRAs). At retirement, the accumulated balances could be annuitized or withdrawn in a series of payments by the accountholders. Social Security benefits would be reduced by 75 cents for each dollar withdrawn from the accounts. Thus, retirees would keep 25 percent of their PRA savings on top of the Social Security benefit that they would receive under current law.

The proposed tax credit for investments in PRAs would cost roughly $800 billion over the next 10 years. The reduction in Social Security spending under the proposal would be small in the same period--totaling less than $10 billion--because workers who retire by 2008 will not have accumulated much in their PRAs. Beyond the 10-year projection period, budgetary pressures arising from the retirement of the baby-boom generation and the continued growth in health care costs are projected to create federal deficits. The proposed tax credit would add to those deficits.

The Feldstein proposal differs from plans that privatize Social Security in an important respect. The current Social Security system carries a large unfunded liability, and most proposals to privatize Social Security would shift some of that liability from future to current generations by cutting Social Security benefits. Without a promise of government support in retirement, people must save on their own for their retirement. By contrast, the Feldstein proposal guarantees the current retirement benefits and does not reduce the government's overall liabilities.

Any economic analysis of a proposed policy change must compare the policy with an alternative. CBO compares proposals with current law. That approach provides a clearly defined alternative and does not require CBO to make assumptions about the policy decisions that the Congress and the President would make if the proposal was not enacted.

Under current law, the government saves projected budget surpluses and uses them to reduce the debt held by the public. Compared with that alternative, the Feldstein proposal would increase future budgetary pressures because it cuts taxes on current workers. Payroll taxes essentially decline by an amount equal to 0.5 percent of payroll because the government transfers about 2 percent of taxable payroll income to workers but recovers only 75 percent of it later. Because the tax cuts must be financed one way or another, the policy would implicitly increase the tax burden on future workers if no further adjustments were made on the spending side of the budget.

Compared with current law, the proposal would also most likely reduce national saving: government saving would decline because the tax credit would reduce the surplus and, in later years, increase the deficit. Although private saving would rise, it would not rise enough to completely offset the loss in revenue. The proposed tax credit would have to be deposited in a retirement account, but even so, it is unlikely that private saving would go up by the full amount of the credit. Other saving most likely would fall because PRAs would increase workers' resources and would reduce their need to save in other forms for their own retirement.

Feldstein argues against using current law as the alternative. By his reckoning, the projected surplus will encourage spending increases or tax cuts that would save little for the future. Comparing Feldstein's proposal against that alternative, however, requires making assumptions about how the Congress and the President would otherwise use the surplus. The surplus could be spent on other tax cuts or for mandatory and discretionary programs. The choice matters because those alternative uses of the surplus would have different effects on the economy. Certain tax cuts, for example, would encourage saving and reduce economic distortions.

To illustrate Feldstein's argument, CBO compares Feldstein's proposal against an alternative in which the budget surpluses would be used to increase government purchases of goods and services. To keep the analysis simple, CBO assumes that those purchases would not affect future output and would be consumed right away.

Compared with spending the projected surpluses on government purchases, the Feldstein proposal could reduce future budgetary pressures. Because the budget surpluses would be saved in personal retirement accounts, those funds would not be available to finance other government activities. Once people start withdrawing their PRA balances, Social Security's outlays fall by 75 cents for each dollar withdrawn from PRAs, alleviating some of the pressure on future taxpayers. The exact long-run budgetary savings depend on the reduction in Social Security's outlays compared with the cost of the tax credit.

Setting up PRAs would also increase national saving if the surplus would otherwise be spent on government purchases of goods and services. PRAs would convert much of the surpluses into private saving. By contrast, higher government spending would simply increase current consumption.

Even assuming that budget surpluses would be consumed, the extent to which Feldstein's proposal would raise national saving and reduce budgetary outlays in the long run is uncertain. Those effects depend on the actual number of years with a surplus, the rate of return earned on the PRAs, and other adjustments to spending or taxes once the unified budget slips back into deficit some time after 2020. Unless those other changes are made, the tax credit for PRAs would increase deficits in some future years, diminishing the positive effect of PRAs on saving and budgetary pressures.

A variety of practical issues remain to be addressed by policymakers. Those issues include the administrative costs arising from PRA investments, the regulation of investment choices for accountholders to avoid overly risky investment strategies, and the integration of PRA withdrawals with Social Security.

An attachment to this letter presents a more detailed analysis of the proposal. If you would like further information, please feel free to contact me, or your staff may wish to contact Jan Walliser.

Sincerely,

June E. O'Neill
Director
 

Attachment

cc: Honorable Charles B. Rangel, Ranking Minority Member; Honorable Pete V. Domenici, Chairman, Senate Committee on the Budget; Honorable Frank R. Lautenberg, Ranking Minority Member; Honorable John R. Kasich, Chairman, House Committee on the Budget; Honorable John M. Spratt, Jr., Ranking Minority Member; Honorable William V. Roth, Jr., Chairman, Senate Committee on Finance; Honorable Daniel Patrick Moynihan, Ranking Minority Member
 
 

In two editorials in the Wall Street Journal, an article in The New Republic, and a recent paper written with Andrew Samwick, Martin Feldstein has laid out a proposal to use projected budget surpluses to fund personal retirement accounts (PRAs).⁽¹⁾ Under the proposal, individuals could invest 2 percent of their taxable payroll earnings in PRAs. In its analysis, the Congressional Budget Office (CBO) relied on the more recent publications, which clarified and modified the earlier proposal. The current version of the proposal offers a refundable income tax credit equal to the PRA contribution. In other words, people would be able to reduce their taxes dollar for dollar for the amount they invest in their personal retirement account. Because the credit would be refundable, those who did not owe any tax would be able to claim a full refund for the amount of their contribution.

The PRA balances could be invested in stocks, bonds, money market funds, or mutual funds and would accumulate tax-free. Policymakers could regulate the funds to prevent overly risky investment behavior and reduce administrative expenses. Distributions from the accounts would be made after people reached retirement age. The account balances could be withdrawn as an annuity, or retirees might be permitted to withdraw funds gradually over time.

The proposal would explicitly integrate PRA withdrawals with Social Security benefits. Social Security benefits would be reduced by 75 cents for every dollar an individual withdrew from his or her PRA account. Thus, retirees would keep 25 percent of their PRA savings on top of the Social Security benefit that they would receive under current law.

Other proposals put forward by Members of Congress and the last Advisory Council on Social Security share some of the features of this proposal, especially the idea that earnings from private accounts could partly substitute for Social Security benefits. But whereas the Feldstein proposal funds those private accounts from income tax credits and integrates withdrawals with Social Security benefits, most of the other proposals divert payroll taxes from Social Security to private accounts and cut Social Security benefits. The Feldstein proposal differs from plans that privatize Social Security in an important respect. The current Social Security system carries a large unfunded liability, and proposals to privatize Social Security would shift some of that liability from future to current generations by cutting Social Security benefits. Without a promise of government support in retirement, people must save on their own for their retirement. By contrast, the Feldstein proposal guarantees the current retirement benefits and does not reduce the government's overall liabilities.
 

BUDGETARY AND ECONOMIC EFFECTS

Any economic analysis of a proposed policy change must compare the policy with an alternative. Assuming a continuation of current law provides a useful benchmark for analyzing a proposal because it does not require making assumptions about whether the Congress would alter its decisions on other matters if the proposal was not adopted. That assumption is also consistent with the budget conventions used in CBO's cost estimates. Under that assumption, if the Feldstein proposal was not enacted, the budget surplus would not be spent on other proposals; instead, it would be used to reduce the level of outstanding debt.

However, Feldstein argues that the looming surpluses may induce policymakers to change outlays and revenues in ways that would save little or nothing of that surplus for the future. In his view, by committing policymakers to the tax credit before other spending decisions are made, the proposal could prevent the Congress and the President from spending down the surplus and would thus preserve some of the positive economic and budgetary effects of the surplus. Clearly, the evaluation of a proposal depends on the alternative.

The Feldstein Proposal

Compared with current law, the Feldstein proposal would increase long-run pressures on the budget and reduce national saving. It could also affect labor supply, though the direction and size of that effect is difficult to determine.

Budgetary Effects. The Congressional Budget Office has not prepared a cost estimate of the Feldstein proposal, nor has the Joint Committee on Taxation (JCT) estimated the revenue losses caused by the proposal's tax credit. However, JCT did estimate the revenue losses of an earlier proposal advanced by Feldstein that assumed PRAs would be set up in 1999. Because the current proposal is quite similar to the one analyzed by JCT, the main results of that analysis still apply, even though the exact figures would change somewhat.

The Feldstein proposal would reduce revenues in the near term and would increase budgetary pressures in the long run. In the first 10 years, the cumulative revenue losses would total roughly $800 billion. The reduction in Social Security's outlays as retirees draw down their PRA balances would most likely be small before 2008--cumulating less than $10 billion--but would rise significantly in later years.

The proposal would probably increase the long-run pressure on the budget. Because the trust fund eventually recoups only 75 cents (plus earnings) of each dollar provided by the tax credit, the government is essentially cutting payroll taxes by 0.2 percent of gross domestic product (that is, one-fourth of the tax credit). As a result, the long-run outlook for the unified budget would probably worsen relative to current law. (Under current law, the surpluses would retire federal debt). To be sure, the proposal improves the expected financial condition of Social Security's trust funds, but it does so essentially by shifting tax revenues from other parts of the budget. The trust fund's future outlays fall because of the integration of Social Security benefits with PRA withdrawals, but the rest of the budget loses revenue from the tax credit that financed the creation of the PRAs.

Saving. Creating PRAs would probably reduce national saving if the budget surplus would otherwise be used to retire federal debt. By increasing household wealth, the PRAs create incentives for people to consume more and save less.

Integrating 75 percent of Social Security benefits with PRA withdrawals, as Feldstein proposes, implicitly allocates 25 percent of the money in PRAs to current taxpayers and the other 75 percent to future taxpayers. By contrast, paying down government debt would use the entire surplus, not just the 75 percent implied by the proposal, for the benefit of future taxpayers. The proposal would thus reduce national saving below what it would be if the surplus was left untouched.

The reduction in national saving would depend on how much people could increase their consumption before retirement in response to the tax credit, and how much additional wealth they would bequeath to others. Feldstein argues that most households could not increase spending because the average family's financial assets are worth less than six months of income. However, households may reduce other assets to increase consumption. Median earners hold some wealth in retirement assets and considerable wealth in housing equity.⁽²⁾ Households could increase consumption by decreasing housing equity (taking out a second mortgage, for example), increasing credit card debt, or reducing saving in individual retirement accounts (IRAs) or 401(k) plans. A significant fraction of households have such assets: 45 percent of families either participate in employer-sponsored defined contribution plans or own an IRA, and among families with at least one member in the workforce, that figure is 61 percent.⁽³⁾ Empirical estimates--though highly uncertain in magnitude--generally also confirm that people reduce saving in response to their expectation of receiving higher Social Security payments.⁽⁴⁾

Labor Supply. A person's decisions about whether and how much to work depend, in part, on the tax rates he or she faces on current and prospective income. To be precise, decisions about whether and how much to work reflect the taxes levied on an additional dollar of labor income. Changing the tax burden on labor by offering a tax credit for PRAs can therefore influence the decision of how much to work. Moreover, offering a tax-favored vehicle for retirement saving may affect the decision about when to retire.

The effect of PRAs on work effort is unclear, however. A lower tax rate on an additional dollar of income may entice people to work more or work harder. The tax credit for PRAs would reduce the tax rate on an additional dollar of earnings by 0.5 percentage points for workers with income below the maximum taxable earnings ($68,400 in 1998). Those people would have an incentive to work more. Empirically, spouses of higher-income workers with smaller earnings themselves are most likely to respond to improved incentives to work.⁽⁵⁾ However, a higher after-tax income over a worker's lifetime also increases the incentive to spend more time on leisure activities. For workers who earn more than the maximum taxable earnings, creating PRAs would increase their after-tax income but would not lower the tax they pay on an additional dollar of earnings. Thus, those workers would be unlikely to increase their work effort in response, and they might even work less. On net, the change in work effort for the economy as a whole is unclear and most likely would be small.

The impact of PRAs on retirement behavior is also uncertain. On the one hand, workers may feel richer because of the windfall of reaping an additional 25 percent of PRA income at retirement, and some may decide to retire earlier. On the other hand, since the tax credit reduces the tax burden of working, some people may decide to postpone retirement. The net effect of PRAs on retirement behavior is thus unclear, but it is likely to be small since PRAs cannot be withdrawn before the earliest age for receiving Social Security benefits.

Spending the Surpluses in Other Ways

Feldstein and others argue that the predicted surpluses may entice policymakers to increase government spending or reduce taxes such that the projected surpluses never materialize.⁽⁶⁾ The surpluses could be spent in a number of other ways. They could be used to finance additional discretionary spending, to increase mandatory spending, or to reduce taxes. Compared with paying down the debt under current law, those ways of spending the surpluses would also increase future budgetary pressures and reduce national saving.

The exact impact of other changes to outlays and revenues on the budget and national saving would depend on what changes were made. Some ways to spend the surplus may simply allow current generations to consume more; others may have long-run benefits that partially offset the direct budgetary costs. For example, increasing government consumption would most likely reduce national saving by the same amount. A tax cut, however, could entice households to increase saving somewhat and would therefore have a less detrimental effect on national saving.

To simplify the analysis in this section, CBO assumes that the projected budget surplus would be spent by increasing government purchases of goods and services and that those purchases would be consumed right away and provide no lasting benefit to the economy. That alternative is extreme but provides a useful benchmark: compared with other ways to spend the surplus it has the most negative impact on future budgetary pressure and national saving.

Budgetary Effects. Eliminating the expected surpluses by raising government purchases implies an increase of government outlays (including the effects on debt service) of approximately $1.5 trillion between 1999 and 2008. Government spending would remain higher than under current law until some time around 2020, when the budget surpluses are expected to disappear as a result of the retirement of the baby boomers and rising health care costs.

The higher spending in the near term would also have adverse long-run effects. Under current law, projected surpluses would reduce outstanding federal debt, easing future budgetary pressures by reducing interest payments. By contrast, if the surpluses were spent, future interest spending, deficits, and debt would be higher. For example, spending the projected surpluses between 1998 and 2020 would increase the ratio of debt to gross domestic product (GDP) by about 40 percentage points in 2020. Accordingly, the projected deficits after 2020 would be larger than under current law if budget surpluses are spent in ways that preserve nothing for the future.

Saving. Under current law, the surpluses would add to national saving. By contrast, spending the surpluses by raising government expenditures would reduce national saving compared with current law. The lower national saving (and higher consumption) would reduce domestic capital investment, increase indebtedness to foreigners, and slow the growth of the economy.

Labor Supply. Increasing government purchases may not have a significant effect on households' decision to work and how much to work. Such decisions can be changed if a policy affects the relative price of labor and leisure or changes household wealth. Spending the surplus on government purchases does not alter the current tax burden placed on an additional dollar of labor income. However, future tax burdens may rise because the additional government spending must be financed. The effect of those future taxes on today's work effort is likely to be small.

Comparing the Feldstein Proposal with Other Ways to Spend the Surplus

Comparing Feldstein's plan with other proposals requires that those proposals be clearly specified. As discussed earlier, different ways of spending the surplus would affect saving and consumption differently. Thus, without a clearly specified alternative, CBO cannot make a comparison of the economic impact of the proposals. In this section, CBO assumes that the surplus would be spent in ways that do not raise the saving of the private sector--for example, by increasing the government's purchases of goods and services. Even if the Feldstein proposal raised national saving more than such an alternative, it might not compare as favorably against other options.

Budgetary Effects. Before 2020, the Feldstein proposal would cost less than the alternative on average, although in some years it would cost more. Moreover, 75 percent of the money in the PRAs would eventually return to the federal budget (with interest). Thus, the saving in PRAs during that early period would add resources to the budget in later years when workers who have accumulated PRAs begin to retire. The cash return to the federal budget, however, would be small for many years because people retiring only a few years after the program starts will not have built up large personal retirement accounts.

After about 2020, the federal surplus is projected to disappear as the retirement of baby boomers and the expected rise in health care costs increase pressure on federal spending. At that point, the money transferred into PRAs from federal revenues would simply increase the deficit and federal debt. Although three-quarters of it would eventually be returned to the federal budget as workers retire, one-quarter would be a long-run cost to the budget. Nonetheless, if the economy expanded sufficiently under the Feldstein proposal, the increased budgetary costs would be offset by higher revenues, and long-run budgetary pressures would most likely be smaller under the proposal than under a policy that consumed the surplus.

Saving. The Feldstein proposal would increase national saving compared with an alternative of simply consuming the surplus. Although government saving would be about the same under both options through 2020, private saving would be much higher under the Feldstein proposal because simply spending the surplus to finance government purchases of goods and services would not encourage private saving to the same extent.

Under the Feldstein proposal, workers would have an incentive to save three-quarters of the funds in their PRAs. Those funds would be needed for retirement because their Social Security benefits would be cut by 75 cents for every dollar they withdraw from their PRAs. However, they would probably consume the other 25 percent of the money transferred into PRAs.

After 2020, the effect of the proposal on national saving, compared with other ways of spending the surplus, depends on the amount of additional capital created under the Feldstein proposal in the earlier years. CBO's calculations suggest that the proposal would raise the capital stock by about 15 percent of GDP by 2020. That additional capital would increase gross domestic product and tax revenues. The additional tax revenue may raise government saving by enough to offset the additional consumption by the private sector that the tax credit for PRAs induces.

Labor Supply. Neither spending the surplus nor setting up PRAs would have large effects on labor supply. As discussed earlier, PRAs may change somewhat the incentive to work and retire, but the overall magnitude of the change is likely to be small. Therefore, it is probable that labor supply effects play almost no role in comparing the economic effects of Feldstein's plan with spending the surplus.
 

FELDSTEIN-SAMWICK ANALYSIS OF THE PROPOSAL

In a recent paper, Feldstein and Samwick have analyzed the effect of creating PRAs on the U.S. economy.⁽⁷⁾ Their calculations use an accounting model developed for the analysis of Social Security privatization.⁽⁸⁾ Feldstein and Samwick find that the proposal would substantially raise saving and GDP--GDP would be 3.2 percent larger in 2030 and 6.7 percent larger in 2070. Moreover, according to the calculations, the proposal would be self-financing after 2030 and would reduce Social Security's outlays such that the payroll tax could remain at 12.4 percent indefinitely.

The analysis by Feldstein and Samwick relies on a number of assumptions that would lead to an overstatement of the gains from creating PRAs. Most important, the analysis treats all money accumulated in PRAs as new saving. Other assumptions relate to the allocation of saving to different sectors and the rate of return people receive on their PRA balances and annuities.

First, Feldstein and Samwick consider funds accumulated in PRAs as new wealth, implying that the surplus would be used to increase government spending if PRAs were not set up. Moreover, for PRAs to be additional saving, it is also necessary to assume that policymakers would make certain (unspecified) cuts in spending or increases in taxes whenever the projected budget surplus was insufficient to finance the tax credit for PRAs (after accounting for the changes in tax revenues caused by PRA savings). That assumption ensures that the tax credit for PRAs would not add to the deficit. However, there is no basis for making such an assumption: it reflects a judgment about how future Congresses will respond to increasing budgetary pressure. Unless those adjustments are made, the tax credit would increase the federal deficit and become a drag on national saving. As a result, capital would accumulate at a significantly slower pace than calculated by Feldstein and Samwick.

Second, Feldstein and Samwick assume that people are unable to reduce their private saving outside the PRAs. As outlined in the previous section, people would be inclined to reduce other saving because PRAs increase their wealth. For example, because many households own IRAs or participate in 401(k) plans, they would most likely be able to reduce their other saving to a certain extent. Those reductions in private saving would diminish capital accumulation in the Feldstein-Samwick analysis and reduce the positive effect on output.

Third, they assume that all of the new saving generated by PRAs would raise both the corporate capital stock and federal corporate tax revenue. Indeed, under their assumptions, corporate tax revenue would finance the entire tax credit for PRAs after 2030. However, it is unlikely that all additional saving would be invested in the corporate sector. As Feldstein and Samwick acknowledge, additional saving might be funneled into the housing sector, where it would earn a much lower rate of return and generate much lower tax revenues than in the corporate sector. In addition, some money may be invested in noncorporate businesses, where it might also generate lower rates of return and less in federal taxes than in the corporate sector.

Fourth, the Feldstein-Samwick calculations assume that the equity premium can be exploited to increase the well-being of future generations, but a more detailed analysis suggests that well-being may not increase because higher returns also expose those generations to more risk. Over the past century, the equity premium--the difference between the real return on equities and Treasury bills--has been about 6 percent. In their model, the authors assume that people invest in a portfolio of stocks and bonds that generates an expected real rate of return of 8.5 percent for the economy as a whole (some of which goes to the government in the form of corporate taxes), whereas the government pays a real rate of return of 2.3 percent on its debt. Therefore, shifting funds into PRAs generates a faster accumulation of capital than paying down the debt under current law. However, those higher returns are also riskier. The proposal would pass that risk to future generations because they would be responsible for keeping the trust fund financially viable.⁽⁹⁾

Analysts disagree over whether taxpayers would be better off on a risk-adjusted basis. Although the equity premium compensates for the riskiness investors face in the stock market, some studies suggest that it is larger than can be explained by risk alone.⁽¹⁰⁾ However, analysts are uncertain whether a large equity premium will persist. The equity premium is not well understood, and it may be unwise to assume that the government can get a "free lunch" by simply investing in the stock market.

Finally, Feldstein and Samwick assume that all retirees annuitize their PRA savings and continue to receive a real rate of return of 5.5 percent after retirement. That return reflects a portfolio of bonds and stocks and is risky. Currently available annuities can either be fixed (paying a constant stream of income) or variable (paying a stream of income that fluctuates with market returns). Fixed annuities that resemble Social Security's income stream are generally backed by fixed-income instruments that offer a real return much lower than 5.5 percent. In their analysis, Feldstein and Samwick sidestep the difference between fixed and variable annuities by assuming that realized and expected returns are identical. In other words, their analysis also assumes that the equity premium can be exploited during the withdrawal of PRA balances.
 

OTHER ISSUES

Feldstein's plan to establish personal retirement accounts does not address certain issues--administration, investment and risk, the risk allocation between the government and beneficiaries, and withdrawal and integration of benefits--that would be critical for implementing the proposal. CBO's analysis of those issues has flagged the following questions.

Administration

Feldstein does not lay out an administrative framework for the PRAs. A number of questions arise that have repercussions on the costs of setting up a new system. Who would be responsible for collecting and transferring the funds for PRAs? Would employers send the money directly to private investment companies, or would they transfer it to the Social Security Administration, which would act as the financial intermediary? Who would bear the additional administrative burden--the employer or the government? Even if the responsibility for collecting and distributing funds lay with the employer, government oversight would probably be necessary to ensure compliance.

Costs could differ substantially depending on how policymakers structure the administrative framework of PRAs. For example, if PRA investors could choose among a variety of private investment funds, the administrative burden for the employer could be fairly large if it was the firm's responsibility to transfer the money. High administrative costs would would lower the funds being funneled into PRAs, diminishing their effect on national saving.

Investment and Risk

What types of investment would be available? The proposal assumes that the regulation of PRAs would resemble the current rules for IRAs, permitting investment in a large variety of mutual funds, stocks, and bonds. By contrast, many existing employer-sponsored plans, like the federal government's Thrift Savings Plan, limit investment to a few funds.

Regulating investment choice has important repercussions for risk and costs. If policymakers allowed investment in a variety of financial instruments, that could lead people to indulge in risky investment behavior. Such behavior would be costly to the taxpayer, who is a 75 percent partner in the PRAs without a direct vote on how they are invested. However, limiting the number of investment options to a few funds would reduce the opportunity for people to seek out investments that match their personal willingness to undertake risk.

Moreover, allowing many options also tends to increase administrative and management costs, possibly making the investment in PRAs unattractive for low-income and part-time workers with small account balances. In 1994, almost 42 million workers--32 percent of all workers paying payroll taxes in that year--had covered earnings of $8,400 or less, many of them part-time workers who did not work the entire year and would have qualified for a tax credit of less than $170 under the proposal.⁽¹¹⁾ Currently, many mutual fund companies require a minimum investment of from $500 to $1,000 for IRAs. If PRAs were created, those companies might be willing to open smaller accounts but would probably require a minimum annual administrative fee, reducing the benefit of the PRA tax credit.⁽¹²⁾ Therefore, those low-income and part-time workers might be less inclined to invest in PRAs and take advantage of the tax credit.

Risk Allocation Between Government and Beneficiaries

The rules on integrating PRA withdrawals with Social Security benefits would determine the allocation of risk between the government and future beneficiaries. Since future Social Security benefits would be cut in line with the actual performance of PRAs, Social Security's outlays would vary with the return on PRA investments. If PRAs underperformed, future taxpayers would have to step in to pay for the promised Social Security benefits.⁽¹³⁾

The implicit guarantee afforded by the continuation of Social Security benefits could increase the incentive to invest PRAs in risky assets. Moreover, some investors might seek highly speculative investments because they could keep additional returns once the PRA balance was sufficient to replace their entire Social Security benefit. Such risky investment behavior would increase the risk for government finances, possibly creating a new liability.

Withdrawal and Integration of Benefits

The proposal would allow retirees to withdraw their funds either in a series of periodic payments or as an annuity.⁽¹⁴⁾ However, without appropriate regulation, annuities that replace Social Security's inflation-indexed income stream may be costly or unavailable. Existing annuities markets suffer from imperfections because people who expect to and indeed do live longer than average are more likely to purchase annuities than those who do not have those expectations. Also, fixed annuities--annuities that pay a constant stream of income similar to Social Security--generally receive lower rates of return than Feldstein and Samwick assume in their analysis. The rates are lower because the annuity insurer bears the risk of fluctuation in interest rates. Moreover, inflation-indexed annuities are currently available only for investors in the College Retirement Equity Fund (CREF); unless those annuities became available more widely in the future, PRA annuities--unlike the current Social Security benefits--would be subject to inflation risk.

It might be difficult to integrate Social Security benefits with PRAs. For example, if retirees chose to withdraw their PRA funds in periodic payments, by how much should their Social Security benefits be cut? Would a retiree receive the full Social Security benefit if he or she outlived PRA savings? Those problems would be eliminated if retirees were forced to purchase a fixed inflation-indexed annuity with survivor benefits, because such annuities mirror those paid by Social Security. Integration could also be difficult and costly to administer. Without Social Security or some other government entity collecting earnings records and keeping track of investment returns to PRAs, it would be difficult to enforce the integration of PRAs with Social Security.
 

CONCLUSION

Compared with using the projected surpluses to retire the national debt, the Feldstein proposal would increase future budgetary pressures and most likely reduce national saving because it would not reduce future entitlement spending by the full amount saved in the proposed personal retirement accounts. If, however, the surpluses that are currently projected never materialize because their prospect encourages policymakers to increase spending or cut taxes in ways that would not help the economy in the long run, creating PRAs might preserve a substantial part of the increase in national saving that a surplus naturally provides.

A variety of practical issues remain to be addressed by policymakers. Those issues relate to the administrative costs, regulation of investments, integration of benefits, and the distribution of risk between the government and owners of PRAs.

1. See "Don't Waste the Budget Surplus," Wall Street Journal, November 4, 1997; "Let's Really Save Social Security," Wall Street Journal, February 10, 1998; "Savings Grace," The New Republic, April 6, 1998; and Martin Feldstein and Andrew Samwick, Two Percent Personal Retirement Accounts: Their Potential Effects on Social Security Tax Rates and National Saving, Working Paper No. 6540 (Cambridge, Mass.: National Bureau of Economic Research, April 1998). All articles are also available on the World Wide Web (http://www.nber.org).

2. Alan L. Gustman and others, Pension and Social Security Wealth in the Health and Retirement Study, Working Paper No. 5912 (Cambridge, Mass.: National Bureau of Economic Research, February 1997).

3. CBO calculation based on the 1995 Survey of Consumer Finances, a triennial survey conducted by the Board of Governors of the Federal Reserve System. Information about the survey and survey data can be found on the World Wide Web (http://www.federalreserve.gov/boarddocs/surveys/).

4. See Martin S. Feldstein and Anthony Pellechio, "Social Security and Household Wealth Accumulation: New Microeconometric Evidence," Review of Economics and Statistics, vol. 61, no. 3 (August 1979), pp. 361-368; and Congressional Budget Office, Social Security and Private Saving: A Review of the Empirical Literature, CBO Memorandum (July 1998).

5. Congressional Budget Office, Labor Supply and Taxes, CBO Memorandum (January 1996).

6. See "Overselling the Surplus," The Economist, May 30, 1998, p. 22, for a concurring opinion.

7. Feldstein and Samwick, Two Percent Personal Retirement Accounts.

8. Martin Feldstein and Andrew Samwick, "The Economics of Prefunding Social Security and Medicare Benefits," in Ben S. Bernanke and Julio Rotemberg, eds., 1997 NBER Macroeconomics Annual (Cambridge, Mass.: MIT Press, 1997), pp. 115-147.

9. Kent A. Smetters, "Investing the Trust Fund in Equities: An Options Pricing Approach," Technical Paper 1997-1 (July 1997), available from CBO's Macroeconomic Analysis Division.

10. Narayana R. Kocherlakota, "The Equity Premium: It's Still a Puzzle," Journal of Economic Literature, vol. 34, no. 1 (March 1996), pp. 42-71.

11. Social Security Administration, Annual Statistical Supplement to the Social Security Bulletin (1997), Table 4.B7.

12. A number of mutual fund companies currently waive their low-balance fees for IRAs.

13. Smetters, "Investing the Trust Fund in Equities."

14. For a discussion of the options for the withdrawal phase and the importance of annuities markets, see Congressional Budget Office, Social Security Privatization and the Annuities Market, CBO Paper (February 1998).