Entitlement Reform

Henry J. Aaron
Henry J. Aaron The Bruce and Virginia MacLaury Chair, Senior Fellow Emeritus - Economic Studies

July 27, 2000

Mr. Chairman:

Thank you for the invitation to testify today on entitlement reform. This topic raises so many issues that the limits of time demand that one be selective.

I should like to start with some observations with which I think everyone on this committee and all those sitting before you agree.

  • First, current long-run projections indicate that neither Social Security nor Medicare will have enough revenue under current law to pay for all the benefits promised under current law. Each faces a long-term financing problem, and the sooner Congress acts to deal with those problems the better.

Second, the long-run projections of both programs have improved in recent years—dramatically in the case of Medicare and significantly, but less dramatically in the case of Social Security.

Third, both programs are now running sizeable cash flow surpluses and these surpluses are currently projected to continue throughout the forthcoming decade and beyond. Social Security has sufficient revenues to pay all currently promised benefits for the next thirty-seven years, Medicare for about the next twenty-five. These facts mean that talk of “crisis” is hyperbolic nonsense. But they do not contradict the existence of a projected, long-run financing problem or excuse Congress and the next president from moving expeditiously to solve it.

Fourth, whether one favors or opposes the diversion of part of the current payroll tax to underwrite the creation of individual savings accounts, reducing revenues flowing to Social Security will increase the size of the cuts in Social Security benefits necessary to restore balance in that program. This is a matter of simple arithmetic. We may disagree on the likelihood that balances accumulated in such individual accounts will compensate workers for the cuts in Social Security benefits. That is an issue to which I shall return presently.

I should also like to comment on two other matters. The first is the claim that Social Security and Medicare reserves are just accounting mechanisms, that all it holds are “paper” assets, and that the existence of large trust fund balances does not have any impact on the Government’s ability to pay benefits. This view is simply wrong, and I shall explain why. The final issue that I shall address in my testimony is the legislative proposal, now under discussion, to repeal the 1993 legislation mandating couples with taxable incomes above $44,000 and single persons with incomes above $34,000 to include 85 percent of their Social Security benefits in taxable incomes, revenues from which are now deposited in the Medicare Trust Fund. I shall explain why repeal of this tax has no analytic justification.

The Financial Condition of Social Security and Medicare

The long-run financial projections for both Social Security and Medicare indicate that the programs face sizeable projected long-term deficits. This fact is well-known. What is less well known is that these projections have been quite volatile and that further sizeable adjustments can be expected. Table 1 illustrates both the existence of financial problems and the volatility of projections. The projected long-run deficit in Medicare is now 72 percent smaller than it was just three years ago. The Balanced Budget Act of 1997 explains most of the change, but the projected Medicare deficit today would be more than twice as large as it is, were it not for other changes. The improvement in the financial status of Social Security is less dramatic, but still significant; the projected long-term deficit is 15 percent smaller than it was three years ago.

Even more striking has been the inaccuracy in projections of balance in the fund in specific years. I have picked the year 2000 to illustrate the problem. Just three years ago, Medicare was projected to run a $32 billion deficit in 2000. It fact, it will run a $28 billion surplus, a $60 billion swing in just three years. The Social Security surplus is also about $60 billion larger than projected just three years ago, and without significant legislative change. These errors illustrate that even very skilled professionals fail to make accurate projections, even in the near term balance of these two programs. These projections also illustrate that both programs are now collecting much more than they are spending and that results are better than anticipated just a few years ago. To label this situation a crisis makes little sense.

As one looks into the more distant future, the uncertainty of projections increases. The reason is that long-term projections depend on extrapolations of assumed growth rates. Small errors in assumptions regarding compound growth rates cumulate into massive errors after periods as long as the seventy-five year projection periods used for Social Security and Medicare. If real wages were to grow half a percent a year faster than assumed by the actuaries—a rate that is below the actual record of the last three years—real earnings seventy-five years hence would be 45 percent higher than the current projections assume and the projected deficits in Social Security and Medicare would be 27 percent and 14 percent smaller, respectively, than current official projections indicate. On the other hand, sharper decreases in mortality rates than now assumed could result in significantly larger deficits than current projections indicate. The simple fact is that we do not know how to make accurate forecasts, over very long periods, of any of the variables on which Social Security and Medicare projections depend—birth rates, death rates, productivity growth, disability rates, immigration rates, real interest rates, the rate and character of advance in medical science, or the evolution of institutions to hold down medical costs. If you doubt me, I invite you to examine previous projections of each of these variables contained in past Trustees Reports. Current projections may be too optimistic. They may be too pessimistic. But they will assuredly be wrong—despite the best efforts of some of the most competent and dedicated professionals working in the pension and health insurance fields.

This fact does not mean that they are biased or that we should ignore long-term projections. They are based on reasonable, if unreliable, assumptions regarding key variables that lie well within the rather wide range of estimates of responsible analysts. The projections are signals that problems may well lie in the future. Given the length of pension promises and the need for gradualism in modifying those promises, we should gradually introduce changes when long-run projections indicate that problems probably lie ahead. It would be imprudent in the extreme to make abrupt changes based on long-term financial projections, particularly when, as now, financial balances are currently favorable and expected to remain that way. But it would also be imprudent to ignore the warning signals and do nothing now. We should act promptly to phase in changes to close projected long-term deficits. And we should recognize that as more information becomes available, we may undo those changes or we may do more.

The Simple Arithmetic of Diverting Payroll Taxes to Individual Accounts

Many people favor the creation of individual accounts as a partial or complete substitute for Social Security. Some propose to fund these accounts out of general revenues. When some part of the pensions based on these individual accounts is used to reduce Social Security benefits, this approach can indirectly reduce the projected long-term deficit in Social Security. This is the approach used, for example, in the Archer/Shaw bill.

Other so-called “carve-out” plans, such as those of Senator Kerry and Governor Bush, would divert part of the current payroll tax from the Social Security system. Their plans would carve out part of the payroll tax, which would then be directed to individual accounts. They would cut Social Security benefits enough to restore projected long-term balance.

  • The first point to recognize is that by subtracting revenues from the Social Security system, these plans force larger cuts than would otherwise be necessary to restore financial balance in that system. On the other hand, pensioners would have the balances in their individual accounts with which they could (or, in some plans, would have to) buy annuities.

This trade raises several practical questions:

  • Will the individual-account-based pensions fully compensate pensioners for the Social Security cuts?

Will the individual-account-based pensions be inflation protected?

Will individual account holders be required to convert their accounts into annuities? If not, what happens to those who are imprudent or unlucky, exhaust their accounts, and find themselves dependent on much-reduced Social Security benefits.

The Bush Plan as Illustration

As far as Governor Bush’s proposal is concerned, I have no idea about how he would answer the second and third questions because his statements so far have been confined to broad principles and do not address many of the difficult technical questions with which, to their credit Senator Kerry and Representatives Archer and Shaw have grappled. But some simple arithmetic suffices to answer the first of these questions “What will be the effect of the proposal on retirement income?” Tables 2, 3, and 4, from a report by Alan Blinder, Alicia Munnell, Peter Orszag and me, provide the answer.

Average Benefit Cuts

  • If one were to use benefit cuts to close the gap, proportional cuts relative to current law in Social Security benefits of 41 percent for all workers would restore long-term financial balance (Table 2).

We derived this conclusion from the following basic arithmetic. The cost of the benefits that Governor Bush does not promise to leave unchanged is 9.2 percent of taxable payroll. The current deficit is 1.89 percent of payroll. If one diverts to individual accounts 2 percentage points of the payroll tax starting in 2002, the size of the deficit rises to 3.8 percent of payroll—3.8/9.2 = 0.41.

Phased Benefit Cuts

The foregoing estimate assumes that benefits are cut abruptly and equally for all workers who are covered, even briefly, under the new individual account system. Such a policy would be unfair, however. Older workers would have little time to build up individual account balances and would suffer major reductions in their pensions. Younger workers would fare better because their individual accounts would have more years to build up.

Accordingly, we calculated a phased-in reduction in Social Security cuts, so that the change in the combined Social Security benefit and individual-account-based-pensions would be the same for workers of all ages. Table 3 shows the cuts in Social Security benefits for the average earner.

  • Under this more realistic schedule, Social Security benefits would have to be cut 25 percent for 55-year-old workers and 54 percent for workers age 30 or younger.

Overall Effect

Table 4 shows the combined effect of the Social Security benefit cut and the partly offsetting pension that could be financed by the individual account.

  • Total benefits are cut 20 percent relative to current law for the average single earner who earns the average rate on individual accounts assumed by the Bush advisers.


Average earners who happened to experience the lowest rate of return actually observed over a thirty-five year historical period (1947-1981) would have earned less in their individual accounts and would experience an overall reduction of 38 percent relative to current law. Average earners who received the highest rate of return actually observed over a thirty-five year historical period (1965-1999) would have earned more on their individual accounts and would have no cut in benefits relative to current law.

All of the foregoing statements apply only to average single earners. Table 5 shows the effects of the partial shift to individual accounts on married earners and on workers who have above- or below-average earnings. Married workers experience larger cuts in their combined benefits because their Social Security benefits—and hence their benefit cuts—are larger absolutely than those of single workers, yet their individual accounts will be the same. Low earners experience larger cuts in their combined benefits because the Social Security benefit formula favors low earners while individual accounts do not. Under the Bush plan, cuts in combined Social Security and individual account benefits for married, low-earners who receive lower-than-average returns on their individual accounts could approach 50 percent.

Can the Disabled, Current Retirees, and Older Workers be Protected?

Despite claims in the Bush plan that the disabled, current retirees, and those near retirement would be spared all benefit cuts, it is difficult to believe that Congress—or, indeed, Mr. Bush on fuller consideration—would decide to cut retirement benefits for younger workers by 50 percent or even more and leave the disabled, older workers, and current retirees wholly insulated from benefit reductions. I believe that few elected officials would think it fair to subject some Americans to large benefits cuts yet spare others from any cuts at all.

Can General Revenues Soften the Blow?

Benefit cuts would be smaller than I have indicated if a plan transfers general revenues to the Social Security fund. Based on revenue and expenditure projections based on current law, official projections indicate that there will be sufficient general fund revenues to support sizeable transfers—$2.2 trillion over the next decade alone, according to the most recent CBO projections.

However, these projections misstate the budget situation for several reasons. First, as this committee knows well, this projection assumes that growth of discretionary spending will not exceed inflation. Neither party has shown a willingness to live within such tight constraints. Neither Republicans nor Democrats, as groups, have recently shown a willingness to hold discretionary spending growth as low as the rate of inflation.

Second, both parties have agreed that cash flow surpluses in Social Security should not be used to justify spending increases or tax cuts. The logic is that these reserves (and more) will be needed to pay for future benefits. Yet the same logic applies to Medicare reserves and to balances accumulating in the Civil Service Retirement system, both of which are now counted toward the projected budget surpluses. Both should be removed. Doing so would reduce the projected surpluses by approximately $500 billion over the next decade, reducing the projected surpluses to $1.7 trillion. Adjusting the AMT for inflation and various other tax extenders will reduce revenues by about $150 billion over the next decade, leaving a projected surplus of about $1.55 trillion. As it happens $1.55 trillion exactly matches the cost of the tax cut that Governor George W. Bush has proposed, as estimated by the Joint Committee on Taxation, plus added interest costs that would be generated by the tax cut.

  • In short, if the candidate for president who supports individual accounts is elected, his other policies are adopted, and one uses plausible budget projections, there would be no funds to transfer to Social Security, unless the deficit financing pays for the transfers

Martin Feldstein has argued that establishing individual accounts would boost national saving and that the corporate profits taxes generated by a larger capital stock could be transferred to Social Security to reduce the size of benefit cuts that would otherwise be necessary. The claim that individual accounts would boost national saving is without foundation, however. As indicated, the general revenue transfers would have to be financed by borrowing from the public; and each dollar of payroll taxes transferred to individual accounts would force the government either to borrow $1 more or pay down the federal debt $1 less for each dollar deposited in individual accounts, a wash transaction that would not tend to boost saving at all.

If, on the other hand, large tax cuts are not enacted, the general fund is likely to generate some surpluses—although not as large as current CBO or OMB projections would lead one to think—and resources would be available to support general revenue transfers to Social Security.

Why General Revenue Transfers Make Sense

General revenue transfers to Social Security do make sense. The program’s unfunded liability is more properly viewed as an obligation of the American people as a whole than of future workers based on their earnings.

Early Social Security beneficiaries received benefits worth far more than the payroll taxes they and their employers paid. Money to pay these extra benefits came from the payroll tax collections of still-active workers. The period when cumulative benefits to new retirees will be worth more than the payroll taxes paid by them and their employers is coming to an end. Current retirees and those who will retire in the future will, on the average, receive benefits worth no more than the taxes they have paid, cumulated at a modest real rate of return. Thus, Social Security will not be generating new unfunded liabilities for future retirees.

Whether or not one thinks that the payment of comparatively generous benefits to early Social Security retirees was a good or a bad idea, that action cannot now be undone. The reserves not accumulated to support benefits of future retirees is an obligation that we must meet, one way or another. The question is: who should meet it? Under current law, the cost of paying for this unfunded liability falls on workers, in proportion to their earnings. The rationale for this policy is difficult to comprehend. They will be receiving in benefits no more than they and their employers will be paying in taxes. The unfunded liability, like the national debt, should be recognized as a general obligation of the American people. To be sure, workers are a large part of the American people and their earnings are a large part of the overall tax base. But I know no one who would suggest financing the pay-down of the national debt or interest payments on that debt exclusively from the payroll tax, and there is no good reason for distinguishing Social Security’s unfunded liability from general obligations of the Federal government.

Social Security Trust Fund:—Phony or Real?

Some analysts have claimed that Social Security and Medicare reserves are just accounting mechanisms, that the Trust Funds only hold “paper” assets. They sometimes claim that the accumulation of large trust fund balances does nothing to improve the Government’s ability to pay future benefits. This view is simply wrong.

One should begin by acknowledging that government accounting provisions contain many arbitrary conventions and that if different conventions had been adopted, budget accounts would look different from the way they do now. Professor Laurence Kotlikoff, among others, has contributed greatly to our understanding of these anomalies by pointing out these problems in a series of articles in economics journals. But the issue here is not whether government accounts are logically consistent constructs. The issue is whether a policy of collecting more in taxes earmarked for Social Security than is paid in Social Security benefits contributes to the nation’s and the government’s capacity to meet future benefit obligations. The answer to both questions is “yes,” and the issue is not even close.

The first step is to recognize that the direct effects on private investment of adding $1 billion to Social Security reserves or to individual accounts are identical, as shown in Table 6. Given government spending and revenues outside Social Security, a $1 billion cash flow surplus in Social Security and $1 billion of private saving directly add to funds available for private investment in exactly the same way and in the same amount. In each case, the return to the nation is $1 billion multiplied by the private, marginal productivity of capital. Table 6 demonstrates that the answer to the first question I posed—does the accumulation of Social Security reserves increase the nation’s capacity to pay pensions in the future—is a clear and unambiguous “yes.”

The accumulation of reserves also shifts the asset position of the federal government. The accumulation of $1 billion in Social Security reserves means that future taxpayers will be spared $1 billion in taxes to pay for any given level of future benefits. By paying more in taxes today, we shall have to pay less taxes in the future. To be sure, some form of financial transaction will be necessary to pay for those future benefits, but that is true everywhere and always when savers cash in assets to pay for something they want to buy. Private savers must reduce future saving or increase borrowing when they cash in assets they have accumulated. The Social Security administration will have to do the same.

The statement that Social Security reserves are only “paper assets” is true at an insignificant level that has no significance, and is false in substance. Neither Social Security nor private financial savers, including individuals and pension funds, hold “real” assets in their accounts. Both hold IOUs—paper promises of some private or public entity to pay interest or dividends. In each case, the assets are only as good as the willingness of someone to redeem the assets or buy them before maturity. In each case, any future need to cash in reserves to meet current obligations would reduce national saving. The only difference between reserves of Social Security and those of private savers is that Social Security’s reserves consist entirely of “gild-edged” federal securities, because federal law restricts Social Security trustees to invest only in securities guaranteed as to principle and interest by the federal government, while private savers can invest in assets in private securities, which carry higher yields because the companies issuing them face some risk of bankruptcy. Social Security reserves are as real as the reserves of any private pension fund, personal brokerage account, or corporate reserves.

This view that Trust fund assets are not real confuses two distinct questions: whether trust fund accumulation adds to national saving, investment, and the capacity to pay future pension benefits; and whether government budget operations on accounts other than Social Security add to national saving, investment, and the capacity to pay future benefits. As noted, additions to Social Security reserves add to national saving and the capacity of the government to meet future pension obligations in precisely the same sense that additions to private savings accounts add to national saving and the capacity of savers to meet their debts.

On the other hand, simultaneous deficits in the non-Social Security budget can subtract from national saving. From fiscal year 1983 through fiscal year 1999, Social Security ran surpluses—thereby adding to national saving—but deficits in the rest of government operations subtracted from national saving. From 1983 through 1997, the deficits on non-Social Security accounts exceeded Social Security surpluses so that the federal government as a whole ran deficits, thereby reducing national saving. In 1998 the Social Security surpluses exceeded the deficit on the rest of government operations. And starting in 1999 the federal government began to run surpluses both in Social Security and in the rest of government operations. In no case, however, does the fact that non-Social Security operations of government are in deficit contradict the fact that additions to Social Security reserves add to national saving, productive capacity, and the government’s balance sheet, thereby increasing the capacity of federal government and of the nation to meet future pension obligations.

Taxation of Social Security Benefits

In 1983, President Reagan signed into law a bill under which only half of Social Security benefits would be included in income subject to tax and only to the extent that couples’ incomes exceeded $32,000 and single persons’ income exceeded $25,000. The revenues were to be transferred to the OASDI trust funds. In 1993, President Clinton signed into law a provision that 85 percent of Social Security benefits would be included in income subject to tax, but only to the extent that couples’ incomes exceeded $44,000 and single filers’ incomes exceeded $34,000. The revenues were to be transferred to the Medicare trust fund. There would be no income thresholds below which Social Security income would be exempt.

The rules applied to taxing Social Security are patterned on, but are more lenient than, those applied to taxation of contributory private pensions. Pensioners are required to include private pensions in income subject to tax pensions only to the extent that they represent the repayment of contributions out of previously taxed income. If the same rules were applied to Social Security, workers would be required to include in income subject to tax all Social Security benefits in excess of a portion equal to their own payroll tax payments, but the rest of benefits would be subject to tax.

In 1979, I chaired the Advisory Council on Social Security which reported that if that rule were applied to workers retiring at that time, less than 15 percent of benefits would be excluded from income subject to tax for any worker and the percentage would be lower for most workers. That meant that 85 percent or more of Social Security benefits should be included in income subject to tax if they were to be treated in the same way as contributory private pensions.

Even after the 1993 legislation, Social Security benefits are treated more favorably than are contributory private pensions. On grounds of tax policy, there is no basis for repealing the tax enacted in 1993. It is needed for the proper definition of an income tax base. To be sure, there is no particular reason for allocating revenue from the taxation of Social Security benefits to either the Social Security or the Medicare trust fund any more than there is justification for transferring revenues from taxing private pensions to private pension funds. But, as I have noted, there is a good case for general revenue transfers to Social Security; and the same logic applies to Medicare. Since both programs face projected long-term deficits and since the current tax treatment of Social Security is still more favorable than that of contributory private pensions, I believe that there is no analytical justification for reducing this tax at this time.

Return to the ‘Table 1’ citation.

Return to the first ‘Table 2’ citation.

Return to the second ‘Table 2’ citation.

Return to the first ‘Table 3’ citation.

Return to the second ‘Table 3’ citation.

Return to the first ‘Table 4’ citation.

Return to the second ‘Table 4’ citation.

Return to the ‘Table 5’ citation.

Return to the first ‘Table 6’ citation.

Return to the first ‘Table 6’ citation.