4. Social Security, Medicare and Medicaid

"Good luck! Continued success!" — Participant

Brooks Jackson, senior CNN correspondent directs a participant’s question to the panel of experts.

Barry Rogstad, President, American Business Conference, answers questions during the Exercise.

The Committee for a Responsible Federal Budget is a bipartisan, non-profit, 501 (c)(3) educational organization committed to educating the public regarding the federal budget process and issues that have significant fiscal policy impact. The Committee conducts analysis and research, organizes educational symposia, and prepares and distributes educational materials to Congress, the Administration, the media, and the public. The Committee board consists of former members of Congress, former directors of the Office of Management and Budget and the Congressional Budget Office, and other economic and fiscal policy experts.

The committee is supported entirely by contributions from businesses and business organizations, foundations, and individuals.

Washington DC

Major Conclusions

Strengthening Budget Enforcement and Scorekeeping

Budget Concepts and Economic Assumptions

Language Matters

Example: The Budgetary Impact of Trust Fund Investments in Private Securities

Appendix 1: Trust Fund Accounting and the Scoring Rules for Proposed Reforms of the  Social Security System, Rudolph G. Penner

In a perfect world, policy decisions—especially decisions as big as those involved in Social Security reform—would be based on the merits of competing proposals and the potential for each to fulfill overall policy objectives. In the real world of politics, how policy decisions will be recorded on the government’s "books" can be as contentious as any substantive policy argument. Most seriously, budget accounting and scorekeeping can influence the outcome of policy debates.

The Committee for a Responsible Federal Budget presents the following conclusions about budget concepts, scorekeeping, and enforcement issues as they pertain to Social Security reform. The answers to these difficult issues are not always crystal clear. Different answers to key conceptual issues could produce different budget outcomes. Although these findings do not always represent unanimity among our Board, they are consensus views.

• Budget scorekeeping and enforcement measures should be rigorously observed in the short term. To provide effective fiscal constraint over long-term policy choices and to further inform the policy process, these measures must be extended and supplemented. Budget scorekeeping and enforcement measures impose important short-term discipline. Scorekeeping and enforcement, however, are not designed to support longer-term fiscal objectives. To understand and manage outcomes beyond the budget "window," policy-makers will require additional analytic perspectives and budget enforcement tools based on fundamental budget concepts and designed to gauge likely economic impacts.

• Budget concepts and economic assumptions must be interpreted and applied consistently to provide "apples to apples" comparisons of complex proposals. The conceptual issues are difficult. The Committee for a Responsible Federal Budget sought advice from other experts and reached our own conclusions with regard to some of the most significant of these issues including: where to draw the line between tax-financed public programs and private activities; what is the right baseline; how Social Security’s off-budget status affects analyses; and how the types of assets held by Social Security Trust Funds affects the budget and the overall economy.

Our conclusions are summarized below and discussed in further detail in the following pages. Some of these issues are not susceptible to bright lines and clear distinctions. The delineation between public and private falls along a spectrum. While some experts will disagree about where specific proposals fall along the continuum, our Committee concludes:

• Language matters. The Social Security vocabulary complicates reform. Terms like "trust fund" and "bankruptcy" mean one thing in ordinary usage and another when applied to public programs. The resulting misunderstanding makes it more difficult to forge consensus. Policy-makers should avoid using such misleading terms. Where that is not possible, they should take care to explain what the terms mean.

• Proposals to invest trust fund assets in marketable securities take advantage of current scorekeeping rules and thus exemplify the importance of broader budget and economic analyses. On the one hand, this approach could forestall other changes that actually could make it more difficult for government to address long-term challenges (e.g., using short-term surpluses for consumption-oriented tax cuts or spending increases). On the other hand, this proposal would not ease the burden on government and the economy when the baby boom generation begins to retire and, by making the Social Security Trust Fund appear to be "more solvent," could erode support for reforms that really would. Policy-makers should understand that there is less to this proposal than first meets the eye.

In addition, to the conclusions summarized above, we offer the following advice:

Social Security is a pay-as-you-go system. From the program’s inception, taxes paid into the system were used to pay current benefits. That is how a pay-as-you-go system works. An accounting device, called a "trust fund", was set up to compare taxes paid to government (and the interest earned on balances of taxes not immediately paid out) with commitments to pay benefits. Currently, Social Security is characterized as being actuarially out of balance. That means income projections under existing tax rates will be insufficient to cover benefits in the future. Any meaningful reform needs to address this imbalance and do so without impeding the government’s ability to satisfy other policy objectives. Current scorekeeping and enforcement measures (which have a near-term focus) simply were not designed to support all of these objectives.

• Scorekeeping measures the impact of policy change on the budget. Scorekeeping estimates extend up to 10 years. Proposals that may look benign over the next 10 years could worsen the government’s long-term financial problems and nonetheless escape scorekeepers’ scrutiny. Proposals that look costly in the short term could improve significantly government’s long-term prospects, but suffer disadvantage due to scorekeeping analyses.

• Enforcement disciplines the annual budget process. Pay-as-you-go, or "paygo," requires any change in law that would reduce revenues or increase direct spending to be offset by revenue increases or direct spending reductions. Paygo expires in 2002 and, in any case, does not apply to Social Security. Separate House rules inhibit consideration of legislation that would reduce Social Security reserves or worsen the program’s actuarial condition over a 75 year period. In the Senate, there must be 60 votes to waive an objection and pass a budget resolution that would reduce the Social Security surplus based on the allocation to the trust fund, in the budget year or over five years, or to pass legislation that would reduce the surplus compared to budget resolution.

• The 1990 Budget Enforcement Act reaffirmed off-budget status for Social Security receipts and expenditures. Social Security surpluses are to be excluded from calculations of surpluses or deficits for the purposes of annual budget legislation. Summary presidential and congressional budget materials regularly overlook or ignore that requirement.

Within 15 years, Social Security benefit payments will exceed cash income. That will put tremendous pressure on all other government revenue and spending choices and on budget aggregates. Some propose to solve these problems by adjusting program benefits and payroll tax revenues. Other proposals would go further by converting Social Security from a largely pay-as-you-go system to one that is partially- or wholly-advance-funded through mandatory deposits to individual accounts. Each proposal would affect the federal budget and, over the long-term, could affect the entire economy. It is difficult to estimate the scope, incidence, and magnitude of such changes.

First, the issues are difficult conceptually. Second, analyses rest on economic assumptions; and someone must make judgment calls to decide what assumptions to use.

Most Social Security reform proposals include new individual retirement savings accounts. Some accounts would be voluntary. They would create incentives to encourage individuals to invest in the new accounts. People would not have to participate. Like Individual Retirement Accounts (IRAs), these voluntary individual accounts (if they are truly voluntary) would be non-budgetary.

The preponderance of proposals for private accounts would be mandatory. Individuals, and employers on behalf of their employees, would be required by law to contribute to the accounts. The conceptual question is: would such accounts be budgetary? Are mandated deposits an exercise of government’s sovereign power to tax? If mandatory contributions are governmental receipts, would deposits to the accounts be counted as outlays simultaneously, or would expenditures be reflected in the budget when beneficiaries withdraw resources from the accounts?

The Committee concludes that all of the mandatory individual or "private" account proposals we have examined are tax-financed federal programs. (See Appendix 2 for summaries of representative proposals.) They would create financial transactions that would not occur absent substantial federal intervention, compulsion, and control. We characterize the financial flows into such accounts as budgetary. We conclude that the accounts themselves could be private. In those instances, receipts and outlays, should be recorded simultaneously as resources flow from individuals and employers into the accounts. This is the case whether or not the resources would flow through Treasury or directly to financial intermediaries. In other cases, the characteristics of the accounts would trigger on-budget treatment for the accounts themselves.

There is substantial debate about whether the cost of mandates is ever budgetary. The government imposes mandates to protect the public and to minimize the social costs of private activities. While there is no bright line separating government mandates from taxes, at some point, a mandate can cease to regulate private activity and become a means of carrying out a governmental program through a backdoor means.

Because taxes are not popular, politicians may want to "have their cake and eat it, too." Keeping accurate track of the size and scope of government is important to political accountability. To determine when a proposal crosses over from a regulation to a government program, experts examine a number of characteristics. Most of these characteristics focus on the degree of government compulsion and control.

In the case of mandatory deposits to individual retirement accounts, the issue is: do these transactions look more like a tax-based system designed to allocate individual’s current income to uses the government deems socially desirable; or are they more like regulatory mandates that protect public health and safety and limit social costs (e.g., catalytic converters and auto insurance)? The former are considered budgetary, while the latter are not.

• Tax equivalence—Mandatory accounts would be different from regulatory mandates. Everyone who works would be compelled to contribute to mandatory individual retirement accounts—just as anyone who works is subject to payroll taxes. The policy objective would be to provide retirement income, just as payment of Social Security taxes provides retirement benefits. Some proposals even would use a common collection mechanism, making deposits to individual accounts and payroll tax payments virtually indistinguishable.

• Financial transactions—Mandatory deposits to individual retirement accounts would be financial transactions. Like payroll taxes, contributions would be fixed in law as a percentage of earnings. By contrast, the costs of converters and insurance are not fixed in law. Government is indifferent to how much individuals pay to comply—it only cares that we have catalytic converters and auto insurance. Policy focuses on ends, not means. In the case of mandatory individual retirement accounts, the means are the ends. Government would care very much about the size of contributions. The policy goal would be to ensure that mandated financial transactions occurred: first, that individuals deposit a fixed percentage of earnings into accounts established pursuant to law for that purpose; and second, that once workers retire, the funds are disbursed over the balance of their lives. The larger goal may be increased national saving—specifically increased individual saving for retirement—but people would be forced to comply with the mandated financial transactions whether or not such aggregate increases actually occur.

Enforcement—If Congress and the President enact mandatory accounts, they would have a keen interest in assuring that individual contributions take place in the prescribed manner. One simple way to do so is to "piggy back" on mechanisms already in place to track and enforce employer payroll tax withholdings. Using the payroll tax system may be a rational reaction to concerns about administrative costs and potential for employer abuse. But funds do not have to flow through Treasury to achieve this monitoring, reporting, and enforcement capability. Whatever the mechanism that would be employed to answer such concerns, it seems certain to be very different from that employed to meet IRA requirements.

Having concluded that mandatory contributions are taxes and should be recorded as governmental receipts when collected from workers and their employers, the question arises: would outlays be recorded when deposited to accounts or when disbursed to individuals? That is, when would the funds leave government? The answer depends on the point at which the resources become private.

• On the one hand, the accounts appear to be privately-owned. The individual who has an account would get the economic benefit of the account—he or she would get back the taxes paid plus the earnings. That being the case, outlays should be recorded simultaneously with receipts. Investment earnings and losses and withdrawals from the accounts would be non-budgetary since they would involved adjustments to private assets.

• On the other hand, government would maintain substantial control over how funds are invested and when they can be accessed by individuals. Due to those controls, characterizing the accounts as private actually is a harder call than identifying the deposits into the accounts as taxes. Indeed, some argue that the nominal individual owners would have so little control over their accounts, it would be farcical to call the accounts "private" and the accounts should be included in the budget.

• Would government investment restrictions be designed to serve fiduciary purposes (to ensure that resources are managed for the best interest of beneficiaries) or to serve larger public policy purposes? For example, government directed investments into targeted companies or industries—or investments withheld from companies deemed to be socially irresponsible—would seem more public than private.

• Does government guarantee a statutorily-defined level of income from the accounts? If government assumes the risk and guarantees a minimum return, then the accounts clearly should be categorized as public, not private, and should be included in the budget. If, on the other hand, individuals bear the risks of poor investments and low rates of return, the accounts would appear to be private.

• Would government employ prohibitions or penalties to enforce account transactions? For example, individuals may withdraw funds early from IRAs or the federal employee Thrift Savings Plan (TSP). Doing so triggers a penalty, but participants in these accounts can decide to access funds early. Both IRAs and TSP accounts are considered private. Most mandatory individual account proposals would prohibit access to the funds until the individual reached retirement age. The more proscriptive the account requirements, the less private they look.

The budgetary treatment of individual accounts is not always an easy call. Various proposals have different characteristics and detailed transactions are significant. Experts must track carefully the resources reformers propose to allocate to Social Security and any successor publicly-sponsored retirement program(s). The following complications arise:

• Simple rules are not easy to apply consistently. For example, the "Gramm/Feldstein" would make the amount of Social Security benefit payments dependent on individuals’ withdrawals from private accounts. The private accounts would not be compulsory; but a dollar for dollar tax credit would create a compelling incentive for individuals to participate in them. The Social Security benefit reduction, effectively, would permit government to recapture three-fourths of amounts accumulated in private accounts. Thus, some experts argue the accounts are a government benefit and should be reflected in the budget. This is a strong case. Ultimately, the determination would depend on how the "claw-back" mechanism operates. In the case of this proposal, we believe the accounts would be private.

• Practicality matters. Administrative complexity, while not determinant, argues against including the accounts in the budget. We observe that administrative complexity alone is not a sufficiently compelling reason to decide this issue; but we also shudder at the complex analysis which would be required accurately to reflect these accounts in the budget. The astronomical number of transactions budget analysts would have to track to measure accurately the aggregate value of all accounts is mind boggling. All other things being equal, we would decide not to incur such administrative complexity. But we do not believe all else is equal.

The extended current law baseline projects unified budget surpluses for the next 10-20 years, followed by rapidly escalating deficits and debt. This baseline creates two major problems.

• Experts believe that a combination of policy and economic changes will reduce short-term surpluses and prevent escalating long-term deficits. The economy simply would not sustain debt rising to the indicated levels (200 percent of GDP by 2050). Some experts question the usefulness of measuring policy changes against baselines that become progressively more unlikely over time. They are concerned that unrealistic baselines either could skew results or foster indifference to unsustainable trends. Yet, for Social Security reform, it is particularly important to assess the impact of proposed changes many decades in the future.

• Politicians have become accustomed to policy changes that improve upon baseline projections. Until recently, baselines projected "deficits as far as the eye can see." Now, it is far more difficult to adopt policies to better current law baseline projections. The current law baseline assumes 100 percent of short-term surpluses reduce outstanding public debt, lower federal interest costs, and add to national savings. Policy alternatives are not likely to achieve the same salutary short-term budget and economic impact. Most would look worse relative to the current baseline.

One solution to these problems is to construct different baselines. For example, an alternative base could assume somewhat smaller short-term surpluses and lower long-term deficits or a permanently "balanced budget" (no short-term surpluses or long-term deficits).

The Committee agrees that alternative baselines could provide additional perspectives, but we oppose substituting such baselines for the current policy base. In no case, should baselines be used selectively to portray individual proposals in the best light.

• The current law baseline represents the best picture of what is required by an extension of current policies and law indefinitely into the future. However, the end result of this extrapolation is an outcome that simply is not feasible. What the current law baseline helps to identify is the magnitude of the adjustment that will be required.

• Alternative baselines could provide additional information about policy impacts, but multiple bases also could confuse. Supplemental analyses should be made available for all major reform proposals. That would allow decision-makers to compare all options under comparable sets of assumptions. Policy-makers should specify which changes to current policies should be assumed in alternative bases.

• Similarly, static analyses require fewer choices about assumptions than dynamic projections. Because one of the major objectives of Social Security reform is to increase savings, investment, and growth, additional policy simulations that estimate such changes could prove very informative. Again, for comparative purposes, we urge that major policy options be simulated using uniform sets of assumptions based on guidance from policy-makers. These analyses should supplement, not replace, static projections.

Social Security is "off-budget". The goal is to insulate Social Security from pressures to reduce spending or raise revenues to reduce the deficit, balance the budget, or pay for other priorities. Taking Social Security "off-budget" does not take it out of government.

"Off budget" status can confuse and mislead. Social Security is a part of government. It is the largest single government program. In fact, Social Security’s financial status depends on government solvency. Current year tax collections are used to pay current benefits. In 2013 and beyond, projected benefits will exceed projected Social Security tax receipts. Taxpayers will have to provide additional resources in the form of higher Social Security or other taxes, public borrowing, or reductions in other programs. Or, there will not be sufficient cash to meet current law benefit commitments.

The Committee for a Responsible Federal Budget recommends a consolidated budget approach to analyze and compare reform proposals, but we support excluding Social Security surpluses from near-term fiscal targets. This will permit "apple to apples" comparisons of competing reform proposals and will promote short-term fiscal discipline by highlighting the extent to which unified surpluses are due to Social Security surpluses.

• A consolidated budget approach is essential to the analysis of Social Security reform. It produces "apples-to-apples" comparisons of various proposals and captures intragovernmental impacts.

The language and vocabulary of the Social Security debate may be the most significant single impediment to successful reform. Terms that mean one thing in private usage mean something entirely different in the context of government. For this reason, political leaders should exercise caution when using misleading terms, avoid them entirely where possible, and explain them where it is not.

It provides a means to associate earmarked receipts with program expenditures. This connection links work to retirement, but is an accounting device. The Social Security Trust Fund, like other federal trust funds, is a creation of federal law. Congress and the President could pass a law to increase or decrease trust fund balances, even to eliminate entirely the "unfunded" liability. This would change the level of earmarking of anticipated receipts and would uncouple payroll taxes from promised benefits, but it would have no real economic impact.

"Trust fund" does not mean that the government is holding individual payroll tax payments until that person withdraws them as benefits. Trust fund balances are not assets to the government as a whole because they are offset by Treasury liabilities. When payroll taxes and other income become insufficient to pay benefits, Social Security will cash in its Treasury certificates, giving it additional spending authority without the need for Congressional or Presidential action. Until then, the trust fund balances have political, not economic significance. Once the balances are gone, projected current law revenues would be sufficient to cover about 75 percent of promised benefits. At that point, government will borrow, raise taxes, or reduce other federal spending, or it will have to reduce Social Security benefits.

Payroll taxes and benefit levels, not trust fund balances, have real impact on the economy. Trust fund balances only become real when government has to provide the cash to redeem Treasury securities held by the trust fund to meet benefit promises.

• "Bankruptcy" is misleading when applied to the Social Security Trust Fund. Social Security cannot go bankrupt any more than national defense can go bankrupt. Unlike private entities, the government only has to authorize more spending authority for benefits. Absent policy change, there eventually will be a gap between resources earmarked to pay Social Security benefits and promised benefit levels. Hypothetically, government would be in a quandary. One law would entitle individuals to benefits. Another would prohibit the government from expending funds in excess of the budgetary resources provided. But political pressure would force Congress and the President to act before government officials got sent to jail or individuals had to sue for full benefits.

• Individual "rates of return" imply that Social Security is a funded pension program. It is not. It is a public social insurance program. In addition to retirement benefits, it redistributes income, provides income to spouses and workers’ survivors, and supports disabled workers. Individual benefit levels are determined by law and earnings history, not by contributions and investment earnings. For that reason, early beneficiaries had phenomenal "rates of return." They received far more in benefits than they paid in payroll taxes. As the system matured, "rates of return" had to fall.

Social Security "rates of return" vary depending on factors such as marital and health status, income levels, and numbers of children. Many of those differences result from conscious policy choices made during the design of a public program. Consequently, "rates of return" have about as much meaning for Social Security as they do for other government programs such as Medicare, Medicaid, and family assistance payments.

The simple truth is: maintaining Social Security’s current structure over the long-term requires additional income or benefit reductions. This fact has prompted some reformers to propose investing Social Security trust fund balances in private equities and debt investments. They are willing to accept more risk in hopes of increasing Social Security’s investment income relative to earnings from Treasury securities. From the narrow perspective of the Social Security program, this approach looks attractive. From overall federal and economic perspectives, it likely would fail to make Social Security any more affordable to future taxpayers.

The proposal would not create new savings or real economic gain, only paper shuffling and portfolio shifts.

• Treasury would have to issue new debt to the public to pay off Treasury securities redeemed by the trust fund. The debt held by the public would increase and government’s net interest expense would rise. The trust fund would invest in private securities. Trust fund returns would be equal to dividend and interest earnings and trading gains or losses.

• Higher Treasury borrowing from the public would exert upward pressure on interest rates. Social Security investments would exert upward pressure on market prices and downward pressure on returns.

• Gains to the trust fund largely would be offset by higher interest expenses to the Treasury and reduced returns to private investors. Government’s net position would not improve. New bond purchasers would be better off; holders of old bonds would be worse off. Stock purchasers would pay higher prices and returns on their investments would go down; stock holders would be better off. Portfolio values would shift, but no new savings or investment would occur.

Some proponents of this proposal understand that it would not produce economic benefits. Their objective is more practical. Under current scorekeeping rules, government investments in private securities would show as outlays equal to the purchase price. Those outlays could eliminate short-term unified budget surpluses. That would prevent the use of surpluses to offset other spending or tax cuts. Proponents argue that given a choice between using surpluses for consumption-oriented spending increases and tax cuts or purchasing equities for the trust fund, purchasing equities is better. Relative to the alternative, it would increase national savings.

Policy-makers may decide that that this approach indeed is the lesser of two evils. But they should not be misled into thinking that this proposal would make Social Security more affordable over the long term. Social Security cannot achieve financial stability at the expense of the overall budget or other sectors of the economy.

Introduction

The increase in life expectancy at age 65 and the imminent retirement of the baby boomers has provoked a vigorous debate about Social Security reform. An even greater budget burden will be imposed by Medicare and the long-term care component of Medicaid. But this paper will focus only on Social Security issues.

Thus far, the debate over Social Security has focused on the substantive merits of various reform proposals. It has not considered how various proposals would be treated by the rules governing the budget process. The debate has also ignored how various proposals would be treated by traditional budget accounting rules.

That is as it should be. The debate should concentrate on the substantive merit of different proposals. That is very different from concentrating on how proposals are scored. Indeed, the following analysis hopes to show that the budget rules in use since 1990 are not appropriate for judging Social Security reform and that traditional budget accounting procedures provide a misleading impression of the effects of certain proposals. Problems arise because some scoring rules are short term in nature when Social Security reform should be judged by its effects over the very long run. In addition, scoring rules and accounting concepts are largely based on cash flows and that often gives an erroneous impression of the effects of certain proposals on public saving. Scoring also generally operates as though the estimates are certain, whereas many policy proposals differ mainly in the degree of risk that they impose on taxpayers. Last, it is very important to judge the effects of different proposals on economic growth. Traditional scoring procedures ignore such effects for good practical reasons. Moreover, growth effects are usually quite unimportant over the short-time horizon used for scoring on-budget transactions.

It is not the intent of this paper to evaluate various proposals. In the interest of full disclosure, it should be noted that the author was a member of the Center for Strategic and International Studies (CSIS) National Commission on Retirement Policy that proposed a particularly appealing plan. But in what follows, remarks biased in favor of this plan or against others will be held to a minimum.

Before discussing the possible impact of scoring rules, I shall digress into a description of the mysteries of the Social Security Old Age and Survivor Insurance (OASI) trust fund. The language generally associated with trust funds can be terribly confusing and misleading. It is clear that the public is not well informed on what the trust funds do and how they do it. The resulting confusion often distorts the debate over Social Security and makes it more complex than it should be.

A Primer on Trust Fund Operations and Accounting Practices

The Social Security trust funds are called trust funds, primarily because the law calls them trust funds. The OASI trust fund does not, as many in the public seem to believe, hold an individual’s payroll tax payments in trust in an individual account until the person retires. There might be less misunderstanding of the trust fund’s role if the word "trust" was not used in its name. Perhaps, it should be called a "financing account" or even a "rainy day fund". But unfortunately, it is difficult to find a simple term that describes its operations, because the nature of those operations has changed over time.

When Social Security was first designed, there was much debate over whether benefits should be pre-funded by investing payroll tax payments in a way that would finance future benefits or whether it should be a pay-as-you-go system in which payments of payroll taxes would be used immediately to fund the benefits of retirees. By 1939, the advocates of pay-as-you-go financing clearly won the battle, as benefits were then raised and eligibility was expanded to roughly absorb the revenues flowing into the funds.

In a pay-as-you-go system, it would be a colossal accident if payroll receipts exactly equaled benefit claims each and every year. It was, therefore, decided to have a fund into which payroll taxes would be paid and the fund was allowed to build up before the first monthly benefit was paid in 1940. The fund would act as a cushion and build up in good years and shrink in bad years. Payroll tax rates and the generosity of benefits would be continually adjusted, so that inflows matched outflows over the very long run. A rule of thumb has often been stated that suggests that the fund would be safe if it contained the equivalent of one year’s benefits, but historically the fund has seldom approximated this goal. Its balance has either been far less or far greater than one year’s benefits.

The amounts in the trust fund are invested largely in non-marketable Treasury securities. The Treasury pays an interest rate on these securities that is determined by a formula that reflects market rates of interest.

The cash proceeds received by the Treasury when it sells securities to the trust fund can be used to finance an excess of spending over receipts in the rest of government or it can be used to retire ordinary debt held by the public. In fiscal 1997, the OASI trust fund had a surplus of about $68 billion consisting of an excess of payroll tax receipts over benefits of about $29 billion and investment income of about $38 billion (see figure 1)7. The Disability trust fund had a surplus of about $13 billion. The rest of government had a deficit of about $104 billion, which includes the cost of interest paid to the trust funds by the Treasury.

The Social Security trust funds are technically off budget. Although their off-budget status has important procedural implications, it has had little effect on aggregate fiscal policy. The Congress has set its deficit goals based on the unified budget, which combines Social Security outlays and receipts with those of the rest of government. With a few minor exceptions, the unified budget focuses on the government’s transactions with the public, that is to say, it measures outlays going to individuals and firms and receipts received from the public in the form of taxes and user charges, and from sales of services and assets. The difference between outlays and receipts is the unified deficit or surplus. This budget balance provides a good indication of how much the government will have to borrow from the public or how much debt will be retired. Payments of interest from the Treasury to the trust funds and the proceeds of securities sold to the trust funds are not counted in determining the balance in the unified budget, because they reflect cash flows from one pocket of the government to another.

The off-budget status of Social Security has also not hidden its size relative to total government outlays or the significance of payroll taxes to total receipts. Social Security outlays and receipts are typically included along with other spending programs and receipts in OMB and CBO budget displays.

In the early 1980s, an excess of benefit payments over payroll tax receipts caused the amounts in the trust funds to shrink to the point that promised benefits could no longer be funded. The Greenspan Commission was appointed to recommend ways of restoring balance to the system. The Commission’s recommendations balanced outlays and receipts for 50 years, but the Congress wished to achieve balance for the traditional time horizon of 75 years. Therefore, the Congress added an increase in the normal retirement age to the Commission recommendations.

As a result, the 75-year fix, enacted in 1983, implied that the Social Security system would enjoy an era of large surpluses while Depression babies were retiring and baby boomers were paying payroll taxes. But soon after, the trust funds would sink into deficit when the baby boomers ultimately retired and labor force growth slowed because of the baby dearth that followed. During the era of prosperity, the trust funds would accumulate huge balances, far in excess of one year’s benefits. Balances would then plummet relative to outlays about a decade after the baby boomers began retiring (see chart 1). The trustees of the system currently estimate the trust funds will be exhausted about 2032. The Congress’ 1983 reforms did not succeed in fixing the system for 75 years, because the economic and demographic assumptions underlying the design of the reforms were slightly too optimistic.

Although the 1983 reforms caused the system to depart from a pay-as-you-go philosophy, it is also far from being fully funded. It is a peculiar hybrid. We are now experiencing an era of prosperity for Social Security as Depression era babies are retiring and the trust funds are accumulating balances at a rapid rate. The surplus in the OASI trust fund has helped to produce a surplus in the unified deficit, although in asking how much the trust funds have helped, it is important not to count the payment of interest from the Treasury to the trust funds. That addition to the trust fund’s surplus is an obligation for the rest of government. To understand the OASI trust fund’s contribution to the current unified surplus, it is better to look at the excess of payroll tax receipts over benefits paid, currently running at about $29 billion. An amount equal to a portion of the income taxes paid on benefits, amounting to about $6 billion in 1997, are transferred to the trust fund.

It is often said that the trust fund will be "bankrupt" or "insolvent" around 2032. Like so many other terms surrounding trust fund accounting, these are extremely misleading. The U.S. government can print money and compel people to pay taxes. Under these circumstances, the possibility that a portion of the government will become "bankrupt" is highly remote. If the trust fund exhausts its assets, the Congress will have to change the law to ensure that promised benefits are paid. It can, at that time, cut benefits, raise payroll taxes, or use general revenues to pay benefits. Even if the Congress was unable to agree on a new law, the Social Security system could continue to pay benefits to the extent that they are financed by incoming payroll taxes. As late as 2050, over 70 percent of benefits would be covered.

When the trust fund begins to run a deficit, it will present some of its holdings of Treasury securities for redemption. At that point, the government will be forced to raise taxes, cut spending, or issue bonds to the public in order to redeem the bonds of the trust fund, so that the trust fund can finance benefit payments. Alternatively, the Treasury could issue bonds to the Federal Reserve in return for new money, but that is a horror to contemplate, since it would be highly inflationary.

The most important point, however, is that the existence of the trust fund does not, by itself, move resources from the present to the future. If the securities in the trust fund are redeemed using tax increases, most of the resources going to benefits will come out of the consumption of those working at that time. If the redemption is financed by issuing debt, domestic investment will be crowded out and future genrations will have a lower standard of living. If other expenditures are cut, the distributional effects will depend on the type of program that is curtailed. If the redemption is financed by creating inflation, it will mainly be the holders of money balances and debt instruments that lose.

It would be possible to run the Social Security system without a trust fund and that might be less confusing to the public. Benefits could be calculated as they are currently and the benefit structure and payroll tax rates could be continually adjusted to maintain a balance over a 75-year time horizon.

It would even be possible, as some have suggested, for the government to buy equities and to take account of the estimated profits in computing the relationship between future benefits and receipts. Computations would have to be repeated from time to time depending on how well the government actually did on the stock market.

It is always tempting to exploit the government’s low borrowing rate and to earn a profit by buying private securities. Alternatively, the government’s low borrowing rate can be used to subsidize direct lending programs. Such arbitrage could be applied in many other areas without using trust funds, or could spread to other trust funds, such as the highway trust fund. Senators Domenici and Gramm have recently proposed earning a profit by investing the surplus in corporate securities rather than using it to retire government debt. Taken to its limit, the argument would seem to justify the government buying the whole private sector, but it will be noted later that there is some doubt that the government as a whole actually profits from such transactions. Moreover, government purchases of equities create other dangers that will not be discussed here. The point of this discussion is that equity purchases can be contemplated whether or not there is a related trust fund.

Scoring Rules

The rules used to discipline budget policies are mostly the result of the budget agreement of 1990. They divide the budget into three categories, each of which is subjected to its own rules. Entitlements, or more accurately, mandated spending and taxes are subjected to "pay-as-you-go" rules (PAYGO). In the House, this means that any benefit increase or tax cut has to be paid for by a benefit cut or tax increase in the first year and on average for five years. The Senate has a longer time horizon. There, the policy change must be paid for in the first year and also in the first five years and in the second five years. Proposals that do not satisfy the PAYGO rule are subject to points of order that require 60 votes to be overridden in the Senate. At the end of the year, if the PAYGO rules are not satisfied, enough spending is sequestered to eliminate the excess spending. The order in which programs are cut is quite complicated and need not be described here. Low income programs are generally exempted from cuts, as are Social Security benefits. Indeed, only a small portion of total mandated expenditures are subjected to a sequester -- less than $30 billion in fiscal 1997.

Discretionary spending is subjected to outlay caps that are divided each year among subcommittees of the Appropriations Committee. Caps are applied both to budget authority, which allows agencies to make spending obligations, and to outlays, which occur when the cash is paid to settle an obligation. The Balanced Budget Act of 1997 establishes caps through 2003. If the caps are violated, discretionary spending is subjected to an across-the-board sequester.

Because the OASI and DI programs, except for their administrative expenses, are off-budget, they are not subjected to caps and PAYGO rules. They have their own set of disciplining rules. In the House, anyone proposing a specific benefit increase or payroll tax cut must propose benefit cuts or future tax increases that leave the long-run actuarial balance of the system unchanged. Violations of the rule are subject to points of order, but unlike the rules governing on-budget programs, they are not ultimately enforced by using a sequester of spending. The rules are somewhat more lenient in the Senate and focus on the short run. Changes in OASI and DI cannot increase the deficit over five years.

The off-budget status of OASDI is very important procedurally. Changes in the programs cannot be part of a reconciliation bill. This means that they must stand alone and cannot be considered as part of a complicated package of budget reforms. It also means that, unlike a reconciliation bill, an OASDI policy change can be filibustered in the Senate.

Scoring Reform Proposals

Buying Equities for the Trust Funds - It has been recommended that the Social Security trust funds be allowed to buy equities. Because equities are assumed to earn a higher rate of return than government debt, the average rate of return on trust fund investments would be assumed to be higher. If this increase in the rate of return materialized, future benefits would have to be cut less or taxes increased less to maintain balance in the system. The trust fund investments would, of course, face considerably more risk. With the exception of the scoring of credit programs, changes in governmental risk bearing are not considered anywhere by current scoring rules.

Purchases of assets have traditionally been considered to be outlays in the unified budget. Consequently, the purchase of equities by trust funds would appear to raise outlays and reduce the budget surplus. This is because the unified budget tracks cash flows and makes no attempt to differentiate capital transactions from those associated with the operating expenses of the government. Usually, the cash basis of the unified budget serves us pretty well, but occasionally, it presents a highly misleading picture of what is going on.

This is an instance in which the unified budget is highly misleading. If a trust fund sells bonds in order to purchase equities, the government’s balance sheet is not affected. An asset has been acquired by increasing liabilities by the same amount. Similarly, nothing has happened to national saving as the result of swapping these pieces of paper. Yet, at a superficial level, it appears as though public saving has declined, because in the first instance the surplus has fallen.

The recorded outlays associated with the proposal would not, however, be subject to the outlay caps imposed by the Balanced Budget Act of 1997. That is because it would be the trust fund making the outlays and the Social Security trust fund is technically off budget.

Recently, there have been some hints that scoring procedures will be changed for this type of transaction. A recent proposal to use the budget surplus to buy commercial paper rather than redeem government bonds was scored as a credit program by CBO. However, the government’s profit from the transaction was not recorded, and therefore, it could not be spent as though it was a negative outlay. A negative outlay would make room under the caps on discretionary spending and allow spending increases elsewhere. If it were recorded as a PAYGO item, the profits could finance a tax cut or entitlement increase.

Some have suggested that purchases of equities be handled the same way, if the government does not use its equity holdings to control the management of a firm. In that case, it is a purely financial transaction.

There are good reasons to consider the purchase of equities by the trust fund a credit transaction. It is also probably appropriate to avoid recording the profit accruing to the trust fund from the arbitrage implied when debt is sold to buy equities. Initially, the transaction simply moves income and risk from the private sector to the public sector. The public has to be paid to give up the present value of whatever benefits they saw in the combination of risk and income that they invested in before the transaction. The public’s reward will take the form of a rise in the rate of return to bonds and a rise in the price of equities. The rise in bond rates will cost the government significant amounts in the long run -- about $300 million per basis point -- as it refinances its debt and this cost will not be scored. It may also lose tax revenue as income is moved from the private sector to the trust fund, depending on the relative tax burdens born by the owners of the equities purchased and the bonds that are sold. Once the costs imposed on the rest of the government are considered as well as the trade-off between income and risk faced by the trust fund, it is not clear that the public sector has really gained anything as a result of the transaction. Calling the transaction a wash may be more accurate than recording a significant profit.

Mandated Individual Saving Accounts - Many proposals mandate that individuals save a certain portion of their income or their payroll taxes in individual retirement accounts. Mandates are extremely difficult to handle in the budget process. Mandates are often proposed solely to keep government activities off budget and their costs hidden. This frustrates budget analysts who often would like to put such activities on budget. CBO concluded that President Clinton’s health plan, which mandated the purchase of health insurance by the private sector, should be on budget, because government controlled essentially all the details of the program. However, I shall argue below that putting mandated deposits on budget results in a number of distortions and inconsistencies as well as many practical problems, and I conclude that they should be kept out of the budget.

The main argument for on-budget treatment is that a mandated deposit into a retirement account is like a tax, because there are many who would rather spend the money on consumption. They will suffer because of the requirement to make a deposit. Moreover, the accumulated funds are likely to be subjected to many regulations. The individual will not be able to withdraw money until a certain age and upon withdrawal, many proposals force some annuitization. Before withdrawal, there are likely to be some restrictions on how the funds can be invested. In other words, the funds are not completely private property. They cannot be disposed of freely by the holder. On the other hand, they are like private property in that they can be bequeathed; they would be shared in divorce proceedings; and there is no doubt who ultimately owns them.

The main reason for keeping the deposits off budget is that there are thousands of other regulations that are off-budget, but that prescribe actions and goals that could be implemented equally well through taxing and spending policies. For example, anti-pollution scrubbers could be purchased and installed by government. Instead, we force businesses to purchase and install them. We do not put this transaction on budget, even though the cost to the business is a much more painful tax than a mandated deposit in an account. It is more painful, because the business gets no direct benefit from the scrubber. The benefits provided are public in nature whereas the ultimate benefits provided by a mandated account are much more private. In my view, that makes the case for putting the scrubber on budget much more persuasive in my view than the case for putting mandated deposits on budget. But I would leave both off budget.

Historically, there has been a clear distinction between budgetary programs that almost always involve cash flowing in and out of the Treasury and regulatory programs. There may be a case for putting all regulatory programs on budget, although that would be extremely difficult practically and would radically change the nature of the budget and its meaning. There may be an even stronger case for having a separate regulatory budget as has often been suggested. That budget would try to estimate the economic costs of different types of regulation. But there is not a strong case for putting some regulatory programs on budget and not others. And even if it were decided to put some regulatory activities on budget, it would be odd to start with a regulatory initiative like mandatory accounts that involves such large elements of private ownership and ultimately conveys benefits that are mainly private rather than public.

Perhaps even more important with regard to the Social Security debate, the inclusion of mandatory programs on budget will result in some very odd comparisons. For example, Senator Gramm has introduced a version of the Feldstein plan where deposits are legally voluntary. According to the consensus described in the foregoing report, this voluntary plan should be off budget, whereas a mandatory plan such as proposed by the NCRP should be on budget. However, the acquisition of the deposit is free in the Gramm plan, because it is offset with a one hundred percent tax credit. Thus, 100 percent of all workers should "volunteer" to make deposits if they have any sense at all. The cash flows and all economic effects will be identical to what occurs in a mandated program and yet it will get radically different budget treatment according to the consensus described above.

Although it is not a persuasive reason to keep mandates off budget, it can be noted that one can mandate deposits in certain accounts, but that is very different than mandating extra saving for retirement. We might like to do the latter, but that mandate is easily avoided. More affluent people can shift funds from other saving to satisfy the mandate and some will find it easy to borrow more. In other words, the mandate may fail in its purpose of forcing people to save more in order to increase their private retirement income. With an ordinary tax, avoidance becomes apparent because the tax raises less revenues than if everyone paid it. With mandated saving, the degree of avoidance will not be apparent. Recording the total inflow into mandated accounts as a tax under the assumption that there is no avoidance would be terribly misleading. On the other hand, estimating an amount of avoidance would be difficult and would result in a very complex and not easily understood budget concept.

Let us assume for the moment that the mandated deposit is recorded as a budget receipt. Is it an on-budget receipt subject to the PAYGO rules or is it an off-budget receipt because it bears some relationship to the trust fund? Since I will argue later that PAYGO rules should not apply during the debate on Social Security reforms, this is not a big issue for me. It will be a big issue for those who disagree. Since the concept of Social Security being off-budget is difficult to rationalize, I see no logical way of resolving the matter.

When does an outlay occur if the mandated deposits are considered to be a receipt? In the case of health insurance, the mandated payment or tax was immediately spent on an insurance policy. Therefore, outlay and receipt effects balanced. There was no effect on the unified deficit.

It is argued in the foregoing report that the same treatment should apply to mandated accounts. That is to say, budget outlays should be recorded at the same time as the receipts associated with the mandated deposit. The argument is that the receipts are going into a private account that is not useable by the government. Why then record the receipts as public receipts in the first place? The only reason for recording the receipts as being budgetary is that the regulations or mandates controlling them give them a public or non-private element. Certainly, the mandate that prohibits the monies from being used until retirement is as important as the mandate to make the deposits in the first place. I do not see how this restriction can be ignored. In my view, the decision to put the deposit on budget implies that the outlay should not be recorded until benefits are paid.

But then the mandate would appear to increase the budget surplus initially. Moreover, if the "receipt" were subject to PAYGO rules, the Congress would have some extra money to spend on tax cuts or entitlement increases. In addition, the earnings on the funds should also be government receipts in this case. These might be difficult to estimate in plans where the accounts are held in private investment funds. In short, this fully on-budget treatment of the deposits creates a mess that is better avoided.

The Financing of Mandated Accounts - From the point of view of budget accounting, it is important whether the mandated account is added on to the existing Social Security tax and benefit structure or whether it is offset by a payroll or income tax cut.

If the mandated payment is treated as a tax receipt, a mandated add-on plan would obviously increase budget receipts. There would be no effect on the unified surplus if the outlays were recorded immediately. The unified surplus would initially rise if the outlays were recorded when benefits were paid.

Under these circumstances, a mandated payment offset by a payroll or income tax cut would not affect total receipts. If the outlay were recorded immediately, the transactions would appear to reduce the budget surplus. If the outlay were recorded when benefits are paid, the transactions would not affect the recorded surplus immediately. Again, there are arguments for and against recording the outlays immediately in a situation in which the deposit is recorded as a tax. The decision is difficult and yet, it will have an important impact on how the proposals appear. One’s evaluation of the policy should not be influenced by such cosmetics.

Voluntary Accounts - Senators Moynihan and Kerrey have proposed payroll tax cuts that can voluntarily be placed in a personal account. Payroll tax increases and slowdowns in benefit growth finance the payroll tax cut in the long run. The budget rules are quite clear for this proposal. The initial payroll tax cut would reduce the unified budget surplus, but would not be subject to short-run PAYGO rules. It seems obvious that the voluntary accounts should not be included on budget, even though they would be subjected to some government restrictions.

Appropriated individual accounts - Representative Kasich has proposed appropriating funds to individual accounts that are administered in a manner similar to the thrift accounts of the civil service. In the March 18, 1998 version of the plan, the total appropriation would equal 80 percent of the unified budget surplus. A similar plan, using only one-half the surplus, has been introduced by Senator Roth.

When the civil service thrift program was set up, it was decided that voluntary payments to the accounts and account earnings should be non-budgetary, even though the accounts offer limited investment choices and are regulated as to when they can be withdrawn. The accounts seem to be sufficiently private to justify this decision.

The Kasich plan involves contributing budget resources to a thrift-like account, but the account is somewhat more severely regulated. When eligible to withdraw funds, the individual must take funds out in equal payments based on expected life or buy an annuity, whereas thrift participants have many more choices in how funds are withdrawn.

Even though the Kasich plan imposes more restrictions on the individual accounts than apply in the thrift plan, I would leave the accounts off budget, just as I would leave mandated accounts off budget. The appropriation to move funds to these off-budget accounts then results in an immediate surplus reducing outlay. If it were decided to keep the accounts on budget, it would make sense to have the outlays occur when funds were withdrawn from the accounts, either in level payments or in the lump sums required to buy annuities from private institutions. Without any change in the rules, the outlays would be considered to be in a PAYGO account and would have to be paid for. This would make it very difficult to enact the proposal.

Dealing With Uncertainty

Traditional scoring rules are mostly applied as though the cost and revenue estimates underlying them are made with complete confidence. Although the CBO has recently applied "probabilistic scoring" to deal with the uncertainty inherent in some proposals, it has been applied in a very limited fashion.

Many of the proposals for Social Security reform involve either the government or individuals investing greater amounts in equities, that is to say, moving toward riskier portfolios. The proposals differ significantly in the degree to which individuals bear the greater risk and the degree to which risk is left with the government as a whole, i.e., born by all present and future tax payers.

If the trust fund buys equities, obviously one hundred percent of the greater risk is absorbed by the government. Various proposals for individual accounts differ in the degree to which a minimum return is guaranteed by government or the accounts are buttressed by an income-tested welfare system.

It is extremely important to consider the impact of different plans on economic growth. As noted previously, the only real resources available to pay for benefits in the future will be those produced at the same time as the benefits are owed. If more resources can be created by enhancing economic growth, future transfers from workers to retirees will be less painful.

Traditional scoring practices do not consider a policy’s effect on economic growth in computing future outlay costs and tax revenues. There is a good practical reason for that. Estimates of GDP in the future play an important role in estimating all future revenues and entitlement outlays. When numerous policy changes have to be considered simultaneously, as in a reconciliation bill, each analyst cannot be allowed to make up their own GDP estimates, because changes in the GDP assumption will affect the program cost estimates of all other analysts. When hundreds of policy initiatives have to be evaluated very quickly, as is usually the case, there would be chaos. Usually, the lack of a growth estimate is not important, because few policies have a significant impact on growth within the short time horizon used for scoring purposes.

It would be practical to look at the growth effects of individual policy initiatives presented separately, but then there is a strong incentive to unbundle growth enhancing initiatives while leaving growth detracting initiatives buried in complex legislation that makes numerous changes in policies. However, these considerations are mainly important for the enforcement of PAYGO rules. If the rules are suspended, there is little harm done in considering growth effects and there is no reason to estimate them precisely. The analysis can be done qualitatively.

Assessing the growth effects of various proposals involves both political and economic judgments. Affects on both government and private behavior must be analyzed. Given that economists disagree on the effects of initiatives like mandating saving, the Congress will have to act like a jury of laymen examining various technical judgments. CBO can summarize the literature on such matters.

Political judgments will also be important in evaluating options. With regard to proposals to have the trust fund buy equities, it is important to ask whether it is practically possible to save the surpluses of the trust fund, or will balancing the unified deficit continue to be the goal of the Congress? With regard to proposals that would cut payroll or income taxes to fund mandated accounts, it is important to judge how the surplus might be used otherwise. In particular, would it be used to finance other growth enhancing policies or would it be used for consumption enhancing activities?

For policies that advocate add-on mandatory accounts, the main economic issue is how much extra saving is created. It was noted previously that they are easily evaded. However, they cannot be easily evaded by people who have few other assets or who find it expensive or impossible to borrow. Some extra saving will come from such individuals. Some extra saving may also be generated from people who could avoid the mandate, but appreciate the discipline that it provides. There is some evidence that people do not approach saving rationally and go to great lengths to create disciplining mechanisms that force saving.

Voluntary accounts could also significantly enhance saving, although presumably by less than mandated accounts. The Moynihan-Kerrey approach would provide a strong incentive, because employers would be expected to match employee contributions up to one percentage point of the payroll tax.

For policies that reduce the growth of future benefits, the main question is whether individuals will increase their private saving to replace their lost benefits. Rational individuals should do so, but it was just noted that people are not entirely rational in their saving behavior. A cut in benefits combined with a mandate may provide both the incentive and the discipline to save more.

The results of the analysis regarding scoring and accounting are summarized in the following matrix. Current accounting and budget process rules would create major biases against specific program options that are not warranted by their substantive merits.

Using traditional rules, the purchase of equities by the OASI trust fund would appear to reduce the budget surplus and reduce public saving, even though the plan does not affect the net wealth of government. Absent a change in the rules, plans that offset contributions to mandated accounts with income tax credits against payroll tax payments would be subject to PAYGO rules and become difficult to pass, while plans that achieve almost identical goals by directly cutting payroll taxes would not face this hurdle. Plans that would appropriate funds into individual accounts would be handicapped by outlay caps that were not designed for a day when the main problem involves deciding what to do with a surplus.

It is also vital that the debate over policy options considers their differing impacts on economic growth and on revenues and outlays in the very long run. There is no place in current budget rules for explicitly taking account of such effects when making cost and revenue estimates. Similarly, different proposals affect the value of government assets and contingent liabilities very differently. The practical difficulty of estimating these effects and the degree of confidence that can be placed on the estimates will also vary from proposal to proposal. Such problems are not considered in applying traditional scoring rules.

For such reasons, I believe that the debate over Social Security options should proceed unencumbered by the usual budget rules. Some may fear that the Congress will proceed irresponsibly to raise Social Security benefits and/or cut taxes. There may be no harm in retaining the House rule that Social Security policy proposals should not increase the long-run actuarial deficit of the system and ideally it would be expanded to cover the Senate. However, short-run PAYGO rules and outlay caps should definitely be suspended. This creates a danger that while the rules are suspended, the Congress may attempt to sneak in other program changes that are far removed from Social Security. This will have to be prevented, much as the "Byrd Rule" prevents issues unrelated to the budget from being introduced into reconciliation bills. Depending on the nature of the Social Security system that emerges from the reform effort, the Budget Enforcement Act may have to be amended to impose future discipline.

Many reform proposals involve some form of mandatory individual savings account. Mandates provide a means of capturing resources for some public purpose without recording the costs in the budget. Moreover, the mandated payment will be a considerable burden for some people that looks and feels like a tax. But it is also a tax that is easily avoided by others, and if every dollar going into an account is recorded as though it is a tax, the burden will be greatly overstated. It is also not clear how the outlay counterpart of the "tax" should be recorded – when the money is deposited in the account or when the benefit is paid. To me, it is more logical to conclude the latter if the deposit is considered to be a tax, but that results in an immediate increase in the measured surplus that is illusory. The "tax" revenues carry with them an obligation to pay the money out eventually. My conclusion is that the mandated accounts should not be included in budget totals mainly because most of the dollars flowing into the accounts will substitute for other saving. This will not be true if the mandate is combined with a Social Security benefit cut and people choose to replace lost benefits with personal saving. However, it is then the benefit cut that causes the saving and not the mandate.

The term "privatization" describes several, very different approaches, in the context of Social Security reform.

• Investing the trust fund in private securities. "Privatization" can describe investment of a portion of trust fund reserves in private securities rather than U.S. Treasury obligations. This approach is intended to increase trust fund income by capturing higher rates of return typically associated with private investments. The federal government would retain ownership and control over investments and trust fund assets, including private investments. The higher rates of return would compensate the government for the increased risk of private investments.

• Individually-owned accounts. In other proposals, "privatization" refers to the ownership of retirement program assets. This approach would direct retirement resources largely into privately-issued securities (although investment in government bond funds could be allowed), and allocate the investments to individual accounts. Individuals would own the accounts. Proposals differ as to the degree of control owners would have over contributions to, investments of, and withdrawals from such accounts.

Most proposals would create individual accounts without raising taxes—account deposit would be "carved out" of the existing 12.4 percent Social Security payroll tax. In other proposals, account deposits would be "add-ons" to existing payroll tax rates. New resources would come from payroll tax rate increases or tax incentives would generate contributions to new individual accounts.

To determine the budgetary treatment of privatization approaches, we extend indefinitely the baseline assuming current tax and spending policies. Under CBO and GAO long run simulations, current policies produce consolidated budget surpluses until sometime between 2010 and 2020. Thereafter, deficits and debt grow rapidly as a percentage of gross domestic product (GDP). There are many possible benchmarks to choose from, but we choose the one most familiar to most policy-makers.

Key Assumptions

Proposals add incrementally to, or subtract from, Federal receipts, outlays, consolidated surpluses/deficits and public borrowing.

• Unless the proposal specifically states something to the contrary, we assume that it adds to, or subtracts from, consolidated budget aggregates. For example, if a proposal would increase payroll taxes, we assume total receipts go up by the same amount. If program expenditures go up, we assume total outlays reflect the full increase.

• We do not assume offsetting tax or program cuts elsewhere in the budget unless those changes are spelled out in the proposal.

• Individual behavior is notoriously hard to predict. Economists assume individuals act rationally. The evidence, however, indicates that individuals are myopic, particularly where retirement savings are concerned. Thus, it is difficult to conclude that individual savings will increase dollar for dollar as a result of mandates. For all the reasons Penner articulates so well, it is difficult to estimate the magnitude of the change that would occur.