ACCOUNTING FOR SOCIAL SECURITY AND ITS REFORM

The financial problems of Social Security are frequently debated on editorial pages and in policy circles.[1] For almost a decade, politicians and analysts have been warning of a looming crisis, and numerous commissions and pamphleteers have advanced a host of solutions to the system’s problems, from reductions in benefits to increased taxes to a range of privatization proposals.[2] Though much ink has been spilled and many speeches given, Social

Security has undergone no fundamental changes in more than twenty years.[3] As of this writing, there is no prospect for serious reform until at least the completion of the next presidential election and, most likely, not even then.

For a variety of reasons, politicians have difficulty addressing problems like Social Security reform. The stakes are high, both financially and politically. The primary beneficiaries of the program are the elderly, who are both politically active and well-represented.[4] The most obvious solutions—raising taxes or cutting benefits—are politically unpalatable, perhaps even toxic.[5] While stock market investments appeared to offer a painless fix for at least part of the system’s difficulties a few years ago, subsequent volatility in stock prices has deflated public support for reforms of this sort.[6] For politicians, Social Security presents a problem with substantial political risk and no easy answers.

But the current accounting system for Social Security also contributes to the current political gridlock. Almost all public discussion of Social Security finances are based on annual reports that the Social Security Board of Trustees publishes each spring.[7] The trustees’ reports are, in essence, statements of annual cash flows, comparing the system’s annual cash receipts to its yearly payments for benefits and administrative expenses. This format highlights the system’s current annual “surpluses” and its mounting reserves. These favorable financial reports undermine efforts to address Social Security reform proposals by locating the Social Security crisis far in the future. If, however, the financial statements of the Social Security system were prepared in accordance with the principle of accrual accounting—under which commitments to make future benefit payments are recognized when those obligations are incurred—the public would understand that promises being made today to future retirees are the cause of Social Security’s financial problems.

Consider, for example, the reported financial performance of the Social Security trust funds in 2002. According to the trustees’ report, the combined Social Security trust funds ran a cash-flow surplus of $165.4 billion in 2002

and accumulated record levels of reserves, equal to $1.4 trillion, by the end of the year.[8] These financial facts, which are widely reported in the popular press and well known to political leaders,[9] imply that the financial position of the Social Security trust funds at the end of 2002 was quite good.[10] While the trustees’ longer-term projections suggest that Social Security will face serious problems several decades in the future, the short-term outlook is so favorable that there is little incentive for political leaders to expose themselves to the potentially painful choices necessary to address the system’s problems by altering benefits or raising revenues.

If, however, the trust fund accounts were presented under a system of accrual accounting, public perceptions of the problems facing Social Security would change dramatically. Restated under accrual accounting, the Social Security trust funds would have had to report a loss on the order of several hundred billion dollars for 2002 and comparable or even larger annual losses for much of the last decade.[11] A Social Security trust funds balance sheet prepared in accordance with principles of accrual accounting would show accrued liabilities of approximately $14 trillion at the end of 2002—that is, more than ten times the system’s current reserves of $1.4 trillion.[12] In other words, as of December 31, 2002, the Social Security trust funds had unfunded accrued liabilities of $12.6 trillion, or 122% of the country’s gross domestic product (GDP).[13] Even if one factors in the present value of excess taxes that current participants will pay into the Social Security trust funds over the balance of their working lives, the system’s current unfunded obligations are on the order of $10.5 trillion.[14]

Moving Social Security to an accrual-based accounting system would greatly enhance the quality of debate over Social Security reform proposals. In a variety of ways, cash-flow accounting biases public debate over Social Security reform proposals, encouraging politicians to obscure the extent of Social Security’s mounting liabilities and favoring certain kinds of reform, particularly those that increase short-term cash-flow surpluses while simultaneously increasing the system’s long-term fiscal obligations.[15] Cash-flow accounting also makes certain kinds of reform proposals—notably some individual-account proposals and any reforms that include investments in private capital markets—look prohibitively expensive.[16] An accrual-accounting sys-

tem would clarify both the current status of Social Security’s finances and the relative merits of competing reforms.

Another benefit of adopting accrual accounting for Social Security is the effect that such a change would have on the country’s overall fiscal policy. Currently, the federal government regularly includes annual trust-fund surpluses in the budgetary aggregates reported to the general public. Even though the trust funds are supposed to be “off-budget,” current practice allows the federal government to use annual trust fund surpluses to support other spending programs and disguise the growth in financial obligations that future generations will have to shoulder.[17] Were Social Security to restate its accounts in accordance with the principles of accrual accounting, politicians would be less likely to use trust funds’ reserves in this way and the public would gain a clearer picture of the country’s mounting financial commitments.

This Article is divided into four Parts. Part I presents a critique of the current system of accounting for Social Security and advances the claim that accrual accounting offers a valuable alternative framework for evaluating the annual performance and overall solvency of the Social Security system. The argument focuses first on the traditional accounting treatment of the Social Security trust funds and then considers the relationship between those trust funds and the overall federal budget. In both contexts, current practices are inherently misleading and distort the debate over Social Security reform proposals in important ways.

Part II presents an alternative framework for Social Security accounting based on principles of accrual accounting. After reviewing the different kinds of assets and liabilities that might be recognized on a set of financial statements for Social Security, this Part sketches out a system of balance sheets and income statements for the trust funds in 2002 using an accrual-based system of accounting. It then compares this system of accrual accounting with alternative measures of long-term solvency that experts sometimes use to assess Social Security solvency, and explores the relative merits of various approaches. Based on this analysis, it proposes a modified system of accrual accounting of Social Security that incorporates the strengths of both Generally Accepted Accounting Principles (GAAP)-style accrual accounting and alternative measures of long-term solvency. Part II concludes with a brief discussion of how such a modified system of accrual accounting

BOX ONE

THE RISING CONCERN OVER THE IMPLICIT DEBT OF THE FEDERAL GOVERNMENT

Over the past year, commentators and government officials have begun to take a greater interest in the implicit obligations of the federal government. For example, in a recent study, Gokhale and Smetters have estimated the government’s fiscal imbalance to be $44.5 trillion at year-end 2002, of which they attribute a $7.0 trillion long-term fiscal imbalance to Social Security. See Jagadeesh Gokhale & Kent Smetters, Fiscal and Generational Imbalances: New Budget Measures For New Budget Priorities (AEI Pamphlet No. 4, Apr. 2003); see also Liqun Liu et al., Meaningful Measures of Fiscal Deficit and Debt: The Case for Incorporating Entitlement Debts (Texas A&M Private Enterprise Research Paper No. 2010, May 2002). The federal government itself has begun to compile and report similar measures of long-run shortfalls, see 2002 Fin. Rep. of the United States 62–65 (reporting $4.6 trillion unfunded obligations of Social Security over the next seventy-five years); Analytical Perspectives: Budget of the United States Government: Fiscal Year 2004, at 45–49 (similar), and even the most recent Social Security trustees’ report has supplemented its reports to include additional measures of the system’s long-run unfunded obligations. See infra text accompanying notes 166–185. These recent developments reflect concerns for fiscal imbalance similar to the ones that motivated this Article. These broader analyses also indicate that the problems of Social Security are not unique. Other federal programs, most notably Medicare and other retiree health programs, are also seriously underfunded over the long run. Accordingly, aspects of the accrual accounting proposals for Social Security advanced in this Article might profitably be extended to other contexts.

This Article differs from these other writings in its focus on the significance of accrual accounting for a single, extremely important federal program. Unlike other work on fiscal imbalance, the emphasis here is arguing why a system of accrual accounting would provide a more accurate overview of the financial obligations of Social Security than do its current statements based largely on the presentation of annual cash flows. Also in contrast to other work on the subject, this Article analyzes the manner in which the current Social Security trustees’ reports and most other official writing on the subject distort public understanding of the Social Security system’s financial condition and needlessly complicate the debate over Social Security reform. While other studies offer a birds-eye view of the federal government’s overall fiscal imbalance, this Article provides a ground-level assessment of how a new system of financial reporting could improve our understanding of one major social insurance program and the prospects for its sensible reform.

for Social Security might be integrated with other aspects of the federal budget to produce a more accurate and useful presentation of the government’s overall financial position.

Part III explains how a modified system of accrual accounting could improve the ongoing debate over Social Security reform. Most obviously, a system of accrual accounting for Social Security—with substantial annual losses and growing unfunded obligations—would increase pressure on the political leadership to address the problems of Social Security in a timely manner. In addition, a system of accrual accounting would clarify the nature of the choices being presented to the American public and improve the incentives for political leaders to adopt responsible reform proposals that promote the long-run stability of Social Security.

Part IV concludes with a postscript on normative baselines for this proposal and argues that its approach is more consistent with these baselines than is the current system of cash-flow accounting. As explained below, the essential problem of accounting for Social Security is that the government first makes commitments to Social Security participants while they are working and contributing payroll taxes and then, many years later, must honor these commitments by making payments after the workers retire. In other governmental contexts where there is a substantial temporal disconnect between commitment and payment, government accounts are increasingly expressed on an accrual basis. What this Article proposes is simply an extension of this trend to our most important social-insurance program.[18]

Before turning to the substance of the argument, a few additional introductory points are in order. While issues of governmental accounting will undoubtedly strike many as mundane and technocratic, this Article presents a radical, in some circles even heretical, proposition for a social-insurance program such as Social Security. If adopted, this proposal could have a profound effect on the way the country thinks about Social Security and perhaps other social-insurance programs for the elderly, such as Medicare.

This Article does not, however, advance two more far-reaching propositions. First, it does not argue that the accrual-based statements should be the only financial statements prepared for the Social Security trust funds. Cash-flow statements and long-term projections of liquidity also provide valuable insight into the financial position of the Social Security system and should continue to be prepared. Accrual accounting should, however, be the dominant lens through which the Social Security system presents its financial posture to the world. The burden is on critics of this proposal to explain why accrual accounting should be the dominant format of financial presentation

in so many other areas of financial and economic analysis, but not for the Social Security trust funds.

The second proposition from which this Article disassociates itself is the notion that the movement toward an accrual accounting system necessarily implies adoption of full, advance funding for Social Security: that each generation should pre-fund the full value of that generation’s retirement benefits. It is perfectly possible to imagine a Social Security system with partial pre-funding and financial statements prepared on an accrual basis. In a growing economy with an expanding population, it may make perfect sense for future generations to pay for a portion of the cost of the retirement benefits of current generations, particularly if current generations have borne the cost of retirement for generations of past workers. It is, however, critical to keep control over the magnitude of liabilities that are being passed on to future generations. One of the virtues of accrual accounting is that it offers a perfect tool for monitoring the size of these liabilities and therefore provides a superior framework for ensuring the degree of pre-funding necessary to keep the growth of unfunded Social Security obligations in balance with other economic and social goals.

I. A Critique of the Current Accounting System for Social Security

Social Security historically has employed a cash-flow method of accounting. In part, this presentation reflects the traditional operating philosophy of the program. For many years, Congress and the Social Security Administration ran the program on a pay-as-you-go basis. The lion’s share of annual revenues—raised primarily in the form of payroll taxes paid by current employees—were immediately transferred to current benefi-ciaries.[19] As long as inflow was adequate to meet outflow, the system was considered to be in balance. Indeed, for many decades, if the inflow exceeded the promised benefits, Congress interpreted the cash-flow surpluses as a sign that benefits should be raised so as to make use of the system’s excess cash reserves.[20]

While generally consistent with traditional federal government accounting practices,[21] cash-flow accounting leaves much to be desired in the context of public retirement programs such as Social Security. This Part briefly summarizes current accounting practices for the Social Security program and then explains how these practices distort public understanding of

the program. It begins with the Social Security trust funds themselves and then turns to the relationship between the trust funds and the unified federal budget.

A. Analysis of Accounting for the Trust Funds

Since the Social Security program’s early days in the 1930s, Social Security revenues have been placed in “trust funds.”[22] These trust funds have little in common with traditional trusts. Assets are not legally segregated for particular classes of beneficiaries, nor do Social Security trustees have any legal obligation to protect the interests of program participants.[23] Rather, the trust funds are accounting entries in the federal budget to which Social Security contributions and certain other accruals and payments are attributed and from which Social Security benefits and certain other expenses are paid.[24] The trust funds are, however, the principal vehicles through which politicians and the public analyze the program’s financial status and therefore provide a useful starting point for analysis of the program’s current accounting treatment.

1. Standard Presentation

Each year, the trustees of the Social Security system produce an annual report of the trust funds’ financial condition. A cursory review of the trustees’ most recent report—for the calendar year ending December 31, 2002—reveals two basic financial messages. In the short term, Social Security is doing quite well, but over the longer run, this program is on a course for financial ruin.[25]

a. Short-Range Projections

According to the 2003 Trustees Report, the short-range prospects of Social Security are good because the combined Old Age, Survivors, and Disability Insurance (OASDI)[26] trust funds have more than a trillion dollars of reserves deposited with the U.S. Treasury and are projected to take in more in income than they put out in expenditures for at least the next decade. Table One provides the kind of typically reassuring presentation that one finds in the opening pages of the most recent Trustees Report.[27] It shows a rosy picture of current operations. The combined OASDI trust funds started the year with some $1,212.5 billion in assets deposited with the U.S. Treasury. Over the course of the year, the funds took in $532.5 billion in payroll taxes,[28] another $13.8 billion in taxes on Social Security benefits,[29] plus $80.4 billion

from interest on trust fund assets deposited with the Treasury,[30] for a combined income of $627.1 billion for the year. Charged against this inflow, as seen in Table One, were some $461.7 billion in expenditures, consisting almost entirely of benefit payments. Overall, there was a “net increase in assets” of $165.4 billion during the year, which increased the funds’ total assets to $1,378 billion at year end.[31]

Another way that the trustees commonly describe the favorable short-term outlook for the Social Security system is by reference to a “trust fund ratio.”[32] This statistic is the ratio of the trust funds’ total assets at the beginning of the year to the projected total expenditures over the course of the year.[33] In assessing the trust fund ratio, the trustees have set 100% as a minimum acceptable ratio on the theory that, as long as the trust funds had at least a year’s worth of expenditures on hand, Congress would have time to respond to any unexpected short-term crisis in the system’s financing; the trust funds could receive no additional revenues for a year and still be able to pay benefits at current levels until the end of the year.[34]

Measured in terms of trust fund ratios, the 2003 Trustees Report also shows a carefree short-term picture, as presented in Figure One. Based on the trustees’ intermediate forecasts,[35] this chart shows the trust fund ratio increasing from 263% at the end of 2002 to a high of 452% a decade later—that is, cash reserves equal to nearly four and a half years of projected expenditures in 2012:[36] In the trustees’ words, “[B]ecause the trust fund ratio for the combined funds is estimated to remain above 100% under the intermediate assumptions, the combined funds meet the short-range test of financial adequacy.”[37] Even under the pessimistic, high-cost estimates, the trust funds are projected to remain above the 100% threshold throughout the coming decade.[38]

b. Long-Range Projections

When discussing the trust funds’ long-range prospects, the trustees’ reports become decidedly more pessimistic. Notwithstanding a much-publicized bipartisan effort in the early 1980s to produce a permanent solution to the problem of Social Security,[39] the trustees have for some time now been warning of serious dangers confronting the system in the next decade.[40] The 2003 Trustees Report sounded the same alert:

Based on these projected long-range shortfalls, the trustees regularly urge Congress and the President to revisit the issue of Social Security and make appropriate adjustments to forestall the looming crisis.[42]

The 2003 Trustees Report includes several different accounting presentations to support the report’s claims of long-range insolvency. One presentation, summarized in Figure Two, compares long-range estimated income rates to cost rates over the next seventy-five years, a period chosen because it includes the life expectancy of almost all current workers and retirees.[43] The rates represent the system’s projected annual cash inflows from taxes and outflows for expenses and are expressed as a percentage of taxable wages. The income rates rise fairly modestly over the period, as payroll taxes are not scheduled to increase beyond their current 12.4% of taxable payroll, and income generated by the taxation of Social Security benefits has only a modest effect on overall inflow.[44] The system’s cost rates, in contrast, rise dramatically in the second and third decades of the century, surpassing income rates (under intermediate assumptions) by 2018 and generating an annual combined trust fund deficit of 6.71% of payroll by 2077.

The substantial imbalance between projected income and benefit rates lies at the heart of the long-range crisis facing the Social Security system. It is, however, a somewhat complicated phenomenon to comprehend because it combines the short-range surplus that will accumulate between now and some time in the next decade with longer-term deficits. Admittedly, a quick review of charts such as Figure Two suggests that the long-term deficit is, crudely speaking, bigger than the short-range surplus. Such eyeball assessments are unreliable, however, particularly when dealing with a period extending seventy-five years into the future. Moreover, a simple comparison of income and benefit rates does not take into account the current reserves of the combined trust funds (that is, the more than $1,378 billion in assets deposited with the Treasury as of December 31, 2002) or the interest that can be expected to accrue on these reserves.[45] Recognizing that simply comparing cost and income rates yields an incomplete picture, the trustees’ reports offer two different ways of conceptualizing the long-range imbalance of the combined OASDI trust funds.

The first, which is illustrated in Tables Two and Three, presents what are known as summarized income and cost rates. In essence, these summarized figures attempt to normalize projected costs and benefits over three time

periods (the following twenty-five, fifty, and seventy-five years), generating what are, in effect, average income and cost rates for the three periods.[46] The summarized presentations offer a more complete picture of income and expenses than do the simple comparisons of annual income and benefit rates discussed above. As can be seen from Table Two, these presentations include information not just about the payroll tax and the taxation of benefits but also about the annualized value of the current trust fund reserves, spread over the relevant period, plus the cost of building up a final trust fund balance equal to 100% of benefits projected for the year following the end of the period.

The value of summarized revenue and cost rates is that they allow analysts to compare the relative magnitudes of various components of the trust funds’ long-term balance. These tables show, among other things, the relatively minor long-run significance of the trust funds’ current reserves of $1,378 billion.[47] An analysis of the summarized components also confirms

the accuracy of an eyeball assessment of Figure Two—the long-run deficits of the out years substantially outweigh the funds’ near-term surpluses in all but the trustees’ low-cost estimates.

The statistic that encapsulates all of this information can be found in Table Three, showing summarized and combined income and cost rates. The table restates the total income and cost rates shown in Table Two and then indicates the difference between those two rates for each period of analysis. The figure that the arrow points to—minus 1.92—is the one that policy analysts often employ as a shorthand measure of Social Security’s long-term problems. This measure, sometimes referred to as the “actuarial deficit,” reflects the difference between seventy-five-year revenue and cost rates under the trustees’ intermediate assumptions. The actuarial deficit is often described as the amount that the payroll tax rate would have to be increased (starting today and continuing for the full seventy-five-year period of analysis) to bring the Social Security trust funds into long-range balance.[48] Typically, reform measures—ranging from reductions in benefits to increases in

the taxation of Social Security benefits to gains from improving the yield on trust fund assets—are measured in terms of their effect in reducing or eliminating the current actuarial deficit.[49]

According to standards of long-range “close” actuarial balance that the trustees have established, a 1.92% actuarial deficit in the combined rate for the seventy-five-year horizon is unacceptable. According to the trustees’ standards, the combined ratio for this period should be no greater than 5% of the summarized cost rate for the same period, that is, 5% of 15.70%, or 0.79%.[50] The intuition underlying this standard is that over the long run the trust funds should be considered in actuarial balance only if projected revenues meet 95% of projected costs.

Another way of presenting the trust funds’ long-range solvency is to use charts showing the projected trust fund balances in future years.[51] In essence, these charts reflect a continuation of those in Figure One, which was limited to ten years of trust fund ratios. Figure Three is an example of this sort of long-range presentation. This figure reveals that (under intermediate estimates) the trust fund ratios will begin to decline in approximately 2016, will pass the trustees’ minimum prudent reserve level (100% of projected annual benefits) in 2039, and will be entirely depleted of resources in 2042.[52] At that point, if this scenario were to come to pass and no intervening legislation had been enacted, the combined trust funds would have insufficient resources to honor the projected cost of all promised benefits. In fact, the funds’ projected income for that year (roughly 13.26% of taxable payroll) could cover less than three-quarters of projected benefits (roughly 17.80% of taxable payroll).[53] By the end of the seventy-five years, the trust funds’ projected coverage would be about two-thirds of projected benefits (that is, 13.42% of taxable payroll in income versus 19.92% of taxable payroll in benefits).[54]

2. Criticisms of the Standard Presentation

Although the information embodied in the trustees’ reports is well presented and of considerable value, there are serious limitations in their standard presentations. Put simply, the optimistic short-range projections, and the relatively pessimistic long-range projections, are both too optimistic. Moreover, in many respects, these presentations distort public debate over competing reform proposals.

a. Short-Range Projections

As explained above, the take-home message of the trustees’ short-range projections is that the next ten years appear favorable for the Social Security trust funds. To be sure, there is some truth to this message, at least in the sense that the trust fund reserves on deposit in the Treasury will grow during the coming decade. But this assessment is based exclusively on an analysis of cash flow. It takes no account of changes in the system’s liabilities over the course of each year—either liabilities that were liquidated (benefit promises fulfilled) or the new liabilities generated (new benefits promised).

Indeed, the trustees’ reports have a somewhat curious attitude with respect to the system’s liabilities. Reading through reports, one is constantly warned that the primary problem facing the Social Security system is the retirement of the baby-boomer generation starting a decade or two in the future.[55] At this point, the reports advise, the system will owe this generation a substantial level of benefits, which it will be unable to pay even after drawing down substantial reserves that will have accumulated over the intervening years.[56] Thus, the short-range surplus will be overwhelmed by liabilities arising fully formed several decades in the future.

The dichotomy between short-range feast and long-range famine is inaccurate and misleading. Social Security faces a long-term crisis because it has already made substantial pension promises to current workers and retirees, and each year promises to make additional pension payments to most current workers. When the 2003 Trustees Report suggests that 2002 was a good year for the Social Security system and emphasizes a $165.4 billion increase in the system’s assets in that year,[57] the data presented give no sense of how the growing cash surpluses compare to other aspects of the system’s operations during the year. To be sure, an astute reader might infer that all was not well with the system from the report’s ample evidence that current surpluses will be insufficient to meet long-run benefit commitments. But it is all but impossible to determine from the 2003 Trustees Report whether these long-run problems are the consequence of promises made before 2002 (that is, promises made to current retirees and older workers) or promises to be made to baby boomers and others at some point in the future. In short, the 2003 Trustees Report gives little sense of whether, all things considered, 2002 was a good year for the Social Security trust funds.

In referring to liabilities of the Social Security system, it is important to recognize that the character of Social Security’s pension promises is complex and contestable. As a purely legal matter, Social Security benefits do not constitute binding obligations on the part of the federal government.[58] Because Social Security beneficiaries have no constitutionally protected right to receive any particular level of Social Security benefit, it remains problematic to define how individual benefits accrue over time. While some might argue that the absence of enforceable legal rights on the part of Social Security beneficiaries wholly justifies the formats adopted in the trustees’ reports,[59] there are several strong grounds for rejecting this approach.

BOX TWO

THE LEGAL STATUS OF SOCIAL SECURITY BENEFITS

The Supreme Court has addressed the question of whether Congress has the authority to reduce or eliminate Social Security benefits for individuals who have paid Social Security taxes under a statutory regimen that provides for a certain level of benefits. In Fleming v. Nestor, 363 U.S. 603, (1960), the Court considered whether Social Security participants have a property interest in Social Security benefits that, under the Fifth Amendment of the U.S. Constitution, cannot be diminished without just compensation. The Court ruled that “a person covered by the [Social Security] Act has no[] . . . right in benefit payments as would make every [statutory] defeasance of ‘accrued’ interests violative of the Due Process Clause.” Id. at 611.

The Fleming case involved a Bulgarian immigrant named Ephram Nestor, who had arrived in the United States in 1913 and paid Social Security taxes from 1936 until his retirement in 1955. Id. at 605. In 1954, Congress had amended the Social Security Act to deny retirement benefits to any beneficiary deported from the United States on the grounds of being or having been a member of the Communist party, and Nestor was deported on that basis in 1956. Id. Nestor, who would otherwise have been eligible for benefits given his years of payment of Social Security taxes, had his benefits revoked, and he challenged the revocation as violating the Fifth Amendment. Id. at 606. Writing for a majority, Justice Harlan rejected Nestor’s claims, emphasizing that giving Congress flexibility to recalibrate Social Security benefits from time to time is a practical necessity and that the Social Security Act expressly reserves for Congress the right to “alter, amend or repeal any provision.” Id. at 611 (quoting 42 U.S.C. § 1304). Though occasionally criticized and distinguished in subsequent decisions, the core holding of Fleming retains vitality: Congress can reduce or eliminate accrued Social Security benefits without having to provide just compensation to participants or other beneficiaries whose entitlements are thereby diminished. Section 1304 of the Social Security Act still explicitly preserves this authority by providing that “the right to alter, amend or repeal any provision of this Act is hereby reserved to the Congress.” 42 U.S.C. § 1304 (West. 2000).

The only significant constitutional protection afforded Social Security participants is that individualized decisions to reduce or eliminate benefits must be conducted in accordance with statutory standards and effected through procedures meeting the requirements of the Due Process Clause of the Fifth Amendment. For a review of the doctrine in this area, see Julie A. Nice & Louise G. Trubek, Cases and Materials on Poverty Law: Theory and Practice 284–374 (1997).

For one, an accounting treatment that reflects the accrual of benefits is more consistent with the statutory framework for the payment of Social Security benefits. As defined under the Social Security Act, an individual’s Social Security benefits are based on the number of years the individual participates in the system. Individuals become eligible to receive retirement benefits once they have made contributions for ten years, and their statutory entitlements grow larger as they make contributions to the system.[60] The amount of benefits ultimately paid out of the trust funds is a function of a participant’s lifetime earnings, based on the highest-wage years over the course of his or her working career.[61] When economists want to calculate the value of Social Security benefits, they routinely look to these statutory formulas to estimate how large a benefit a particular individual has accrued at a particular time or at the individual’s retirement.[62] The trust funds’ financial statements, however, make no effort to quantify this gradual accretion of statutory entitlements for individual beneficiaries.

The trustees’ reports also fail to reflect the political reality of the trust funds’ benefit structure. According to Social Security lore, President Roosevelt wanted Americans to think of themselves as having contributed to individual Social Security accounts—that is, accounts held in trust and funded through individual contributions—so as to create a sense of personal connection to the Social Security program and thereby increase public support for the system.[63] If this was, in fact, President Roosevelt’s intention, history has seen his vision realized. Public opinion surveys confirm that older workers and retirees do feel that the federal government has an obligation (that is, has incurred a liability) to pay Social Security benefits as promised.[64] Although it is difficult to say exactly when that obligation arises (that is, accrues), there is a national consensus that somewhere after entry into the workforce and before the age of retirement, workers and their families do accrue—politically, if not constitutionally—an entitlement to receive Social Security benefits at approximately the levels indicated by the statutory structure.[65]

The trustees’ reports are also inconsistent with methods used to account for analogous private pension plans. Under Generally Accepted Accounting Principles (GAAP), private firms are not permitted to ignore retirement benefits promised to current workers.[66] Whether offering pension or retiree health benefits, companies must accrue the costs of retiree benefits while employees are still in the workforce. The required rates of accrual are not based simply on the legal entitlements of workers, but rather on the projected level of benefits that workers are actually expected to receive—levels of projected benefits that are often much higher than workers’ legal entitlements.[67] The premise of this GAAP requirement is that retiree benefits should be recognized on corporate financial statements as soon as it is reasonably clear that the benefits will have to be paid.[68] The financial statements of Social Security disregard this central GAAP principle.[69]

Finally, the inclusion of presently incurred promises is consistent with accounting’s own definition of liabilities. Liabilities are defined as “probable future sacrifices of economic benefits arising from present obligations of a particular entity to transfer assets or provide services to other entities in the future as a result of past transactions.”[70] This definition is intentionally expansive. The commitments need not be definite, so long as obligations are probable—“that which can reasonably be expected or believed on the basis of available evidence or logic but is neither certain nor proved.”[71] Moreover, such commitments need not be legally enforceable obligations; rather, the term “obligations” as used in the definition of liabilities is broader, and includes duties that “one is bound to do by contract, promise, moral responsibility, and so forth.”[72] The trustees’ reports fail to recognize in a timely manner the substantial financial commitments that the trust funds will in all likelihood need to honor as a result of past contributions to the system.

This notion of timeliness is of critical importance to the construction of financial accounts. Liabilities are recognized when they become probable

economic commitments so that managers and other readers are made aware of the creation of financial commitments at the time those obligations are incurred and can be controlled.[73] As a matter of political reality and individual equity, the Social Security system’s obligations to pay retirement benefits are incurred during the working life of individual participants, which is when accrual accounting would recognize those obligations. The trustees’ reports, however, recognize those commitments much further in the future, after the participants have retired and begun to receive payments. At that point—regardless of the constitutional status of Social Security benefits—the federal government has long since assumed the practical responsibility to honor these commitments.

b. Long-Range Projections

Although the trustees’ long-range projections quite properly warn of a looming crisis, these projections are also misleading and incomplete. Recall for a moment the basic message of this aspect of recent trustees’ reports: over the next seventy-five years, the projected revenues of the trust funds will be insufficient to cover projected costs. The summary statistic provided is the 1.92% actuarial deficit (marked with the arrow in Table Three) for the trust funds’ combined rate. This statistic is usually interpreted as the amount by which payroll taxes would have to be raised throughout the seventy-five-year period to bring the system into perfect balance. Converting these taxes to present value implies that if the Social Security trust funds had received a lump-sum contribution of roughly $3.8 trillion on December 31, 2002, the program’s long-range problems would be solved.[74]

Assume for a moment that the federal government were to raise this staggering sum through a special assessment of some sort. Would the long-range problems of Social Security be solved once and for all? The trustees’ long-range projections suggest that they would, but this is false. The trust funds would begin to fall out of actuarial balance again the very next year. Consider the long-range estimated income and cost rates shown in Figure 2. In 2078, at the end of the long-range projection period, the income and liabilities of the Social Security system will be substantially out of balance. Even

if the long-range Social Security problem defined in the 2003 Trustees Report were “solved,” a new long-range problem requiring a similar solution would emerge in a decade or two.[75]

The trustees’ long-range projections suffer from the same defect as their short-range projections in failing to account for the accrual of future obligations. As described earlier, the long-range projections are designed to take into account all benefits paid and revenues received from current and future participants over the projection period.[76] Thus, the long-range projections encompass the trust funds’ obligations to these individuals inasmuch as all of the obligations are liquidated over the seventy-five-year period. What the long-run projections fail to consider are the trust funds’ accrued liabilities to generations of workers who will receive benefits after this period. After seventy-five years, at the end of 2077, millions of workers and retirees will have substantial claims on the system, and the trust funds will have neither reserves to honor these claims nor revenue streams to support them. The actuarial deficit measure, large though it is, underestimates the true long-range shortfall of the system, because it fails to account for the future accrual of Social Security promises to individuals who have not yet entered the workforce but will have substantial claims on the system at the end of seventy-five years.[77]

c. Biases with Respect to Reform Proposals

The way in which the trustees’ reports present the financial problems of Social Security also significantly distorts the public debate over reform proposals. To begin with, the take-home message of these reports—short-range surplus and long-range deficits—invites politicians to postpone the difficult and politically charged problems of Social Security reform. The trustees’ reports’ repeated dichotomy between short- and long-term prospects suggests that the system is currently making some sort of profit—which it is not—and can safely be left alone until the real problems appear somewhere down the road. Experts in the field know this is not the case, and regularly warn that the system’s problems will be much harder to solve if reform ef-

forts are delayed.[78] But the trustees’ reports undercut this warning by emphasizing the magnitude of the system’s short-term cash surpluses.

Consider, for example, the press coverage that followed the release of the 2003 Trustees Report. On the day after the report’s release, for example, the Wall Street Journal reported:

In many respects this account is typical of the press coverage that the trustees’ reports generate, portraying the problem, to the extent that it exists, as located far in the future.[80] The Journal’s own conclusion that the funds’ long-term financial health improved in 2003 helps explain why release of the 2003 report “slightly diminished the political pressure to shore up Social Security.”[81]

The trustees’ principal measure of long-range actuarial deficit—the negative 1.92% of total payroll highlighted in Table Three—introduces another more subtle but equally pernicious bias in the reform debates, best illustrated by a report of the Social Security Advisory Council released in early 1997.[82] The report included three different proposals for solving the financial problems of Social Security, each supported by a different coalition of council members. The requirement for each solution was that it eliminate the (then-estimated) 2.17% actuarial deficit over the seventy-five-year long-

range projection period.[83] Not surprisingly, all of the Council’s proposals included elements that increased short-term costs for the trust funds, and one group proposed a series of reforms that would substantially increase the trust funds’ deficits for the next few decades, only to recoup these losses with new (and unspecified) tax increases starting far in the future.[84] One of the unfortunate consequences of focusing on long-range actuarial deficits is that it encourages irresponsible recommendations of this sort, which depend on future sacrifices of an unrealistic magnitude.[85]

As President Bush’s Social Security Commission discovered, cash-flow accounting can also understate the effectiveness of certain solutions. A key component of the Bush Commission’s proposals was a reduction in Social Security benefits to offset voluntary contributions to individual retirement accounts.[86] Under cash-flow accounting, payments into the privatized accounts constitute an immediate diminution in trust fund resources and an immediate expense. The reduction in benefits—that is, future outlays—associated with these contributions, however, is not reflected until the benefits are actually due to be paid. Typically these benefits will not be payable until many years in the future, and oftentimes outside of the seventy-five-year estimation period. Accordingly, under cash-flow accounting, the fiscal benefits of the Bush Commission’s proposal were substantially understated.

B. Consolidation Within the Federal Budget Process

This Part reviews the manner in which the finances of the trust funds are currently integrated into other federal budgetary accounts. After reviewing current practices, it offers several criticisms of the present approach and then discusses its distorting effect on public debate of Social Security reform. Finally, it reveals how accrual-based accounting for Social Security would reduce the incentive to obscure budget realities by integrating annual trust fund revenues with other budgetary aggregates.

1. The Current Approach to Consolidating Social Security Trust Funds

The federal government routinely misleads the public about the government’s fiscal posture by including trust fund surpluses in budgetary aggregates. Consider, for example, excerpts from the opening paragraphs and accompanying summary table (reproduced in slightly altered form in Table Four) from an August 1998 Congressional Budget Office (CBO)[87] report, which was issued when the federal government was just beginning to report the surpluses of the late 1990s:[88]

This artfully drafted prose encapsulates a common approach to discussing budgetary aggregates in Washington, D.C.

The first sentence provides the report’s bottom line: the federal government was projected to run up a $63 billion surplus in fiscal year 1998.[90] This figure is repeated on numerous occasions throughout the rest of the report and was cited in numerous press accounts heralding what was widely reported as the first federal budgetary surplus since 1969.[91] Most readers could, therefore, be excused for not lingering over the third sentence of the opening paragraph, in which the report notes a distinction between off-budget surpluses and on-budget deficits for the year. What that convoluted construction explains is that certain sources of revenue—that is, those associated with Social Security and the U.S. Postal Service—are “legally classified as off-budget,” meaning that Congress has voted to exclude these items from budget aggregates.[92] The sentence further informs that if those revenues were so excluded, something called the “on-budget” would show a deficit for 1998 and four of the five following years.[93] This on-budget figure was projected to reach a surplus on the order of $60 billion, but not until ten years into the future, in 2008.

Since 1998, the overall budgetary picture of the United States has followed a tortuous path, first improving through 1999 and 2000 and then, starting in 2001, deteriorating significantly.[94] For a time, particularly during the final years of the Clinton Administration, budget officials became some-

what more circumspect in their casual combination of Social Security trust fund surpluses and those of the on-budget accounts.[95] However, confusion and obfuscation still prevailed. The repeated and often incoherent references to Social Security lockboxes in the 2000 presidential campaigns are striking illustrations of this confusion,[96] and even publications of budgetary agencies included many of the problems noted above.[97]

With the reemergence of budgetary shortfalls, budgetary documents became even more contorted in their treatment of Social Security surpluses. Reproduced below is the opening paragraph of the CBO’s January 2002 report, and an accompanying summary table (Table Five):

In many respects, the January 2002 report is even more misleading than prior reports. Whereas prior reports were precise, even in summary statements, to speak in terms of “total” surpluses or deficits (which in turn were decomposed into on-budget and off-budget components), the January 2002 report is more casual in its terminology, referring simply to “the budget surplus” in its opening sentence. When the introductory paragraphs eventually turn to the off-budget component of the total-budget picture, the presentation is even more convoluted. The text explains that off-budget surpluses run throughout the period but that on-budget surpluses do not “reemerge” until 2010, three years before the end of the ten-year projection.[99] The astute reader might infer that the on-budget accounts must therefore be in deficit before 2010. But what the summary doesn’t say—and what many readers would likely find salient—is that without Social Security cash-flow surpluses, the on-budget deficits for 2003 and 2004 were projected to be in the range of $180 to $190 billion.[100] This information is contained in tables appearing elsewhere in the 2002 report (summarized in Table Five) but does not factor into the introductory paragraphs.[101]

2. Criticisms of Standard Presentation

Apart from the political appeal of trumpeting higher budgetary surpluses or lower budgetary deficits, total-budget figures are admittedly useful statistics to compile. Several dozen pages into its analysis, the CBO August 1998 Update explains the importance of this budgetary measure, namely its relationship to the amount of government debt held by the public:

The compilation of and emphasis on total-budget aggregates are thus justified on the grounds that they represent the net amount that the federal government must borrow from (in years of deficits) and can repay to (in years of surplus) the general public. By this logic, Social Security trust fund surpluses are properly included in this aggregate because these funds reduce the government’s need to obtain funds from the private sector, both decreasing the government’s net borrowing and “freeing up” financial resources for other investments.[103]

To the extent that the purpose of budgetary aggregates is to communicate to Treasury Department officials the net amount of federal bonds to be sold to (or redeemed from) the general public in the course of a year, the current emphasis on total budget deficits (and occasionally surpluses) is fully justified. The management of public debt auction is not, however, the most important function of federal budgetary aggregates. The principal reason for compiling these budgetary figures is, in my view, to communicate to the general public and their representatives in Washington the amount by which

overall financial claims on the federal government increased or decreased over the course of the reporting period.[104] Measured against this standard, the current practices of the federal government fall far short of the ideal. In particular, the government’s total-budget surplus or deficit is not the most salient number to present to the general public as the best measure of the change in the government’s overall financial obligations over the course of the year. At a minimum it would be preferable for reporting agencies to emphasize what the CBO refers to as the “on-budget” aggregates. However, it also would be preferable for federal budgetary reports to highlight the annual growth in both the total explicit government debt and the implicit debt associated with government programs such as Social Security.

a. Misleading Features of the Standard Presentation

A threshold complaint about the CBO presentation of budgetary aggregates is that it is arguably inconsistent with federal law. Several times in the 1980s and again in 1990, Congress, with some fanfare, voted to move Social Security off-budget, and from time to time, politicians point to this fact with pride.[105] The CBO alludes to this fact in its August 1988 report when it mentions parenthetically that the trust funds are legally required to be off-budget.[106] Throughout the report, however, it speaks almost exclusively in terms of total-budget figures that consolidate off-budget accounts. At a minimum, this approach seems unfaithful to congressional intent.[107]

To make matters worse, the CBO further obscures the significance of trust fund surpluses through a misleading aggregation of another off-budget entity—the Postal Service—with the Social Security trust funds.[108] A reader of these references to the Postal Service might reasonably infer that, because the CBO gives both entities the same degree of prominence, both entities must make roughly comparable contributions to the total-budget aggregates. The memorandum items set forth in Tables Four and Five—information typically not included in the CBO’s own summary tables—reveal that the Postal Service has no meaningful effect on budgetary surpluses.[109]

b. Familiar Criticisms of Relying on Total-Budget Aggregates

This Part focuses on more substantive complaints regarding the CBO’s and other government accounting agencies’ reliance on total-budget aggregates that consolidate Social Security’s cash-flow surpluses as the principal summary statistic measuring the government’s overall operations for any given year.

i. Proper Characterization of Trust Fund Reserves

For those who follow federal budgetary policy closely, complaints about the consolidation of Social Security surpluses with on-budget accounts are familiar.[110] Although the trust funds are statutorily required to invest their surplus reserves in government securities, the funds are not available to finance general federal expenditures in the same sense that income-tax revenues are. Social Security reserves are earmarked for the payment of Social Security benefits, and the amount of benefits is expressly tied to trust fund balances.[111] Further, the manner in which the Social Security trustees account for these reserves in their annual reports suggests that these funds have

already been factored into the long-range estimates for trust fund revenues.[112] The Social Security surpluses have, in essence, already been allocated for a particular use—one with a high degree of political support. Accordingly, this critique runs, it invites misunderstanding for the CBO to trumpet a budgetary aggregate that absorbs the trust funds’ current surpluses—and it encourages politicians to behave as if they had several hundred billion extra dollars to spend. The Clinton Administration, for example, appeared to fall into this trap in the late 1990s, when it called for total-budget surpluses to be used to solve the problems of the Social Security system.[113]

In essence, this critique rests on a challenge to the way in which the federal government currently classifies the ownership of trust fund assets. The premise of a unified federal budget—that is, a budget that turns on total-budget aggregates—is that intragovernmental transactions should be consolidated to get a complete picture of the government’s cash flows and hence of its financing needs.[114] Critics of current practices dispute this premise, arguing that Social Security trust fund reserves are fundamentally different from the accounts of other departments maintained with the Treasury and even from other kinds of federal trust-fund accounts, such as the government’s highway trust reserves, which are deposited with the Treasury pending passage of appropriation bills and distributed periodically to the states

and other recipients.[115] Critics of the consolidated approach regard the government’s obligation to pay Social Security benefits as among the most substantial of our government’s commitments[116] and, for this reason, regard obligations to the Social Security trust funds as equivalent, for accounting purposes, to debt held by the general public. As such, Social Security trust funds should not be consolidated with other intergovernmental transfers and on-budget aggregates should replace total-budget aggregates as the principal measure of fiscal balance.[117]

ii. The Economic Realities of Social Security

Total-budget aggregates are also inconsistent with the economic realities of Social Security. A fair reading of the CBO reports described above and similar publications suggests that the decision to include Social Security surpluses in budgetary aggregates is not simply based on a formal classification of the trust funds as governmental entities but also reflects a desire to present a complete picture of the government’s financial obligations and the net effect of governmental borrowing from capital markets. Even if this is true, it is not necessarily accurate to view the transactions between the trust funds and the rest of the federal government in a given year as reducing the government’s borrowing or enhancing the availability of capital to other borrowers.

Consider the consequences of the $165.4 billion net increase in trust fund assets deposited with the Treasury in 2002.[118] The Treasury is required to repay these funds, with interest, just as it would have to pay back interest and principal on securities issued directly into the capital markets. As discussed above, the trustees of the Social Security system are counting on accumulated surpluses and the interest payments thereon to provide for more than five percent of the trust funds’ summarized income rates over the next seventy-five years.[119] If the trustees were to distribute the entire accumulated reserves of $1.4 trillion to currently covered workers and retirees in the form of a single government note issued to each participant beneficiary and make an offsetting deduction in the benefits of these individuals, there would be no change in the financial status of the trust funds (although the funds’ liquidity would be entirely eliminated). This distribution would, however, unambiguously transform Treasury fund obligations to the trust funds into debt held by the public, suggesting that, as a practical matter, Social Security trust

fund assets deposited in the Treasury are functionally equivalent to debt held by the public.[120]

It is also debatable whether the current operations of the Social Security system should be understood to increase the availability of capital for other forms of investment. For some time, economists have theorized that pay-as-you-go public pension systems such as Social Security might actually reduce the supply of capital, as individuals reduce their private savings for retirement in anticipation of receiving benefits promised from their public pensions.[121] There is a growing body of theoretical and empirical literature estimating the actual value of the retirement income of Social Security benefits and measuring the effect of these benefits on other types of individual savings.[122] While the results of this research are not without ambiguities, the weight of current evidence does come down on the side of Social Security promises’ somewhat depressing private savings rates.[123] If this is true, the CBO’s belief that Social Security’s current operations “release resources for private investment”[124] is inaccurate.

c. A Deeper Critique

Although the familiar criticisms of total-budget aggregates outlined above are valid, proponents of these criticisms have not pursued them to their logical conclusions. Generally speaking, the issue has been framed as whether or not annual Social Security operating surpluses should be consolidated with the rest of the government’s operating budget. The sole remedy sought by critics of current practices is the removal of these off-budget sur-

pluses from budgetary aggregates. For a variety of reasons, this response is incomplete and unsatisfactory.

Even in a year in which both the on-budget aggregates and the Social Security trust funds are in perfect balance,[125] it would be incorrect to conceptualize the federal government’s overall financial posture as being in equipoise with respect to the Social Security system. Admittedly, the trust funds would have made no net deposits with the Treasury in such a year. However, it is equally clear that the amount of pension benefits promised by the Social Security system might have increased. As described above, critics’ objections to total-budget aggregates turn on the special nature of Social Security benefits, not on the contractual arrangements governing trust fund deposits with the Treasury. Even though the level of the trust fund reserves may not have increased, covered workers and their families could have accrued additional benefits and the value of the Social Security system’s “sacred obligation” to the American people might well grow.[126] Simply moving the trust fund surpluses off-budget fails to reflect this basic truth. Those who want federal budget aggregates to reflect the special nature of Social Security’s obligations faithfully must do more than move the trust funds off-budget. They must support an accounting system that more accurately estimates the size and growth of the system’s actual obligations to pay future benefits.

Similarly, to the extent that one is looking to federal budgetary aggregates to get a better sense of the net effect of government operations on capital markets, simple off-budget accounting treatment for Social Security is inadequate. In a year in which the trust funds and the federal budget are in perfect balance, the net sale of government securities to the capital markets would, of course, be zero. But if covered workers and retirees continue to accrue pension benefits from the Social Security system, economic theory predicts that there will continue to be some sort of offsetting reduction in the retirement savings that these individuals would otherwise make. In the extreme, if there were a one-for-one reduction in other savings, for each dollar of newly accrued Social Security benefits, there would be a dollar reduction in private savings.[127]

In short, simple off-budget treatment of Social Security obscures two important and related effects. Properly constructed budgetary aggregates cannot ignore changes in the overall financial status of the Social Security system—particularly increases in unfunded liabilities. Further, increases in the system’s liabilities (that is, its promised benefits) are what make private individuals change their savings patterns and thereby reduce funds available for other forms of investment. The only way that budgetary aggregates can reflect these factors is to rely on financial accounts that reflect the accrual of Social Security liabilities.

d. Biases for Reform Proposals

Beyond the failure of current budget presentations to provide a complete picture of the financial health of the Social Security system, the current primacy of total-budget aggregates distorts the public debate over Social Security reforms in numerous ways.

i. Perverse Budgetary Incentives for Social Security Reforms

To begin with, the time horizon for federal budgetary politics is much shorter than the seventy-five-year perspective that the trustees of Social Security consider in their annual reports. Because budgetary calculations typically project only five or ten years, the effect of reform proposals over this horizon is most salient for budgetary politics. Accordingly, reform initiatives that create short-range cash surpluses and long-range growth in expenditures are favored in budget-driven compromises. Illustrations of this phenomenon are manifold in the budget wars of the past fifteen years. For example, reform proposals often entail the extension of Social Security participation to all state and local workers, many of whom are currently exempt.[128] Such reforms bring an immediate infusion of payroll taxes into the Social Security system—and additional total-budget surpluses for the federal government—while postponing increases in benefit payments until later years. From the perspective of the federal budget, this is a highly attractive reform, irrespective of whether the reform actually improves the long-term viability of Social Security.[129]

To make the same point in slightly more abstract terms, consider the desirability of doubling the size of the Social Security system immediately. Having worked their way through the discussion thus far of the long-range deficit problems of the Social Security system, most readers would readily agree that doubling benefits and payroll taxes would be a very dubious choice for the federal government. But such a move would, if viewed through the lens of total-budget aggregates, generate the positive outcome of decreasing the current total-budget deficit by roughly $150 billion. An accounting convention that rewards such a reform is dangerously misleading.

ii. Distorting Effects on Reforms That Change Trust
Fund Operations

A separate problem with total-budget aggregates is that they generate substantial budgetary consequences for relatively modest reforms in the operation of the trust funds. The perceived budgetary consequences of certain reforms are so great, in fact, that some proposals are summarily rejected as politically infeasible or too costly.

A good example is a proposal to invest some portion of trust fund assets in the stock market.[130] There is much to be said for and against such a reform, but the merits of the idea are not of concern here. Under a total-budget approach, moving trust fund assets into the stock market has a significant budgetary cost. If the trust funds were to move $500 billion into the capital markets over the next five years—a not implausible figure—the full amount would be deducted from total-budget aggregates, as these funds would no longer be invested in intragovernment accounts.[131] The accounting effect would be to exacerbate total-budget deficits over the five-year-period, which would be a distortion in that the wealth of the trust funds would not be diminished. And from a macroeconomic perspective, the transaction is largely a wash, albeit one with potential distributional consequences.[132]

A similar kind of distortion could follow from changes in the structure of the trust funds themselves. Suppose that Congress were to decide that the Social Security program should be more independent of the executive branch so as to ensure, for example, that the collection of trust fund taxes or the

investment of trust fund assets would not be used to manipulate budgetary aggregates.[133] If the management of trust assets were moved to truly independent trusts bound by traditional trust principles, the trusts could well lose governmental status for budget-scoring purposes, with the result that the trusts would be entirely off-budget entities even for purposes of budgetary aggregates. Because this change in classification would remove Social Security’s cash flow surpluses from total-budget aggregates, there would be a large apparent cost to the reform, a cost with little relation to the reform’s actual significance.

iii. Generating Reform Proposals with Ulterior Motives

Another problem with total-budget aggregates is their tendency to generate reform proposals designed primarily to have effects on other parts of the budget. Perhaps the best example of this phenomenon is former Senator Moynihan’s perennial suggestion that Social Security trust fund taxes be cut to eliminate their distorting effect on total-budget aggregates.[134] Moynihan, who was one of Congress’s leading experts on Social Security and a strong supporter of the benefits it provides, perceived other members of Congress to be using the Social Security surplus—as reflected in total-budget aggregates—as a vehicle for cutting income taxes, thereby using regressive payroll taxes as a substitute for progressive income taxes. While his first-order preference likely would have been to move to off-budget treatment for Social Security and to keep payroll taxes and income taxes at their then-current levels, he believed this solution to be politically impossible. Therefore, forced to choose a tax cut, he preferred a cut in Social Security taxes over one in income taxes, notwithstanding the deleterious effect on the system’s long-range solvency.[135] Whatever one thinks of the wisdom of Senator Moynihan’s political judgments, his strategy revealed the twisted effect of our current reliance on total-budget aggregates as the principal measure of fiscal balance.

II. An Accrual Accounting System for Social Security

This Part sketches out an accrual accounting system for Social Security. It begins with a brief discussion of the assets and liabilities that might in theory be recognized under a system of accrual accounting. It then describes a set of GAAP-style accrual-based financial statements for Social Security, including rough estimates of what such a system of accounts would have looked like for the calendar year ending December 31, 2002.[136] Next it compares this system of accounting to other measures that academic writers and other experts sometimes use to assess the long-term solvency of Social Security and explains the principal differences between these measures and accrual accounting. It then discusses how a modified form of accrual accounting, incorporating some of the valuable attributes of alternative solvency measures, could be used to frame and enhance public debate over proposals to reform the Social Security system. The Part concludes with an exploration of how such a modified system of accrual accounting might be consolidated with federal budgetary aggregates.

A. Possible Components of an Accrual-Based System of Accounting

Before constructing a system of accrual accounting for Social Security, one must consider which liabilities and assets such an accounting system should recognize. For this purpose, it is useful to distinguish among three different groups of participants: retirees who have completed their working careers; workers who are in mid-career; and future workers, born and unborn, who have not entered the workforce. Box Three summarizes their positions with respect to both contributions to the Social Security system and statutory entitlements to receive benefits. Current retirees have paid most of their taxes and earned their full statutory entitlements to benefits.[137] Workers in mid-career have paid some, but not all, of their taxes and earned some, but not all of, their statutory entitlements to benefits. Future workers have not yet started to pay payroll taxes or earn statutory entitlements.

Box Three
Contributions and Statutory Entitlements of
Social Security Participants

Participants	Contributions	Statutory Entitlement to Benefits
Retirees	Most Taxes Paid	All Entitlements Earned
Workers in Mid-Career	Some Taxes Paid Some Taxes to be Paid	Some Entitlements Earned Some Entitlements to be Earned
Future Workers –Over Next 75 years –After 75 years	All Taxes to be Paid. All Taxes to be Paid.	All Entitlements to be Earned. All Entitlements to be Earned.

Participants

Contributions

Statutory Entitlement to Benefits

Retirees

Most Taxes Paid

All Entitlements Earned

Workers in Mid-Career

Some Taxes Paid

Some Taxes to be Paid

Some Entitlements Earned

Some Entitlements to be Earned

Future Workers

–Over Next 75 years

–After 75 years

All Taxes to be Paid.

All Entitlements to be Earned.

The basic presentation of the trust funds’ current financial position in the Trustees Report (illustrated in Table One, showing a net increase in assets of $165.4 billion in 2002) recognizes only a limited number of the elements shown in Box Three. Specifically, this basic presentation reflects all taxes that retirees and mid-career workers paid during the year. Past tax payments are also reflected in the trust funds’ accumulated reserves, which accrue interest during the year.[138] In terms of benefits, the basic presentation of the trustees’ report explicitly recognizes only benefit payments that are paid during the course of the year.[139] Accordingly, the basic presentation reflects only a fraction of the statutory entitlements that retirees have earned to date, and essentially none of the statutory entitlements earned to date by mid-career workers.[140] These recognition principles largely follow the logic of cash-flow accounting, under which the transfer of cash is the essential requirement of income or expense recognition.[141]

Were one to model accounting of Social Security on principles of accrual accounting developed under GAAP for the private sector, the system’s financial statements would have to recognize substantially more liabilities for the systems’ obligations to pay future benefits. Under GAAP, liabilities are

probable future economic sacrifices based on past transactions or events. Inherent in the definition is a degree of subjectivity, as there is room for disagreement regarding the probability that benefits promised under the current statutory formula will be paid in the future. The Social Security Act currently requires those payments, but Congress is constitutionally free to alter benefit formulas by statutory amendment. For reasons explained above,[142] the likelihood of full payment of Social Security benefits for current retirees seems sufficiently high to warrant recognition of liabilities for all future benefits of retirees under GAAP. Moreover, the probability of payments to mid-career workers is sufficiently likely to warrant some recognition of obligations to these participants under GAAP, with a larger fraction of benefits recognized for older mid-career workers and a smaller fraction of benefits for younger workers. A GAAP-style accrual accounting system would recognize the system’s obligations to mid-career participants in a gradual manner over the course of their working life, reflecting the growing probability that benefits will be paid in accordance with applicable statutory standards as workers approach retirement age. Box Three and subsequent discussions refer to a benefit as “earned” if there is a sufficient probability that the benefit would be recognized as a liability under a GAAP-style accrual accounting system.

In the field of social insurance, financial analysts have not followed GAAP accounting. Indeed, the dominant presentations of the financial posture of the Social Security system are the ones outlined and critiqued in Part I of this Article, under which benefits are not recognized until paid. Within limited circles of experts, however, accrual-like concepts have been employed to estimate the present value of the system’s total obligations.[143] Following the logic of cash-flow projections emphasized in the trustees’ reports, these alternative measures typically focus on the amount by which the present value of the system’s future benefit payments exceeds the present value of future revenues and current reserves.[144] Unlike GAAP-style principles of accrual accounting, these measures do not seek to assess the likelihood that these future taxes will be contributed or future benefits paid. Rather, these

measures recognize the present value of all future taxes and projected benefits of covered participants under current statutory rules. These measures are thus broader than GAAP-style accrual accounting.[145] On the liability side, these alternative approaches recognize the present value of both benefits that have been earned to date and benefits that will be earned in the future. On the assets side, they recognize the present value of contributions to be made in the future plus, in some instances, residual reserves representing the excess contributions over benefit payments that have been made in the past.[146]

By convention, there are two populations for which these broader measures are calculated: the closed-group and the open-group. The closed-group is the group of people already within the Social Security System: current retirees and mid-career workers (typically defined as everyone currently fifteen years or older).[147] The open-group is the closed-group plus all future workers.[148] In the past, Social Security’s Office of the Chief Actuary has typically calculated the open-group system over a seventy-five-year time horizon,[149] but recently independent experts have pushed for the measure to be calculated over an infinite horizon, thereby taking into account all future contributions to the trust funds and all future benefit payments.[150]

B. A GAAP-Style Accrual Accounting System for Social Security

The bare bones of an accrual accounting system for Social Security are straightforward. They should, at a minimum, include an annual balance sheet, which lists the total assets and liabilities of an institution as of a specified date, and an income statement, which gives operating results for a specific period, such as one year.[151] This Part explains how these elements can be roughly approximated from publicly available information and then presents an estimated GAAP-style balance sheet and income statement for Social Security for the calendar year 2002.

1. Balance Sheet

The chief difficulty in constructing a GAAP-style balance sheet for Social Security is deriving an estimate of the system’s accrued liabilities—that is, the present value of all benefits to be paid to retirees and the portion of benefits the mid-career workers have earned to date. While the Social Security Administration does not publish its own estimate of the system’s accrued liabilities, its Office of the Chief Actuary does report a statistic from which accrued liabilities can be derived. Each spring, the Chief Actuary calculates an estimate of the system’s “maximum transition cost” as of the beginning of the year.[152] This measure, estimated to be $12.6 trillion as of December 31, 2002, represents an estimate of the amount by which the system’s accrued liabilities—that is, its benefits earned to date for both retirees and mid-career workers—exceed the system’s accumulated reserves. The Chief Actuary’s methodology for accruing benefits for mid-career workers under the maximum transition cost measure is relatively conservative. Projected benefits are accrued on a forty-year straight line basis reflecting the typical working life of participants.[153] While one could propose alternative formulas for accruing benefits to mid-career participants,[154] the Chief Actuary’s approach is plausible and close enough to other reasonable systems of accrual to provide useful illustrations for purposes of this Article. Were Social Security to move formally to a system of accrual accounting, more care would need to be given to the precise accrual formula used. But for current purposes, the principles of accrual implicit in the Chief Actuary’s maximum transition cost estimate are acceptable.

BOX FOUR
HOW SHOULD SOCIAL SECURITY BENEFITS ACCRUE?

Accrual accounting for Social Security would spread the costs of retirement benefits over the working lives of participants. Under such a system, retirement benefits should be fully accrued when workers retire, but there are a range of methods under which the growth of benefits could be recognized over the working lives of participants.

The actual statutory requirements of the Social Security Act, for example, could be used to determine the accrual formula. Under this approach, the trust funds would recognize retirement benefit obligations when participants have earned the legal right to receive those benefits under current statutory requirements. This approach, however, would produce a number of anomalies. Forty quarters—that is, ten years—of participation is required before participants are entitled to retirement benefits. So under this formula, workers would not, strictly speaking, earn any benefits until they had worked for ten years. On the other hand, because Social Security benefit formulas are highly progressive, participants earn relatively large amounts of benefits once they pass the ten-year vesting threshold and relatively smaller amounts of benefits in their later years of work.

The Chief Actuary’s estimate of maximum termination cost uses a forty-year straight line method of accrual, in which projected benefits are allocated evenly over an assumed forty-year working life for participants. The illustrations of accrual accounting presented in this article are based on this methodology. Compared to the Social Security statutory formula, the forty-year straight-line approach accrues benefits at a faster rate during the first ten years of participation and at a slower rate thereafter. While not tracking statutory requirements, this accrual formula provides a defensible allocation of costs over time. But one could also imagine accrual formulas with shorter periods of accrual or more complex accrual formulas.

For purposes of developing a sensible formula of accruing Social Security benefits, a number of considerations would be relevant. In addition to the statutory formula and the actual experience of participants, one would also want to recognize that benefits accrued under this formula might become more difficult for Congress to change than benefits that have yet to accrue. In addition, special rules of realization might be appropriate for disability and other non-retirement benefits paid out from the trust funds.

With the Chief Actuary’s estimates of maximum transition costs, it is a relatively straightforward matter to sketch out a GAAP-style accrual-based balance sheet. The system’s only material assets are the reserves of the trust funds, which at year-end 2002 equaled about $1.4 trillion. The system’s accrued liabilities equal the sum of its current reserves ($1.4 trillion) and the Maximum Transition Cost ($12.6 trillion), and therefore reached $14.0 trillion as of December 31, 2002.[155] Based on these estimates, Figure Four presents a balance sheet as of December 31, 2002, with one slight embellishment. Liabilities have been segmented into classes based on the age cohort of the participants with whom the accrued benefits are associated. Based on information reported in the Social Security Administration’s own annual report (as opposed to the trustees’ annual reports discussed above), about one-third of the system’s current liabilities or roughly $4.6 trillion of claims are estimated to represent accrued obligations to participants at or above the age of sixty-two,[156] the age at which Social Security retirement benefits are generally available.[157] In other words, the Social Security system owes current retirees accrued benefits with a present value equal to more than three times the system’s current reserves. The remaining accrued liabilities—roughly $9.4 trillion in accrued claims—consist of obligations to younger participants. A full-blown system of accrual accounting could easily distinguish among other classes of claimants by, for example, separating out accrued liabilities by generational cohorts.[158]

Figure Four conveys an important message, and one strikingly at odds with the basic presentation found in the trustees’ reports. It indicates that the accrued commitments of the Social Security system to current retirees and mid-career workers ($14 trillion) is ten times the value of accumulated reserves ($1.4 trillion). The difference between these two figures, $12.6 trillion, is the amount of funding required to honor Social Security benefits that have been earned to date, using the definition of accrual implicit in the maximum-transition-cost estimate. Were Social Security to conform to GAAP accounting requirements, the system’s unfunded accrued liability would have a magnitude approximately equal to the one depicted in Figure Four.

2. Income Statement

Faced with a balance sheet reporting Social Security’s substantial unfunded accrued liabilities as of year end 2002, one might naturally ask whether the system’s overall solvency has improved or worsened in the course of the past year. An income statement for the past year is the accounting format best suited to answer this question. Table Six presents an accrual-based income statement for the Social Security system for the past year, comparable to the balance sheet present in Figure Seven.

Under accrual accounting, an entity’s income during a period is equal to the difference between its net worth at the beginning of the period and its net worth at the end of the period (adjusting for dividends and capital contributions, which are not at issue here). As mentioned, the Social Security system’s net worth at the end of 2002 was negative $12.6 trillion, and the system’s net worth a year earlier was negative $12.2 trillion.[159] Thus, simple accounting

logic dictates that the system suffered a loss of some $467.5 billion during the course of the year.[160]

With this estimate of the system’s annual loss for the 2002 calendar year, one can construct a rough approximation of the system’s income statement as well. The system’s revenues for the year are known and consist of payroll taxes ($532.5 billion), tax receipts ($14.2 billion), and the yield on assets in the trust fund ($80.4 billion), for total revenues of $627.1 billion. Estimating the system’s expenses on an accrual basis is a bit more complex but possible. Two expenses of the system are known: the administrative expense of operating the system during the year ($4.2 billion) and a limited number of other expenses ($3.7 billion).[161] However, because the system lost $467.5 billion in 2002, it must have incurred some $1,086.7 billion in additional expenses during the year.[162] Conceptually, these expenses must be divided principally between the accrual of new benefit promises and the accrued interest on previously accrued benefit promises. Because the amount of pension promises that were outstanding at the beginning of 2002 and the amount of pension promises that were liquidated in the course of the year are known, it is possible to estimate the interest cost of the Social Security system as $786.9 billion in 2002.[163] This means that the amount of newly accrued benefit promises must have been on the order of $299.8 billion.[164]

Based on the same methodology, Table Seven presents a series of income statements for the trust funds for the past five years. This series of statements reveals that the $467.5 billion loss in 2002 was not an anomaly. Indeed, when this loss is compared to the system’s results for the past half decade, the 2002 performance was a relatively good year. On average, the Social Security system lost $623.7 billion a year between 1998 and 2002 on an accrual accounting basis.

Another way to present the financial position of the Social Security trust funds over the past five years is to chart the system’s unfunded accrued liability over that period. Table Eight offers such a presentation, first providing the level of unfunded deficit at year end in current dollars and then presenting the deficit figure as a percentage of GDP.[165] Under both measures, the system’s financial posture has clearly deteriorated. In the five years, between year-end 1997 and year-end 2002, the unfunded accrued liabilities of the system increased by more than $3 trillion from $9,512 billion at year-end 1997 to $12,629 billion at year-end 2002. Even expressed as a percentage of GDP, the deficit increased from 116.1% to 122.2% of the GDP.

Subsequent discussions address the difficult normative question of whether this magnitude of unfunded accrued liabilities should be viewed as problematic. For current purposes, it is important to note the stark contrast between the direction Social Security appears to be going under an accrual-based accounting system and that indicated by the trustees’ reports. Under the latter approach, the system has been adding approximately $150 billion to its cash reserves for each of the past five years. Under a GAAP-style accrual accounting, the system’s annual returns have been at least $600 billion lower, incurring losses in excess of $450 billion a year.

C. Comparing Accrual-Based Financial Statements with Other Measures of Long-Term Solvency

Academic writers and experts sometimes use other summary statistics to estimate the long-term solvency of the Social Security system. Because these measures sometimes factor into expert discussion and have recently been reported as supplementary information in the trustees’ reports,[166] they are important enough to merit comparison with the GAAP-style accrual-based system. This Part engages in such a comparison and concludes by demonstrating how some of the advantages reflected in these alternative measures might be grafted onto a modified accrual-based system.

1. Closed-Group Unfunded Obligation

One common measure of the Social Security system’s long-term solvency is the closed-group unfunded obligation or closed-group liability.[167]

This measure equals the present value of future contributions that mid-career participants will make to the system, plus the current reserves in the trust funds, minus the present value of all payments to be made to current retirees and mid-career participants in the future.[168] This measure assumes that no new participants are allowed to enter the system, and that current participants continue in the system under the current rules for the rest of their and their beneficiaries’ lives.[169] In other words, the system’s closed-group liability represents what the net obligations due under the current system would be if the system were closed to new entrants. For this reason, the measure is sometimes referred to as the closed-group transition cost.[170] This measure would be appropriate if one were trying to evaluate the solvency of a system under which all current participants were to be grandfathered into the existing system and all new workers were to be steered into a new system.

Figure Five shows the Office of the Chief Actuary’s estimate of the closed-group unfunded obligation of the Social Security system over the past few decades. Also included are estimates of the system’s accrued unfunded liabilities for the years in which such estimates are available. As of December 31, 2002, the system’s closed-group unfunded obligation was $10.5 trillion. This is how much the government would have to raise today to finance fully a grandfathered Social Security system for current retirees and mid-career workers. Notice that this figure is roughly $2 trillion less than the system’s unfunded accrued liability of the same date. This difference represents the amount by which the present value of future taxes to be paid by mid-career workers (and some retirees) exceeds the present value of additional entitlements that mid-career workers will accrue in the future.[171] The system’s overall deficit shrinks a bit if all current participants are required to stay in the system for the rest of their lives. Put slightly differently, one could say that this $2 trillion represents the excess taxes that current participants are going to pay into the Social Security system for the privilege of earning additional benefits from the system over the balance of their working lives.

2. Open-Group Unfunded Obligation: Seventy-Five-Year Horizon

Another summary statistic representing the long-term solvency of the trust funds—and the measure most commonly cited by the Social Security Administration[172]—is the open-group unfunded obligation estimated over a seventy-five-year horizon or, for short, the seventy-five-year open-group liability.[173] Like the closed-group measure, this open-group measure reflects the difference between the present value of future tax revenues and the present value of future benefits.[174] The open-group measure is, however, broader than the closed-group measure in that it includes taxes paid by and benefits paid to future participants insofar as such payments are made over the next seventy-five years.[175] The seventy-five year open-group liability is closely

related to the trust fund actuarial deficits previously discussed. The actuarial deficit (negative 1.92% of payroll as of year-end 2002) is the amount by which payroll taxes would have to be increased over the next seventy-five years for the system to meet its projected obligations. The seventy-five-year open-group liability is the net present value of these additional taxes.

As of December 31, 2002, the system’s seventy-five-year open-group liability was estimated to be negative $3.5 trillion.[176] Figure Six reports the open-group obligations over the past few decades. Notice that around 1983, the seventy-five-year open-group liability fell to zero. This is because, after

the 1983 reforms, the system was projected to meet its obligation for the relevant period of reference: seventy-five years.[177] After 1983, the open-group unfunded liability began to grow. This reflects the cliff effect[178] as well as unrelated changes in actuarial assumptions.[179] After 1983, the system’s seventy-five-year open-group liability increases, because the period of analysis includes an increasing number of years in which cash payments to the system will substantially exceed cash inflows.

Figure Six also includes estimates of the system’s unfunded accrued liabilities for the years in which these estimates are available. Notice the substantial difference between the unfunded accrued liabilities and the seventy-five-year open-group liability. As of year-end 2002, the system’s unfunded accrued liabilities ($12.6 trillion) exceeded its seventy-five-year open-group liability ($3.5 trillion) by about $9.1 trillion. Because we know from our analysis of closed-group liability that current participants will contribute excess taxes of about $2 trillion over the balance of their lives (predominantly within the next seventy-five years), the remaining $7.1 trillion must come from the excess of the present value of taxes to be paid by future participants over benefits to be paid to future participants in the next seventy-five years. Even with these excess taxes, the seventy-five-year open-group liability is negative $3.5 trillion. The foregoing discussion reveals that the relatively small size of the seventy-five-year open-group liability measure is largely explained by excess tax revenues to be contributed by future participants over the next seventy-five years.

3. Open-Group Unfunded Obligation: Infinite Horizon

The seventy-five-year open-group liability measure is, perhaps, an understandable extension of traditional Social Security accounting. The trustees’ reports’ long-term projections go out only seventy-five years, the same time frame as the open-group liability measure. For the reasons outlined in Part I, however, limiting analysis to cash flows over seventy-five years is inherently misleading. It ignores all of the benefits future workers should receive at the end of the seventy-five year horizon and it produces the cliff effect. For this reason, many academic writers prefer a long-term balance measure that looks beyond the seventy-five-year horizon.[180]The infinite-horizon open-group liability measure offers such a tool. It continues the open-group methodology indefinitely, factoring in all future tax revenues and all future benefits.[181] The Office of the Chief Actuary has not yet developed a

times seriesof infinite-horizon open-group liability estimates. The 2003 Trustees Report did, however, for the first time, include a single estimate of the system’s infinite-horizon open-group liability as of December 31, 2002.[182] The estimate was $10.5 trillion—exactly the same as the Chief Actuary’s estimate of the system’s closed-group liability.[183] This equality implies that the present value of benefits to be paid to future participants is equal to the present value of the future taxes that these participants will pay. Other academic experts have estimated that the present value of taxes to be paid by future participants is $3.5 trillion greater than the present value of their benefits, implying an infinite-horizon open-group liability of about $7 trillion.[184] The distance between these estimates comes from differences in demographic projections about future life expectancies and economic assumptions about discount rates.[185] But the main point is that both estimates indicate that the infinite-horizon open-group liability is much larger than the Chief Actuary’s estimate of the seventy-five-year open-group liability.

4. GAAP-Style Accrual Accounting vs. Alternative Measures of Long-Term Solvency

Putting aside superficial differences in presentation format,[186] there are two important theoretical differences between GAAP-style accrual accounts and alternative measures of long-term solvency. First, the GAAP-style measures distinguish between benefits that have been earned (or accrued) to date and benefits to be earned (or accrued) in the future, whereas the alternative measures factor in all benefits to covered participants, regardless of when they are earned. Second, the alternative measures factor in taxes to be paid in

the future, whereas GAAP-style accrual accounts do not. These differences reflect the goals that the two approaches are designed to achieve, and each approach has its advantages.

a. Distinguishing Benefits Earned-to-Date from Total Benefits

Derived from traditional long-term cash-flow analyses of payments, the alternative measures are concerned with estimating the present value of all payments projected to be made for the relevant time horizon to covered participants, assuming the continuation of current statutory benefit and tax formulae.[187] Therefore, there is no reason for them to distinguish between benefits earned before or after the date of the forecast. The rules of accrual under GAAP, however, recognize liabilities only when they arise out of past transactions or events and have a sufficient probability of being paid. For obligations to make payments arising from earnings records stretched over multiple decades, GAAP-style accounting demands differentiation between benefits earned to date and benefits earned in the future.

The key question, of course, is which approach is more appropriate. In contrast to alternative measures, GAAP-style recognition principles privilege benefits earned to date. Is this privileging appropriate for the financial statements of Social Security?

The statutory requirements for private pension plans may offer some guidance. ERISA prohibits employers from reducing pension benefits that have accrued; however, employers are free to change or even eliminate the accrual of additional benefits in the future.[188] The justification of this restriction is that workers should be entitled to rely on an employer’s promises to pay benefits that have accrued. While Congress remains constitutionally free to impose retroactive reductions in the benefits of Social Security, the rules governing private employers may still provide useful normative guidance. At least for purposes of accounting recognition—where a collective estimate of the trust funds’ obligations is made—the government arguably should be held to at least as high a standard for benefit obligations as it imposes on private employers for their pension promises.

Another justification for recognizing only earned benefits is the underlying logic of GAAP recognition rules. Earned benefits are the ones that, in all likelihood, will result in future payment, and GAAP-style financial statements recognize such commitments because their principal purpose is to demonstrate the relationship between an entity’s current economic resources and its current obligations.[189] Arguably, the financial statements of Social Security should also highlight the difference between the system’s current

assets and the value of all benefits the payment of which is sufficiently probable to warrant recognition under GAAP for private firms. Under this principle, whenever the commitments become probable obligations, that moment is the moment when they should appear on the financial statements of Social Security and be included in other federal budgetary aggregates. To the extent that financial statements are intended to help monitor and control the creation of financial obligations, this feature of accrual accounting has undeniable appeal.

Finally, there is the issue of transitional costs. In reform discussions, particularly those involving the suggestion of the partial privatization of Social Security benefits, analysts sometimes overlook the accrued claims of current workers and retirees.[190] Requiring the trust funds to produce financial statements that recognize these accrued liabilities would make it much less likely that these important obligations would be ignored.

b. The Inclusion of Benefits-To-Be-Earned and Future Taxes

The case against privileging earned benefits over to-be-earned benefits and for relying on the broader alternative measures described above has two separate strains: one political and one financial.

The political argument rejects the proposition that benefits to be earned in the future are any less binding than benefits that have been earned to date.[191] Often advanced by defenders of a strong social welfare safety net,this view regards the government’s obligation to pay future generations’ obligations as indistinguishable from its commitment to pay the next check to a current retiree.[192] Sensing perhaps that distinguishing between earned benefits and benefits-to-be-earned will begin to unravel the traditional Social Security system, proponents of this view prefer alternative measures of long-term solvency that treat all benefit payments the same.[193] Irrespective of one’s views of the merits of the political prediction regarding the implications of adopting a

system of accrual accounting for Social Security, one could still object to this line of argument as unpersuasive in its claim that the government’s obligation to make payments to future generations under current statutory formulae are appropriately considered to be equivalent to the government’s obligations either to current retirees or to mid-career workers nearing retirement age.

The financial argument against privileging earned benefits proceeds from the claim that the logic of GAAP-style accrual accounts does not apply to government programs such as Social Security.[194] With private employers, GAAP is critically concerned about unfunded accrued liabilities for fear that unfunded promises will not be paid.[195] With public pension programs, on the other hand, the amount of unfunded accrued liabilities is not as important because the government’s power to tax can cover the shortfall or alter benefits. Accordingly, the alternative measures of long-term solvency are more appropriate for public pension plans because they factor in the economic resources—tax contributions in the future—that are intended to finance benefits earned to date and those earned in the future.[196] Under this reasoning, the closed-group and infinite-horizon group liability measures are better presentations of the system’s financial posture than GAAP-style accounts.[197]

In contrast to the political argument, this financial argument is well-founded and important. By reflecting future excess tax revenues, the alternative measures of long-term solvency do have some advantages. The question then is how GAAP-style accounts might be modified to incorporate these advantages while, at the same time, retaining the clarity of GAAP accrual accounting, which highlights the value of accrued liabilities.

5. Modified Accrual Accounting

Augmenting GAAP-style accrual accounting statements of Social Security to recognize as a “quasi” asset an amount equal to the present value of excess revenues to be contributed by system participants over the additional benefits that they will accrue during the balance of their working lives would resolve this dilemma. Such a system does not diminish the privileged status of earned benefits as only benefits accrued to date would appear on the liabilities side of the Social Security trust fund balance sheet, but it would reveal how much of those accrued liabilities is matched by the present value of excess contributions. The difference between the system’s accrued liabilities and the sum of its reserves and excess tax contributions could be de-

nominated the system’s implicit debt: the amount of additional government resources required to honor the system’s accrued liabilities over and above current reserves and excess future tax revenues already committed under current law.

Figure Seven presents a modified accrual balance sheet for December 31, 2002. The balance sheet’s assets consist of the value of current reserves ($1.4 trillion) plus the present value of excess future tax contributions of current participants ($2.1 trillion). The amount by which the system’s accrued liabilities ($14.0 trillion) exceed the sum of these assets represents the trust funds’ implicit debt as of December 31, 2002, and equals $10.5 trillion. This figure is the same as the closed-group liability of year-end 2002, as only the excess tax contributions of current participants, and not future participants, were included. From this balance sheet, one may readily ascertain the system’s unfunded accrued liabilities: the $12.6 trillion difference between accrued liabilities ($14 trillion) and current reserves ($1.4 trillion), which is the deficit featured in a GAAP-style accrual accounting statement. This format would also reveal the amount by which accrued liabilities are expected to be reduced by excess future taxes.

Following similar logic, one can also construct a modified income statement for the trust funds in 2002, as is done in Table Nine. The only difference between this income statement and the GAAP-style income statement presented in Table Six is the inclusion of income items reflecting the increase in projected net tax contributions ($97.1 billion) resulting principally from the addition of a new annual cohort of participants—those turning fifteen in 2002. Even with the recognition of this additional future net tax revenue, the trust funds still showed a $370.4 billion loss in 2002.

One could imagine amplifying these modified financial statements further, adding in the net tax contributions of future participants and aligning the fiscal imbalance with the infinite-horizon open-group measures. For largely pragmatic reasons, that additional step will not be taken here. First, on the practical level, we lack time series for the infinite-horizon open-group measure, and so it is not possible to construct income statements or multiple-year estimates of fiscal imbalance. Also, on a practical level, the Chief Actuary’s most recent estimate of the infinite-horizon closed-group liability is the same as the closed-group liability, so there does not seem to be much to be gained from using another measure that will generate roughly the same results.[198] Finally, projections of infinite-horizon open-group liability are inherently more speculative than closed-group estimates, which are themselves more speculative than estimates of unfunded accrued liabilities. While all

estimates of long-term solvency entail some degree of speculation and approximation, the infinite-horizon open-group measure is at the extreme.

D. Accrual Accounting and the Unified Federal Budget

Imagine, for a moment, that future trustees’ reports of the Social Security system were revised along the lines outlined above, using a modified system of accrual accounting as the principal presentation format. For the foreseeable future, the system would report annual losses on the order of several hundred billion dollars and the implicit debt of the trust funds would annually mount, both in absolute terms and probably as a fraction of current GDP. Both of these facts—annual losses and mounting implicit debt—would present major challenges for budgetary officials when the time came to aggregate the finances of Social Security with other federal budgetary accounts.

Two basic approaches could, however, accomplish this aggregation of accounts, and both would have important implications for budgetary politics. The first is a fully consolidated approach; the second a fully unconsolidated approach. The consolidated and unconsolidated approaches represent more radical alternatives than do the total-budget and on-budget aggregates that are currently used when Social Security cash-flow surpluses are combined with or kept separate from on-budget accounts. They also offer more accurate presentations of the true impact of Social Security on public finances.

There is considerable uncertainty as to how politicians would react to changes in accounting systems. Some argue that reported levels of budgetary aggregates have little effect on politicians’ substantive choices.[199] In addition, it is possible that even if trustees’ reports adopted a modified system, politicians would still rely on the system’s current cash-flow surpluses for discussing federal budgetary aggregates. Nevertheless, it is plausible that changing the public’s understanding of the financial status of the Social Security trust funds might substantially impact the manner in which the trust funds are combined with federal budgetary aggregates. This Part’s goal is to explore what that impact might be.

1. Full Consolidation Within the Federal Budget

One approach to combining accrual-based Social Security with the broader federal budget would be to establish a system of full consolidation. Under this technique, the profit or loss of the Social Security system each year would be included in the unified budget. Under this approach, Social Security would have put a $370.4 billion drag on the unified budget in 2002 (as

opposed to the $160.3 billion positive effect under current cash-flow accounting).[200] Thus, the consolidated federal deficit for 2002 would have been negative $688.6 billion (rather than negative $157 billion as reported).[201] While this deterioration of more than $500 billion is jarring, it is an accurate reflection of the amount by which the federal government’s express promises to public debt holders and implicit unfunded promises to Social Security beneficiaries increased over the course of the year. To give readers a sense of the magnitude of this change, Table Ten presents restated budgetary aggregates for the federal government from 1998 through 2002.

If one extended the logic of consolidation to the federal government’s balance sheet, one might also present the growth of the trust funds’ implicit obligations alongside the growth of the government’s debt held by the public. Once Social Security’s accrued liabilities are recognized on the government’s balance sheets, it is logical to include these obligations if one opts for a consolidated presentation of public debt. Such a presentation is offered in Figures Eight and Nine, first with debt figures measured in current dollars

and next as measured as a percentage of the GDP. These figures reveal two important points. First, the magnitude of Social Security’s implicit trust fund obligations is much greater than federal debt outstanding to the general public.[202] Second, over the past two decades, the growth in accrued liabilities of Social Security greatly surpassed increases in publicly held debt, and more than offset the small reductions in public debt outstanding from 1998 to 2001, when the total-budget accounts were in surplus.

2. An Unconsolidated Approach to Social Security Accounting

Politicians, undoubtedly, would find full budgetary consolidation of Social Security, under the terms proposed in the preceding paragraphs, an unpalatable option. Indeed, it could be politically costly to explain to the public why the levels of reported federal deficits are so much higher than previously stated. Some might also object to combining explicit and implicit debt to consolidate aggregates.[203] Accordingly, a more plausible and, in many respects, preferable solution would be to use a fully unconsolidated approach. Treating Social Security as an unconsolidated entity is, in a sense, an extension of the off-budget treatment that the system purportedly, but in fact only episodically, enjoys today.[204]

Before going further, one must understand how this fully unconsoli-dated treatment would work.

a. Budgetary Aggregates

If Social Security trust funds were treated on a fully unconsolidated basis, the annual operations of the trust funds would not be included in the federal budgetary aggregates. This approach would be similar to the Clinton Administration’s efforts in 1999 and 2000 to focus attention on what is conventionally referred to as on-budget figures.[205] There would be less incentive, however, for politicians to revert to using a consolidated approach, because as described above, the effect of consolidation would be to detract substantially from total budget aggregates.

b. Interest Payments and Other General Revenue Contributions

Under an unconsolidated approach, transactions between Social Security and other federal budget accounts would be treated in the same way as transactions between the federal government and unrelated third parties. As a result, all payments to the Social Security trust funds, including interest payments on federal bonds held by Social Security, plus any contributions to Social Security from general revenues, would count as expenses of the federal government in the year they were paid.[206] In contrast to the rhetoric of lockboxes, this approach would clarify the extent to which general revenues were actually used to enhance Social Security’s solvency.

c. Measures of Public Debt Outstanding

Under a fully unconsolidated presentation of Social Security finances, bonds held by the trust funds would be included in the calculation of total federal debt outstanding. Not only would this reflect the political reality that the federal government will undoubtedly honor its commitment to redeem these bonds, but the approach also presents a more accurate picture of the federal government’s future burden of explicit debt.[207] Figures Ten and

Eleven provide a graphic presentation of the relative size of these debt holdings over the past twenty years, and reveal the increasingly important share of public debt held by the Social Security trust funds. Figure Ten presents the series in current dollars, whereas Figure Eleven presents the series as a percentage of GDP.

d. Separate Financial Statements for Trust Funds

Finally, the Social Security trust funds would maintain their own separate financial statements—both balance sheets and income statements—prepared on the basis of modified accrual accounting. An important element of these financial statements would be the prominent reporting of annual increases in the system’s implicit debt. Table Eleven presents the summary statistics that Social Security financial statements, presented on a fully unconsolidated basis, would provide.

III. Accrual Accounting and Social Security Reform

Beyond presenting a clearer picture of the true financial posture of the Social Security trust funds, modified accrual accounting statements of the sort outlined above could have a profound impact on the public debate over Social Security reform. This Part sketches out the potentially salutary ways in which an accrual-based accounting system might affect public policy in this area. For purposes of this Part, the reader should assume that the modified system of accrual accounting is the dominant format for presenting Social Security finances. If the bleak financial presentations of Figure Twelve, rather than the current trustees’ reports, framed public discussions of Social Security finances, what would be the effect on the debate over Social Security reform?

A. Clarifying the True Nature of Social Security Finances

1. Debunking the Debilitating Myth of Current Surpluses

Perhaps the most important impact of restating the financial posture of Social Security on the basis of accrual accounting would be to alert the general public to the true financial posture of the program. Public debate would be dramatically different if it were framed by reports of annual losses on the order of several hundred billion dollars a year rather than trustees’ reports locating the trust funds’ difficulties several decades in the future. In addition, publicizing both the trust funds’ mounting accrued liabilities—in excess of $14 trillion—and the $10.5 trillion by which those liabilities exceed both current reserves and the present value of future excess taxes of current participants would likely have a profound effect on public discourse, by stiffening the resolve of politicians to address the system’s problems expeditiously and the willingness of the general public to accept at least some modicum of pain in reform proposals.

2. Focusing Public Attention on the Trust Funds’ Mounting
Implicit Debt

By recognizing the trust funds’ current obligations, a modified accrual accounting system would also focus public attention on the implicit debt that the Social Security system is imposing on future generations. Absent changes in benefit formulae or other structural reforms, future generations will have to bear these obligations through higher direct contributions to the Social Security trust funds or general revenue support.

For example, revised trustees’ reports under this new regime would presumably include projections of the future trends in implicit trust fund debt.

Figures Thirteen and Fourteen project a plausible path of growth of implicit trust fund debt in the next decade, both in absolute terms and as a percentage of GDP.[208] Also included in these figures are levels of total public debt—that is, debt held by the general public plus debt held by the Social Security trust funds. Once Social Security’s finances are restated in accordance with modified principles of accrual accounting, such comparisons between implicit trust fund debt and explicit public debt will be both inevitable and useful, as the general public will quickly learn that mounting financial commitments of Social Security are both larger and growing faster than the public debt, even once restated to include debt held by the trust funds.

3. Developing Appropriate Targets for the Implicit Debt of Social Security

A further advantage of highlighting the size of Social Security’s implicit debt and that debt’s relationship to explicit federal debt is the possibility that it would prompt a sensible national debate over the appropriate level of these obligations. While economists often speak in terms of acceptable level of explicit federal debt relative to the size of the economy,[209] similar discussions about the appropriate levels of other kinds of governmental obligations, particularly public pension obligations—at least in the United States—are rare.[210] Once the implicit debt of the Social Security system is quantified through accrual accounting, one might reasonably expect such discussions to begin. While this Article will not engage in such speculation,[211] it is useful to imagine how discussions of Social Security reform might proceed were such a consensus to emerge. Suppose, for example, that economists generally agreed that the implicit debt associated with Social Security should be limited to no more than 50% of GDP. Suppose further that a bipartisan consensus were reached that Social Security reform legislation should attempt to get the system’s implicit debt to at least 75% of GDP over the next ten years—roughly halfway to the 50% target from the current level of Social Security implicit debt, which is slightly more than 100% of

the GDP. Such an agreement would provide a baseline for evaluating various Social Security reform proposals.

Figure Fifteen illustrates this hypothetical path for reform. The line with circles reflects a ten-year path for reform necessary to achieve this goal of reducing implicit trust fund debt to 75% of the GDP at the end of ten years. If this framework for reform were adopted, one of the criteria for evaluating reform proposals would be to assess whether they reduced the system’s implicit debt within the timeframe specified. Essentially, one could use figures similar to Figure Fifteen to map the projected financial impact of competing reform proposals.

B. Enhancing the Quality of Public Debate over Reform Proposals

As compared with the seventy-five-year actuarial deficit (negative 1.92% of payroll) that currently frames most reform discussions, modified accrual accounting statements and a target path for reducing implicit trust fund debt as a percentage of GDP would offer a far preferable framework for public discussions.

A modified system of accrual accounting would make considerably more transparent how various elements of reform proposals would affect the implicit trust fund debt. Consider the five basic ways in which the financial condition of Social Security might be improved. (1) Reductions in benefits to be accrued in the future would appear initially as increases in the “quasi asset”

representing excess tax contributions on the trust fund balance sheet, and later as smaller increases in the system’s accrued liabilities as those benefits were accrued. (2) Higher payroll taxes to be imposed in the future would also appear first as increases in the “quasi asset” representing excess future tax contributions and later as greater trust fund reserves once those taxes were paid into the system. (3) Direct contributions of general revenues would be recognized in the year in which the contributions were made, presumably as increases in trust fund reserves. (4) Similarly, higher returns from trust fund reserves—through stock market or other investments—would be recognized in the year those returns were earned. (5) To the extent that reform plans included reductions in benefits accrued to date for retirees or mid-career workers, those reductions would be immediately recognized as reductions in the system’s accrued liabilities. Each of these five mechanisms for restoring Social Security solvency would have a clear and distinctive impact on accrual-based balance sheets and income statements.

1. Exposing Short-Range Cash-Flow Effects

One certain advantage of accrual accounting is that it would expose the limitations of reform proposals that rely principally on short-range cash-flow effects. For example, bringing new participants into the Social Security system always has a positive cash-flow effect in the short term because the new participants initially pay large amounts of taxes and receive few benefits. On a modified accrual-accounting basis, and in terms of economic reality, the gains from such proposals are typically much smaller and may in fact be negative, depending on the new entrants and the kinds of benefits they will enjoy. For example, pools of new low-income workers can detract from the system’s solvency in the long term, even though their participation might have a positive cash-flow effect in the short term. Whatever the long-term impact, under accrual accounting, there would be no short-term advantage in terms of either trust fund finances or federal aggregates to pursuing such strategies.

2. Recognizing the Value of Reforms with Future Cash-Flow Effects

Conversely, some reform proposals have a positive economic effect on the trust funds’ solvency, but their benefits are understated in the current accounting system because their cash-flow effects occur in the future, even beyond the seventy-five-year long-term perspective. A good example of this phenomenon can be found in the proposals of the President’s Commission on Social Security Reform. A principal feature of the Commission’s proposals was the imposition of an offset in traditional Social Security benefits for contributions made to individual accounts.[212] Under accrual accounting, such offsets would reduce the trust funds’ accrued liabilities in the year of the

contributions, whereas under traditional Social Security accounting, they would not be recognized until partially beyond the seventy-five-year projection period.

3. Flagging Reform Proposals That Expand Accrued Liabilities

A further advantage of accrual accounting is its ability to flag reform proposals that include retroactive benefit enhancements.[213] Though there is much to recommend these proposals on the merits and though the ultimate solution to the problems of Social Security will almost certainly include a combination of sticks and carrots, the current accounting system makes it too easy to slip in expensive improvements because the yardstick for evaluating reform proposals is their long-range impact on cash-flow solvency. An accrual accounting system would highlight such proposals—at least when applied on a retroactive basis—as immediate increases in the system’s accrued liabilities. An advantage of accrual accounting is that it would reflect the true costs of such reforms and facilitate a more informed debate on their merits.

4. Eliminating the Cliff Effect

A further advantage of accrual accounting is that it would eliminate the cliff effect, which plagues reform proposals evaluated solely on the basis of the actuarial deficit. The cliff effect arises because the trust funds’ tax base is much lower than projected expenditures at the end of the seventy-five-year projection period.[214] A reform proposal can eliminate the seventy-five-year actuarial deficit for the current measurement period, only to have insolvency return to the system a few years later, as the period of analysis shifts forward to include additional years of cash-flow imbalance. Proposals that bring the Social Security system’s implicit debt to a sustainable ratio to the GDP will be much more likely to achieve long-range balance than will proposals that eliminate only the seventy-five-year actuarial deficit.

5. Identifying Back-loaded Reform Proposals

Another way in which accrual accounting would improve the quality of debate over reform proposals would be by unmasking reform proposals that are substantially back-loaded. A good case in point is one of the reform options that the 1996 Advisory Council proposed. While the proposal purported to eliminate the long-range actuarial deficit (then estimated at 2.17% of payroll), the elimination was accomplished in part through a new tax to be imposed five decades in the future.[215] If this proposal were evaluated through

the lens of accrual accounting and pro forma income statements for the near term, it would be readily apparent that these future taxes did nothing to reduce the system’s accrued liabilities. The only impact would be on the amount of the system’s “quasi-asset” for excess future tax contributions and the amount of the impact would likely be smaller than the Advisory Council estimated, as the impact of the increased taxes could be offset to some degree by increases in benefits to be accrued in the future.

6. Flagging the Amount of Excess Future Taxes

A final advantage of the modified system of accrual accounting is its capacity to demonstrate the extent to which the system’s accrued liabilities are being funded with excess taxes to be charged to participants in the future. While there is nothing inherently wrong with levying on participants’ taxes that exceed the value of benefits to be accrued in the future, there is presumably some limit to the extent to which a social insurance system should rely on such excess taxes, lest public support for the system be undermined in the future. A reform proposal that relied on higher payroll taxes in the future might substantially reduce the Social Security system’s implicit debt, but it might be a proposal with significant long-term political weakness. A modified accrual accounting system would expose concerns of this sort.[216]

C. Altering the Optics of Reform Proposals

In addition to the direct benefits of accrual accounting described above, this mode of analysis could bring a number of additional—albeit somewhat more subjective—benefits to reform debates.

1. Distinguishing Accrued Rights of the Elderly and Other Participants

An essential element of accrual accounting would be the recognition of the benefits that participants have accrued to date. These liabilities would be recognized on an annual basis on the system’s income statement and reflected on the system’s balance sheet at the end of each year. In a sense, accrued benefits are privileged over benefits that accrue in the future.[217] If accrued benefits were presented in this way, long-term participants in the system—particularly retirees and near retirees—might be persuaded that their interests

would not be adversely affected by reform proposals. One could imagine policy analysts or politicians using Social Security balance sheets as a tool for persuading nervous constituencies that their expectations for retirement income will not be compromised. In other words, analysts would be able to distinguish between the fully accrued rights of retirees and the only partially accrued rights of other participants. The confusion of the interests of these two groups often complicates current discussion of reform proposals. Accrual accounting offers a possible path out of this quagmire.

2. Differentiating Prospective and Retroactive Effects

A related point is the capacity of accrual accounting to offer a new way to differentiate between the retrospective and prospective effects of reform proposals. Given the current system, changes usually apply to both previously accrued and to-be-accrued benefits. For example, the 1983 change in retirement age applied to covered workers both for benefits accrued before 1983 and for those that would accrue thereafter.[218] Similarly, enhancements in benefit provisions, such as linkage of benefit increases to productivity improvements in the 1970s, also apply both retroactively and prospectively.[219] One of the advantages of accrual accounting is that it offers reformers more options in the way they structure changes to the system, and having more options may be necessary to develop a politically viable reform package.

a. The Possibility of Prospective Benefit Cuts

Consider if benefit cuts were imposed, not on an all-or-nothing basis, but rather only on benefits that would accrue in the future. For example, imagine that, rather than switching over fully to cost-of-living indexation of initial benefit levels, a reform proposal were structured to preserve the existing productivity-adjusted formulas for accrued benefits and establish cost-of-living indexing only for benefits that accrue in the future. Such a reform package could be characterized as honoring existing commitments and expectations while still offering immediate improvements in the system’s financial posture through the reduction of the rate of accrual of benefits in all future years. This option is both preferable to and more equitable than the current practice of imposing benefit cuts with effective dates that are postponed for years but then imposed on a fully retroactive basis.

b. More Modest Methods of Enhancing Benefits

A similar point could be made about benefit enhancements. Accrual accounting would invite benefit improvements to be imposed only with respect to benefits accrued in the future. This option would allow politicians a means to address the significant weaknesses in the Social Security system without imposing on it the substantial costs of fully retroactive application.

3. Changing Expectations of the General Public

The general point that runs through all of these examples is that accrual accounting offers a way to change the expectations of the general public in a subtle, but important, way. Through a complicated combination of factors, American workers have been encouraged to think of themselves as being entitled to the level of retirement benefits authorized under current law, even though their retirement may not occur for many years or even many decades.[220] The implicit understanding seems to be that, as long as participants make their payroll-tax contributions as currently structured, they are entitled to receive their benefits as scheduled. The current system of Social Security accounting reinforces this understanding, because it is designed to measure the balance between projected benefits and projected taxes. This linkage is even more apparent in the personal statements that the Social Security Administration sends out to all participants, estimating their projected level of future benefits on the assumption that they maintain their current level of employment until retirement. Lobbyists and politicians regularly encourage this understanding by opposing any change in promised levels of benefits. Accrual accounting offers a different perspective on Social Security benefits. It privileges a subset of benefits—those that have already accrued—and invites participants to ratchet downward their sense of entitlement to benefits that have not yet accrued.

D. Accrual Accounting and Individual Accounts

The reforms proposed in this Article could also affect the policy debate surrounding the creation of individual accounts in a number of ways, although the overall direction of the effects is ambiguous.

1. Posing the Problem of Transition Costs

Perhaps the most significant effect would be to make the magnitude of unfunded accrued liabilities of the trust funds more prominent. Some proponents of individual accounts emphasize the higher returns that participants could realize on these accounts without addressing the question of how the un-

funded liabilities of the current system would be addressed. Indeed, one of the problems with the public debate over individual accounts is the difficulty of comparing the rates of return under traditional Social Security, which address a portion of the system’s unfunded accrued liabilities, and the rates of return on individual account proposals, which leave the question of unfunded transition costs unresolved. If individual account plans were vetted through the screen of accrual accounting, the issue of accrued unfunded liabilities would be unavoidable.

2. Eliminating Budgetary Anomalies

Accrual accounting would also eliminate a number of budgetary accounting anomalies that can cast individual account proposals and certain other reforms in an unfavorable light. Under current accounting rules, funds transferred to individual accounts are treated as expenditures, as are investments of trust fund assets in any sort of financial asset other than government bonds. These conventions mean that the creation of individual accounts and proposals to invest trust fund assets in the stock market all give rise to the appearance of budgetary costs. Given current budgetary imperatives, these anomalies make it difficult for politicians to adopt such reforms, regardless of their substantive merit. If the Social Security trust funds were accounted for on the fully unconsolidated basis outlined above, anomalies of this sort would disappear. Neither the creation of individual accounts nor the transfer of trust fund assets into the stock market would have a budgetary effect.

3. Comparing Traditional Benefits and Individual Accounts

A final benefit of accrual accounting would be the enhancement of the comparability of traditional Social Security retirement benefits and individual accounts. A problem in the current debate over individual accounts is the difficulty that both experts and the general public face when comparing the relative merits of these two benefit structures. Individual accounts are usually valued in terms of the current amount of assets in a particular individual’s account. Traditional benefits, by contrast, are typically described as future payments, often made on a monthly basis, beginning at a point in the future, for example, the participant’s sixty-fifth birthday, and continuing for the remainder of the combined lives of the participant and the participant’s spouse with various adjustments for inflation as well as spousal and survivors’ benefits. To a large degree, proponents of individual accounts are asking the public to accept the first kind of benefit as a substitute for the second. Comparing these two is extraordinarily difficult. If we began to evaluate Social Security benefits on an accrual basis, however, restating the benefits of individuals as the net present value of expected benefits for each individual would be a fairly simple step. The widespread disclosure of such individual restatements—perhaps as part of the annual statements sent to each participant—could

greatly improve the public’s understanding of traditional Social Security benefits and strengthen public debate over the relative merits of individual accounts.

E. Social Security Reform, the Federal Fisc, and National Savings

Adopting a system of accrual accounting for Social Security and implementing this system on a fully unconsolidated basis could also have important benefits for federal budgetary politics and, by implication, for national savings. While the gains to be had in this area are highly speculative, there are good reasons to believe that the impact would be positive.

1. Budgetary Effects of Fully Unconsolidated Treatment

An initial impact of moving the Social Security trust funds to a fully unconsolidated basis would be to shift the attention of politicians and analysts to what is now referred to as the on-budget surplus or deficit. Unlike the current budgetary treatment, which encourages politicians to resort to total-budget aggregates, eliding distinctions between on-budget and off-budget accounts, this Article’s proposed reforms would discourage the combination of annual trust fund results, because on an accrual-accounting basis the trust funds experience substantial annual losses. By keeping national attention on on-budget aggregates, this proposal would increase the likelihood that the on-budget accounts remain in or near balance.[221] In effect, this would mean that cash-flow surpluses on Social Security trust funds would be used to decrease the amount of public debt held by the general public and thereby increase national savings.

2. Budgetary Effects of General Revenue Contributions

This Article’s proposals would also untangle the budgetary confusions that surround proposals to allocate additional general revenues to the Social Security system. As explained above, one of the great confusions of the lockbox debates of 1999 and 2000 was the question of how to characterize and account for general revenues committed to Social Security.[222] Under this Article’s proposals, their accounting would be quite simple. Any contribution to the trust funds would be treated as an on-budget expense in the year the contribution was made. Were on-budget accounts otherwise in balance for that year, the contribution would move the on-budget accounts into deficit. If Congress wished to make such a contribution without borrowing additional

funds from the general public, new general revenues would need to be raised or other expenses reduced. This treatment would, in my view, impose appropriate discipline on the federal budget process and discourage the federal government from honoring its commitments to Social Security simply through the issuance of new debt to the general public.

3. Tracking Gross Public Debt and Unfunded Trust Fund Liabilities

A final advantage of this Article’s proposal is that it would encourage the presentation of a more accurate picture of the overall size of federal obligations. The approach would highlight the gross amount of public debt—that is, the combination of debt held by the general public and debt held by the trust funds. This figure would be a more accurate representation of the debt burden that taxpayers will have to shoulder in the future and therefore the one that politicians and analysts should monitor. Of course, accrual accounting would also highlight the implicit trust fund debt of the Social Security system, another form of public obligation passed on to future generations and one with significant implications for overall national savings. Bringing these combined measures of public obligation into the limelight would further enhance public debate on the issues.

F. Substantive Implications of Accrual Accounting

Some readers will no doubt be curious about the substantive implications of restating Social Security finances on the basis of accrual accounting. The impact here is ambiguous. Consider, for example, the hot-button issue of privatization. Defenders of traditional Social Security benefits recoil from estimates of the system’s accrued liabilities out of fear that reports of unfunded obligations in excess of $10 trillion would weaken political support for the system. On the other hand, supporters of individual accounts resist references to accrued liabilities because they would highlight the transition costs of moving to a fully privatized Social Security system. It is not clear which of these groups would fare better under an accrual-accounting system, but the general public would be in a better position to evaluate the merits of the traditional system as compared to those of a system of privatized accounts if the comparison were made in terms of accrual accounting.

Another important substantive dimension of Social Security is the system’s role in redistributing wealth from rich to poor and from young to old. Some fear that accrual accounting would more clearly expose Social Security’s redistributive elements.[223] Conceivably, such information could diminish public support for Social Security and its redistributive role. Recent studies,

however, suggest that many younger workers believe that they will never receive any Social Security benefits.[224] Presumably, these participants think that they get no benefit from their annual Social Security contributions. A fully implemented system of accrual accounting would reveal to these individuals that their annual contributions are generating real expected benefits in terms of retirement security, as well as life insurance and disability benefits. Perhaps for these individuals, accrual accounting would increase support for the system.

Finally, objections to accrual accounting on the basis of substantive concerns are, in the end, deeply undemocratic. Such objections reduce to arguments that we should not present the finances of the Social Security system in the most realistic manner because the general public would react badly. The moral weakness of such claims aside, intentional obfuscation of government finances cannot be the right way to build sustained public support for such an important social insurance program.

IV. Postscript on Normative Baselines

The accrual accounting proposal for Social Security outlined in this Article generates surprisingly impassioned responses in some circles.[225] The source of this passion is not always clear. Sometimes, the criticism seems to be based on an intuition that cash-flow accounting is the only appropriate method for recording government programs.[226] Other times, defenders of the status quo stress that Social Security is a social insurance program and assert that this characterization constitutes a complete explanation of current accounting practices.[227] This postscript responds to these reactions, and explains why this Article’s recommendations are consistent with recent trends in government accounting, including recent reforms in federal accounting standards for social insurance programs.

Box Five
The Normative Implications of Accrual Accounting: Funding and Profitability

Once the financial statements of Social Security are restated in terms of accrual accounting, one might reasonably inquire whether this accounting format has normative implications for the system’s funding or profitability. In particular, does accrual accounting imply that Social Security should be fully funded—that is, that the system’s accrued liabilities should not exceed its assets—or that its annual income statement should ordinarily show a profit or at least the absence of substantial losses? The answer to both of these questions is no.

Although this is principally a question of public finance, there is no reason to suppose that a public pension system needs necessarily to be fully pre-funded or even that its accrued liabilities need not exceed the sum of current reserves plus the net present value of excess taxes for current participants—that is, that the system’s implicit debt be zero. What is important, however, is that the system’s level of implicit debt not be allowed to grow in an unbounded manner in comparison to overall growth in the economy. Ideally, that level of this implicit debt should stay within some target ratio to GDP. Perhaps the current level of implicit Social Security debt, equal to slightly more than 100% of GDP is appropriate, although I expect that most experts would prefer a lower target, such 75% or 50% of GDP, targets illustrated in Figure Fifteen.

The appropriate annual level of profit or loss for a Social Security system following principles of accrual accounting follows from the target level of implicit debt for the system. If the system were at the target level, then it would be acceptable for the system to report annual losses as long as those losses did not cause the system’s implicit debt to grow faster than the overall economy. If, on the other hand, the system’s implicit debt were greater than target levels, better performance on annual income statements would be appropriate. Even under these conditions, however, annual profitability would not be required as the implicit debt to GDP ratio of the trust funds will decline as long as the system’s implicit debt grows more slowly than the overall economy. For example, the hypothetical reform path illustrated in Figure Fifteen implies adjusted annual losses for the system on the order of $200 billion over the next decade. Even with this level of annual loss, the system’s implicit debt will gradually decline as a percentage of GDP over the coming decade.

It then presents several additional normative perspectives that support the claim that accrual accounting is the most nearly accurate way to present the finances of the Social Security system. First, it reviews how the problem of unfunded public pension plans is generally discussed in

the international context. When multinational organizations, such as the International Monetary Fund (IMF), and other analysts examine the financial posture of public pension programs in other countries, they generally speak in terms of accrued liabilities as well as the sort of long-term cash-flow projections that dominate the trustees’ reports.[228] While various measures of accrued liabilities are employed, accrual accounting is the principal metric against which the solvency of public pension schemes is tested. Second, this Part offers a brief sketch of how public policy analysts value Social Security benefits in a variety of other academic settings. When trying to assess the economic impact of Social Security, economists regularly use a form of accrual accounting to estimate the value of benefits.[229] If these experts believe that accrual accounting is the best way to estimate the value of participants’ benefits in Social Security, then they should see that accrual accounting is also the best way to estimate the liabilities associated with the obligations of the trust funds to pay those benefits.[230]

A. Accrual Accounting and the Federal Government

Although cash-flow accounting is the norm in the federal budget, the notion that elements of accrual accounting might be grafted onto federal accounting standards is not new,[231] and examples of accrual accounting in current federal budgeting and accounting standards are increasingly common.

1. Accrual Accounting in General

One prominent illustration of accrual accounting in the federal budget is the Federal Credit Reform Act of 1990 (FCRA),[232] which established accrual accounting for a wide range of federal programs, including loan and credit-guarantee programs.[233] With the passage of the FCRA, Congress recognized

that, for some kinds of government programs, cash-flow accounting offers a misleading picture of true government costs. For example, with a loan program, cash outflows in the year the loans are made tend to overstate the cost to the federal government because many loans will be repaid in future years. Conversely, guarantee programs may have no cash outflows in the year a guarantee is made, or even positive cash flows, if a guarantee fee is charged, even though the issuance of a guarantee can represent a significant liability for the government because payments may have to be made in the future when the guarantee comes due. The FCRA requires the government to recognize the expected cost of government credit programs in the year in which the obligations are incurred.[234] Thus, the FCRA mandates accrual, as opposed to cash-flow, accounting for an important segment of the federal budget.

Federal insurance programs, such as deposit insurance or flood insurance, are expressly exempted from the coverage of the FCRA.[235] Nevertheless, the accounting challenges of public insurance programs are quite similar to those of credit programs. When underwriting insurance, the government receives payments in the current period in exchange for a commitment to shoulder costs in the future. Accrual accounting is a natural way to account for these obligations as they arise. As a result, over the past fourteen years, a number of government studies have called for the extension of accrual accounting concepts to this area. For example, in a 1997 report, the GAO explored the extension of accrual accounting treatments to federal insurance programs, such as federal deposit insurance and other insurance programs run by the federal government, but not to Social Security.[236] The GAO generally endorsed such an expansion, with the caveats that, in many areas, the development of accrual accounting systems would be complex and that, as an initial matter, supplemental reporting of risk estimates should be undertaken.[237] Representatives of the GAO recently reiterated their support for this expansion of accrual accounting,[238] and a proposal to account for the retirement benefits of federal employees on an accrual accounting basis is currently being debated in Washington.[239]

Accrual accounting is not inherently inappropriate for government programs.[240]

2. Accrual Accounting for Social Insurance

As a social insurance program, Social Security has typically been excluded from past proposals for accrual accounting within the federal government. Government accountants, however, have addressed accounting statements for social insurance in a recent initiative. This initiative came from the Federal Accounting Standards Advisory Board (FASAB)—the board responsible for developing generally accepted accounting practices for federal entities.[241] In the mid-1990s, FASAB began an extensive review of the appropriate accounting treatment of social insurance, paying particular attention to Social Security, the federal government’s largest social insurance program. The process included a series of exposure drafts and public comments, culminating in the August 1999 Statement on Social Insurance,[242] which established important new standards of disclosure for social insurance programs, including Social Security.

Although little known outside of government accounting circles, the FASAB Statement on Social Insurance represents an extensive and sophisticated consideration of the special problems of accounting for social insurance. While some participants in the FASAB process opposed any use of accrual accounting in financial statements for social insurance programs,[243] others argued for reforms that are quite similar in spirit to, albeit less extensive than, this Article’s proposal.[244] The arguments advanced in favor of recognizing some form of accrued liability on the balance sheets of Social Security were quite similar to the ones made in this Article,[245] and they stressed the fact that existing financial statements are “inherently misleading” because they “fail to quantify the size of the promise that is continuously being made and on which people are being told they can rely.”[246] Both to be consistent with generally accepted accounting standards for the private sector

and to give the public a more nearly accurate picture of the financial posture of the Social Security program, proponents argued, Social Security financial statements should include some actuarial estimate of the future value of benefits.

After extensive debate, FASAB crafted a compromise between a form of accrual accounting and current practice. The final FASAB Statement on Social Insurance mandates that financial statements of government insurance programs include an elaborate system of supplementary information, known as Required Supplementary Stewardship Information (RSSI), which includes specific disclosures about the actuarial value of future benefits as well as a substantial amount of additional material about program sustainability.[247] With respect to Social Security, FASAB rules require disclosure of the actuarial present value of all future benefits payable to participants eligible to receive retirement benefits (those sixty-two years of age or older).[248] The Social Security RSSI also must include separate disclosures of the actuarial present value of future benefits to be paid and taxes to be received from those currently in the system (those fifteen to sixty-one years old) and those not yet in the system but projected to join the system over the next seventy-five years.[249] Taken together, these numbers sum to what was described above as the open-group liability of the system, and with the component line-item entries, users can calculate a number quite similar to the closed-group liability measure.[250] In the summer of 2003, FASAB revised its rules governing social insurance to place the information in a new Statement on Social Insurance, which will have greater prominence than the earlier RSSI format.[251]

As the FASAB establishes generally accepted accounting standards for government entities, its statement on social insurance is binding on the Social Security Administration. Indeed, if one consults the annual financial statements of the Social Security Administration—as opposed to the trustees’ annual reports—one can find the supplementary statements of net present values of future benefits and taxes clearly disclosed along the lines FASAB requires.[252] For example, the most recent report estimates the present value of benefits payable to participants sixty-two years old or older to have been $4.4 trillion as of January 2, 2002.[253] Curiously, budgetary presentations of the OMB, though purporting to comply with FASAB standards, do

not include RSSI information for Social Security or net present values of benefits and taxes.[254]

While the FASAB Statement on Social Insurance clearly does not go as far as the reforms this Article proposes, its provisions indicate that applying accrual accounting concepts to Social Security is not an entirely heretical idea. In the debate leading up to the adoption of the statement, industry experts made arguments similar to the ones advanced in this Article.[255] When FASAB began the rule-making process that ultimately moved the supplementary information to an earlier section of agency financial statements, it emphasized the importance of these disclosures.[256] Reading between the lines, one senses that the Board’s unwillingness to advance reforms even closer to those proposed herein is dictated more by political resistance in Washington than by a conviction that a more nearly complete system of accrual accounting for Social Security would not be appropriate.

More important for purposes of this Article’s argument is that the FASAB statement is based on a normative vision of the purpose of governmental accounting practices that is quite similar to the one described here. Drawing on its prior concept release on the Objectives of Federal Financial Reporting, the Board emphasized the relevance of the nation’s financial condition for the financial statements of governmental entities underwriting social insurance. The Board called for “[F]ederal financial reporting [to] provide information that helps the reader to determine whether the government’s financial position improved or deteriorated over the period.”[257]

This Article’s argument is that the current accounting presentation of Social Security does a very poor job in these dimensions and that the proposed alternative approach would represent a substantial improvement. In particular, the current financial statements of the trust funds are ineffective in explaining whether the system’s financial position improved or deteriorated during a particular year.[258] Moreover, the absence of any measure of the system’s accrued liabilities makes it quite difficult to tell whether budgetary resources will be able to sustain the current level of promised benefits.

In conclusion, Social Security’s status as a social insurance program is not a strong argument against reflecting the system’s mounting liabilities in its financial statements. Indeed, under generally accepted accounting standards for federal entities, these liabilities must currently be reported in supplementary notes, and a number of commentators have correctly argued that these liabilities should also be reflected on the balance sheets of social insurance programs.[259] In other words, the experts on government accounting have adopted an accounting standard that is substantially different from the one that the trustees follow in their annual reports and that dominates public debate over Social Security financing. There is strong intellectual support within the accounting community for the kinds of reforms proposed in this Article.

B. The Estimate of Public Pension Plan Obligations in Other Contexts

Another way to approach the question of how the obligations of the Social Security system should best be estimated is to consider how the issue is addressed in other contexts. As explained below, public policy analysts—principally economists—often need to calculate the size of public-pension-plan obligations. This Part reviews two prominent illustrations. The first is the public-finance literature dealing with the size of public-pension-plan obligations in various countries—often, but not exclusively, developing countries. The second context is a separate set of economic writings in which economists attempt to value Social Security benefits of workers, typically either to assess the impact of Social Security on other forms of savings or to present a full picture of household wealth. In both contexts, the standard approach is to estimate the present value of benefits to be paid in the future. Sometimes the calculation presents a net-present-value figure that includes both accrued and to-be-accrued benefits minus taxes to be paid. Increasingly, however, analysts are employing a benefits-accrued-to-date formulation similar to unfunded accrued liabilities measures that GAAP-style accrual accounting would highlight.

While these analogies are not directly related to financial accounting or government budgetary issues, the methodologies employed in these other areas are relevant to this Article’s arguments. In all of the following examples, analysts are trying to estimate the economic reality of public pension obligations—whether in terms of the obligation they impose on the public fisc or in terms of the amount of value they add to individual wealth. In these contexts, analysts invariably resort to net-present-value calculations, often limited to benefits accrued to date. Within the literature, these measures have emerged as the most appropriate way to value Social Security obligations. That the traditional system of accounting for Social Security financing neglects comparable measures is noteworthy and deeply troubling.

1. Measures of National Public-Pension-Plan Obligations

Over the past decade, as the crisis of Social Security financing has emerged as a major issue of public policy in the United States, a similar debate has emerged in a number of other countries, where the problem of unfunded public pension plans and often the private-pension-plan system is even more severe than in the United States. Many countries with the most seriously underfunded pension plans are in the developing world; other industrialized nations are not, however, devoid of serious deficits. Since the mid-1990s, a number of economists have written about the problem of underfunded public pension plans and have offered a variety of ways to address distressed systems, ranging from privatization to more incremental solutions. What is important about this literature is not the specific reforms advocated but, rather, the manner in which analysts quantify the magnitude of underfunded pension obligations in various countries and then present cross-country comparisons of pension underfunding.

For many years, the leading article on this subject was one written by two economists with the International Monetary Fund: Sheetal K. Chand and Albert Jaeger. Their 1996 working paper proposed techniques for estimating the public-pension-plan liabilities and then applied the techniques to the public systems of eight industrialized nations, including the United States.[260] The approach begins with a measure of accrued pension obligations for all pensioners and current workers.[261] These accrued liabilities, in Chand and Jaeger’s terminology, constitute “recognition bonds”—the amount it would cost a government to terminate its public pension program (analogous to the Social Security system’s maximum termination cost).[262] The Chand and Jaeger framework continues by adding benefits that will accrue in the future (to generate a “gross pension liability” figure) and then netting off projected contributions (to generate a “net pension liability” figure analogous to the open-group liability number discussed above). The balance of the Chand and Jaeger analysis relies principally on the net-pension-liability figure to evaluate various reform proposals.

As a seminal work on public pension financing, the Chand and Jaeger paper is an important precedent for evaluating the appropriateness of the traditional approach to Social Security financial statements. While elements of the Chand and Jaeger framework overlap with aspects of long-range financing estimations included in the annual trustees’ reports, their organizing principles are strikingly different. Chand and Jaeger’s approach is squarely grounded in present value calculations of future benefits, and it begins with a measure of the net present value of benefits accrued to date of the

sort this Article proposes. Subsequent studies of public pension systems routinely adopt a similar perspective.[263] The trustees’ reports traditionally have been wholly devoid of net present value calculations, and even the 2003 report includes these measures in only a limited manner, located some sixty pages into the document and entirely absent from introductory materials.[264] Not surprisingly, general press coverage of the 2003 trustees’ report made no mention of these measures of insolvency.[265]

Another important aspect of the public-finance literature on unfunded pension obligations is the practice of comparing the size of these liabilities to the GDP of particular countries and then making explicit comparisons of this ratio with the ratio of traditional public debt to GDP. For example, in the Chand and Jaeger paper, the level of accrued liabilities of the U.S. Social Security system is reported as 108.3% of GDP in 1990 (31.7% representing obligations to retirees and 76.6% representing accrued obligations to workers).[266] Elsewhere, the paper aggregates the U.S. government’s net public pension plan obligations in 1990 (25.7% of GDP) with its other public debt (63.3% of GDP) to come up with a “combined” net debt liability (89.0% of GDP). Other writers in the literature use a similar approach.[267] As explained

above, one of the principal advantages of recognizing the actual present value of unfunded public pension obligations is that doing so facilitates comparisons of this sort and, presumably, better-informed discussions of public policy. With the traditional approach to Social Security accounting, such comparisons are extremely difficult. The system’s failure to report net public pension liabilities as a percentage of GDP is, at a minimum, curious, because when analysts want to compare the financial status of the public pension systems of other countries to that of the United States, this is the measure typically employed.

Within the community of public economists, the principal issue of disagreement is not whether a country’s unfunded pension liability should be measured on a present-value basis but rather which kind of present-value measure should be used. As explained above, the initial Chand and Jaeger paper used a net-pension-liability measure, decomposed into accrued and to-be-accrued components. This is reminiscent of the supplementary materials that FASAB requires, but it more sharply distinguishes accrued obligations to current workers and is substantially more informative than the single open-group liability figure that U.S. officials tend to report when required to estimate Social Security’s unfunded obligations.[268] Within the community of policy analysts, however, there are those who think that the Chand and Jaeger net-pension-liability measure should be dropped as the principal summary statistic for measuring a country’s unfunded public pension obligations and replaced with a figure that estimates the country’s unfunded accrued obligations to date—that is, the measure of unfunded accrued liabilities highlighted in GAAP-style accrual accounting.

A forceful advocate of this view is the U.K.’s Richard Disney, who writes about the measures of pension liabilities in the context of the European Union, where the debt burden of member states is an important issue.[269] He summarizes his argument as follows:

The views expressed in this excerpt closely track those of this Article, particularly in their recognition that the failure to acknowledge the magnitude of accrued pension obligations expressly biases consideration of certain reform proposals, such as pre-funded individual accounts.[271]

In short, the financial statements of the Social Security trust funds deviate from emerging international standards for the evaluation of the solvency of public pension plans. In addition, an important camp in these debates, epitomized by Disney, advocates solvency measures strikingly similar to the ones this Article proposes.

2. Economic Estimates of the Value of Social Security Benefits

Yet another way to estimate the extent of Social Security obligations is to consider the value of benefits due to participants. After all, every liability of the Social Security system reflects an asset of a participant or beneficiary. If the system truly had no outstanding liabilities—the fiction upon which the financial statements of the system now rest—then the interests of participants and beneficiaries must have no true value. This is clearly not the case. As a matter of political reality, Americans have a strong sense of entitlement to their Social Security benefits.[272] More importantly, when economists model consumer behavior, they routinely classify Social Security benefits as assets

and estimate their value based on the present value of expected benefits.[273] If the economists are right that Social Security benefits are best characterized as financial assets of individual participants, then the obligations of the Social Security system to honor those benefits are best characterized as liabilities of the system.

Examples of economic valuations of Social Security benefits for individuals abound, so this Part simply summarizes some prominent examples. In all cases, the goal of the analyst was to estimate the true economic effect of retirement benefits and the analyst employed some sort of discounted-value technique, typically using projections based on actual or assumed contributions to the Social Security system.[274]

a. Life Cycle Savings Literature

A familiar illustration of this approach to valuing Social Security benefits is a series of papers in which economists attempt to estimate the effect of Social Security benefits on individual savings. Many economists believe that public pension programs such as Social Security reduce other forms of savings and a number of economists have done empirical work exploring the relationship between Social Security benefits and savings. In a recent survey of literature on the subject, the CBO described the basic research methodology for the largest group of these studies:

This approach is telling. When trying to estimate the significance of Social Security for individual behavior, economists routinely invoke discounted cash flow analyses to estimate the Social Security wealth of individual recipients. Their regression models are exploring the relationship between this asset—Social Security wealth—and other forms of private savings.

If one were to analogize this methodology to the earlier discussion of Social Security’s unfunded obligations, the approach used in these studies is conceptually similar to the closed-group-liability measure, which is also the measure of implicit trust debt reflected in the modified accrual accounting approach. The study is limited to current participants in the system, and combines the present value of accrued and to-be-accrued benefits and then deducts the present value of to-be-contributed pay-roll taxes. If one were to aggregate this measure of Social Security wealth over all current participants and beneficiaries—that is current worker, retirees, and their beneficiaries—one should in theory generate a level of assets equal to the trust funds’ closed-group liability or implicit trust fund debt ($10.5 trillion as of December 31, 2002).[276] If, as this literature implies, Social Security wealth is an appropriate way to estimate the value of Social Security benefits to current participants and beneficiaries, then why is a comparable methodology not appropriate for the liabilities of the Social Security trust funds?

b. Household Wealth Literature

A separate body of economic literature explores the composition of individual wealth in the United States. A prominent example of this literature is the 1992 Health and Retirement Study of a nationally representative sample of households, which includes considerable information on Social Security benefits.[277] The study presents several different valuations of the Social Security wealth of study participants. All of the study’s valuations represent the present value of expected Social Security benefits discounted by an appropriate interest rate and adjusted for the life expectancies of beneficiaries. The study’s first valuation technique limits itself to the present value of benefits that individuals have accrued as of the date of the study.[278] Two subsequent estimates represent the value of the participants’ total projected

benefits—either to the age of sixty-two or to the individuals’ normal retirement age.

When economists want to explore the wealth of individual households—for example, to consider variation in wealth across income levels—they routinely rely on data sources such as the 1992 Health and Retirement Study, and they routinely include measures of Social Security wealth as important components of overall household wealth. Typically, researchers take one of two approaches to estimate the value of Social Security wealth.

First, some economists use a measure of Social Security wealth that is similar to the one used in the life-cycle savings literature—that is, a measure based on the present value of accrued and to-be-accrued benefits minus the present value of to-be-paid payroll taxes.[279] As explained above, this approach is analytically similar to the closed-group liability figure and the implicit trust fund debt measures of modified accrual accounting. It includes within the definition of Social Security wealth the present value of accrued benefits plus the net present value of benefits expected to accrue in the future and payroll taxes to be paid in the future.[280]

A second approach is to estimate Social Security wealth based solely on accrued benefits to date.[281] These estimates of the value of Social Security benefits are analogous to GAAP-style accrual accounting and are sometimes used to estimate both asset values for accrued Social Security benefits and also annual Social Security accruals as a component of overall household income.[282]

These surveys on household wealth are noteworthy in two respects. First, they demonstrate that economists clearly regard Social Security benefits as financial assets that are appropriately combined with other assets, such as private pensions, bank accounting, and home ownership. Second, the technique these economists use to value Social Security benefits is highly comparable to the valuation techniques this Article recommends for valuing Social Security trust funds liabilities. A review of these and many other economic studies [283] reveal that it is common practice for economists to attribute real eco-

nomic value to Social Security benefits and to use net present value techniques to estimate those values. Again, the question for defenders of current Social Security accounting practices is why similar techniques are not used to estimate the liabilities of the Social Security trust funds.

[*] Finn M. W. Caspersen & Household International Professor of Law, Harvard Law School. M.B.A. & J.D., Harvard University, 1982; A.B., Brown University, 1976. E-mail: hjackson@law.harvard.edu. I would like to thank Robert Clark, Peter Diamond, Elizabeth Garrett, Michael Graetz, Daniel Halperin, John Langbein, Jeffrey Liebman, Elena Kagan, Louis Kaplow, Maya MacGuineas, Olivia Mitchell, Peter Orszag, Robert Pozen, Mark Roe, Sylvester Schieber, Dan Shaviro, Kent Smetters, A. Haeworth Robinson, Sheila Weinberg, and David Wilcox for helpful comments and suggestions, as well as participants in the Law & Economics Seminar at Harvard Law School, the Shirley A. Webster Lecture at the University of Iowa School of Law, the Regulatory Policy Seminar at the Harvard Kennedy School of Government, and the Stanford Institute for Economic Policy Conference on U.S. Budget Policy and Practice. Steven Goss and his staff at the Social Security Administration’s Office of the Chief Actuary were extremely generous in supplying data for this Article. Tony Feuerstein, Charles Gray, and Wang Xin provided valuable research assistance. The Olin Center for Law, Economics and Business and the Harvard Law School Summer Research Program provided funding for research on this Article. An abbreviated and simplified version of some of the ideas presented here appeared in Howell Jackson, It’s Even Worse Than You Think, N.Y. Times, Oct. 9, 2003, at A37. A .pdf version of this Article is available at http://www.law.harvard.edu/students/orgs/jol/.
[1] Compare Felix Rohatyn, Free, Wealthy & Fair, Wall St. J., Nov. 11, 2003, at A18 (“Our dependence on foreign capital and our budget deficits must be reduced by greater fiscal discipline at home including the politically painful reform of entitlements such as Social Security.”), with Robert M. Ball, Social Security Needs Tweaking, Not ‘Reform,’ Boston Globe, Oct. 11, 2003, at A19 (“Social Security doesn’t need to be ‘reformed’ because it hasn’t failed. The system just needs some timely maintenance work.”).
[2] Two prominent reform proposals, discussed at various points in this Article, appeared in 1994–1996 Advisory Council on Soc. Security, Report of the 1994–1996 Advisory Council on Social Security: Findings and Recommendations (1997) [hereinafter 1994–1996 Advisory Council Report] and The President’s Comm’n to Strengthen Soc. Security, Report of the President’s Commission: Strengthening Social Security and Creating Personal Wealth for All Americans (2001) [hereinafter Bush Commission Report]. For an overview of various approaches to Social Security reform, see generally Gen. Accounting Office, Social Security: Different Approaches for Addressing Program Solvency (July 1998); see also Kathryn L. Moore, Privatization of Social Security: Misguided Reform, 71 Temp. L. Rev. 131, 139 (1998).
[3] For an overview of the political process that led up to 1983 reforms, which included a combination of both benefit reductions and tax increases, see generally Paul Light, Artful Work: The Politics of Social Security Reform (1985).
[4] See John Lloyd, ‘Sans Nothing,’ Fin. Times, Apr. 19, 2003, at P1 (noting the political power of elderly voters and the organizations that represent them).
[5] See, e.g., Susan Page, Social Security Debate May Be Ready to Ignite, USA Today, Dec. 3, 2002, at A11.
[6] See, e.g., Claudia Deane & Dan Balz, GOP Puts Stock in ‘Investor Class,’ Wash. Post, Oct., 27, 2003, at A1 (“Bush’s plan for partial Social Security privatization, for example, was blunted by the drop in the market, and many congressional Republicans remain wary of promoting the idea . . . .”); Janet Kidd Stewart, Jilted Shareholders Rethink Faith in Stocks, Chi. Trib., Sept. 30, 2001, at 1.
[7] See, e.g., Kathy M. Kristof, Averting the Crisis in Social Security, L.A. Times, Mar. 30, 2003, at C3 (citing most recent Trustees Report for information on financial condition of Social Security); John Attarian, Time is Running Out to Fix Medicare, Social Security, Detroit News, June 26, 2003, at 17 (same).
[8] See 2003 Trs. of the Fed. Old-Age & Survivors Ins. & Disability Ins. Trust Funds Ann. Rep. 2 [hereinafter 2003 Trustees Report]. See infra text accompanying notes 26–31.
[9] See, e.g., Allan Sloan, Washington Talks a Blue Streak As Social Security’s Red Ink Runs, Wash. Post, Mar. 25, 2003, at E3 (citing 2003 Trustees Report for level of current reserves and annual cash flow surpluses).
[10] See infra text accompanying notes 26–38, note 80.
[11] See infra text accompanying notes 159–165.
[12] See infra text accompanying notes 152–158.
[13] See infra text accompanying notes 152–158.
[14] See infra Part II.C.5.
[15] See infra text accompanying notes 128–129.
[16] See infra text accompanying note 86. Cf. Robert C. Pozen, The Virtues of Moving From Cash to Accrual Accounting for Social Security, 41 Harv. J. on Legis. 199 (2004).
[17] See infra text accompanying notes 87–101.
[18] This Article is not, moreover, the first to explore this territory. Several years ago, the accounting oversight body responsible for developing generally accepted accounting practices for governmental entities adopted new rules that require social-insurance programs, such as Social Security, to disclose certain accrual-accounting elements similar to, though not nearly as extensive as, the ones recommended in this Article. See infra text accompanying notes 242–257. The logic underlying these reforms in government accounting supports the proposals offered here.
[19] See Eric M. Patashnik, Putting Trust in the US Budget 72 (2000) (describing benefits increases and payroll tax reduction designed to reduce size of projected reserves).
[20] See Light, supra note 3, at 33–57; see also Merton C. Bernstein & Joan Brodshaug Bernstein, Social Security: The System that Works 34–35 (1988).
[21] As described below, the federal budget is increasingly making use of accrual accounting for certain programs. See infra Part IV.A. Indeed, the traditional presentation of Social Security finances in the trustees’ annual reports no longer complies with generally accepted accounting standards for government entities. See infra Part IV.A.2.
[22] The trust funds were not included in the original Social Security legislation of 1935 but were added in 1939 amendments. See Patashnik, supra note 19, at 67–69 (explaining that concerns over constitutional challenges prevented initial legislation from including trusts).
[23] See 42 U.S.C. § 401(c) (2000) (“A person serving on the Board of Trustees shall not be considered to be a fiduciary and shall not be personally liable for actions taken in such capacity with respect to the Trust Funds.”).
[24] 42 U.S.C. § 401 (2000).
[25] See 2003 Trustees Report, supra note 8, at 2. See also 2002 Trs. of the Fed. Old-Age & Survivors Ins. & Disability Ins. Trust Funds Ann. Rep. 3 (2002) [hereinafter 2002 Trustees Report].
[26] Throughout, this Article refers to the combined Old Age, Survivors, and Disability Trust Funds. In fact, there are two separate trust funds—one for Old Age and Survivors Insurance and the other for Disability Insurance—and their individual financial status differs somewhat from that of the combined fund. See 42 U.S.C. § 401(a)–(b) (2000). For the purposes of this Article, however, these differences are not significant.
[27] 2003 Trustees Report, supra note 8, at 4 tbl.IIB1.
[28] These revenues resulted from a levy of 12.40% imposed on taxable wages up to $87,000 in 2003 and divided evenly into employee and employer shares of 6.20% each. The level of covered payroll is adjusted annually to reflect cost-of-living increases. See Soc. Security Admin., 2002 Social Security Changes, at http://www.ssa.gov/cola/colafacts2003.htm (last visited Nov. 17, 2003).
[29] This source of revenue comes from funds raised through the federal income tax system. A portion of Social Security benefits is treated as taxable income under the federal income tax, and a portion of the taxes levied in this manner is redirected to the trust funds. See I.R.C. § 86 (2000); 2003 Trustees Report, supra note 8, at 18–19.
[30] See 42 U.S.C. § 401 (2000). The effective interest rate on OASI trust fund assets in 2002 was 6.4% while the effective interest rate on DI trust fund assets was 6.3%. See 2003 Trustees Report, supra note 8, at 21, 25.
[31] The trust fund expenditures include administrative costs ($4.2 billion) and transfers to the Railroad Retirement program ($3.6 billion).
[32] See 2000 Trs. of the Fed. Old-Age and Survivors Ins. and Disability Ins. Trust Funds Ann. Rep. 15 (2000) [hereinafter 2000 Trustees Report].
[33] Id. In 2002, for example, the trust fund ratio for the combined funds was 263%: total assets at the beginning of the year of $1,212.5 billion divided by total expenditures during the year of $461.7 billion. See 2003 Trustees Report, supra note 8, at 41.
[34] See 2003 Trustees Report, supra note 8, at 32.
[35] The trustees’ reports typically present three alternative cost estimates: high, low, and intermediate. Unless otherwise indicated, this Article will use the intermediate estimates, as these are the ones most commonly used in public discussions of Social Security. To the extent feasible, figures and tables will report all three estimates. For a discussion of the assumptions and methods underlying the actuarial estimates, see id. at 74–128.
[36] See id. at 41.
[37] 2000 Trustees Report, supra note 32, at 15.
[38] See 2003 Trustees Report, supra note 8, at 7.
[39] See Light, supra note 3, at 115–228.
[40] See 2000 Trustees Report, supra note 32, at 3 (“On a combined basis, the OASDI program is not in ‘close actuarial balance’ over the next seventy-five years. In addition, the individual OASI and DI Trust Funds are not in close actuarial balance. These conclusions are the same as those shown in the 1999 Annual Report.”); see also 2003 Trustees Report, supra note 8, at 7–10.
[41] 2003 Trustees Report, supra note 8, at 16. See also 2002 Trustees Report, supra note 25, at 18.
[42] See 2000 Trustees Report, supra note 32, at 30–31 (“In view of the size of the financial shortfall in the OASDI program over the next 75 years, we again urge that the long-range deficits of both the OASI and DI Trust Funds be addressed in a timely way.”); see also 2003 Trustees Report, supra note 8, at 17; 2002 Trustees Report, supra note 25, at 18; 2001 Trs. of the Fed. Old-Age & Survivors Ins. & Disability Ins. Trust Funds Ann. Rep. (2001) [hereinafter 2001 Trustees Report].
[43] See Stephen C. Goss, Measuring Solvency in the Social Security System, in Prospects for Social Security Reform 16, 19 (Olivia S. Mitchell et al. eds., 1999).
[44] For example, under the trustees’ intermediate projections, the combined income rate for OASDI funds increases by only 0.73 percentage points during the seventy-five-year period, from 12.70% of taxable wages in 2003 to 13.43% in 2080. See 2003 Trustees Report, supra note 8, at 47.
[45] In other words, the income rates reported in presentations such as Figure Two do not include income in the form of interest payments from the trust fund reserves. This interest does, however, figure into the annual report of trust fund activities summarized Table One.
[46] The summarized data reflect not an arithmetic mean, but an average of the discounted value of various components over the relevant period. Thus, the summarized rates take into account the time-value of money as well as the projected accrued interest on trust fund reserves. See Goss, supra note 43, at 20.
[47] For example, if one were to assume (under the intermediate-cost estimates) that the entire current reserves ($1,378 billion) would be amortized over the next twenty-five years, the reserves would be equivalent to an increase of 1.48% in payroll taxes, and, if adjustment were made for the 0.56% cost rate associated with building an adequately funded reserve at the end of the period, the increase would be only 0.92%. Over longer periods, the current trust reserves have even less significance.
[48] See 2003 Trustees Report, supra note 8, at 16 (“This deficit indicates that financial adequacy of the program for the next 75 years could be restored if the Social Security payroll tax were immediately and permanently increased from its current level of 12.4 percent (for employees and employers combined) to 14.32 percent.”).
[49] For example, President Bush’s Commission on Social Security used this measure as a benchmark for evaluating various reform proposals. See Bush Commission Report, supra note 2, at 69–70. See also 1994–1996 Advisory Council Report, supra note 2, at 11 (using the same measure of insolvency to assess proposed reforms).
[50] See 2003 Trustees Report, supra note 8, at 63–66. In their assessment of combined rates, the trustees actually use a sliding scale, under which combined rates must precisely match cost rates over a ten-year period (that is, over the short range) and are allowed to deviate gradually by up to 5.0% over the full seventy-five-year horizon. Id.
[51] See id. at 15.
[52] See id. at 16. Annual estimates of projected trust fund ratios are available from the SSA Office of the Chief Actuary. SSA Office of the Chief Actuary, Estimated Trust Fund Ratios: Calendar Years 2003–80, at http://www.ssa.gov/OACT/TR/TR03/lr4B3.html (last visited Oct. 20, 2003).
[53] See 2003 Trustees Report, supra note 8, at 47 tbl.IV.B1 (projected cost and income rates for 2040 and 2075). These ratios reflect the differences in projected income and benefit rates shown in Figure Two (intermediate estimate). Annual estimated projections of cost and income rates are available from the Office of the Chief Actuary. SSA Office of the Chief Actuary, Estimated Annual Income Rates, Cost Rates, and Balances: Calendar Years 2003–80, Intermediate Assumptions, at http://www.ssa.gov/OACT/TR/TR03/lr4B1-2.html (last visited Oct. 20, 2003).
[54] See 2003 Trustees Report, supra note 8, at 47 tbl.IV.B1.
[55] See 2003 Trustees Report, supra note 8, at 2 (“The OASI and DI Trust Funds, individually and combined, are adequately financed over the next 10 years under the intermediate assumptions.”).
[56] See 2003 Trustees Report, supra note 8, at 3 (“The combined OASDI Trust Funds are projected to become insolvent in 2042 under the long-range intermediate assumptions.”).
[57] See Table One.
[58] See Box Two.
[59] For example, in a recent exposure draft regarding the accounting treatment of Fiduciary Activities, the Federal Accounting Standards Advisory Board (FASAB) proposed applying the new standards only to fiduciary activities in which non-federal parties have an “ownership interest that the Federal Government must uphold,” thus exempting social insurance programs such as Social Security. See Accounting for Fiduciary Activities, Statement of Federal Financial Accounting Standards Exposure Draft, (Fed. Accounting Standards Advisory Bd. 2003) [hereinafter Accounting for Fiduciary Activities], available at http://www.fasab.gov/pdf/trustf1.pdf (last visited Oct. 17, 2003). For further discussion of FASAB pronouncements in this area, see infra Part IV.A.2.
[60] The rules governing Social Security benefits are extraordinarily complex. The key statutory entitlement for participants is the primary insurance amount, defined in 42 U.S.C. § 415 (2002). For a good non-technical explanation of Social Security benefits, see C. Eugene Steuerle & Jon M. Bakija, Retooling Social Security for the 21st Century 75–79 (1994).
[61] See Steuerle & Bakija, supra note 60, at 75–79.
[62] See infra text accompanying notes 275–283.
[63] See Steuerle & Bakija, supra note 60, at 25–27.
[64] See Lawrence R. Jacobs & Robert Y. Shapiro, Myths and Misunderstandings about Public Opinion Towards Social Security in Framing the Social Security Debate, in Framing the Social Security Debate: Values, Politics, and Economics 355, 365–74 (R. Douglas Arnold et al. eds., 1998).
[65] See Jacobs & Shapiro, supra note 64, at 359 n.14 (“70 percent to 80 percent of Americans agreed that ‘everyone who pays into Social Security should receive it, no matter what other income they have.’”).
[66] See Employers’ Accounting for Pensions, Statement of Financial Accounting Standards No. 87, 15–19 (Fed. Accounting Standards Advisory Bd. Dec. 1995) [hereinafter Employers’ Accounting for Pensions], available at http://www.fasb.org/pdf/fas87.pdf (last visited Nov. 17, 2003).
[67] John H. Langbein & Bruce A. Wolk, Pension and Employee Benefit Law 372 (3d ed. 2000).
[68] See Employers’ Accounting for Pensions, supra note 66 at 15–16.
[69] As discussed infra text accompanying notes 231–240, federal government budgeting procedures increasingly are making use of accrual techniques to account for government activities that entail long-term commitments. Recognizing the value of accrual accounting, the Bush Administration recently proposed that all retirement benefits for government employees also be accounted for in this way, at least for internal management purposes. See Congressional Budget Office, The President’s Proposal to Accrue Retirement Costs for Federal Employees (June 2002), available at http://www.cbo.gov/showdoc.cfm?index=3580&sequence=0 (last visited Oct. 17, 2003) [hereinafter President’s Proposal].
[70] See Accounting Standards, Statements of Financial Accounting Concepts 191 (Financial Accounting Standards Bd. 1999).
[71] Id. at n.21.
[72] Id. at n.22.
[73] See Gen. Accounting Office, Budgeting for Federal Insurance Programs 8 (Apr. 23, 1998) [hereinafter Budgeting for Federal Insurance Programs] (citing among the benefits of accrual-based accounting, the provision of “an opportunity to control costs before the government is committed to making payments”).
[74] See 2003 Trustees Report, supra note 8, at 61 (explaining components of this estimate). This figure is slightly higher than the estimate of the open-group unfunded obligation discussed below ($3.5 trillion), see infra text accompanying notes 172–179, because this estimate factors in the additional present value cost of building up a reserve at the end of the seventy-five-year projection period ($280 billion). To put this number in context, the total revenues of the United States government in fiscal year 2002 were just $1.85 trillion. See Congressional Budget Office, The Budget and Economic Outlook: Fiscal Years 2004-2013, at 4 tbl.1-2 (Jan. 2003) [hereinafter CBO January 2003 Report].
[75] See Bush Commission Report, supra note 2, at 69–70. This defect in traditional social security accounting is sometimes referred to as the “cliff” problem. Id. at 70. A good example of this phenomenon is the growth of the Social Security trust fund deficits in the years immediately following the 1983 reforms. At the beginning of this period, when the reforms had just been enacted, the system was in long-range balance, although it quickly fell into long-range imbalance again. See infra Figure 5.
[76] See supra text accompanying notes 40–54.
[77] Admittedly, the long-range measures of actuarial balance do contemplate the accumulation of trust fund reserves at 100% of projected annual expenditures at the end of the period. This level of reserves, however, does not ensure long-run solvency. After all, the trust funds currently have reserves in excess of 200% of annual expenditures, see supra text accompanying notes 26–31, and the system is widely perceived to be on the edge of crisis.
[78] See, e.g., Bush Commission Report, supra note 2, at 66.
[79] See John D. McKinnon, Medicare Outlook Appears to Worsen, Wall St. J., Mar. 18, 2003, at A3 (reporting on both Social Security and Medicare Reports).
[80] While the discussion in the text focuses on the Wall Street Journal account, other press coverage of the 2003 Trustees Report was generally comparable. See, e.g., Vicki Kemper, Benefits Outlook Is Mixed Bag, L.A. Times, Mar. 18, 2003, at A20 (“The long-term financial outlook for Social Security is somewhat brighter than a year ago . . . .”); Janelle Carter, Medicare Funds to Run Out Sooner But Social Security on Better Footing, Fund Trustees Say, Boston Globe, Mar. 18, 2003, at A2 (“The Social Security trust fund is slightly stronger than it was a year ago . . . .”); Larry Lipman, Outlook Better for Social Security Fund, Atlanta J. Const., Mar. 18, 2003, at A1 (“Updating their projections, federal officials forecast Monday that Social Security will stay solvent a year longer than expected . . . .”). To be sure, some journalists do a better job dissecting the Trustees Report, see, e.g., Sloan, supra note 9 (offering a more critical analysis of the 2003 Trustees Report); Marie Cocco, We’re Sailing Right into a Fiscal Hurricane, Newsday, May 27, 2003, at A23 (similar), but the 2003 Trustees Report was generally portrayed in contemporaneous press accounts as reporting good news, at least for the short term.
[81] McKinnon, supra note 79, at A3.
[82] 1994–1996 Advisory Council Report, supra note 2.
[83] Id. at 11. To be fair, the trustees’ standard for long-range actuarial balance also requires that certain standards be met for interim periods beginning ten years from the beginning of the period of analysis. See 2003 Trustees Report, supra note 8, at 63. Under these more complex requirements, a lump-sum future payment would not suffice. Reform proposals, however, routinely are evaluated solely in terms of their effect on the seventy-five-year actuarial deficit, confirming the basic point that the seventy-five-year projection tends to become the dominant measure of long-range solvency.
[84] 1994–1996 Advisory Council Report, supra note 2, at 30–33 (Option III).
[85] Admittedly, there are legitimate grounds for structuring reform proposals that gradually come into effect. Indeed, one of the advantages of accrual accounting is that it facilitates transitional reforms that are more difficult to impose under current practices. See infra text accompanying notes 217–270. At some point, however, reform proposals are so severely backloaded as to exceed any legitimate claim of transitional relief, and the advisory report recommendation discussed in the text arguably crosses this boundary line. In any case, regardless of whether one agrees with this assessment, it is unquestionably true that the actuarial deficit measure for reform proposals fails to distinguish between changes instituted in the near term from those scheduled to be implemented at the end of the seventy-five-year period.
[86] See Bush Commission Report, supra note 2, at 25–41.
[87] Although several federal agencies routinely prepare information on the federal budget, this Article focuses on the presentations of the CBO. A similar analysis would follow if the work of the executive branch’s Office of Management and Budget (OMB) or the independent General Accounting Office (GAO) were used instead. See generally Social Security: Long-Term Financing Shortfall Drives Need for Reform: Hearing Before the House Comm. on the Budget, 108th Cong. (2002) (statement of David M. Walker, Comptroller General) [hereinafter GAO Testimony]; Office of Mgmt. & Budget, Mid-Session Review: Fiscal Year 2002 (July 15, 2002) [hereinafter 2002 Mid-Session Review] (OMB budgetary analysis).
[88] Several times a year, the CBO prepares reports of this sort for the Senate and House Committees on the Budget. In recent years, the CBO has produced a substantial report on the budget and economic outlook in January with an update in August. For a list of recent CBO budget documents, see Congressional Budget Office, Congressional Budget Office—Publications, at http://www.cbo.gov/byclasscat.cfm?class=0&cat=0 (last visited Nov, 18, 2003).
[89] Congressional Budget Office, The Budget and Economic Outlook: An Update ix (Aug. 1998) [hereinafter CBO August 1998 Update].
[90] Ultimately, the total surplus for Fiscal 1998 was $69 billion. See Office of Mgmt. & Budget, Budget of the United States, Fiscal Year 2003: Historical Tables 22 tbl.1.1 (2002), available at http://www.whitehouse.gov/omb/budget/fy2003/pdf/hist.pdf) (last visited Oct.17, 2003) [hereinafter FY 2003 Historical Tables].
[91] See CBO August 1998 Update, supra note 89, at xiii. See also Christopher Georges,Surplus Could Hit $60 billion this Year, Wall St. J., May 5, 1998, at A2.
[92] See Stanley Collender, The Guide to the Federal Budget 22–23, n.11 (1995) (describing section 13301 of the Budget Enforcement Act of 1990, as amended).
[93] See infra text accompanying notes 110–127 (exploring problems with this formulation).
[94] As it turned out, the total federal surplus was $126 billion in fiscal year 1999, $236 billion in fiscal year 2000, and $127 billion in fiscal year 2001. See FY 2003 Historical Tables, supra note 90, at 22 tbl.1.1.
[95] For an interesting discussion of efforts to clarify the distinction between on-budget and off-budget surpluses during the final years of the Clinton Administration, see Douglas W. Elmendorf et al., Fiscal Policy and Social Security Policy During the 1990s 44 (Nat’l Bureau of Econ. Research, Working Paper No. 8488, 2001).
[96] For a discussion of the lockbox debate, see infra note 113 and text accompanying note 222.
[97] See Congressional Budget Office, The Budget and Economic Outlook: Fiscal Years 2002–2011, at xiii (Jan. 2001) (reporting a “total surplus [that] will reach $281 billion in 2001” as well as a smaller “on-budget surplus” of $125 billion that is “[p]erhaps more important to some policymakers”).
[98] Congressional Budget Office, The Budget and Economic Outlook: Fiscal Years 2003–2012, at xiii (Jan. 2002) [hereinafter January 2002 CBO Outlook] (citation omitted).
[99] Id.
[100] These projections substantially exceeded actual results. See CBO January 2003 Report, supra note 74, at xvi (reporting on-budget deficits of $317 billion in 2002 and total budget deficits of $158 billion); Congressional Budget Office, Monthly Budget Review: August 2003 (2003) [hereinafter CBO August 2003 Update] (projecting total budget deficits of $455 billion in 2003), available at ftp://ftp.cbo.gov/44xx/doc4491/08-2003-MBR.pdf.
[101] See also CBO January 2003 Report, supra note 74, at xvii–xix (failing to distinguish carefully on-budget and total budget deficits in summary paragraphs on budget outlook). Press coverage immediately following the CBO’s August 2003 report estimating total budgetary deficits of $401 billion in FY 2003 and $480 billion in FY 2004, see Congressional Budget Office, The Budget and Economic Outlook: An Update (Aug. 2003), available at http://www.cbo.gov/showdoc.cfm?index=4493&sequence=0, uncritically accepted the CBO’s projected total deficits, which included Social Security cash-flow surpluses. See, e.g., $1,400,000,000,000, St. Louis Post-Dispatch, Aug. 31, 2003, at B2; A Dismaying Report on Federal Deficits, Minneapolis Star Trib., Aug. 31, 2003, at 12AA; Associated Press, The Markets; Stocks & Bonds, N.Y. Times, Aug. 28, 2003, at c7; Deficit Delusions, Wash. Post, Aug. 29, 2003, at A22; Linda Feldmann, How ‘Pushback’ Plays for Bush on Iraq, Christian Sci. Monitor, Aug. 28, 2003, at 2; David Lazarus, Burning Money in Iraq, S.F. Chron., Aug. 29, 2003, at B1; R.C. Longworth, The Long-Term Cost of the Deficit, Chi. Trib., Aug. 31, 2003, at C1; Making the Kids Pay, L.A. Times, Aug. 28, 2003, at 16. In only one newspaper article that I found—a Saturday editorial—was it noted that the projected deficit in FY2003 would have been $562 billion if limited to the government’s operating budget. See The Deeper Deficit, Boston Globe, Aug. 30, 2003, at A10. During the late 1990s, when the federal government began to run total budgetary surpluses, the press seemed to do a slightly better job of noting the extent to which those surpluses depended on cash-flow surpluses in the Social Security trust funds. See, e.g., Bleeding Social Security, L.A. Times, Sept. 29, 1998, at B6 (“[T]he surplus consists almost entirely of Social Security payroll taxes. Take away those revenues and federal budgets over the next five years are expected to show deficits totaling $137 billion.”); Steven Thomma, Federal Budget in the Black, Fort Worth Star-Telegram, Oct. 1, 1998, at 1 (“Without the Social Security money, the government would have a deficit of $32 billion.”).
[102] CBO August 1998 Update, supra note 89, at 33.
[103] It was for reasons of this sort that a presidential commission under President Johnson recommended that Social Security be brought on-budget. See President’s Comm’n on Budget Concepts, Report of the President’s Commission on Budget Concepts 26–27 (Oct. 1967) [hereinafter Budget Concepts Report].
[104] Cf. David M. Walker, Truth and Transparency: The Federal Government’s Financial Condition and Fiscal Outlook, Speech at the National Press Club (Sept. 17, 2003) (“When it comes to the U.S. Government’s financial condition and fiscal outlook, the federal government’s current measurement and scorekeeping approaches leave much to be desired. The result is an incomplete and misleading picture of the federal government’s current financial condition and future fiscal outlook, as well as a delay in the timeliness in which we address important issues.”), available at http://www.gao.gov/cghome/npc917.pdf). To the extent that the composition of the federal government’s financial obligations changed over the course of the year, federal budgetary statements should also communicate these changes. For this reason, I favor more comprehensive measures of the government’s fiscal imbalance along the lines proposed in Jagadeesh Gokhale & Kent Smetters, Fiscal and Generational Imbalances: New Budget Measures For New Budget Priorities 7–15 (2003). Infra Part II.C.5 explains how an accrual-based system of accounting for Social Security could help compare the system’s mounting levels of implicit debt to the growth of the federal government’s explicit debt.
[105] For a good summary of this history, see generally David S. Koitz, Social Security and the Federal Budget: What Does Social Security’s Being “Off Budget” Mean? (Congressional Research Serv. Report No. 98-422, 2001).
[106] See supra text accompanying notes 87–89.
[107] Congress’s own record in this regard is hardly exemplary. During the 1960s, Social Security was added to the unified federal budget in response to recommendations from the President’s Commission on Budget Concepts. See Budget Concepts Report, supra note 103, at 26–27. Through the 1980s, when public concern over mounting federal deficits led to the passage of the Gramm-Rudman-Hollings Act, Pub. L. No. 99-177 (1985), Social Security surpluses were included in budgetary aggregates, although Social Security expenditures were largely exempt from the Act’s automatic sequestration procedures. See Koitz, supra note 105, at 12. In 1990, with the passage of the Budget Enforcement Act, Pub. L. No. 101-508 (1990), Social Security was officially moved off-budget. 2 U.S.C. § 900 (2000).Deficit targets were, however, statutorily adjusted downward (that is, higher total-budget deficits were permitted) in recognition of the fact that the movement of Social Security off-budget would otherwise impose increased pressure for on-budget accounts.
[108] See supra text accompanying notes 87–89.
[109] Cf. CBO August 1998 Update, supra note 89, at 33–34 (acknowledging minor role of Postal Service surpluses).
[110] See Alan J. Auerbach et al., The Budget Output and Options for Fiscal Policy, 91 Tax Notes 1639, 1639 (2002); see also Bob Davis, Clinton’s Simple Goal, Saving Social Security, Involves Complex Plan to Cut Debt, Boost Savings, Wall St. J., Feb. 2, 1999, at A9 (“The [FY 2000 proposed] budget reiterates Mr. Clinton’s pledge to devote 62 percent of the budget surplus over the next 15 years to bolstering Social Security. But that is less definitive than it sounds. Most of the budget surplus already comes from the payroll taxes that fund Social Security payments. The administration is counting the same money twice, many critics charge.”); David Wessel, The Outlook: Social Security With Subtitles, Wall St. J., Dec. 7, 1998, at A1 (“For the next few years, the surplus exists only because Social Security is collecting more taxes than it’s paying in benefits; the rest of the federal budget isn’t projected to go into the black until 2002.”).
[111] See 42 U.S.C. § 401(h) (2000).
[112] See supra Table Two (showing that under intermediate estimates, current trust fund balances contribute a summarized revenue rate of 0.69% of taxable payroll over the next seventy-five years).
[113] See Elmendorf et al., supra note 95, at 43 (discussing internal debate within Clinton Administration on this point). The 2000 presidential campaign devolved into dueling lockbox proposals, which became fodder for late-night-television comedy routines. The lockbox concept was never clearly defined. Often, the term seemed to refer to the allocation of Social Security surpluses to the trust funds. See id. at 73. At other times, the term seemed to be used to refer to the allocation of a portion of on-budget surpluses to the Social Security trust funds. See id. Exactly how this second formulation was to have been implemented is unclear. Conceivably, retired public debt could have been allocated to the trust funds. Or, to similar effect, general revenues could have been directed to the trust funds and then used to purchase government bonds in the secondary market. Ultimately, the unified surplus disappeared before any lockbox concept could be implemented and references to lockboxes have largely fallen out of the public debate. See Douglas W. Elmendorf & Jeffrey B. Liebman, Social Security Reform and National Savings in an Era of Budget Surpluses, in 2 Brookings Papers on Economic Activity 11–18 (2000); see also Rudolph G. Penner et al., Saving the Surplus to Save Social Security: What Does It Mean? (Urban Inst. Brief Series No. 7, 1999).
[114] See Budget Concepts Report, supra note 103, at 26–27. When this approach is applied, the critical question is whether a particular entity should be characterized as governmental or nongovernmental. The notion that the Social Security trust funds should be classified as governmental is not altogether implausible. As mentioned above, the trusts are an accounting fiction managed by a group of six trustees, four of whom are ex officio government officials and two of whom are public representatives appointed by the President and confirmed by the Senate. See 42 U.S.C. § 401(c) (2002). The putative beneficiaries of the trust funds are current and future participants and their family members. But the legal right of these beneficiaries to trust assets are not robust, particularly with respect to prospective changes in benefit formulas and rates of taxation. See supra text accompanying note 59. It is for this reason that supporters of consolidation of Social Security surpluses can credibly assert that the trust funds should be considered governmental agencies.
[115] See Patashnik, supra note 19, at 113–34.
[116] See supra text accompanying notes 64–65.
[117] Note that “debt held by the general public” is the standard against which government accounting officials measure aggregate borrowings of the federal government. As currently used, this formulation does not include federal debt held in the Social Security trust funds.
[118] See supra Table One.
[119] See supra Figure Two. Under intermediate assumptions over the seventy-five-year period, the beginning fund balance accounts for summarized income rate equal to 0.69% of payroll as compared with a total summarized income rate of 13.78% of payroll.
[120] To be sure, one could resist this point by arguing, once again, that obligations to the trust funds are different from other government securities because the government has the legal right to reduce the amount of Social Security benefits or increase the amount of Social Security payroll taxes, thereby eliminating the need for the Treasury to honor its obligations to the fund. In my view, this is a dubious point, given the magnitude of changes necessary to obviate the trust funds’ need for current reserves and projected operating surpluses over the coming decade.
[121] For a summary of this literature, see generally Congressional Budget Office, Social Security and Private Savings: A Review of the Empirical Evidence (July 1998).
[122] Id.
[123] Id. at 3.
[124] See CBO August 1998 Update, supra note 89, at 33. A few more words about the relationship between pre-funding and Social Security’s negative effect on savings may be in order here. The basic point in the text is that the promises that Social Security makes to U.S. workers each year diminish other forms of savings to some degree. Having a portion of the Social Security system’s new commitments funded each year (through the purchase of government securities) to some degree offsets the system’s negative effect on savings. So determining the net effect of Social Security on national savings would require a comparison of the extent of pre-funding and the size of the system’s negative impact on savings each year. Because the annual accrual of obligations of the system are so much larger than the annual increase in its funding, the negative effects would likely outweigh the positive effects of pre-funding, but this is simply a conjecture, which is subject to empirical validation or rebuttal. Even without such further inquiry, however, the basic point that annual gross increases in funding likely overstate the net savings effect remains valid.
[125] As revealed in Figure Two, the revenue and cost rates of the trust funds are expected to be in balance at some point during the next decade.
[126] Press Release, Senator John Ashcroft (May 2, 1999) (on file with author) (“Honoring the commitment to retirement security for the American people must be a sacred obligation of the government. Promises made must be promises kept, for this generation and for future generations. I welcome the House’s decision to end once and for all Washington’s indefensible practice of dipping into Social Security to pay for irresponsible new spending, higher deficits, or tax cuts.”).
[127] Of course, the actual effect is unlikely to be this strong. In a 1974 study, Martin Feldstein originally estimated the reduction in private savings to be on the order of 30% to 50% of Social Security benefits. See Martin Feldstein, Social Security, Induced Retirement and Aggregate Capital Accumulation, 82 J. Pol. Econ. 905 (1974); see also Martin Feldstein, Social Security and Private Savings: Reply, 90 J. Pol. Econ. 630 (1982). See also infra Part IV.B.2.a (review of the life-cycle literature). However, the direction of the effect, not its magnitude, is what is important here. For a recent review and critique of related literature on the overall impact of private pensions on overall savings, see generally William G. Gale, The Effects of Pension on Household Wealth: A Reevaluation of Theory and Evidence, 106 J. Pol. Econ. 706 (1998).
[128] See, e.g., 1994–1996 Advisory Council Report, supra note 2, at 19–20.
[129] To assess such a proposal’s long-term effect, one would have to analyze the wage structure and other demographic facts about the new state and local workers brought into the system. Conceivably, if a sufficient percentage of these workers were low-income, the progressive nature of Social Security benefits could make their participation a long-run net loss for the system. Measured in terms of near-term (five- to ten-year) total-budgetary effects, however, the inclusion of these workers would be a clear benefit. For an otherwise careful study of the inclusion of state and local workers in the Social Security system, see Alicia H. Munnell, The Impact of Mandatory Social Security Coverage of State and Local Workers: A Multi-State Review (AARP Public Policy Institute, Paper No. 2000-11, Aug. 2000) (based on an analysis limited to the seventy-five-year time frame, reporting that the inclusion of state and local workers would reduce the system’s actuarial deficit by ten percent)
[130] See, e.g., 1994–1996 Advisory Council Report, supra note 2, at 25–62.
[131] A similar budgetary effect would occur if trust fund assets were transferred to individual accounts. See Congressional Budget Office Paper, The Budgetary Treatment of Personal Retirement Accounts (Mar. 2000).
[132] For an overview of the economic effects of stock market investments, see Peter A. Diamond, The Economics of Social Security Reform, in Framing the Social Security Debate: Values, Politics, and Economics (R. Douglas Arnold et al. eds., 1998).
[133] Such a scenario is not entirely far-fetched. In a budgetary controversy a few years ago, when Congress’s failure to increase the limit on the national debt was seen to give the legislative branch a tactical advantage over the executive, the Secretary of the Treasury delayed investments of other trust funds in government securities in order to keep the debt within legal limitations and thereby diminish the bargaining power of Congress. See Use of Trust Funds to Avoid Default: Hearing Before the House Comm. on Banking and Fin. Servs., 104th Cong. (Dec. 13, 1995) (statement of Robert E. Rubin, Secretary of the Treasury) (describing past techniques to avoid debt limit). Increasing the independence of the Social Security trust funds is one way to prevent such behavior with respect to Social Security reserves in the future.
[134] See Patashnik, supra note 19, at 88–90.
[135] See id.
[136] I am extremely grateful to the Office of the Chief Actuary and, in particular, Stephen Goss, the Chief Actuary of the Social Security Administration, for making this information available to me.
[137] Retirees have paid most, but not all, of their taxes, because Social Security trust fund revenues include some income taxes paid on Social Security benefits. See supra text accompanying notes 26–31. Retirees pay some of these taxes after leaving the workforce and ceasing to pay payroll taxes.
[138] As explained in the 2003 Trustees Report, the current value of the trust funds equals “the accumulated value of past OASDI taxes [plus accrued interest] less cost.” 2003 Trustees Report, supra note 8, at 62.
[139] Implicitly, past benefits payments are also recognized in the sense that they have been offset against past tax contributions.
[140] I use the qualifier “essentially” because some mid-career workers will have been paid temporary disability benefits and perhaps other ancillary benefits.
[141] The basic presentation is not fully consistent with cash-flow accounting because it recognizes the accrual of interest on reserves even though there is no corresponding cash payment.
[142] See supra notes 64–65 and accompanying text.
[143] See, e.g.,Goss, supra note 43; Gokhale & Smetters, supra note 104, 25–28; 2003 Technical Panel on Assumptions & Methods, Report to the Social Security Advisory Board 87–88 (Oct. 2003) [hereinafter 2003 Technical Panel Report], available at http://www.ssab.gov/2003TechnicalPanelRept.pdf (last visited Nov. 22, 2003). See also infra text accompanying notes 260–271 (discussion of international comparisons of the implicit debt of public pension plans).
[144] See, e.g., Goss, supra note 43. See also infra text accompanying notes 241–257 (discussion of FASAB requirements for social insurance). Accrual accounting also employs present-value concepts for liabilities, such as retirement benefits, that are to be paid in the future. The present value of the benefit is recognized at the time the obligation is incurred, and then each year the obligation accrues an interest charge until the obligation equals the full amount of the benefit at the time of payment. The illustrative income statements presented below distinguish between the initial accrual of benefit obligations and the annual accrual of interest on previously accrued benefits. See infra Part II.B.2.
[145] Accrual accounting, thus, occupies an area between the trustees reports’ basic presentation, illustrated by Table One, which recognizes a narrower set of taxes and benefits, and the alternative measures of long-term solvency, which reflect a broader range of taxes and benefits.
[146] See 2003 Technical Panel Report, supra note 143, 87–89.
[147] Id. at 87.
[148] See, e.g.,Goss, supra note 43, at 32.
[149] See id.
[150] See, e.g., Gokhale & Smetters, supra note 104, at 44.
[151] A balance sheet offers a picture of an organization’s total financial resources (assets) and obligations (liabilities) at a specific point in time. If its assets exceed its liabilities, the organization has a surplus; if its liabilities exceed its assets, it has a deficit. An income statement illuminates the extent to which an organization’s operations increased or decreased its net assets during the period in question. Profits increase an organization’s surplus or decrease its deficit. Losses decrease an organization’s surplus or increase its deficit.
[152] What this measure reflects is the amount that the federal government would have to pay if the Social Security system were to be shut down at the point of measurement. The maximum transition cost reveals how much money it would take beyond the amounts currently held in the trust fund reserves to pay all future payments of participants and beneficiaries accrued to date. In policy debates, this measure is sometimes described as the cost of completely shutting down the Social Security system. The measure is sometimes rejected as irrelevant to public discussion of Social Security finances because it is widely agreed that there is no political support for entirely ending Social Security. For my purposes, however, the maximum transition cost is a most useful statistic because it is quite similar to the measures of accrued liabilities that GAAP requires of private pensions. The maximum transition cost makes no allowance for future revenues nor does it consider benefits that would accrue in the future. It simply represents the difference between the total value of Social Security benefits that have earned to date and the current level of trust fund reserves.
[153] See, e.g.,Goss, supra note 43, at 34.
[154] See Box Four.
[155] This estimate probably understates the level of accrued liabilities that would be reported if the financial statements of Social Security were constructed in accordance with the actual GAAP rule for private pension plans. The Office of the Chief Actuary’s forty-year accrual period is longer than the thirty-five-year period under which benefits accrue under the Social Security Act. See Social Security Handbook, How are the Average Monthly Earnings (AME) or the Average Indexed Monthly Earnings (AIME) Computed? (Last revised Mar. 2001), available at http://www.ssa.gov/OP_Home/handbook/handbook.07/handbook-0701.html (explanation of how benefits are calculated). See also Steuerle & Bakija, supra note 60, at 76. It is also longer than comparable periods of accrual for purposes of private pension plans. See Dan M. McGill et al., Fundamentals of Private Pension 213–14 (7th ed. 1996). Shorter periods of accrual tend to raise the net present value of accrued liabilities.
[156] See Soc. Sec. Admin., FY 2002 Performance and Financial Report 91 (2002) (estimating the actuarial present value of estimated benefits for participants 62 years of age and older to be $4,401 billion as of January 1, 2002, which was 32.9% of the system’s total projected benefits at the time: $13,374 billion).
[157] See Social Security Handbook, Entitlement to Retirement Insurance Benefit (Last Revised Mar. 2001), available at http://www.ssa.gov/OP_Home/handbook/handbook.03/handbook-0301.html (entitlement to retirement insurance benefit).
[158] These distinctions could be useful in explaining the differing impacts of various reform proposals.
[159] This is the Chief Actuary’s estimate of the maximum transition cost as of December 31, 2001.
[160] To be more precise, the trust funds’ negative net worth was $12,629.3 billion at the end of 2002 and $12,161.8 billion at the end of 2001.
[161] The other expenses consist principally of annual contributions to another pension system, one for retired railway workers.
[162] The $467.5 billion in annual loss is explained above as the amount by which the system’s maximum transition cost (or negative net worth) increased in 2002. This implies that the system’s expenses must have exceeded its revenues by $467.5 billion. Total revenues have been calculated above to be $627.1 billion. This means that total expenses were $1,094.6 billion (the difference between positive $627.1 and negative $467.5). So far, I have only identified $7.9 billion in expenses in 2002 ($4.2 billion in administrative expenses plus $3.7 billion in other costs). Therefore, $1,086.7 billion of expenses remain to be identified.
[163] At the beginning of 2002, the trust funds had $13,374.6 billion of total accrued liabilities ($1,212.5 billion of which were supported with cash reserves and the remaining $12,161.8 billion of which were unfunded liabilities). According to the 2003 Trustees Report, $454 billion in benefits were paid during the course of 2002, implying that $12,920.5 billion of the accrued liabilities remained outstanding. See 2003 Trustees Report, supra note 8, at 27. A 6.09% interest rate—the ultimate valuation interest rate used in the estimates of Office of the Chief Actuary—suggests an interest charge of $786.9 billion for the year.
[164] This figure represents the difference between the amount of expenses and the $786.9 billion in interest charge derived in the preceding footnote. As a residual category, this net accrual of liabilities incorporates a number of factors (such as changes in a variety of technical estimates) and thus can fluctuate considerably from year to year. A better estimate of the annual rate of accrual of net liabilities can be obtained by averaging the net accruals over a number of years. See Table Seven (presenting five years of income statements).
[165] By construction, the decline in the system’s unfunded accrued liabilities each year is the same as the trust funds’ operating losses for the year. So, the difference between the system’s accrued liability between year-end 1997 ($9,512 billion) and year-end 1998 ($10,170 billion) was $658 billion, the same as its net loss for 1998 as reported in Table Seven.
[166] See 2003 Trustees Report, supra note 8, at 63–65.
[167] Under generally accepted accounting principles for government entities, the Federal Accounting Standards Advisory Board requires that this measure, and its component parts, be included as supplementary information. See infra text accompanying notes 241–259. This information now appears in the Financial Report of the United States and the Social Security Administration’s annual financial statements. Id. In the 2003 Trustees Report, the closed-group unfunded obligation was also reported in a new section of the report entitled Additional Measures of OASDI Unfunded Obligations. See 2003 Trustees Report, supra note 8, at 63.
[168] See 2003 Technical Panel Report, supra note 143, at 88.
[169] Id.
[170] See Goss, supra note 43, at 33 tbl.3.
[171] One could decompose this figure into the net present value of future taxes and the net present value of future benefits. As of December 31, 2001, the present value of future contributions from current participants was in excess of $13 trillion, and the present value of benefits to be earned by current participants in the future, over and above the benefits accrued to date, was approximately $11 trillion. See Soc. Sec. Admin., Performance and Accountability Report: Fiscal Year 2002, at 77 (2002) [hereinafter 2002 Performance Report], available at http://www.ssa.gov/finance/2002/fy02PAR.pdf.
[172] See 2003 Trustees Report, supra note 8, at 3, 61–62. Occasionally, this measure of solvency is even picked up by the general press. See, e.g., Cocco, supra note 80, at A23 (“The Social Security trustees report that in today’s dollars, a transfer of $3.5 trillion from general government revenues, if made now, would enable the retirement system to pay the Baby Boomers their promised benefits.”).
[173] A more complete description of the open-group unfunded obligation, as well as of the other measures of accrued liability discussed in this Part, may be found in Goss, supra note 43, at 31–33.
[174] See 2003 Technical Panel Report, supra note 143, at 87.
[175] See Goss, supra note 43, at 33 tbl.3.To a limited degree, the seventy-five-year open-group liability measure is also narrower than the closed-group measure. First, some current participants—i.e., the very young mid-career participants—will receive benefits more than seventy-five years in the future. These payments factor into the closed-group liability measure but not the seventy-five-year open-group liability measure. Also, the current reserves of the trust funds are typically deducted from the closed-group liability measure on the assumption that they would be used to finance the benefits of current participants if these participants were grandfathered under current law. See 2003 Trustees Report, supra note 8, at 63 tbl.IV.B8. Because the open-group measure is generally considered to be an estimate used to evaluate the solvency of the system on an on-going basis, the reserves are not fully available to finance benefits over the next seventy-five years. See supra text accompanying notes 153–159. At least a portion of these benefits must be set aside to maintain minimum acceptable trust fund ratios.
[176] See 2003 Trustees Report, supra note 8, at 61.
[177] See Goss, supra note 43, at 33 tbl.3 (reporting positive seventy-five-year open-group surplus at year-end 1983).
[178] See supra note 75 and accompanying text; see also infra Part III.B.4.
[179] Cf. 2003 Trustees Report, supra note 8, at app. B (reviewing changes in actuarial balance since 1983).
[180] See generally Gokhale & Smetters, supra note 104.
[181] See 2003 Technical Panel Report, supra note 143, at 87–88.
[182] See 2003 Trustees Report, supra note 8, at 63.
[183] Id.
[184] See Gokhale & Smetters, supra note 104, at 25–28.
[185] Id.
[186] Although the alternative measures are typically expressed in terms of a single statistic (the closed-group liability or infinite-horizon open-group obligation), one could decompose these measures into a series of assets—current reserves plus the net present value of future tax revenues from various groups of participants—and liabilities—such as the present value of future benefits to various groups of participants. As explained below, FASAB’s requirements for social insurance require these elements to be reported in the Required Supplementary Stewardship Information section of financial statements for government entities that seek to comply with GAAP for federal entities, although FASAB does not require these elements to be included as balance sheet entries. See infra Part IV.A.2. The trustees’ reports do not attempt to comply with FASAB requirements. In the summer of 2003, FASAB amended its rules governing the placement of this information, calling for a more prominent display of the information in a Statement of Social Insurance, or SOSI. See Reclassification of Stewardship Responsibilities and Eliminating the Current Services Assessment, Statement of Federal Financial Accounting Standards No. 25 (Fed. Accounting Standards Advisory Bd. July 2003), available at http://www.fasab.gov/pdf/sffas-25.pdf (last visited Nov. 21, 2003) [hereinafter Reclassification Statement]. Unlike the prior supplementary information, the SOSI is considered to be a basic financial statement and subject to audit requirements. It remains to be seen how this change will affect the Trustees Report and other public disclosures regarding Social Security.
[187] See supra note 144 and accompanying text.
[188] Employee Retirement Income Security Act § 204(g), 29 U.S.C. § 1054(g) (2000).
[189] See Gerald I. White et al., The Analysis and Use of Financial Statements 73 (2d ed. 1997).
[190] See, e.g., Michael Tanner, The Better Deal: Estimating Rates of Return under a System of Individual Accounts 15–16 (Cato Institute Social Security Paper No.31, 2003) (erroneously assuming the seventy-five-year open-group actuarial deficits as reported in the Trustees Report to be a reasonable estimate of transition costs and then discounting the need to account for transition costs on the grounds that these costs should be covered through a reduction in federal spending), available at http://www.socialsecurity.org/pubs/ssps/ssp31.pdf.
[191] For a good example of this view, see Peter A. Diamond & Peter R. Orszag, Assessing the Plans Proposed by the President’s Commission to Strengthen Social Security Reform, 96 Tax Notes 703, 705–06 (2002) (defending the use of “scheduled benefits” as the appropriate baseline for comparing reform proposals).
[192] Id.
[193] As far as I know, this precise objection to accrual accounting has not been made in print. Indeed, defenders of cash-flow accounting sometimes object to accrual accounting on the grounds that it might make it more difficult to reduce accrued benefits. See Peter A. Diamond & Peter R. Orszag, Accrual Accounting for Social Security, 41 Harv. J. on Legis. 173 (2004). The same analysts, however, then insist on scheduled benefits—which include accrued as well as to-be-accrued benefits—to be the appropriate baseline for analyzing reform proposals. See Diamond & Orszag, supra note 191, at 705–06.
[194] See Diamond & Orszag, supra note 193.
[195] See Employers’ Accounting for Pensions, supra note 66.
[196] If the goal here is to calculate the net value of future taxes, alternative measures must account for both future taxes and future benefit payments.
[197] The seventy-five-year open-group measure is not defensible on this ground because it does not provide an accurate measure of net tax contributions, truncated, as it is, after seventy-five years.
[198] See supra notes 182–183 and accompanying text.
[199] For a discussion of the literature on this point, see Kent A. Smetters, Thinking About Social Security’s Trust Fund, in Prospects for Social Security Reform 201, 203–05 (Olivia S. Mitchell et al. eds., 1999).
[200] See CBO January 2003 Report, supra note 74, at 148. In this Part, measures of profit and loss will be based on the modified accrual accounting approach. Were the losses based on a GAAP-style accrual accounting system, without current recognition of future excess tax contributions, losses would be approximately $100 billion higher each year.
[201] See Table Ten. If the 2002 Social Security performance under this approach were consolidated with the on-budget performance of the federal government in Fiscal Year 2003, the consolidated deficit would have been roughly $930 billion. See Jackson, supra note * .
[202] Under principles of consolidation, inter-government transactions are netted out. See generally Budget Concepts Report, supra note 103. For that reason, trust fund reserves are excluded fromthe calculation of public debt and are not counted against trust fund accrued liabilities. Thus, the implicit trust fund obligations (consolidated) presented in Figures Eight and Nine are equal to the closed-group liability plus trust fund reserves.
[203] In my opinion, this combination is not obviously inappropriate, as the accrued liabilities presented in Figures Eight and Nine are, by definition, obligations that the federal government is, in all probability, going to honor. One third of these liabilities are owed to retirees, and are therefore politically sacrosanct. The balance are due to mid-career workers, but have been recognized under an accrual formula that reflects the probability that they will be honored. See supra Box Four.
[204] See supra text accompanying notes 88–101.
[205] See Elmendorf et al., supra note 95, at 41–44.
[206] An open question is how income taxes currently allocated to the Social Security trust funds should be treated. As described above, the trust funds’ revenues include a portion of the income taxes imposed on certain Social Security benefits. Although these payments are not typically characterized as the allocation of general revenues to the trust funds, that is, arguably, what they are.
[207] When Social Security was first included in the unified budget in the late 1960s, the commission recommending the change contemplated that debt issued to the trust funds would be reported in an aggregate measure of gross federal debt. See Budget Concepts report, supra note 103, at 85. Over time, public accounting has focused nearly exclusively on federal debt held by the general public and has rarely given measures of gross federal debt much prominence. See, e.g., January 2002 CBO Outlook, supra note 98, at xv summary tbl.2.
[208] For purposes of this projection, the critical unknown is the future rate of growth of implicit trust fund debt. These figures are based on the assumption that the debt will exceed GDP growth by the same percentage it has exceeded GDP growth (seventeen percent) over the past five years. In earlier periods this relationship does not always hold, but the difference is difficult to interpret because of changes in actuarial assumptions and, in some years, program design. See Cong. Budget Office, The Budget and Economic Outlook: Fiscal Years 2004–2013, at xvi, 43 (2003) (providing rates of GDP and public debt growth).
[209] See, e.g., CBO August 2003 Update, supra note 100, at 2 tbl.1-1 (reporting debt held by the public as a percentage of GDP).
[210] As discussed below, economists do, however, routinely estimate and criticize the unfunded pension obligations of other countries. See infra Part IV.B.1.
[211] Acceptable levels of implicit debt for Social Security would likely depend on a number of factors, including the size of other entitlement programs (most notably Medicare and other health benefits for the elderly), projected growth rates of the economy and population, and a host of other technical factors.
[212] See Bush Commission Report, supra note 2, at 90.
[213] Examples include improved benefits for elderly widows and various efforts to increase the level of benefits for participants who have had lower wages.
[214] See supra note 75 and accompanying text.
[215] See 1994–1996 Advisory Council Report, supra note 2, at 90.
[216] This again raises an interesting question of public finance: the extent to which accrued liabilities of social insurance programs should be funded by excess payroll tax contributions as opposed to other potential sources of support, such as general revenues. For an interesting discussion of this issue, see generally Peter A. Diamond, Social Security, The Government Budget, and National Savings (2003) (unpublished manuscript, on file with author).
[217] For private pension plans, federal statutes prohibit the elimination of all accrued interest in pensions and thus also privilege accrued benefits over benefits that will accrue in the future. See supra text accompanying note 188.
[218] See Robert L. Clark, Liabilities, Debts, Revenues, and Expenditures: Accounting for the Actuarial Balance of Social Security, 41 Harv. J. on Legis. 161 (2004). Although the change did not go into effect until many years after 1983, once it went into effect for a particular age cohort, the change applied to all of that cohort’s benefits.
[219] See President’s Proposal, supra note 69, at 2–4.
[220] See supra text accompanying notes 63–65.
[221] There is considerable uncertainty whether changes in budgetary targets actually affect budgetary decisions, and simply focusing attention on on-budget aggregates does not mean that the political branches will keep these budgetary accounts in balance. See supra Part III.A.3. Moving from total budgetary aggregates to on-budget aggregates, however, should tend to reduce, if not eliminate, federal deficits.
[222] See supra note 113 and accompanying text.
[223] This problem would become more acute if participant benefits were also reported on an accrual-accounting basis. Each participant would then see the net expected tax (or net expected subsidy) of his or her annual participation in Social Security.
[224] See Jacobs & Shapiro, supra note 64, at 355–56.
[225] Compare Pozen, supra note 16, and Elizabeth Garrett, Accounting for the Federal Budget and Its Reform, 41 Harv. J. on Legis. 187 (2004) with Clark, supra note 218.
[226] See generally Budget Concepts Report, supra note 103.
[227] See Research and Dev. Arrangements, Statement of Financial Accounting Standards No. 17, § 26 (Financial Accounting Standards Bd., 1999). Others argue that the pay-as-you-go nature of the Social Security system justifies the system’s current accounting treatment. See id. at § 28. This claim is not particularly strong. First of all, the Social Security system is no longer operating on a purely pay-as-you-go basis. The combined trust funds are now pre-funded to the tune of $1.4 trillion and the level of pre-funding will continue to grow for a number of years. See supra Figure Three. More important, the central problem with Social Security finances is that, even with this substantial amount of pre-funding, the system is promising benefits that will not be sustainable from projected revenues in the future. Given demographic trends, the pay-as-you-go aspect of Social Security financing is more a cause of the system’s problems than a justification of its current accounting practices.
A separate, but equally insubstantial, objection is that accrual accounting is appropriate only for fully funded public pension systems. Nothing about accrual accounting necessitates any particular level of pre-funding. See Box Five. Indeed, Social Security would likely remain substantially underfunded on an accrual-accounting basis. This Article’s argument is that the level of under-funding should be kept apparent.
[228] See infra Part IV.B.1.
[229] See infra Part IV.B.2.
[230] For a description of the accounting treatment of defined-benefit pension plans, see Howell Jackson, A Comparison of Social Security Benefits and Private Pension Plans (Sept. 3, 2002) (unpublished manuscript, on file with author). As explained in that paper, the structure of private pension benefits is more similar to that of traditional Social Security benefits than is often appreciated, and the accounting rules for private pension plans offer a good model for fleshing out this Article’s accrual accounting proposal.
[231] For example, the 1967 presidential commission on budget concepts—the same group that initially recommended that Social Security be brought on-budget—also advocated introducing accrual accounting techniques to certain governmental operations. See Budget Concepts Report, supra note 103, at ch. 4.
[232] 2 U.S.C. § 661c (2000).
[233] For an overview of the FCRA, see Accounting & Info. Mgmt. Subdiv., Gen. Accounting Office, Credit Reform: Great Effort Needed to Overcome Persistent Cost Estimation Problems 1–5 (1998); see also Michael R. Pompeo, Accrual Accounting for Federal Credit Programs: An Evaluation of the Federal Credit Reform Act of 1990, 66 Tax Notes, 257, 257–58 (1995).
[234] See 2 U.S.C. §§ 661a–661f (2000).
[235] Id. at § 661e (2000).
[236] See Budgeting for Federal Insurance Programs, supra note 73, at 4–7.
[237] Id.
[238] See Gen. Accounting Office, Budget Process: Extending Budget Controls (2002) (testimony of Susan J. Irving).
[239] See generally Cong. Budget Office, The President’s Proposal to Accrue Retirement Costs for Federal Employees (June 2002).
[240] The federal government’s adoption of accrual accounting methods is part of a broader trend toward accrual accounting in governments around the world. See generally Int’l Fed’n of Accountants Pub. Sector Comm., Guidelines for Government Financial Reporting (July 1998). As discussed in the IFAC report, accrual accounting is an increasingly prominent alternative form of presentation for public accounting systems with numerous advantages over traditional cash-flow accounting. Over the past decade or two, accounting reforms in New Zealand, the United Kingdom, and Australia have incorporated elements of accrual accounting into their governmental financial statements, some of which extend to public pension systems. See Gen. Accounting Office, Accrual Budgeting: Experience of Other Nations and Implications for the United States 10 (2000) [hereinafter GAO Accrual Budgeting Report] (concluding that “for some activities, such as credit and pension programs, cash-based measurement is incomplete and potentially misleading”).
[241] See Federal Accounting Standards Advisory Bd., Memorandum of Understanding among the General Accounting Office, the Department of the Treasury, and the Office of Management and Budget on Federal Government Accounting Standards and a Federal Accounting Standards Advisory Board (May 7, 2003), available at http://www.fasab.gov/pdf/mou05222003.pdf.
[242] See Accounting for Social Insurance, Statement of Recommended Accounting Standards No. 17 (Fed. Accounting Standards Advisory Bd., Aug. 1999) [hereinafter Accounting for Social Insurance], available at http://www.fasab.gov/pdf/17_ss.pdf (last visited Oct. 14, 2003).
[243] Id. at paras. 65–72.
[244] Id. at paras. 73–79.
[245] See id. at para. 74 (“[A]n accounting liability should be recognized at an earlier point than when payments are due and payable.”).
[246] Id. at para. 79.
[247] Id. at paras. 24–27.
[248] Id. at para. 27(3).
[249] Id.
[250] The closed-group liability number is based on a seventy-five-year projection, as opposed to the one-hundred-year projection typically used by the Office of the Chief Actuary.
[251] See Reclassification Statement, supra note 186, at 2.
[252] See 2002 Performance Report, supra note 171, at 75–78. Starting with the 2003 report, the trustees have begun to report alternative long-term measures of insolvency, but not the line-by-line components that FASAB requires. See 2003 Trustees Report, supra note 8, at 63.
[253] See 2002 Performance Report, supra note 171, at 77.
[254] See, e.g., Office of Mgmt. & Budget, Analytical Perspectives: Fiscal Year 2003, at 32–33 (2002).
[255] A key difference between this Article’s proposals and the most analogous proposals advocated in the development of the FASAB statement is the Article’s recommended distinction between benefits accrued to date and benefits that will accrue in the future. While the FASAB approach employs an accrued-liability concept for participants who are already eligible for retirement, it uses a net present value computation for future benefits and payments for other participants—in effect, lumping accrued benefits with those to be accrued in the future and then deducting future taxes from those amounts. As explained above, there are a number of advantages to distinguishing accrued benefits from those that will accrue in the future. See supra Part III.C.2. Among other things, this distinction reflects the greater moral obligation of the government to honor accrued benefits.
[256] See Reclassification Statement, supra note 186, at 3–4.
[257] See Accounting for Social Insurance, supra note 242, at para. 7.
[258] See supra Part I.A.2.a.
[259] See Accounting for Social Insurance, supra note 242, at para. 78.
[260] See generally Sheetal K. Chand & Albert Jaeger, Aging Populations and Public Pension Schemes (IMF Occasional Paper No. 147, 1996).
[261] Id. at 36.
[262] Id.
[263] A good example of this literature is Robert Holzmann, Financing the Transition to Multi-Pillar (World Bank Soc. Prot. Discussion Paper No. 9809, 1998) (estimating the implicit pension debt of OECD member countries). Other examples abound. See, e.g., Sergio Clavijo, Fiscal Effects of the 1993 Columbian Pension Reform (Working Paper of the Int’l Monetary Fund, IMF. Doc. WP/98/158, 1998); Edward Whitehouse, Pension Reform in Britain (World Bank Soc. Prot. Discussion Paper No. 9810, 1998); Cheikh Kane et al., Brazil: Social Insurance and Private Pensions (World Bank Rep. No. 12336-BR, 1995); China Pension System Reform 125–29 (World Bank Rep. No. 15121-CHA, 1996). See also IMF Fiscal Affairs Dep’t, Draft Manual on Fiscal Transparency 47 (Oct. 19, 1998) (endorsing measurement of “unfunded public pension liabilities alongside public debt in assessing sustainability” of current fiscal policies).
To be fair, other studies of the pension liabilities in the international context employ cash-flow analyses similar to the ones included in the trustees’ reports. See Dave Turner et al., The Macroeconomic Implications of Ageing in a Global Context 7 (Ageing Working Papers, OECD Doc. AWP 1.2 Eng, 1998). See also Louise Fox & Edward Palmer, Latvian Pension Reform29–32 (World Bank Soc. Prot. Discussion Paper No. 9922, 1999); Manfred Koch & Christian Thimann, From Generosity to Sustain-ability: The Austrian Pension System and Options for its Reform 16–23 (Working Paper of the Int’l Monetary Fund, IMF Doc. WP/97/10, 1997). Even Chand and Jaeger have written papers that employ cash-flow analyses. See Sheetal I. Chand & Albert Jaeger, Reform Options for Pay-As-You-Go Public Pension Systems 12–13 (World Bank Soc. Prot. Discussion Paper No. 9927, 2000).
[264] See 2003 Trustees Report, supra note 8, at 61–63.
[265] See supra text accompanying notes 79–81.
[266] Chand & Jaeger, supra note 260, at 27 tbl.16.
[267] A recent paper by a group of World Bank economists on pension reform in China, for example, discusses the implicit pension debt of a wide range of countries measured as a percentage of their GDP. See Yan Wang et al., Implicit Pension Debt, Transition Cost, Options and Impact of China’s Pension Reform 9–13 (World Bank Policy Research Working Paper No. 2555, 2000) (estimating implicit pension debt of the United States at 113% of GDP). See also Richard Disney, OECD Public Pension Programs in Crisis: An Evaluation of Reform Options 4–11 (World Bank Soc. Prot. Discussion Paper No. 9921, 1999); Robert Holzmann et al., Pension Systems in East Asian and the Pacific: Challenges and Opportunities 30–32 (World Bank Soc. Prot. Discussion Paper No. 0014, 2000) (estimating implicit pension debt as a percentage of GDP in Korea, the Philippines, and Thailand).
[268] See supra Parts II.C.2–.3.
[269] See Richard Disney, How Should We Measure Pension Liabilities in EU Countries?, in Pensions: More Information, Less Ideology 95–111 (Tito Boeri et al. eds., 2001).
[270] Id. at 102.
[271] In more recent work, many other economists are using measures of accrued liabilities as the appropriate estimate of implicit pension debt. See, e.g., Yan Wang et al., supra note 267, at 9. For a recent overview endorsing an accrued liability definition of implicit pension debt for purposes of international comparison, see generally Robert Holzmann et al., Implicit Pension Debt: Issues, Measurement and Scope in International Perspective (World Bank Pension Reform Primer, 2001).
[272] See supra text accompanying notes 63–65.
[273] See infra Part IV.B.2.
[274] To be fair, other economic studies use discounted cash-flow systems to estimate the unfunded liability of the Social Security system or to calculate the implicit rate of return on Social Security contributions. For examples of economists’ estimates of the system’s unfunded liabilities, see supra note 127. For a well-known example of economists’ using discounted cash-flow analyses to estimate rates of return on Social Security contributions, see John Geanakoplos et al., Social Security Money’s Worth, in Prospects for Social Security Reform 79–151 (Olivia S. Mitchell et al. eds., 1999). See also 1994–1996 Advisory Council Report, supra note 2, at 15–21 (making extensive use of money’s-worth analysis for various reform options); Steuerle & Bakija, supra note 60, at 106–15 (estimating net transfers from Social Security for various income groups and generational cohorts). Lawrence Kotlikoff’s work on generational accounting is similarly spirited in its comparison of the net present value of Social Security contributions and benefits by generational cohort. See, e.g., Laurence J. Kotlikoff, Generational Accounting: Knowing Who Pays, and When, for What We Spend ch. 5 (1992).
[275] See Cong. Budget Office, Social Security and Private Saving: A Review of Empirical Evidence 10 (July 1998) (reviewing thirty empirical studies of the effects of Social Security on savings).
[276] See 2003 Trustees Report, supra note 8, at 63.
[277] See generally Olivia S. Mitchell et al., Social Security Earnings and Projected Benefits, in Forecasting Retirement Needs and Retirement Wealth (Olivia S. Mitchell et al. eds., 2000) (describing study’s methodology for valuing Social Security benefits).
[278] Id. at 345–50.
[279] See, e.g., Arthur B. Kennickell & Annika E. Sunden, Pensions, Social Security, and the Distribution of Wealth 7–9 (Fed. Reserve Bd., Finance and Economics Discussion Paper No. 97-55, Nov. 6, 1997) (analysis based on Survey of Consumer Finances, a study sponsored by the Federal Reserve Board and Internal Revenue Service triennially).
[280] In the Kennickell and Sunden paper, only employee contributions are deducted (as opposed to combined employer and employee contributions). See id. at 9, n.11.
[281] See, e.g.,Alan L. Gustman et al., Pension and Social Security Wealth in the Health and Retirement Study, in Wealth, Work and Health: Innovations in Survey Measurement in Social Sciences 150 (Robert Willis ed., 1999) (using Health and Retirement Study data to estimate Social Security wealth).
[282] See id. at 162 tbl.2.
[283] There are numerous other cases where economists quantify Social Security benefits in a similar manner. For example, studies that explore the effect of Social Security on retirement decisions often quantify the effect of Social Security by estimating the amount by which the accrued value of Social Security benefits increases over the course of a year. See, e.g., Courtney Coile & Jonathan Gruber, Social Security and Retirement 14 (Boston Coll. Ctr. For Ret. Research, Working Paper No. 2000-11, 2000) (“We initially follow the literature and focus on accrual, the change in [Social Security Wealth] resulting from an additional year of work.”).
Other economic writing looks at the relationship between the accrual of Social Security benefits and the accumulation of pension wealth. See, e.g., Christopher M. Bone, Actuarial Perspectives on Implications of Social Security Reform for Employer Sponsored Pension Plans (Pension Research Council, Working Paper No. 97-01, 1997). See also Peter R. Orszag, Should A Lump-Sum Payment Replace Social Security’s Delayed Retirement Credit? (Boston Coll. Ctr. for Ret. Research, Issue Brief No. 6, Apr. 2001) (estimating the value of delayed retirement by taking the discounted present value of additional benefits accrued).

If you have any questions or comments about JOL, please contact us: hlsjol[at]law[dot]harvard[dot]edu
Last modified: September 16, 2005.

ARTICLE ACCOUNTING FOR SOCIAL SECURITY AND ITS REFORM

Table of Contents

I. A Critique of the Current Accounting System for Social Security

A. Analysis of Accounting for the Trust Funds

1. Standard Presentation

a. Short-Range Projections

b. Long-Range Projections

2. Criticisms of the Standard Presentation

a. Short-Range Projections

b. Long-Range Projections

c. Biases with Respect to Reform Proposals

B. Consolidation Within the Federal Budget Process

1. The Current Approach to Consolidating Social Security Trust Funds

2. Criticisms of Standard Presentation

a. Misleading Features of the Standard Presentation

b. Familiar Criticisms of Relying on Total-Budget Aggregates

i. Proper Characterization of Trust Fund Reserves

ii. The Economic Realities of Social Security

c. A Deeper Critique

d. Biases for Reform Proposals

i. Perverse Budgetary Incentives for Social Security Reforms

ii. Distorting Effects on Reforms That Change TrustFund Operations

iii. Generating Reform Proposals with Ulterior Motives

II. An Accrual Accounting System for Social Security

A. Possible Components of an Accrual-Based System of Accounting

B. A GAAP-Style Accrual Accounting System for Social Security

1. Balance Sheet

2. Income Statement

C. Comparing Accrual-Based Financial Statements with Other Measures of Long-Term Solvency

1. Closed-Group Unfunded Obligation

2. Open-Group Unfunded Obligation: Seventy-Five-Year Horizon

3. Open-Group Unfunded Obligation: Infinite Horizon

4. GAAP-Style Accrual Accounting vs. Alternative Measures of Long-Term Solvency

a. Distinguishing Benefits Earned-to-Date from Total Benefits

b. The Inclusion of Benefits-To-Be-Earned and Future Taxes

5. Modified Accrual Accounting

D. Accrual Accounting and the Unified Federal Budget

1. Full Consolidation Within the Federal Budget

2. An Unconsolidated Approach to Social Security Accounting

a. Budgetary Aggregates

b. Interest Payments and Other General Revenue Contributions

c. Measures of Public Debt Outstanding

d. Separate Financial Statements for Trust Funds

III. Accrual Accounting and Social Security Reform

Two Basic Presentations of Social Security Finances Under Modified Accrual Accounting

A. Clarifying the True Nature of Social Security Finances

1. Debunking the Debilitating Myth of Current Surpluses

2. Focusing Public Attention on the Trust Funds’ MountingImplicit Debt

3. Developing Appropriate Targets for the Implicit Debt of Social Security

B. Enhancing the Quality of Public Debate over Reform Proposals

1. Exposing Short-Range Cash-Flow Effects

2. Recognizing the Value of Reforms with Future Cash-Flow Effects

3. Flagging Reform Proposals That Expand Accrued Liabilities

4. Eliminating the Cliff Effect

5. Identifying Back-loaded Reform Proposals

6. Flagging the Amount of Excess Future Taxes

C. Altering the Optics of Reform Proposals

1. Distinguishing Accrued Rights of the Elderly and Other Participants

2. Differentiating Prospective and Retroactive Effects

a. The Possibility of Prospective Benefit Cuts

b. More Modest Methods of Enhancing Benefits

3. Changing Expectations of the General Public

D. Accrual Accounting and Individual Accounts

1. Posing the Problem of Transition Costs

2. Eliminating Budgetary Anomalies

3. Comparing Traditional Benefits and Individual Accounts

E. Social Security Reform, the Federal Fisc, and National Savings

1. Budgetary Effects of Fully Unconsolidated Treatment

2. Budgetary Effects of General Revenue Contributions

3. Tracking Gross Public Debt and Unfunded Trust Fund Liabilities

F. Substantive Implications of Accrual Accounting

IV. Postscript on Normative Baselines

A. Accrual Accounting and the Federal Government

1. Accrual Accounting in General

2. Accrual Accounting for Social Insurance

B. The Estimate of Public Pension Plan Obligations in Other Contexts

1. Measures of National Public-Pension-Plan Obligations

2. Economic Estimates of the Value of Social Security Benefits

a. Life Cycle Savings Literature

b. Household Wealth Literature

ARTICLE

ACCOUNTING FOR SOCIAL SECURITY AND ITS REFORM

ii. Distorting Effects on Reforms That Change Trust
Fund Operations

2. Focusing Public Attention on the Trust Funds’ Mounting
Implicit Debt