Budgeting for Mandatory Spending: Prologue to Reform

This paper identifies a feature of the federal budget process that, consistent with the findings of behavioral research, increases the difficulty for policymakers of restraining the growth of mandatory spending to sustainable rates: budgetary accounting. Specifically, use of cash&#8208;basis accounting, on&#8208;budget payment accounts, and a narrow definition of debt defers recognition of the cost of mandatory spending until benefits are payable and politically unavoidable. Acting to control &#8220;future&#8221; costs is cognitively more difficult for decision&#8208;makers than addressing cost now as obligated. The paper proposes a trial of alternative budgetary accounting for mandatory spending that saliently recognizes cost as it accrues.


INTRODUCTION
Under current U.S. budget policy, federal debt/GDP is projected to rise without foreseeable limit (Congressional Budget Office [CBO] 2018). This path implies commensurate future net taxes, increases risk of debt-driven fiscal crises, restricts the ability of government to respond to future shocks and changed priorities, lowers standards of living, and increases the chances that government will rely more heavily on newly created money to finance its obligations. Those outcomes threaten economic, political, and social stability (Burman et al. 2010;CBO 2018).
A proximate driver of rising debt under current law is the growth in mandatory spending, 1 which constitutes about twothirds of federal outlays, up from 25 percent in 1962 (Office of Management and Budget [OMB] 2017). Those payments are largely for Social Security, Medicare, Medicaid, federal civilian and military pensions, other post-employment benefits, and various safety net and insurance programs. 2 Mandatory spending grows with the number of eligible beneficiaries and prices, without further action by the Congress. The procedural automaticity of mandatory spending often prompts Members and observers to complain that growth in those programs is "walled off" from budget decisions.
Although the Congress and the Executive Office of the President are aware of the risks of current policy, they procrastinate. Indeed, they are more inclined to cut taxes and increase spending than to address the existing imbalance. Delaying a corrective response is problematic because adjusting fiscal course sooner rather than later would reduce the excess burden of uncertainty from government indecision (Gomez, Kotlikoff, and Viceira 2012), increase efficiency in terms of smoothed tax rates and consumption (Shaviro 2009), and require a smaller change with lower transition costs (e.g., Blahous 2016). In a world of consistently rational, high-energy agents and principals, devoid of weak self-control, present bias, or cognitive limitation, inaction would be an informed, mindful choice. 3 Neither accounting nor other features of the budget process would affect decisions. Agents would "see through" accounting veils and other process distortions to the underlying reality and make the same

APPLICATIONS FOR PRACTICE
Budgeting is a choice process for getting maximum benefit from limited resources.
Informed budget choice requires best feasible estimates of the resource constraint and total expected costs and benefits of alternative beneficial uses of resources at the time of decision.
A budget process lacking this information is likely to underperform.
Cost estimates of deferred payment programs are especially susceptible to misstatement in cash-basis budgetary accounting systems.
Adding, improving, or increasing the salience of relevant information-though itself costly -can improve budget decisions and increase well-being.
1. I use the term "mandatory" spending in the sense of the Budget Enforcement Act of 1990, to distinguish it from "discretionary" spending, which is subject to annual control through legislated appropriations. The defining characteristics of mandatory spending are that existing legislation has provided authority to obligate funds without future legislative action, and that beneficiaries have a political claim to "earned" benefits, even though they do not have a legally enforceable claim, if the authorizing law is changed. Those programs are also referred to as direct spending. For a more detailed, nuanced discussion of mandatory spending, see White (1999).
2. Social Security and Medicare dominate. Together they account for 60 percent of mandatory spending and more than 40 percent of all federal spending.
3. Inaction under current economic conditions may not be strong evidence of irrationality. It may be rational to defer tax increases and spending reductions when interest rates are low (Elmendorf and Sheiner 2016). The issue is rather the fixed effect of process on decisions, including inaction, both when inaction is optimal and sub-optimal.
choices. Beneficiaries would accurately value future gains against short-term losses. Process would not matter.
With less able and energetic decision-makers, however, "process architecture" can nudge decisions toward sub-optimal alternatives, including inaction (Alm and Bourdeaux 2014;Congdon, Kling, and Mullainathan 2011;Kahneman 2011;Thaler 2015;Thaler and Sunstein 2008). 4 In the case of federal budgeting for mandatory spending, several features of the current process have been associated with decision avoidance. For policymakers, those include complexity of choice, absence of salient relevant information, and immediate cost for future benefit. For beneficiaries, inaction may be preferred to near-term losses of sums already credited to beneficiaries' mental accounts, even if the change offers seemingly greater present value.
From an accounting perspective, the current process fails to support budgeting for mandatory spending because of the failure to recognize cost when it is controllable at the decision to incur it. Instead current accounting recognizes cost only when payments are made to or on behalf of beneficiaries, that is, when non-payment would be a grievous breach of acceptable norms of behavior. Simply, current accounting gives prominence in the budget process to a politically sunk, unavoidable measure of cost.
The remainder of this paper is organized as follows: the next section details the current budgetary treatment of three mandatory programs: a defined benefit pension plan; Old-Age, Survivors, and Disability Insurance (OASDI); and the Federal Employees Thrift Savings Plan (TSP). This review highlights a key accounting difference between the defined benefit and OASDI programs on one hand and the TSP on the other. That difference suggests the potential contribution of an alternative budgetary accounting to improved management, planning, and control of mandatory spending, equivalent in expectation, to that of discretionary spending.
Second, the paper reviews and evaluates the principal arguments for current practice. Those include that government has no compelling obligation to make mandatory payments beyond those already paid and that future mandatory spending is so uncertain, most especially for Medicare and other health-care programs, that meaningful long-term budgeting is neither feasible nor would it be fair to future generations (White 1999(White , 2016. The third applies the proposed alternative accounting model, using a federal defined benefit plan and Social Security as examples. A key feature of the alternative model distinguishes previously accumulated, now "earned," benefits from future, to-be-acquired benefits whose value and cost are subject to ex ante adjustment. This section also includes illustrative, order of magnitude estimates, using actual budget results from FY 2016, of the aggregate budget effects of adopting the alternative accounting for federal defined benefit pensions and OASDI. It also addresses some likely objections to the proposal. A concluding section suggests that, while the risks of perverse effects of the proposed alternative appear modest, 4. A classic case is the decision by employees to participate in an employer-sponsored pension plan and the effect of automatic enrollment on that decision (Thaler 2015, Ch. 31). uncertainty about the details of enacting legislation and its effects on decisions argues for a limited pilot trial.

BUDGETARY ACCOUNTING FOR FEDERAL RETIREMENT
FERS is the federal pension plan for most civilian employees hired after December 31, 1983. 5 Adopted in 1986, it re-structured the existing system to align more closely with private plans and to increase the portability of earned benefits for employees moving between federal and nonfederal employment. FERS retained, at a reduced level, the defined benefit of its predecessor, Civil Service Retirement System (CSRS), and assigned its costs more fully to employing agencies, rather than to the general fund of the Treasury (Torregrosa 2003). It also added Old-Age, Survivors and Disability Insurance (OASDI), and a defined contribution component, Thrift Savings Plan (TSP). 6 Policy makers rely on the Congressional Budget Office (CBO) and the Office of Management and Budget (OMB) for relevant budget information about the cost of federal programs. This information is especially salient in estimates of the cost of proposed legislation. Because the budget uses cash-basis accounting for most 7 spending, the literal bottom line of a cost estimate for new legislation is its effect on annual cash revenues, cash outlays, and the resulting cash deficit/surplus for a 10-year budget window.
Although annual budgetary cash flows garner most of the user's attention, the budgetary accounting system also records intra-governmental accrual flows that affect account balances. 8 For example, federal agencies make actuarially determined, accrual payments that approximate the current period net cost to the government of FERS defined benefit pensions.
5. The federal government also provides specialized pension plans based on the occupational requirements of some employees. Those include plans for the military, foreign service, law enforcement, firefighters, air traffic controllers, members of Congress, and Congressional employees. Except for the military retirement system, which has 20-year, cliff vesting for the defined benefit component (to which was added a defined contribution component effective January 2018), most of those are variations on the FERS accounting model (see https://www.opm.gov/retirementservices/fers-information/). 6. Defined benefit plans provide a specified retirement benefit, for example, a fixed percent of the average of highest three years of pay for each year of covered service. Defined contribution plans provide a specified (defined) annual contribution to an employee's retirement account that may be invested in securities of varying risk chosen by the employee. The pension benefit of a defined contribution plan depends on the account's value at retirement and varies with realized returns on amounts contributed by the employee and employer. 7. Exceptions employing an accrual-basis of accounting include interest on debt held by the public, direct loans, loan guarantees, other accounts payable and receivable, and lease-purchases. See Summary Table S 8. In addition, CBO occasionally provides accrual cost estimates for a cohort of new employees as supplementary budget information for some legislation, including alternative options for changing civilian retirement plans (CBO 2017).

Budgetary Accounting for Defined Benefit Pensions
The budgetary accounting model used for the FERS defined benefit plan, and its predecessor, CSRS, is applied to most other deferred payment, mandatory spending. We therefore consider its budgetary treatment in some detail. The Civil Service Retirement and Disability Fund (CSRDF), which serves as the funding account for both the old CSRS, now closed to new employees, and the newer FERS defined benefit plan, is the key budget account. 9 CSRDF is credited with earmarked receipts, including agency and employee contributions, interest on account balances, and other transfers from Treasury. It is charged with outlays for annuity payments and administrative expenses. Figure 1 shows the principal financial flows to and from this account. The horizontal line across the figure is the accounting boundary between the federal entity and the rest of the world. The federal side has three accounts: employing agency; U.S. Treasury; and the CSRD Fund. The relevant non-federal entities are current federal employees (EMP) and annuitants. All cash flows that cross the budget boundary are recognized in the budget when they occur as outlays or receipts and increase/decrease the budget deficit accordingly. Flows between on-budget accounts are intra-governmental; the outlays of one account are receipts of another. The two effects cancel on consolidation of all budget accounts. Intra-governmental transfers, therefore, have no effect on aggregate budget outlays, receipts, or the deficit when they occur.

FIGURE 1
Accounting for Accrual of Defined Benefit Pensions in Budget (Current Practice) 9. The use of a single Fund for substantially different retirement systems makes it difficult to track the cash flows or changes in the fiscal position for either using budget documents. OPM issues an annual report for the Fund that discloses financial information including accrual of benefits for CSRS and FERS separately, but discrepancies exist between the data in the Reports and the budget Appendix volume. In addition, the Annual Report's release of accrued pension benefits lags the release of the budget. The FERS defined benefit plan, by contrast, estimates the annual normal cost of accruing pension claims using an entry age normal method. 10 That cost is 15.1 percent of pay, with employing agencies contributing 11.9 percent of pay for all employees. Employees hired after 2013 pay 4.4 percent-those hired earlier pay less. Thus, total contributions received by the Fund for post-2013 hires are 16.3 percent of pay (Budget of the U.S. Government À Fiscal Year 2018). The excess 1.2 percent (16.3-15.1) of wages and salaries reduces previous CSRDF under-funding.
Contributions to the Fund by federal agencies do not affect budget outlays or the deficit until benefits are paid to annuitants. Employee contributions, however, cross the budget boundary when paid and are shown in the budget as revenues. Similarly, annuity payments by the Fund increase budget outlays and the deficit when paid. Table 1 shows amounts for those flows reported in the budget for FY 2016. 10. The entry age normal actuarial cost method allocates the present value of pension benefits on a level annual basis over the expected service life between entry and exit age. The allocation for a single year is "normal cost." The cumulative excess of past cash inflows over outflows for the CSRD Fund produces a positive balance that is held in notional assets made up of special issue, interest-bearing Treasury securities. 11 Interest on balances provides a third source of spending authority for the Fund and is essential to the determination of periodic actuarial cost. Fund balances held in Treasury securities are not classified as debt held by-or owed to-the public, including federal employees, whether active or retired. Rather, the Fund's claims are reported as intragovernmental debt of Treasury to the Retirement Fund, or more vividly, as "money the government owes itself." By this accounting, the government avoids recognizing in the budget deficit and the debt an obligation for earned and promised defined benefit pensions, until those are paid to beneficiaries.
When cash outflows of the Fund exceed cash inflows, the Fund draws down its accumulated balances to finance the shortfall. This, in turn, requires Treasury to borrow those sums from the public. Only when government draws down fund balances to liquidate obligations to annuitants is Treasury's intra-governmental debt transformed into debt owed the public.
The government's net budget cost of defined benefits for any fiscal year, and the program's reported effect on the government's fiscal position, or deficit, is the Fund's net cash flow with non-federal entities. That is, annuities, expenses, and other distributions paid (outlays) less contributions received from employees (revenues). This measure is problematic for measuring the current period cost of defined benefit pensions because it focuses users' attention on the difference between current period payments to annuitants and contributions from current employees. The first is a legacy cost that cannot be reduced without reneging on an obligation to pay deferred compensation. The second is an earmarked revenue. Neither is a cost of labor services to the government in the current budget period.
Nor are the other intra-governmental inflows to the Fund current period labor costs. Credits from the general fund of the Treasury amortize previous underfunding of the CSRS defined benefit. Interest on Fund balances-corresponding to the time value of money-maintains the present value of past contributions. Both are used to pay promised benefits to annuitants, but neither is a current labor cost (Lucas 2003).
Current accounting also renders budget cost estimates for new legislation largely insensitive to changes in pension benefits. For example, a proposal to increase defined benefit pensions for current employees would have little budget cost until those employees begin to receive payments. Even with a 10-year budget window, cost estimates for changes in long-term benefits for current employees exclude much of the current period cost (or saving) of enabling legislation. 12 Budget planning and management is served poorly by current budgetary accounting for defined-benefit pensions because of the absence of a salient, comprehensive, and timely measure of controllable current period labor costs, and the exclusion of earned defined benefits from government's reported debt until paid. Policymakers could develop more relevant cost estimates 11. Fund balances are budget authority, or the authority to issue obligations that will lead to outlays (Isaacs 2015). 12. CBO (2017) shows 75-year effects for some pension options as supplementary budget information.
for themselves from currently disclosed information, but that would require bootstrap effort and the result would differ from budget scoring.

Social Security
Budgetary accounting for OASDI largely follows the general model applied to FERS/CSRS defined benefits. 13 A significant difference from the FERS defined benefit, but in common with the old CSRS system, is that instead of actuarial estimates of periodic costs, employer and employee contribution rates are established in statute at levels that in aggregate and over time, are less than the cost of benefits. Further, in contrast to the practice of amortizing the accumulating CSRS shortfall over 30 years with transfers from the general fund of Treasury, OASDI has no routine budget provision for correcting underfunding.
The 2016 Annual Report of the Social Security Trustees estimates the employer/ employee combined statutory OASDI rate of 12.4 percent of taxable wages falls short of full funding by an average of 2.66 percent per year for the next 75 years. That rate is the immediate and permanent increase required to balance revenues and outlays over the projection period. Estimates of annual cash shortfalls increase from 1.10 percent of taxable wages in 2016 to 2.92 percent in 2030 and 4.35 percent in 2090 (Social Security Administration 2016, 208). Table 2 shows the financial flows reported in the budget for a combined OASI and DI Fund account. 14 Intra-governmental transfers credited to the fund include agency payments of the employer's contributions and Treasury interest on Fund balances. In addition, intragovernmental transfers from Treasury replace amounts lost to the Fund during the payroll tax "holiday" 15 enacted as economic stimulus in 2010 and credit the Fund with income taxes collected on Social Security benefits.
Social security is a near-universal, federal, defined benefit retirement and disability program. It is only secondarily a component of the retirement program for federal employees. As such, government's financial interests extend to the accumulating shortfall for all beneficiaries, not just for its own employees. From that perspective, Table 2 is missing an important element for budgeting: the unrecognized shortfall in the statutory combined employer/employee contribution rates for all federal and non-federal covered employees. The 13. A minor difference is that instead of a single on-budget Fund for accumulating balances of budget authority and paying benefits, this program uses two accounts, one each for OASI and Disability Insurance. In addition, the OASDI receives most of its funding from outside the government in the form of employee and employer payroll tax contributions. This difference requires the addition of non-federal employers to the EMP box on the non-federal side of the budget boundary in Figure 1.
14. Under current law (Budget Enforcement Act 1990), Social Security is off-budget for all budgetary purposes. CBO and OMB comply with this requirement, without affording the OASDI Fund non-budgetary status, by dividing the unified budget into two parts: on-budget and off budget. The off-budget component consists of OASDI and the U.S. Postal Service. Budget analyses and data provided by the central budget agencies focus on the unified budget, including Social Security. 15. The "holiday" reduced the employee's rate from 6.2 to 4.2 percent of taxable wages.
cash-basis shortfall in 2016 (1.10 percent of taxable wages) was about $67.5 billion. In addition, the budget recognized $810.7 billion in revenues from non-federal employees and employers as inflows of fiscal resources without recognizing any corresponding obligation to pay future benefits.

Thrift Savings Plan
Budgetary accounting for TSP is the most transparent of federal retirement programs. Current accounting for TSP explicitly recognizes new obligations in aggregate budget outlays as they accrue. And, because the Plan has used this accounting from its inception, the total accumulated present value of TSP benefits has been recognized in outlays and the deficit. This result has been achieved by cashing out obligations with transfers from the employing agency to non-budgetary Fund accounts. Those accounts are defined to be on the non-federal side of the budget boundary, even though some are held in U.S. Treasury accounts. In 2016, Federal employees earned and employing agencies were scored with $8.5 billion in outlays credited to the TSP Funds (Budget of the U.S. Government À Fiscal Year 2018, 1234). As illustrated in Figure 2, classifying an account such as the TSP Funds as below the deficit line, as a "means of financing [the deficit] other than borrowing from the public" has the following effects on key budget measures: Agency contributions to the retirement accounts cross the budget line and are included in budget outlays and the deficit as paid; but employee contributions to the non-budgetary retirement accounts are excluded from budgetary receipts; Interest on debt held by the Funds is scored as net interest of Treasury as it accrues rather than as a pension cost when paid.
In terms of administrative law, the non-budgetary status of the TSP Funds is due to provisions in the Federal Employee Retirement Act of 1986 (Federal Employee Retirement Act 1986). That statute specifies that balances in the TSP Fund are to be held by the government in trust for the benefit of plan participants and that balances may not be appropriated for any purpose other than the payment of benefits. 16 No corresponding language is included in current law governing fund balances for defined benefits or Social Security. However, the language in the 1986 Act identifies a clear legislative path to non-budgetary treatment for the CSRD and OASDI Funds, as well as to the fund accounts of other deferred payment programs.
A key difference between defined benefit (DB) pension plans-CSRD and OASDI-and defined contribution (DC) plans such as TSP is that the employer, government in this case, bears no risk for the uncertain value of a defined contribution plan at the employee's retirement or when benefits are otherwise paid, but all the risk of the uncertain cost of a defined benefit plan. That is, the employer cashes out the cost of retirement with the agreed contribution to the (DC) fund, which the employee invests in assets of varying risk. By contrast, the final cost of a commitment to pay a fixed dollar (DB) benefit in the future is uncertain until benefits are paid. Even though the government invests DB balances in low risk Treasury debt, the future value of those balances depends on uncertain future interest and inflation rates. In the event that balances are insufficient, government is responsible for the defined benefits shortfall.
The future cost of mandatory programs promising defined services rather than defined cash benefits, notably health-care services paid for by Medicare, Medicaid, and federal retiree health insurance, is especially uncertain (White 2016). That is because of the difficulty of accurately projecting the cost of medical services when today's working generations reach the age of eligibility. However, this uncertainty is inescapable given the commitment to provide specified services in the future. Under current budget accounting, uncertainty is addressed by deferring recognition of all costs until payment is made and the cost is known.
A more rational response to cost uncertainty would be to form a best-estimate for decision and subsequently revise that estimate as new information arrives. Under the proposed accounting, estimated expected cost would be recognized as claims to benefits accrue, with annual reestimates also recognized as increases or decreases in budget cost. That is, uncertainty would be addressed by systematic integration into the budget process, rather than defaulting to a budget cost of zero (Bhatti and Phaup 2015;Phaup and Torregrosa 1999).

DEFENSE OF CURRENT BUDGETARY ACCOUNTING, REASSESSED
Current budgetary accounting for mandatory spending is often defended on grounds of consistency with the absence of a constitutionally protected right to benefits, if current law were to be changed. 17 That is, elected policymakers could change the law at any time and terminate benefits, including for those already receiving payments. A Supreme Court decision upholding a legislative revocation of benefits is frequently cited to support this position. 18 According to this view, the budget properly delays recognition of current law mandatory obligations until government forgoes exercise of its option to terminate and makes payment.
17. This is a variation of the doctrine of "legislative sovereignty," the dictum that no sitting legislature can bind subsequent legislatures, or otherwise prevent future changes in current law. Use of "legislative sovereignty" in this context seems ironic. Existing mandatory legislation and budgetary accounting demonstrate precisely how one Congress can tie the hands of future legislators. Occasional reductions in net benefits, the third rail of American politics, have been achieved only through extraordinary political effort. This triumph of past legislatures inspired the title of Steuerle's 2014 analysis of mandatory spending growth, Dead Men Ruling.
18. In Flemming v. Nestor (363 U.S. 603;1960), the U.S. Supreme Court, voting 5-4, upheld the denial of Social Security benefits to Ephram Nestor under a 1954 law that withdrew benefits for those deported for membership in the Communist Party. Nestor had been receiving benefits when he was deported in 1956 for membership in the Party 20 years earlier when membership was not illegal. In its decision, the Court opined famously, perhaps recalling the Depression origins of the program as an instrument of depression relief and economic stabilization, that making Social Security benefits an enforceable property right, "would deprive it of the flexibility and boldness in adjustment to ever-changing conditions which it demands." However, for most budget purposes, for example, baselines, cost estimates, the maintained assumption is the precise opposite: that current law spending continues indefinitely. Only when the law is changed is this assumption extended to the new statute.
The use of current tax collections, especially in the early years of Social Security, to pay current benefits, so-called pay-as-you-go financing, 19 is cited as further evidence that obligations arise only as earmarked revenues arrive to pay benefits. 20 Payable benefits are also limited by current law to collections plus balances in the fund account. The OASI Fund holds balances of about $2.8 trillion, or enough to pay about three years of benefits in the absence of additional collections. 21 Finally, some analysts argue that democracy is better served by permitting future generations to determine their preferred level of benefits (White 2016).

Reconsidering This Defense: Criterion for Evaluating Accounting Systems
Accounting systems do not exist in nature; they are entirely a human construction devised to provide information to support decisions. System performance may be evaluated only by the extent to which the needs of purposeful decision-makers are met. For many years, the President's budget proposal has included an extended discussion of accounting principles or "concepts" consistent with the information needs of budgeting. The core of those standards is summarized in the following: The budget needs to measure costs accurately so that decision makers can compare the cost of a program with its benefits, the cost of one program with another, and the cost of one method of reaching a specified goal with another. These costs need to be fully included in the budget up front, when the spending decision is made, so that executive and congressional decision makers have the information and the incentive to take the total costs into account when setting priorities. (Analytical Perspectives À Budget of the US Government À Fiscal Year 2019, 77). Current budgetary accounting for mandatory spending adheres to this standard only if the operative annual budget decision is whether to pay benefits this fiscal year. But that is no choice.
19. "Pay-as-you-go" implies that current spending is financed by some means other than borrowing. But a general definition of government borrowing suggests mandatory spending is financed by new debt. "Government borrowing occurs whenever the government foregoes control over some future flow of resources or benefits in order to acquire resources for current use" (Bifulco et al. 2012, 660). Also see Irwin (2015).
20. The accumulation of Fund balances increased significantly in the 1980s. In earlier periods, funding aimed at providing only amounts sufficient to pay current benefits. 21. However, executive branch budget documents, consistent with current budgetary accounting, caution beneficiaries against relying on receiving benefits from either accumulated fund balances or current period collections. For example, consider the following description of federal trust funds used to finance many mandatory programs: "In contrast [with private trust funds], the Federal Government owns and manages the earnings of most trust funds and can unilaterally change the law to raise or lower future trust fund collections and payments or change the purpose for which the collections are used" (Analytical Perspectives À Budget of the U.S. Government À Fiscal Year 2017, 77).
The political and social consequences of a decision to withhold payment on long-standing commitments render that choice unthinkable. Current budgetary accounting not only fails to support budgeting, it pre-empts choice.
The justification for current practice also provides a convenient rationalization for deferring a decision: when change becomes necessary, future legislators and electorate will have all the necessary "flexibility and boldness in adjustment" that conditions demand. Yet, every generation of policymakers faces a political imperative to "protect benefits." 22 Short of catastrophically failed government, it is not clear how the requisite "flexibility and boldness in adjustment" could materialize.

PROPOSED ALTERNATIVE ACCOUNTING FOR MANDATORY SPENDING
This approach to budgeting for long-term, deferred commitments on which people have been encouraged to rely would recognize new obligations to make mandatory payments as an annual budget outlay and include the stock of accumulated obligations in a more comprehensive measure of existing claims on future fiscal resources, debt owed, rather than the rhetorical misdirection of "held by" the public.
To measure the incremental flow of current period costs in the annual budget, the arbitrary assumption that the government will pay no future benefits is replaced with alternative, arbitrary assumptions more consistent with political constraints and useful to the information requirements of budgeting. Specifically, for budgeting, the proposal would assume: (a) a prorata share of current law benefits is earned in each year of covered employment and (b) benefits accumulated prior to a change in current law will be paid. The key feature of this change is the clear distinction between accumulated, "owned and owed" benefits and prospective "to be" acquired claims (Diamond and Orszag 2003).
Using those assumptions, the proposed accounting could be accomplished in the following steps: Fully fund estimated accrued obligations by intra-governmental transfers from Treasury to the Fund accounts; Include Fund balances in "debt owed the public;" 22. My latest Social Security Statement tells me that I ". . .have earned enough credits to qualify for benefits," provides my estimated monthly payment if I continue to work until age 76, and reports that I have also "earned enough credits for my family to receive survivor benefits" of specified amounts if I were to die this year. An asterisk leads to a notice: "Your estimated benefits are based on current law. Congress [sic] has made changes to the law in the past and can do so at any time. The law governing benefit amounts may change because, by 2034, [the trust fund balance will be depleted and] the payroll taxes collected will be enough to pay only about 79 percent of scheduled benefits." This language is inconsistent with the rationale for the current budgetary treatment because it suggests that a change in current law would restore scheduled benefits, rather than reduce them. Relatedly, CBO's long-term budget projections assume that mandatory benefits are maintained at scheduled levels even if trust fund balances are depleted (CBO 2016).
Re-classify the Fund accounts 23 as non-budgetary; 24 Outlay annual, actuarial normal costs for mandatory programs from on-budget accounts to the non-budgetary fund accounts; and Re-estimate accrued benefits and rebalance the fund accounts annually.
The scoring effects of those changes can be visualized using Figure 2. Governmental transfers to the funds would be scored as budget outlays. Re-balancing transfers would be recognized as outlays or offsetting receipts (depending on the direction of flow). Fund payments to non-federal entities or beneficiaries would not affect the budget aggregates. Earmarked tax revenues would be scored as Treasury receipts as received and simultaneously as budget outlays when credited to the Fund accounts. 25

Feasibility of Accounting Change
Much of the information required for the new accounting is already in the public domain. For example, actuarial measures of the present value of government's outstanding obligation for many mandatory programs are disclosed or recognized in the Financial Report of the U.S. Government, Annual Reports of the Social Security and Medicare Trustees, and the Annual Reports of the largest federal insurance programs. 26 CBO's annual Long-Term Budget Outlook provides cash flow projections of deficits and debt for 30 years. 27 However, disclosure of relevant information is often insufficient to ensure its use in decisions, especially if significant search or analytic effort is required (Coronado et al. 2008;Hearn and Phaup 2016a;Phaup and Kirschner 2010;Thaler and Sunstein 2008;and additional references therein). This accounting change would present mostly existing information to policymakers in a highly salient form to make informed decisions cognitively easier. It is also significant that actuarial accounting models for estimating and reporting the periodic cost of mandatory programs are now available for use (Vidal-Melia, Ventura-Marco, and Gonzalez 2018). 23. In some cases, for example, Federal Employee Retirement health care, mandatory spending is financed from an annually appropriated account. For these programs, it would be necessary also to create a Funding account.
24. Technically, the account is to be moved to the "means of financing the deficit, other than borrowing from the public." One such means is the increased balance in the account to which accrued Treasury interest has been credited but not yet paid to debt holders.
25. This simultaneous receipt-outlay accounting maintains the comprehensiveness of the budget for taxes and prevents earmarked receipts from appearing to provide resources for other uses. Inflows from sources other than the exercise of the sovereign power to tax would be credited to the fund account without affecting the budget aggregates. An example of this receipt-outlay treatment as simultaneous receipts and outlays is in Burman and Phaup (2012 One challenge to the proposal is to specify the discount rates to be used in calculating the present value of accrued, but uncertain future benefits. The choice is between the use of low-risk Treasury rates, currently used by the Social Security actuaries and risk-adjusted or "fair value" rates used by CBO in valuing some federal direct loans, including those issued under the Troubled Assets Relief Program. While the case for risk-adjusted rates is conceptually robust where government-on behalf of taxpayers-is retaining risk for the future value of defined benefits (Geanakoplos and Zeldes 2010), this "fair value" approach to valuation has proved difficult for legislators and some budget practitioners to accept.

EFFECTS OF ACCOUNTING CHANGE FOR CSRSF AND OASDI ON PROGRAM COSTS, DEFICIT, AND DEBT
Identifying the budgetary effects of converting all mandatory spending to the proposed treatment would be a major task for the central budget agencies. Here, I provide only rough approximations of the major budgetary effects of adopting this proposal for the CSRD and OASDI Funds using budget actual numbers for FY 2016. Consistent with Lucas (2003), I find this accounting change would have modest short-term effects on program outlays for defined benefit pensions and Social Security, larger changes in aggregate budget outlays due to a significant increase in Treasury interest and accelerated recognition of interest to "as earned" from "as paid" in pension benefits, and an order of magnitude increase in government debt. The increase in outlays and the deficit from the increase in Treasury interest would not affect the primary deficit, which excludes Treasury interest payments.

CSRSF Defined Benefit Pensions
As shown in Table 3, the proposed accounting would reduce program outlays by $43.4 billion, increase Treasury's net interest outlays by $56.4 billion, and increase the deficit by $13 billion. The decline in budget outlay costs results from substituting the lower cost of FERS defined benefits for the liquidation of CSRS obligations and the conversion of Treasury interest from a program cost to a Treasury outlay. The increase in the deficit is due to the accelerated recognition of Treasury interest outlays from as paid to annuitants to as earned by the CSRS Fund. An order of magnitude change occurs in the government's debt of $1.6 trillion, the sum of amounts held by the Fund before the change, formerly intra-governmental, and the additional Treasury debt required to fund total accrued obligations, about $730 billion. At the average rate of interest paid by Treasury on Fund balances to OASDI in calendar 2016 (3.12 percent), the increase in annual net interest paid to the Fund would be $56.4 billion.

OASDI
The budgetary effects of the change would be similar (adjusted for scale) to those on FERS: program outlays would decline slightly, despite large Treasury outlays to the Fund of employee and employer payroll taxes and interest, primarily because of the attribution of interest costs to Treasury, rather than Social Security. The large increase in the deficit is almost all due to the accelerated recognition of interest from when paid to annuitants to when earned and paid to the OASDI fund and the increase in the interest-earning fund balance. The near tripling of debt owed the public is the result of counting existing Fund holdings of Treasury securities in the debt and recognizing legacy amounts to be paid participants, age 62 and over, in excess of the present value of expected taxes ($12.3 trillion) and (arbitrarily) one third of the PV of the gap between expected revenues and expenditures for those who have not reached that age ($6.4 trillion; Department of the Treasury 2017, 60).

SOME POTENTIAL CRITICISMS OF THE PROPOSAL
Several objections may be raised against this accounting change. Those include that the accounting change would not directly increase the security of accrued benefits, could severely increase the volatility of budget aggregates, and make reductions in previously accrued benefits more difficult to enact. I briefly address those concerns.
The proposal does not directly address the security of promised future benefits. Continuous full funding of accrued obligations with intra-governmental, notional, that is, "fake," assets, does not, per se, increase the government's ability to pay benefits when due. Instead, full funding with Treasury debt makes salient the current value of existing liabilities under current policy. Explicitly recognizing those obligations would better inform policy makers about the magnitude of existing claims on future budget resources. More secure benefits will come about only from a more informed policy path that is consistent with that objective. Re-estimating the accrued cost of mandatory benefits annually could produce large year-toyear changes in budget aggregates, especially in response to changes in interest rates. One means of avoiding this volatility would be to follow the Credit Reform model which excludes the effects of interest rates changes from re-estimates. That practice may be justified on grounds that the effect of interest rate variations on the value of the federal government's vast, heterogeneous assets and liabilities is, to a first approximation, equal. Thus, for example, a decline in rates would increase the value of both assets and liabilities but leave the long-term net position and fiscal gap unchanged. An alternative justification for excluding interest rate changes from re-estimates may be that the added volatility from including the effects of interest rates on the value of mandatory obligations would not provide additional information useful for budget decisions. Finally, the budget aggregates could be smoothed by using a three to five year moving average of interest rates for estimating the value of previously accrued claims.
Enactment of the proposed accounting changes could also make it more difficult to reduce previously earned benefits. Indeed, one goal of this change is to reduce the chances government will cut legacy benefits either by choice or fiscal necessity. On the other hand, the change should facilitate enactment of measures to balance benefits and their financing.

CONCLUDING COMMENT
Humans often exhibit both a present bias in consumption decisions and a tendency to adopt personal and social strategies to restrain that impulse. In addition to private forced saving contracts, for example, life insurance and home mortgages, we also make use of government coercion to smooth life-cycle consumption, for example, Social Security, Medicare. Even though policymakers are themselves subject to human biases and cognitive shortcomings, we often presume that "in their role as agents in a deliberative process, [they] are protected to some extent from committing such errors in an unchecked fashion" (Congdon et al. 2011, 56; but also see Viscusi and Gayer 2015).
In the case of mandatory spending, however, policy makers have inherited a budgetary accounting system poorly suited to managing mandatory spending for sustainability and reliability. That system inhibits the adoption of corrective adjustments by highlighting sunk legacy costs rather than manageable incremental cost.
An accounting that saliently recognizes an accruing, controllable cost of deferred payment obligations could increase social welfare by facilitating legislative management of mandatory spending. It could thereby increase the reliability and value (Luttmer and Samwick 2018) of government's commitments. It also holds promise of reducing dead-weight losses from future tax increases and consumption reductions (Shaviro 2009), and mitigating the risks of social and economic turmoil from a significant cutback in long-standing, strongly implied government commitments.
The proposed changes may have appeal for both progressives and fiscal conservatives. For the former, it signals policymakers' resolve to honor past commitments and mitigates constituents' uncertainty and anxiety about the current law "requirement" that benefits must be reduced to the level of receipts when Fund balances are exhausted. For the latter, it provides a salient measure of current period costs, equivalent to the cost of discretionary spending (Penner and Steuerle 2016). For both, it may create the political space to move to a more sustainable and equitable safety net of social insurance programs.
Nonetheless, one cannot anticipate all the effects on decisions of process change (Manzi 2012) nor of the detailed form in which the proposed change may be enacted (Lucas and Phaup 2008). Accordingly, the proposal could be subjected to a limited pilot trial before consideration of application to all mandatory programs. One possibility, for example, would be to apply the proposed budgetary accounting model to federal defined benefit pensions and an insurance program such as the single-employer insurance of the Pension Benefit Guarantee Corporation. 28