New Ideas for Federal Budgeting:
A Series of Working Papers for the
National Budgeting Roundtable
Page left intentionally blank.
This series of working papers have been written for the National Budgeting Roundtable as part of an ongoing research project at the Centers on the Public Service of George Mason University’s Schar School of Policy and Government.
This research was made possible by a grant from the William and Flora Hewlett Foundation. The Roundtable is co-chaired by Dr. Paul Posner of George Mason University, Dr. Stuart Butler of the Brookings Institution, and Maya MacGuineas, President of the Committee for a Responsible Federal Budget.
The GMU Working Papers have been edited under the direction of Principal Investigator, Paul Posner.
WORKING PAPER #8
Joseph White, Ph.D.
Luxenberg Family Professor of Public Policy
Case Western Reserve University
Budgeting is a process in which decisions about program and policy details – such as spending for the FBI, how much to pay physicians for Medicare services, the top income tax rate and any exclusion of education expenses from income – are related to preferences about totals: total spending, revenue, and deficit or surplus. Decisions about details are constrained by preferences about totals; but the totals are the sum of the parts, so preferences about details shape the totals. For example, policy-makers may decide that a larger deficit is required because of national security threats; or to invest in human capital. Or, they may decide to cut spending or raise taxes in order to hit some deficit target.
The basic challenge of budgeting is how to reconcile conflicting preferences about details and totals. That reconciliation has historically been an annual process (Wildavsky 1978; Caiden 1982). Each year's budget decisions could and should, however, be informed by assessments of how they might affect future totals or details. The OECD network of Senior Budget Officials, for example, has recommended "publishing a report on long-term sustainability of the public finances, regularly enough to make an effective contribution to public and political discussion on this subject, with the presentation and consideration of its policy messages – both near-term and longer-term – in the budgetary context" (OECD 2015a: 10; see also White 2015a). In the United States, such long-term projections have been common since they were created by GAO and taken up by CBO in the 1990s (GAO 1992; CBO 1997a). They were and are used to call for budgetary caution even in good times (GAO 2001).2
Many of the leading figures in debate over the federal budget process recommend going further. They maintain that instead of being judged by the prospective deficit in the next year, the federal budget should be judged by the deficits or debt it is estimated to produce decades into the future. In short, allocation decisions should be driven by long-term goals for totals; and policy details designed and judged based on their estimated effects on budget goals 20, 30, or more years into the future. That is very different from international practice, in which "long-term projections are not used for allocation" (Schick 2009: 17).3
Thus, GAO has proclaimed that, "Long-Term Focus is Critical" (GAO 2004) and that any delay in addressing those concerns would be "destabilizing" and unwise (GAO 2011: 2). It has called for "enforcement mechanisms" to encourage better deficit control for longer periods of time (GAO 2011: 3). In 2008 a group of budget commentators including three former Directors of the Congressional Budget Office proposed thirty-year budget caps for Social Security, Medicare, and Medicaid, enforced by "automatic adjustments in benefits, premiums, provider payments, or other revenues" in order to save the nation's "fiscal future" (Antos et al. 2008: 2). Eleven members of the National Commission on Fiscal Responsibility and Reform ("Bowles-Simpson Commission") in 2010 endorsed a plan that defined the goal of budgeting as to, among other things, reduce federal debt as a share of GDP in 2035 to 40% (National Commission 2010). They clearly believed that budgets should be judged by whether they would (purportedly) hit that target. In 2011 the Committee for a Responsible Federal Budget (2011a) called for a "credible process" for entitlement "reforms," again with binding, automatic consequences if said reforms were not adopted. Alice Rivlin and Pete Domenici (2015: 13) urged reforming the federal budget process so that Congress would "enact explicit long-term budgets for Medicare, Medicaid, and Social Security as well as other mandatory programs" with "limits on automatic spending growth" that would be enforced by "reestablishing and simplifying pay-as-you-go rules for these mandatory spending programs."4
The National Budgeting Roundtable has now generated two analyses calling for enforceable long-term totals for either all or most of the federal budget. Phil Joyce (2016: 4) calls for establishing a statutory fiscal rule to determine long-term totals. Stuart Butler (2016: 3) called for "a 25-year budget for long-term mandatory programs, together with a funding plan." Each proposal says that Congress could make changes to the long-term plan. Each also, however, provides for automatic procedures to make program or revenue details fit the targets as a default. In Butler's (2016: 4) testimony, "(i)f spending or revenues for these programs exceeded or fell short of the corridor established in the original statute, automatic provisions would be triggered to maintain the original long-term budget." In Joyce's version (2016: 5), deviations from the agreed budget path would be countered with sequestration applied to "all spending and revenue changes, including tax expenditures."
Advocates for such proposals may argue they are not really long-term "budgets" because the "enforcement" provisions are not intended to go into effect.5 In this view, setting caps is not really budgeting; a budget must include decisions about budget details. In this paper I agree with the principle – deciding about totals without considering what that means for details is an irresponsible and in some cases dishonest way to budget. Yet enforcement provisions are decisions about details. They are laws that will determine policy if Congress and the president cannot agree on alternatives. This makes them just like any other budget law. As a default, the enforcement has all the power of the status quo in a political system in which legislating can be exceedingly difficult. Moreover, if Congress and the president could agree to waive the details of the enforcement, they also could agree to waive the totals – that is, to have higher deficits than planned. Therefore, proposals which combine long-term decisions about totals with enforcement procedures either should be viewed as long-term allocation decisions to reconcile details and totals – that is, budgets - or should not be viewed as providing serious caps.6 Advocates who propose "enforcement" actions that they do not mean to occur are refusing to take responsibility for their own choices. They are misleading the public, themselves, or both.
In this paper I argue that the long-term budgeting described above represents a narrow, impractical, and dangerous view of how federal budgeting should work. Budgeting based on long-term estimates requires making decisions with particularly unreliable "information" – especially, as we will see, in the case of projected spending on health care programs. It requires substituting current politicians' judgments for those of the voters who will experience policy. In some cases, especially Social Security, the policy design requires long-term commitments; and there is little reason to expect public preferences to change. In other cases, however, future choices and trade-offs logically would depend on information we cannot know. Again, this is especially true of health care. We do not know what benefits medical care will offer in thirty years; we do not know how efficiently those could be provided; and we do not know what future judgments of the relative efficiency and equity of financing health care through government will be. Therefore, we cannot set a share of GDP that government health care programs should consume without substituting our poorly informed judgment for the better-informed views of future voters.
Many of the basic purposes of budgeting, especially for non-entitlement ("bureau" – see White 1998b) programs, are best served by the traditional annual budget process. Other purposes, such as a rough setting of policy priorities and national economic policy, should be subject to change whenever a new government takes office. Any proposal to set economic policy or priorities for longer terms would require making elections and so representative government far less relevant.
Budgeting based on long-term estimates requires making decisions with particularly unreliable "information"
Federal budgeting is not working well at present, but the reasons have little to do with the rules of the budget process, and far more with broader partisan warfare.7 Demanding that budgeting solve supposed problems thirty years in the future will only intensify the blame associated with budgeting, and make it even harder to make decisions (White 2009b). It sets a standard that is both unrealistic and unwise. If policy-makers are concerned about long-term spending trends, they would do better to focus on medium-term policies for the entire U.S. health care system, rather than on long-term, largely mythical, policies for the parts that are on the federal budget.
The balance of this paper will begin with an overview of the relationship between the purposes of budgeting and the time period for which budgets are made. I will suggest why long-term decision-making fits poorly with many of those purposes, while a biennial approach makes sense for a few. Next I will discuss experience with enforcement of "caps" as applied to one portion of the budget, discretionary (bureau) spending. It is not especially encouraging. The following section considers Social Security, the part of the federal budget for which long-term budgeting is, in a sense, already established. The final section considers Medicare, and makes the case for a different approach to Medicare within the context of the U.S. health care system.
II. Annual Budgets and the Goals of Budgeting
Budgeting is supposed to contribute to continuity (for planning), to change (for policy evaluation), to flexibility (for the economy), and to provide rigidity (for limiting spending) … Obviously, no process can simultaneously provide continuity and change, rigidity and flexibility. And no one should be surprised that those who concentrate on one purpose or the other should find budgeting unsatisfactory. (Wildavsky 1978: 501)
Federal budgeting is a complex process because it makes many different decisions and must serve many different goals. Both analytic and prescriptive studies about budgeting provide summaries of those goals, which go well beyond simply "controlling deficits" (Axelrod 1988; Lewis and Hildreth 2013; OECD 2015a; Rubin 1997). Budget processes should fit with representative government by promoting accountability for and transparency of decisions. They should encourage good management and efficient delivery of services. They should add to totals that are affordable (however defined) and good for the economy (by whatever theory is chosen). None of these goals can be attained if budgets are not honest and accurate. Last but hardly least, a budget process should not set standards that lead to unmanageable conflict.
The reasons for annual budgeting. The traditional budget process or schedule in both the United States and other countries has been annual (Wildavsky 1978), but that has never meant that all decisions were made for only the following year (Caiden 1982). Just as families will make long-term commitments such as mortgages, buying cars, and committing to pay a child's college tuition, governments engage in some activities for which they make commitments for multiple years: such as buying an aircraft carrier (they take a while to build), or committing to pay pensions (such as government employee pensions, or Social Security), or promising to pay back loans (much like a home mortgage). Just as people do not normally change jobs every year, governments normally do not change their revenue sources (the tax code) every year. Yet the federal government (and almost all other firms and governments) have had annual budget processes in the sense that each year they reviewed and tried to project what the spending and revenue totals would be if they kept doing what they did before, and considered whether to make changes in budget details in order to achieve different totals.
One reason to focus on annual totals is as a basis for managing cash flow: making sure money is on hand to pay bills. This includes both a planning function (forecasting current trends and making decisions about changes from trends) and a control function (making sure government agencies spend in line with plans, and that revenues are collected as planned).8 Since adoption of the Full Employment Act of 1948 annual budgets have also been viewed as influencing the economy through shaping aggregate demand and potentially interest rates. Changes in the extent to which federal spending and taxing adds or subtracts money from the private economy will influence short-term inflation and unemployment. Therefore, the annual budget is also the logical focus for fiscal policy.
Annual budgeting also fits with concerns for democratic accountability or transparency. Annual statements of agency plans and spending or revenue prospects tell the public, or the section of the public that cares to know, what the government is doing. Publishing plans and promises and then auditing performance provide ways for citizens and their representatives to direct and/or oversee the government. This kind of review, however, must fit within the election cycle.
Annual budgeting processes in the United States also serve to provide oversight of how programs are managed and incentives for efficiency in the provision of government services. In principle, efficiency is a neutral value. From the perspective of either consumers of services or taxpayers, getting more for the money (or the same for less money) are good things.9 Allocations to agencies are justified by plans about what they will do with the money. As the sum of decisions about individual programs, the overall set of budget decisions each year will approach what W. F. Willoughby called a "general financial and work program" for a government and its agencies (Mosher 1984: 21fn6).10 Agency plans are reviewed in two ways: by the Office of Management and Budget that "scrubs the estimates" when the President's Budget is assembled, and by the Appropriations committees when agencies submit detailed justifications of the requests within the President's Budget to Congress. Annual decision-making fits best with this pursuit of efficiency, because many of the factors in agency production of goods and services will change frequently. Input costs will change with factors such as the price of fuel or trends in wages. Demands will change with events in the economy. These factors will change enough over time that plans for longer than a year can easily be superseded by events.
Efficiency is not the same as economy (Simon 1997; Wildavsky 1966). Economy means simply spending less, even if that means eliminating highly useful programs, or cutting program spending in a way that reduces output even more, so gives citizens a worse deal than they received before. Economy is the dominant goal for many participants in budget processes, especially participants who represent those who are likely to pay more of the taxes that finance spending. It is the stated goal of many budget process reforms. As we will see, however, economy is not logically especially related to annual budgeting; in fact it has been pursued in recent years by setting multi-year "caps" on spending. In contrast, annual budgeting's routines for increasing efficiency can help fit details to totals in a way that reduces the conflict between policy goals for programs and the reasons to constrain totals.11
The annual budget serves fiscal policy goals by forcing a review of trends and a statement of the government's approach. Annual budget decisions are not as effective a tool for fiscal policy as was assumed when the Full Employment Act was passed. Legislative responses to economic change are slow and often poorly timed (e.g. the spending occurs when the slump has ended). Therefore, the best response to economic cycles is "automatic stabilizers": programs that change with the economy. A graduated income tax is countercyclical because it reduces revenues in a slump and raises them in a boom. Similarly, spending on Unemployment Insurance and Medicaid grow during a slump and decline during a boom. Legislated fiscal changes in response to economic cycles make most sense in extreme circumstances, such as 2008-09.
For these and other reasons, the traditional annual budget process has persisted in spite of many criticisms. We turn now to those criticisms.
Concerns about annual budgeting. Proposals to budget for longer terms may follow from emphasis on other values, or perceived weaknesses in the current annual process.
One argument has been that the annual budget process tends to run on "automatic pilot" – to be too "incremental." From this view, the purpose of a budget should be to "set national priorities,"12 and too much is normally taken for granted. There are good reasons, however, for most of the budget to be stable from year to year.13 The reasons for one year's decisions normally apply in the next year. Stability in public policy helps businesses and individuals understand their environments. A long campaign against incrementalism by budget reformers has failed because incrementalism is a vital aid to making decisions about details and because the balance of power which establishes priorities rarely changes very much (White 1994; Schick 2009).
A related concern, highlighted in many of the proposals mentioned in this paper's introduction, is that federal budgeting puts specific parts of the budget – revenues but in those proposals especially entitlement (or "mandatory") spending on "automatic pilot." Supposedly, programs like Medicare and Medicaid, or aspects of the tax code, escape budget discipline. Such claims that either entitlements or the tax code are on automatic pilot are grossly exaggerated.14 Nevertheless, there is a distinction between discretionary programs, which receive annual appropriations and cannot function unless those appropriations are renewed, and mandatory programs, which in many cases have budget authority that continues from year to year (White 1998b).15 Congress and the president can choose not to act on most of that law – which means fail to agree on action – without drastic consequences.16
OECD's Senior Budget Officials (2015a: 6) have worried that budget processes may not be sufficiently responsive if a government wants either to change priorities or to alter fiscal policy. They therefore suggest that annual budgets be controlled by Medium Term Expenditure Frameworks (MTEFs). In the Netherlands, for example, budget priorities and targets for budget totals are negotiated as part of the coalition agreement when a new government is formed (Bos 2008). Annual budgets then are to fit into that framework (unless events lead to changes).
These concerns about broad priorities and ability to change entitlements and revenues are addressed by the current federal budget process. Presidents can propose anything they want, and consider the full scope of budget decisions. The congressional budget process includes budget resolutions which set multiyear totals, beyond the length of any presidency (recently five or ten years, though the amount has changed over time; see Heniff 2015). If the political balance allows, changes in entitlements (such as Medicare or Medicaid) or in tax law are encouraged by reconciliation instructions in the resolution. Reconciliation instructions allow congressional majorities to overcome procedural barriers to action (especially the Senate filibuster), and even to bypass uncooperative congressional committees (Kogan 2016; White and Wildavsky 1991). Major reconciliation laws are only enacted at irregular intervals, however, because opinion about taxes and entitlement programs does not change dramatically every year (remember, they don't change all that much for other programs, either). Reconciliation allows a governing majority to change all aspects of budget law – but does not create such majorities.
In practice, major changes in preferences are highly unlikely to occur more often than every two years. The governing coalition, however, could change with each election. Therefore, it would make sense to reform the current process by adopting two-year Budget Resolutions during the first year of each new Congress. These would include reconciliation instructions to fit changes in the preferences of the governing coalition, or respond to new conditions. They would also set targets for the next two years of appropriations legislation – allowing annual decisions that fit the management requirements of agencies.
Although it would address legitimate concerns about establishing a new government's policy priorities and attending to entitlement and revenue choices, making the budget resolution and reconciliation processes biennial will not address the real goals of proponents of long-term budgeting. From their perspective, the real problem with the current process is that it leads to the wrong totals and priorities. In practice, people usually support budget process reforms because they hope to get different budget results. So to clarify thinking, we need to understand different standards for evaluating budget totals. Debate about budget totals in the United States reflects deep disagreements about both the political meaning of debt and deficits and how the budget influences the economy.
American politics includes strong political pressures for the traditional "balanced budget" and deep uneasiness about debt. This demand is based partly on a false household analogy ("I have to balance my budget so the government should have to balance its budget;" (see White 1998a)), and partly on a deeply rooted belief that a government that can create debt is out of control (Savage 1988). Public finance and budgeting specialists generally disagree with both views, as is shown in both the Maastricht standards for participation in the European Monetary Union and in OECD's budgetary recommendations (2015a: 6). This deep distrust of deficits at all times is not really addressed by budgeting for the distant future. It would be better addressed by a constitutional balanced budget requirement.17 The arguments for and against long-term budgeting therefore will depend more on other beliefs about the economy or the role of government in society.
Any argument that there is a "right" deficit total that can be identified far in advance fits poorly with the Keynesian focus on demand management. There is plenty of room for disagreement about totals within that framework, mainly between policy-makers who worry more about preventing unemployment and those who are more worried about inflation (with the latter wanting smaller deficits). In either case, the "right" budget balance should depend on each year's economic conditions, so be set as close to that time as possible.
An alternative view of how government should influence the economy says its major role should be to increase national savings. This idea rose to prominence among American economists in the later years of the Carter administration. This view says budget totals should be managed to increase national savings, so ideally the budget should be in surplus.18 From this perspective, short-term demand management is much less important, and long-term spending control (but not tax reduction) especially important.19 Savings should be increased permanently, so a single year's improvement, from this perspective, is not enough. The savings argument, unlike demand management, sets no standard for totals in any given year.20
The government is too big approach is now dominant among Republican policy-makers and activists. In this view, the difference between spending and revenues is far less important than the totals for each. Lower spending and lower taxes are by definition good for the economy. A telling example was the George W. Bush administration's argument that taxes should not be raised to pay for new "homeland security" spending "because of the economic distortions introduced by the tax system" (White House Office of Homeland Security 2002: 65).21 In this view, budget rules should force reductions in spending or taxes, over any period of time.
A fourth argument emphasizes financial market “confidence” and can be applied either to annual budgets or over a longer period. It maintains that the financial markets will punish the nation for whatever budget totals the advocates for this approach do not like. This argument usually says deficits must be lowered because, otherwise, high real interest rates will torpedo the economy (White and Wildavsky 1991; Woodward 1994). Since participants in financial markets are assumed to be forward-looking, expected future deficits could be as damaging as current ones. There have been variations, such as that Carter-era deficits were fueling inflationary expectations and behavior, or that the Reagan budget package would win market confidence by stimulating expectations of rapid growth (Stein 1984; White and Wildavsky 1991). But assertions that "the markets" demand austerity are the major basis for claims that projected future deficits are a crisis that demands immediate action (CRFB 2011a; 10 Ex-Chairs 2011).22 Contrary experience both in the 1980s (with big deficits, declining inflation and a stock market boom) and in the past seven years (with big deficits and historically low real interest rates) are ignored by promoters of the market confidence view.
In principle the first economic goal, demand management, could be combined with the others, as variations from a baseline. In practice, demand management fits awkwardly with long-term budgeting, because any provisions to allow response to economic distress can be accused of being a way to bust the necessary fiscal constraints. The Peterson-Pew Commission (2011b: 19) framed the problem nicely:
There are questions about how to formulate this long-term rule in a way that is both
transparent and sufficiently flexible. The former is necessary… to ensure public pressure
is brought to bear when the rule is violated. The latter is necessary to accommodate
inevitable short-term shocks that will require deficits and thereby sustain support for the
rule over time. However, experience with cyclically adjusted balanced budget rules is
insufficient to judge whether they are sustainable both economically and politically.
It may be possible to design a fiscal rule that allows for some automatic response to deficits that exceed targets, is suspended under sufficiently dire circumstances, but cannot be manipulated to allow deviation under less dire circumstances. Such a rule has yet to be identified.
Any particular long-term budgeting proposal, then, seeks to favor some specific economic approach over others.23 Advocates for long-term budgeting hope it will help them enact their views of budget totals, and establish them over time through the "enforcement" provisions. Yet there are good reasons for skepticism about the savings, size-of-government, and financial market confidence argument. Therefore, citizens might doubt that budget processes should be designed to favor any of those positions.