Statement of
Douglas Holtz-Eakin
Director

The Economic Costs of
Long-Term Federal Obligations

before the
Committee on the Budget
United States Senate

February 16, 2005

Mr. Chairman and Members of the Committee, I appreciate the opportunity to appear before you today to discuss the budgetary portrayal of the federal government's fiscal outlook. The Congress adopts federal programs to achieve a wide range of objectives and uses the budget to indicate the costs of those policies. At the heart of the Congressional Budget Office's (CBO's) mission is the agency's responsibility to quantify the costs of federal programs and policies. Thus, with the strong caveat that I will be speaking only about costs while ignoring benefits, I want to make the following points in my statement today:

• Under current law, the U.S. government faces enormous foreseeable demands for federal spending over the long term. At the same time, it faces budgetary risks each year that are smaller in magnitude but difficult to predict.
  
  
• The commitment to spend a dollar has a cost. It makes no difference if the payment is charged to the general fund, a trust fund, or an enterprise fund. It does not matter if the dollar purchases goods and services, provides income support, or subsidizes an activity. The commitment has a cost--whether it is for a predictable outlay or for a potential outlay to liquidate a guarantee or to meet some other contingency.
  
  
• The value of government commitments is usually a good measure of the cost of government to the economy--its economic cost--because spending preempts the use of resources by others for other purposes. The dollars measure the value of forgone alternatives for the private sector and within the budget.
  
  
• The budgetary costs of federal commitments should reflect their economic costs. Even though the government commits to future spending in a variety of ways, including social and other insurance, pensions for federal workers, and the support of international organizations, all uses of funds can be compared in terms of their economic costs.
  
  
• It is timely to reassess the principles of federal budgeting to better measure economic costs. CBO has begun to provide long-term projections under a wide range of assumptions and to examine the effects of risks on spending and the costs of federal activities not currently shown in the budget.

Let me discuss each point in turn.
 

The Long-Term Outlook for Government Spending

It will not be news to Members of this Committee that the United States faces severe fiscal demands in the decades ahead. Last year, total federal spending, excluding interest payments, was about 18 percent of the nation's income. CBO projects that, on the basis of current rules for benefits, federal spending is likely to rise to a much larger percentage of the nation's income by 2050 (see Table 1). Little disagreement exists about the cause of that situation. It stems primarily from the interaction of demographic shifts, rising health care costs, and policies aimed at improving the well-being of retirees, the disabled, and the chronically ill.

Table 1.
Alternative Long-Term Paths for Primary Spending
(Percentage of GDP)
			2010		2030		2050
High Spending Path
	Defense		3.7		2.8		2.0
	Social Securitya		4.2		5.9		6.3
	Medicare and Medicaid		5.3		11.5		21.3
	Otherb		4.9		4.3		3.4
		Total	18.1		24.5		32.9
Intermediate Spending Path
	Defense		3.1		2.0		1.4
	Social Securitya		4.2		5.9		6.2
	Medicare and Medicaid		4.8		8.4		11.5
	Otherb		4.9		4.5		4.2
		Total	17.0		20.8		23.4
Low Spending Path
	Defense		3.1		2.0		1.4
	Social Securitya		4.2		5.9		6.1
	Medicare and Medicaid		4.4		5.7		6.4
	Other		4.8		3.5		2.5
		Total	16.5		17.1		16.6
Source: Congressional Budget Office, The Long-Term Budget Outlook (December 2003), p. 7.
Note: Primary spending is the sum of spending for defense, Social Security, Medicare and Medicaid, and other spending (except interest).
a. Minor differences in simulated gross domestic product (GDP) result in small differences among paths in Social Security spending as a share of GDP.
b. Other spending is lower in 2030 and 2050 under the high spending path than under the intermediate path because this category includes premiums paid by Medicare enrollees, which are treated as negative outlays, and those premiums are larger under the high path's assumption of 2.5 percent excess cost growth.

As you know, the growth in spending for health programs is far from certain. Rising health care costs are boosting spending to a greater degree than can be explained by the growth of enrollment and general inflation alone. Since 1970, those factors, as well as policy changes, have caused annual costs per Medicare enrollee to rise about 3.0 percent faster than per capita gross domestic product (GDP), on average--a difference referred to as "excess cost growth." If that growth remained high--at 2.5 percent, for example--the federal government's spending for Medicare and for its share of the joint federal/state Medicaid program would together exceed 21 percent of GDP by 2050 (compared with 4.1 percent in 2004), and total spending would be about 33 percent of GDP. CBO's intermediate projection (presented in Table 1) is based on the Medicare trustees' assumption that excess cost growth declines to 1 percent above the growth of per capita GDP. However, even at that rate, the federal costs of Medicare and Medicaid would climb to more than 11 percent of GDP in 2050.

Other potential demands, such as Social Security, defense spending, homeland security, and environmental cleanup, may also claim a substantial share of society's resources. The cost of meeting those demands is uncertain, but there is little doubt that over the long term, the federal budget faces growing pressures that are likely to absorb an increasing share of the U.S. economy.
 

Contingent Government Spending

In addition, the federal government faces significant fiscal exposure arising from its insurance and guarantee programs and other commitments that are contingent on adverse events that may or may not occur. In many cases, the federal government's commitment to cover at least a portion of a potential loss is unambiguous. In other cases, the extent to which the government would be required to cover a loss is, itself, uncertain. In virtually all cases, the exact amount that the government will be required to pay in any particular year is unpredictable.

Federal credit programs already budget the expected costs of those programs, but the government remains liable for all unexpected losses as well. The fiscal exposure resulting from guaranteed loan programs is exemplified by the mortgage insurance provided by the Federal Housing Administration (FHA) and the Department of Veterans Affairs (VA). The FHA programs provide mortgage insurance to encourage lenders to make credit available to expand home ownership. The VA program provides partial guarantees of residential mortgage loans issued to eligible veterans, reservists, and service members. Together, the face value of the outstanding loans that were fully or partially guaranteed by those programs totaled over $700 billion at the end of 2004.

The federal government also guarantees loans made to students and their parents for higher education. At the end of 2004, the outstanding volume of guaranteed student loans totaled about $250 billion. Smaller programs provide guarantees for a wide range of loans, including guarantees for certain export activities, small business loans, agricultural credit, business and industry loans, and rural housing.⁽¹⁾ At the end of 2004, the outstanding loans made by those programs had a face value totaling about $200 billion.

Through other programs, the federal government provides direct loans for various purposes. The largest such program is the student loan program run by the Department of Education. At the end of 2004, about $100 billion in direct student loans were outstanding. Other large direct loan programs provide lending for rural utilities and housing, various foreign loans, loans to support export activities, and loans to finance spectrum auction sales.⁽²⁾

A number of insurance programs also expose the federal government to risk. The Federal Deposit Insurance Corporation insures deposits in more than 9,000 commercial banks and savings associations. The Terrorism Risk Insurance Act, enacted in 2002, created a temporary federal reinsurance program to absorb most of the risk of financial loss from acts of foreign terrorism in the United States.⁽³⁾ Other programs provide insurance for losses from floods and crop failures caused by bad weather or other natural disasters.

Much attention has been focused recently on the financial condition of the pension plans insured by the Pension Benefit Guaranty Corporation (PBGC). PBGC insures projected benefits in 31,000 defined-benefit plans. For years, the corporation's receipts exceeded its benefit payments, and the budget reflected that positive cash flow instead of the underlying liability. Recently, though, PBGC estimated that insured pension plans were underfunded by more than $600 billion. So the ultimate cost of pension insurance to the government could be significantly larger than PBGC's current deficit of less than $25 billion.⁽⁴⁾

A number of other potentially large contingent liabilities are suggested by recent experience rather than by policies under current law. Although no statutory requirement exists for most federal relief following natural disasters (such as large earthquakes and hurricanes), homeowners, small businesses, and state and local governments frequently anticipate and receive substantial federal assistance for uninsured losses following those events.

The United States also supports multilateral financial institutions (MFIs)--such as the International Monetary Fund and the World Bank's International Development Association and its International Bank for Reconstruction and Development--that lend money to other member countries. Those borrowers have often gone into arrears and sometimes defaulted on their debts to other lenders. They have restructured their debts and, at times, asked for debt forgiveness. Some of the features of world financial markets that have protected MFIs from losses, such as the presumed seniority of their claims over other lenders', may not protect them in the future. The creditor countries in the G-7 are debating forgiving the debts that poor countries owe to the MFIs, and U.S. taxpayers may bear some portion of those costs.⁽⁵⁾

In addition, in financial markets, the government-sponsored enterprises (GSEs) Fannie Mae, Freddie Mac, and the Federal Home Loan Banks are perceived to be backed by an implied federal guarantee of their debt and other financial obligations. That implicit guarantee is communicated to investors through a number of provisions of law that create a perception that the GSEs have enhanced credit quality as a result of their affiliation with the government. Those provisions include a line of credit at the U.S. Treasury and exemption from state and local income taxes. In addition, although federally chartered and federally insured banks face a limit on the amounts that they can invest in other types of securities, that limit does not apply to the GSEs' securities. Taken together, those statutory privileges have been sufficient to overcome an explicit denial of federal backing that the GSEs include in their prospectuses. Assisted by the implied federal guarantee, those three companies have grown into some of the largest financial institutions in the world.
 

The Economic Costs of Federal Spending

As a general rule, the best measure of the economic burden of a government program is its spending. Consider, for example, a discretionary program financed by annual appropriations. Spending by such a program diverts productive resources from private consumption or investment to government use. If the activity replaces private consumption with government consumption, the costs are felt in the present. If, however, the effect of government spending is to displace private investment, the cost is forgone growth in the capacity of the economy to produce--a loss that persists into the future. Federal financing of expenditures, either through taxes or borrowing, reduces the resources available in the private sector, and the people deprived of those resources bear the burden of government spending.

Resources are limited. The use of resources for one purpose necessarily denies them to others--a fact of life that is sometimes easy to forget. For example, much of the discussion about future spending for Social Security and Medicare has focused on whether revenues earmarked for those programs will be sufficient and whether their trust funds will become insolvent. Although those issues may be important, they should not distract from the more fundamental economic consideration: the resources expended on those programs must be financed either by taxes or by borrowing, which implies future taxes. Thus, that spending will be just as costly as any other federal spending.

In economists' jargon, every dollar spent on a government program has an opportunity cost: that dollar is not available to be spent on something else. The cost,then, is whatever is forgone. When, as an individual consumer, I am deciding whether to buy an automobile, I am (at least implicitly) determining whether I would get more value using the money for that purpose than for any other. When, as elected representatives, Members are deciding whether to spend $100 million for a federal program, they are making a similar determination: is that the best use of taxpayers' money, given the possibility of other uses? Even though most such legislative decisions are not directly tied to decisions about taxes, the result is the same: unless other expenditures are reduced, current or future taxpayers will be required to give up the benefits from the use of those funds.

A distinction is sometimes drawn between the economic costs of government activities in which the government directly purchases goods and services, such as military procurement, and other government activities in which the government transfers purchasing power (money) to recipients, such as the Social Security program. In the first case, the government is causing taxpayers to have fewer resources at their disposal so that it can use those resources to purchase specific goods and services. In the second case, the government is reducing the resources available to taxpayers in general and is increasing the resources available to the program's beneficiaries but is not directly purchasing specific goods and services. Recipients can use the resources to buy whatever they want or save them for themselves or their heirs. In both cases, however, taxpayers are giving up control of resources. Whether their tax payments are then used by the government to purchase aircraft or by the recipients to purchase consumer goods or anything else does not affect the cost to the taxpayers.

Although I am stressing spending as a measure of economic cost, it is worthwhile to note an additional cost of public programs financed through tax revenues. The existence of taxes may change the behavior of the taxpayers in ways that reduce their well-being, a cost referred to as the excess burden of taxation. For example, a tax on wages may cause some people to work fewer hours or to retire earlier than they otherwise would have. A tax levied on a good or service will induce taxpayers to reduce consumption of the taxed item to avoid the tax. (Of course, in some cases, the tax is designed to reduce consumption, as with the taxes on alcohol and tobacco, because consumers may not fully cover the costs of their behavior.) Taxes that distort economic decisions thus have two costs: the amount collected and the loss to individuals from induced changes in behavior. The latter cost, however, is quite difficult to estimate, which suggests that focusing in the budget on the direct burden of government spending is the most valuable immediate objective.
 

The Economic Costs of Federal Risk-Bearing

A particularly difficult and increasingly important issue is the treatment of risky activities like providing loans, guarantees, and insurance. Earlier in my testimony, I noted a wide range of economic transactions in which the federal government either directly or indirectly takes on risk associated with bad outcomes. For example, the Federal Deposit Insurance Corporation bears most of the cost of the risk of bank failures.

Such arrangements in effect shift some or all of the risk to the federal government and, through the government, to taxpayers as a group. The risk of a bank failing is not reduced by the government bearing the cost. Indeed, the existence of the insurance may, itself, provide an incentive for organizations or individuals to engage in riskier activities than they would have without that insurance. Even though the government may be in a better position to take on a particular risk, the government's assumption of a risk does not eliminate its cost.

Risk is costly, whether it is borne by an individual or a company. Most of us prefer a sure thing over an uncertain outcome. Individuals are willing to pay to avoid risk. They buy insurance policies because they do not want to bear all of the risk of their house burning down. Companies purchase many kinds of financial instruments to protect themselves from incurring the full cost of losses that would result from circumstances over which they have no control. In each case, the reduction of risk is considered to be something of value, and the individual or the company is willing to pay for it.

The mirror image of this aversion to risk is that individuals typically demand compensation in order to bear risk. In financial markets, a measure of the extent to which individuals dislike risk is the so-called equity premium--the difference between the expected return on riskier equities (stocks) and the expected return on safer short-term Treasury securities. Depending on the time period examined, that premium has ranged from 3 percent to 7 percent in the United States. It arises from the fact that investors are willing to take on the additional risk of stock investments only because the expected return is higher than that from bonds. An investor choosing between a risky stock portfolio and low-risk bonds would almost certainly choose the bonds if the expected return was the same on both. In that case, the price of bonds would be bid up relative to the price of stocks, until investors had no preference between the two--that is, until the additional expected yield on the stocks exactly offset the costs to investors of the additional risk. Consequently, when returns are measured net of the costs of risk, private securities carry the same returns as government securities.

When the federal government assumes a risk--as it does, for example, when it guarantees a loan--the risk of default has been transferred from the lender to the government (and through the government to taxpayers), but the risk has not been eliminated. Instead, the government, rather than the lender, bears the risk. Market behavior, such as that which gives rise to the equity premium, may be used to assess the budgetary value assigned to assuming the costs of risk.

Direct loans and guarantees by the federal government constitute an area of budgeting where the Congress addressed accounting shortcomings through the Federal Credit Reform Act of 1990. Prior to that law, both direct loans and guarantees were treated on a cash basis in the budget. For direct loans, cash flows in any single year consist of outlays for new loans and repayments for some outstanding ones. The net cash flow in any single year, as an amalgam of the amounts of cash in and out, does not provide any meaningful indication of a credit program's long-term costs, although it is useful for the purpose of assessing federal borrowing needs. The cost of new loan programs is especially overstated on that basis because nearly all the cash flows are out in the early years. For guarantees, single-year cash flows are a mix of fee collections, payments for defaults, and inflows from recoveries. Before credit reform, the misstatement of costs for new guarantee programs was especially perverse because cash flows in the early years often were dominated by the inflow of guarantee fees, with few outlays for defaults.

Under the Federal Credit Reform Act, the cost of a direct loan or loan guarantee is the net present value of all expected cash flows over the life of the loan, recognized when the loan is disbursed. Net present value is calculated by discounting cash flows with interest rates on Treasury securities of the same maturity. A much-needed step toward getting the economic cost of credit programs in the budget, credit reform follows the principle of recognizing budgetary impacts at the time loans and guarantees are extended. Grants, direct loans, and guaranteed loans now can be compared on a more level playing field. In my judgment, the budget information now available to the Congress on the cost of credit programs is far superior to what existed before.

With the experience of more than a decade since the enactment of credit reform, however, it may be time to revisit the credit-reform model and its application. One shortcoming of the current approach is that it appears to understate the economic cost of federal credit programs because the discounting of expected cash flows at the government's risk-free borrowing rate ignores certain costs of risk. Private investors, by contrast, require compensation to induce them to bear risks that cannot be eliminated by diversification--for example, market risk. The compensation to investors for market risk comes in the form of an expected return that is higher than the rate on government debt that is used to value loans and guarantees under the credit-reform model.

Turning to other areas involving the federal treatment of risk, I would point out that the current budgetary accounting for federal insurance programs, such as deposit and pension insurance, still falls far short of the objective of assigning full economic costs to those activities. Currently, the costs of those activities are reported on a cash basis, which does not reflect the multiyear nature of the commitments. One result is that the programs report negative spending in most years, suggesting that they provide net income to the government, when in fact they represent a potentially large contingent liability. Consequently, alternative approaches may be needed to recognize the economic costs of insurance programs in the budget, perhaps building on the principles underpinning credit reform.
 

Principles to Guide Budgetary Recognition of Costs

Policymakers constantly weigh the costs and benefits of proposed and existing legislation. Just as markets work best in allocating resources to their highest-valued uses when prices reflect the true costs of goods and services, the Congress is best served when Members have the most comprehensive and accurate information about the costs of legislation. Moreover, because federal budgeting affects the allocation of resources between private and public uses as well as among public uses, the relevant cost is the highest-valued alternative to all other uses, private as well as public. All alternatives can be better compared when budgetary costs reflect economic costs.

Spending is a good measure of cost because it will have to be financed, at least eventually, by taxes. Thus, a guiding budgetary principle should be to recognize in the budget the amount of taxes that will be needed to finance a commitment. Furthermore, at the point when the commitment has been made, its cost should be recognized in the budget, even if the spending will not occur immediately. I acknowledge that it is sometimes difficult to distinguish exactly when a commitment to spend has been made and how durable that commitment may be.

Although a one-time appropriation may reflect a commitment with clear timing and duration, relatively few spending decisions are that straightforward. In fact, many programs that are nominally controlled by annual appropriations are ongoing functions--such as defense, transportation, and education--that the federal government could reasonably be expected to continue, and baseline budget projections reflect that expectation.

The difficulty of determining the timing of commitments is illustrated by a federal policy to provide financial assistance to low-income students enrolled in higher education. That policy might be regarded as a commitment to spend for students who are now eligible and for students who become eligible in the future. However, because the commitment is not contractual for future applicants, the Congress might change the law defining eligibility or substitute a different form of assistance. Clearly, the current program cannot be regarded as irrevocable; therefore, the present value of future assistance should not be recorded in the budget.

Indeed, for social insurance programs, it seems fair to say that although the commitments are clear in current law and are so reflected in baseline budget projections, the government has not firmly committed to paying the current level of benefits to all future generations. In other cases, such as loan guarantees and insurance, the government's commitment to spend may be contractual and firm, but the value of the dollar payments may be uncertain and difficult to estimate.

I suggest that the principle of recognizing the costs of commitments in the budget when they are incurred implies that the mere expectation of future spending is not sufficient to warrant recognition in the budget. The government's obligation in the future must be firm to justify including the costs for it in the budget today. However, I also suggest that the principle of being timely in recognizing costs in the budget never excuses an estimated cost of zero just because the amount is not yet certain.
 

Improving Budgetary Presentations to Reflect Economic Costs

In order to discuss alternative budgetary presentations, it is important to consider the role of the budget alongside other forms of federal financial reporting, most importantly the U.S. Financial Report published annually by the Treasury. Taken together, the financial report and the budget provide a wide array of information to policymakers.⁽⁶⁾

The role of the financial report is to summarize the government's current assets and liabilities and changes in those amounts during the past year. The report uses accrual accounting to summarize current and future cash flows into a single presentation of the government's net financial position. In contrast, the role of the budget is to facilitate decisionmaking by the Congress about how to allocate resources now and in the future. Unlike the financial report, the budget displays economic costs in two ways: it shows the allocation of spending authority--up-front permission to commit the government to future expenditures--and it tracks the actual outlays of funds from the Treasury.

The budget seeks to present future demands on federal resources under current law, as well as proposed changes to those demands. However, current budget presentations may not be sufficiently comprehensive in two important ways: they may not present information over a suitably long time period for certain policy decisions or include all commitments for which the federal government may incur economic costs.

The current budgetary horizon, generally either five or 10 years, is adequate for most policy decisions of government. However, with the pending retirement of baby boomers and the corresponding demands that will arise in the Social Security and Medicare programs, it is not surprising that many federal policy issues today involve much longer time frames.

Federal outlays are expected to grow significantly over the next several decades. To facilitate understanding of the financial consequences of that growth, long-term budget projections extend the budget's horizon for 50 to 100 years to depict the timing and trends that are inherent in current and proposed policies. CBO has also, where feasible, provided information regarding the source and size of uncertainties that surround those projections. The scale, timing, and uncertainty about current-law spending are useful inputs to any proposal to reform policies.

Some observers have suggested presenting long-term information in the budget through summary measures of future spending flows, which are similar in many ways to the accrual measures used in the financial report. Such summary present-value measures can display--in a single number--certain expected long-term obligations. As of 2004, for example, the present value of the future benefits provided under Medicare until 2050 was roughly $20 trillion (on the basis of the assumptions used for the intermediate spending path in Table 1).⁽⁷⁾ The financial report presents such estimates as supplementary information to reflect the potential long-term costs of currently scheduled social insurance benefits.

For decisionmaking, such measures can be very useful to rank the size of specific budgetary imbalances. However, because such measures summarize cash flow budget projections, they are subject to the same uncertainties as those presentations but also include additional uncertainty in the choice of discount rates that are used to produce one summary number. Such measures can also be difficult to interpret, because they often reach trillions of dollars but provide little information about the resources that will be available to pay costs as they occur.

In particular, the term "unfunded liabilities" has been the source of considerable confusion, leading some people to misconstrue the economic significance of trust funds and trust fund balances. The term may also suggest that certain liabilities are funded when, in fact, the real source of all funding is the government's ability to levy taxes.

Cash flow presentations present important information about the expected timing of financial events, which can be critical information for policy development. Many policies could be adopted that would reduce a summary measure of imbalance to zero in 50 or 100 years but leave an imbalance in subsequent years, particularly for the fast-growing health programs. Furthermore, potential policy fixes could be "gamed" by allowing expenditures up front with promises of larger cuts later.

The second potential shortcoming of current budget presentations is the treatment of contingencies. The budget recognizes the losses from contingencies when the government makes cash payments to beneficiaries. For insurance programs, multiyear budget projections attempt to anticipate a "normal" amount of such spending. However, in any budget period, it is possible that no outlays will be made or that income from premiums will exceed outlays. The resulting net cash inflow could cause observers to underestimate the government's exposure to loss--especially when the commitments involve costs associated with infrequent events.

In general, however, presentations of contingencies are incomplete in the federal budget and the financial report. For example, neither presents the federal government's expected responses to natural disasters (for example, earthquakes or floods)--yet federal supplemental emergency appropriations in response to natural disasters averaged more than $5 billion annually in the 1990s, and insurance markets and the public at large clearly expect such support to continue in the future. Nor does current reporting include the financial risks associated with government-sponsored enterprises or with government support of multilateral financial institutions.

Another cause of incomplete reporting on contingent commitments involves the manner in which the federal government now estimates the magnitude of potential claims. Those uncertain future costs appear to be underestimated because no cost is currently recognized for bearing the risk associated with uncertainty.⁽⁸⁾