Mr. Chairman and Members of the Committee:
Thank you for inviting me to testify at this hearing.
My testimony provides perspectives on the emerging federal budget surpluses and examines the case for and against using the surplus for tax cuts versus other goals. My principal conclusions are as follows:
- The federal surplus is a major achievement, but it is only the first step toward long-term fiscal sustainability. The short-term surpluses are an accounting illusion, and the long-term forecast shows a significant fiscal deficit, due primarily to social security and medicare.
- Recent CBO estimates project large surpluses over the next 10 years in both the on- and off-budget portions of the federal budget. Although there is general agreement that the $1.8 trillion in accruing balances in the social security trust funds should be allocated to shoring up social security, there is significant disagreement about how to allocate the other $800 billion in surpluses, with the main candidates being debt reduction, government spending, and tax cuts.
- There is little in the short-term surplus estimates that justifies a tax cut. Over 50 percent of the projected on-budget surpluses are due to accumulations in government pension reserves and the medicare trust fund. These accumulations, like social security, represent resources owed to current workers when they retire, and so should not be spent on tax cuts. Over 75 percent of the on-budget surplus arises from projected cuts in real discretionary spending. These cuts are unlikely and may not be advisable either. The surplus forecasts also assume that almost all of the recent revenue surge will prove permanent, which may prove optimistic.
- The long-term fiscal situation provides even less justification for a tax cut. Over the next several decades, the government is projected to incur large fiscal deficits, not surpluses, due to the rising costs of social security, medicare and medicaid. The long-term fiscal gap is at least 1.5 percent of GDP, even if the entire surplus materializes and is saved over the next 10 years.
- The proposed 10 percent across-the-board cut in income tax rates would require using about $200 billion of the social security trust fund to finance general tax cuts, would provide large benefits to high-income households, but meager benefits to middle- and low-income households, and would require an ill-advised waiver of the budget rules.
- Although aggregate federal revenues are near an all-time high relative to GDP, the evidence shows that families at most points in the income distribution will pay a smaller share of their income in federal taxes in 1999 than in any year since at least the 1970s. The reconciliation of these two facts is that both tax rates and income have risen for high-income households.
- I conclude that the combination of a short-term surplus, a sound economy, and the lowest tax rates for most households in decades provides a rare confluence of good fortune that should be used to address the nation’s pressing long-term fiscal problems related to social security and medicare, rather than being used to finance tax cuts.
In January of this year, the Congressional Budget Office (1999) announced projected federal budget surpluses of almost $2.6 trillion for fiscal years 2000 to 2009. Although budget forecasts have been improving rapidly over the last few years, the most recent forecast is notable for at least two reasons: the magnitude of the estimate surpassed by more than $1 trillion a similar estimate made last August, and, for the first time, the forecast contained a significant surplus in the non-social security portion of the budget.
This forecast, coupled with the release of the president’s long-awaited proposals for social security reform, have led to a veritable explosion of ideas about what to do with the surplus. While there is general agreement that the $1.8 trillion in surpluses accruing in the social security trust fund should be preserved for future social security obligations, views differ considerably about how to allocate the $800 billion in “on-budget” surpluses. Broadly speaking, there are three approaches: use the funds for saving, government spending, or tax cuts. The president has proposed a mixed set of uses that weighs heavily toward saving the surplus. He would allocate to the Social Security system that portion of the unified budget surplus that is attributable to it (although making use of an extremely confusing accounting mechanism to do so) and devote some of the remaining surplus to providing resources for Medicare and government-sponsored 401(k)-like saving plans. He also would allocate some of the surplus to increased defense and domestic discretionary spending.
In sharp contrast, leading Congressional Republicans, including House Budget Chair John Kasich (R-OH), Senate Finance Chair William Roth (R-DE), and Senate Budget Chair Pete Domenici (R-NM) weighed early in with proposals to use the entire on-budget surplus to finance 10 percent across-the-board cuts in income tax rates. Senate Majority Leader Trent Lott (R-MS) then announced a campaign of 150 Republican “town meetings” to popularize the idea (Edsall 1999). Shortly after the first meeting, and in the face of mutinies among fellow Republicans, Republican leaders backed down from advocating the across-the-board cuts (Stevenson 1999a, Pianin 1999).
Nevertheless, the prospect of a large-scale tax cut financed by the surplus is hardly a dead issue, for several reasons. First, tax cuts are a perennial topic, and the debate about the surplus is unlikely to disappear anytime soon. Just as the federal deficit dominated fiscal policy discussions in the 1980s and early 1990s, choices concerning how to allocate the emerging budget surpluses will be the centerpiece of tax and spending debates for the next several years. Second, despite the Republican Congressional leaders’ abandonment of across-the-board cuts, none of the Republican Presidential candidates have abandoned support for the proposition (Harwood 1999). Although George W. Bush has not advocated such a cut, his chief economic adviser, Larry Lindsey, is a strong proponent of across-the-board cuts (Novak 1999, Lindsey 1999). Finally, many of the issues that arise in financing across-the-board tax cuts also arise in the analysis of targeted tax cuts, which have been proposed on both sides of the aisle.
This paper examines the appropriate use of the projected on-budget surpluses, with a particular emphasis on whether the existence of the surplus justifies large-scale, across-the-board tax cuts. We draw several sets of conclusions:
First, the existence of the short-term surpluses provides little justification for across-the-board tax cuts. The case for tax cuts assumes that it is appropriate to use all of the (on-budget) surpluses for tax cuts and that the forecasted surpluses will actually be available for tax cuts. Both assumptions are questionable. For example, over 50 percent of the projected on-budget surpluses over the next 10 years is due to accumulations in the Medicare trust fund and in government pension reserves. The general agreement that the social security trust fund should not be squandered on new spending or tax cuts has implications for the use of the pension and Medicare trust funds. Like social security, pensions and Medicare balances represent resources owed to current workers when they retire. This implies that, like social security, these trust fund reserves should not be raided to finance tax cuts.
The on-budget surplus also depends critically on the assumption that discretionary spending will fall in nominal terms between 1999 and 2002, and will fall by 2009 by 25 percent of its current value relative to GDP. Under reasonable assumptions about the composition of the decline, domestic discretionary spending would fall to its lowest share of GDP since 1962 and defense to its lowest share since 1940. Such drastic cuts already seem unlikely to occur for political reasons (Stevenson 1999b) and may not be appropriate in any case. But even modest adjustments will cost plenty. Just holding discretionary spending constant in real terms—which would still reduce such spending by more than 20 percent relative to GDP—would cost over $600 billion between 2000 and 2009.
The on-budget surplus also depends on the apparent assumption that at least 85 percent of the recent surge in income tax revenues relative to GDP will prove permanent. This assumption may be too optimistic. Taking into account all of these factors suggests that there will be no surplus left over the next 10 years that would be appropriate to use for tax cuts.
Our second set of conclusions relates to the long-term fiscal situation. The short-term surpluses are a highly misleading indicator of the underlying fiscal position of the federal government. Current surpluses exist only because the government’s accounting methods ignore the enormous accruing liabilities of future entitlement benefits. In the long term, when these liabilities begin to mature, the government faces the prospect of sizable deficits as an aging population puts pressure on social security, Medicare, and Medicaid expenditures. The appropriate allocation of the short-term surplus hinges on whether the short-run surpluses outweigh the long-run deficits. Indeed, it is difficult to see how intelligent policy choices can be made at all in these circumstances without an understanding of the longer-run fiscal situation.
Over the next several decades, even if the entire 10-year surplus materializes and is saved (used for debt reduction), the federal government faces a large fiscal deficit. The Congressional Budget Office (1999) estimates that it would require an immediate and permanent increase in taxes or reduction in spending of about 0.6 percent of GDP, or roughly $50 billion in current terms, to maintain the same debt/GDP ratio in 2070 as currently exists. This “fiscal gap” rises to 2.2 percent of GDP if the on-budget surpluses over the next 10 years are returned to households via tax cuts or increases in spending. CBO’s estimates, however, understate the long-term problem because the government in 2070 would be running large deficits. Using a methodology developed by Auerbach (1994, 1997), we find that a permanent and immediate tax increase or spending cut of at least 1.5 percent of GDP is required to maintain fiscal balance in the long-run, even if the surpluses are used for debt reduction. Thus, the long-term fiscal situation provides no justification for a large-scale tax cut.
Our third set of conclusions results from direct examination of the proposed 10 percent rate reduction. We show that the tax cut would cost more than 100 percent of the projected on-budget surplus—that is, it would allocate about $200 billion from the social security trust funds for tax cuts. It would also provide very large benefits to the wealthiest households but very small benefits to low- and middle-income households, and would require what we view as an ill-advised waiver of the budget rules.
Finally, we show that recent claims that American taxpayers are laboring under heavier tax burdens than ever are not correct. While it is true that aggregate federal revenues are close to an all-time high relative to GDP, families at most points in the income distribution will have to forfeit a smaller share of their income in federal taxes in 1999 than at any time in the last 20 to 30 years. These two patterns are reconciled by the fact that burdens have risen among high-income households and, more importantly, that the share of aggregate income going to higher-income households facing higher tax rates has risen over time.
We conclude that the combination of a short-term budget surplus, a strong economy, and a tax system that is imposing the lowest rates for most households in more than two decades provides a rare confluence of good fortune that should be used to address the nation’s long-term fiscal problems, rather than being squandered on tax cuts financed by the on-budget surplus. Although we do not explicitly analyze the allocation of the off-budget surplus that is accruing in social security trust funds, the same line of reasoning as above suggests that tax cuts financed out of social security trust fund accumulations would be even less appropriate than cuts financed from the on-budget surplus.
Section I evaluates the projected surpluses over the next 10 years. Section II analyzes the long-term fiscal imbalance. Section III examines the effects of a large-scale tax cut. Section IV asks whether Americans are overtaxed and explores trends in aggregate and family tax burdens over time. Section V offers some concluding thoughts.
I. Surpluses over the next 10 years
A. Current projections
Table 1 shows the CBO’s January 1999, baseline budget projections. Between 2000 and 2009, the unified budget is expected to accumulate $2.565 trillion in surpluses. The surplus is projected to double relative to GDP, from 1.4 percent in 2000 to 2.8 percent in 2009, and to average about 2.2 percent of GDP.
The off-budget surplus reflects the amount by which social security tax payments and interest earned by the social security trust fund on the Treasury bonds it holds exceeds social security benefit payments and administrative costs. Off-budget surpluses are estimated at $1.777 billion, rising from $127 billion in 2000 to $217 billion by 2009. As a proportion of GDP, the off-budget surplus is relatively constant, rising from 1.5 percent in 2000 to 1.6 percent in later years.
The rest of the budget is projected to run small deficits in 1999 and 2000 and to begin running significant surpluses in 2002. The on-budget surplus gradually rises from about $50 billion in 2002 and 2003 to $164 billion by 2009. As a share of GDP, the on-budget surplus rises from -0.1 percent in 2000 to 1.2 percent in 2009.
If the surplus is maintained, debt held by the public is projected to shrink by two-thirds in nominal terms, and from 41 percent of GDP in 1999 to 8.9 percent in 2009. Relative to GDP, this would be lowest level of public debt since before World War I.
The turnaround in the budget forecasts has been nothing short of astounding. From 1981 to 1995, federal deficits averaged $193 billion per year, or 4.0 percent of GDP. Federal debt held by the public nearly tripled in real terms and nearly doubled relative to GDP. In 1995, the federal deficit stood at $164 billion and deficits stretched “as far as the eye can see.”
Since then, the budget forecasts have improved dramatically. For example, Figure 1 shows that, in March 1995, CBO forecasted a deficit of $472 billion for 2005. By January 1999, the forecast (corrected for policy changes) had changed to a surplus of $259 billion.
The cumulative estimates, shown in Figure 2, demonstrated similar changes. In January 1998, the CBO estimated a 10-year surplus of $660 billion. By July 1998, this was revised to $1.55 trillion, and by January 1999, the figure was revised to $2.565 billion.
In his January 1998, State of the Union address, President Clinton proposed to use the surplus to “save social security first.” At that point, the entire 10-year surplus resided in social security; the rest of the budget showed a substantial deficit over the 10-year horizon (Figure 3). It was not until the most recent forecast that there was a projected on-budget surplus. It is the emergence of the on-budget surplus that has energized the debate about how to use the surplus.
The recent improvements in fiscal position can be attributed to a strong economy, the deficit reduction packages in 1990 and 1993, rising revenues, and a decline in spending as a proportion of GDP. These sources, of course, are interrelated.
The economy Between 1992 and 1998, the economy grew continuously as unemployment, interest rates, and inflation all fell. The improvement in the economy had many sources, including sound monetary policy and some fortuitous events (such as very low energy prices).
Deficit Reduction Packages Some credit should also be given to the deficit reduction packages enacted in 1990 and 1993. George Bush’s willingness to abandon a poorly-conceived “no new taxes” pledge in 1990, and Bill Clinton’s 1993 tax increases, passed without any Republican support, greatly improved the fiscal status. Each of those budget and tax agreements were projected to reduced deficits by about $500 billion in the first five years after their enactment. The tax acts also raised the top income tax rate from 28 percent to 39.6 percent. Thus, when the income of higher-income households grew dramatically in subsequent years, tax revenues rose because of the higher marginal tax rate on those income gains as well as the increase in income itself.
Spending cutbacks Between 1992 and 1998, spending fell by 2.9 percentage points of GDP, from 22.5 percent to 19.6 percent. Defense spending accounted for much of the decline, falling by 1.7 percent of GDP. In addition, domestic discretionary spending fell by 0.3 percentage points of GDP, entitlements and other mandatory spending fell by 0.5 percentage points, and net interest fell by 0.3 percentage points. An important component of the decline in entitlement spending was lower-than-expected outlays for federal health care, mainly Medicare and Medicaid.
Revenue Growth The 1992-98 period also saw robust revenue growth. Revenues grew by 2.8 percentage points relative to GDP, from 17.7 percent to 20.5 percent. Federal revenue growth outpaced GDP growth in every year from 1994 to 1998. Most of the revenue gain was due to individual income taxes. After averaging 8.0 percent of GDP from 1950 to 1990, income tax revenues rose from 7.7 percent of GDP in 1992 to 9.9 percent in 1998. In 1993-5, income tax revenues grew by an average of over 7 percent per year, while nominal GDP grew by less than 5.5 percent per year. In the next three years, income tax revenues grew by more than 11 percent per year, while nominal GDP grew by about 5 percent annually.
While the 1993 tax package raised revenues relative to GDP in 1994, growth of tax revenues relative to GDP since then has been due to four factors, according to the CBO (1999, pp. 48-9). First, capital gains realizations increased by 150 percent between 1993 and 1997. Most of this increase mirrors the growth in the stock market and occurred before the 1997 tax act reduced capital gains taxes. Taxes on capital gains realizations accounted for nearly a third of the growth of tax liabilities relative to GDP from 1993 to 1997.
Second, taxable components of GDP—including wages, interest, dividends, rent, and proprietors’ income—rose relative to GDP. This accounted for about 10 percent or more of the increase. Third, other components of AGI that are not part of GDP also rose. In particular, a rise in retirement income, perhaps due in part to the stock market boom, accounted for about 15 percent of the increase.
Fourth, and most significantly, the effective income tax rate on income other than capital gains rose and accounted for about 40 percent of the increase in revenues relative to GDP growth. But there were no increases in statutory tax rates during this period, and the 1997 tax act actually reduced average tax rates for many taxpayers. Rather, the increase in effective tax rates occurred because, despite the income tax rate hikes on the top 2-3 percent of households in 1990 and 1993, higher-income households had proportional income gains that outpaced other groups.
C. Deconstructing the surplus
i. The underlying economic forecast
One possible concern in any budget projection is the broad economic forecast that underlies the fiscal estimates. On the whole, however, CBO’s economic forecasts appear to be mid-range or conservative relative to other forecasts. For example, for 1999 and 2000, CBO forecasts real growth of 2.3 percent and 1.7 percent, respectively. The analogous Blue Chip consensus estimates are 2.4 percent and 2.3 percent, and the Blue Chip “Low 10” estimates are 1.9 percent and 1.8 percent, respectively.
ii. Projected spending
Federal outlays were 19.6 percent of GDP in 1998, their lowest level relative to GDP since 1974. The budget forecast projects that federal spending will grow by 3.2 percent annually in real terms, but will decline relative to the rest of economy, falling to 17.3 percent of GDP in 2009 (Table 2). In only one year since 1958 has federal spending been a smaller share of GDP.
Components of spending are projected to grow in very different patterns from 1998 to 2009. Net interest payments are estimated to fall by almost two-thirds in nominal terms and from 2.9 percent of GDP to 0.6 percent of GDP. The decline is due to the vastly lower levels of public debt that would occur if the surpluses are maintained.
Social security, Medicare, and Medicaid are projected to grow from 8.1 percent of GDP to 9.6 percent. Although these levels are still manageable, they foreshadow larger increases that will occur when the baby boomers begin to retire en masse.
A key assumption is that discretionary spending will fall from 6.6 percent of GDP to 5.0 percent. The spending caps are assumed to be enforced between 1999 and 2002, and discretionary spending is assumed to stay constant in real terms from 2002 to 2009.
These assumptions may be unrealistic. For example, to comply with the spending caps in the Deficit Control Act, discretionary outlays would have to decline in nominal terms in each of the next three years, from $575 billion in 1999 to $568 billion in 2002. This implies that even if all of the “emergency spending” and IMF funds that were provided last year are discontinued, other discretionary appropriations will have to decline in nominal terms in 2000 by about $13 billion (CBO 1999, p. 64).
Even if the 1999-2002 spending levels comply with the caps, holding discretionary spending constant in real terms from 2002 to 2009 may prove difficult. Table 3 shows trends in the level and composition of discretionary spending between 1980 and 2009. Since 1980, discretionary spending has fallen from 10.2 percent of GDP to 6.6 percent. About half of the decline has occurred in defense, which has fallen from 5 percent to 3.2 percent of GDP. Domestic discretionary spending has fallen by almost as much, from 4.7 percent to 3.2 percent. International spending fell from 0.5 percent to 0.2 percent. Thus, relative to the size of the economy, discretionary spending has already sustained deep cuts over time.
However, virtually all of the reductions in real discretionary spending relative to GDP that have taken place since 1990 have occurred in defense spending (Table 3), where at least some downsizing was inevitable following the collapse of the Soviet Union. But large additional reductions there may prove difficult. If so, then a major portion of future cuts will have to come from domestic spending.
The implications of the budget’s forecasted decline in discretionary spending relative to GDP would be startling. Suppose that all international spending were eliminated, and the rest of the cut were divided equally between domestic and defense spending, so that each was allocated 2.5 percent of GDP in 2009. For domestic spending, this would be the lowest percentage since 1962 (CBO 1999, p. 135). For defense, it would be the lowest percentage since before World War II (OMB 1999, tables 1.2 and 3.1).
Changing the discretionary spending trajectory can have huge effects on future budget outcomes. Table 4 reports the results of various changes in discretionary spending, accounting for the interest costs of the change as well as the change in discretionary outlays. Holding discretionary spending at its current level of GDP would cost $1.4 trillion over the next 10 years. But even more modest changes in the spending trajectory would cut significantly into the surplus. For example, if discretionary spending were held constant in nominal terms from 1999 to 2002 and then held constant in real terms after that, the 10-year surplus would be reduced by $73 billion. If discretionary spending were held constant in real terms from 1999 to 2009, the 10-year on-budget surplus would be reduced by $609 billion. That is, more than three-quarters of the 10-year surplus is based on the assumption that real discretionary spending will fall.
iii. Trust fund accumulations
More than 50 percent of the projected on-budget surpluses are due to accumulations in federal trust funds for pensions and Medicare. Table 5 shows that from 2000 to 2009, these funds are expected to grow by $363 billion and $55 billion, respectively. Together, accumulations in these two trust funds account for 53 percent of the projected 10-year on-budget surplus.
Analysts on all sides recognize that it is inappropriate to use social security trust funds to finance tax cuts or non-social security spending programs. The reason is that government budget accounting seriously misrepresents the long-term costs of the social security. But social security is only the tip of the iceberg when it comes to misleading government accounting. Like the social security trust fund, government pension reserves and the Medicare trust fund represent funds that are owed to current workers when they retire. Thus, it would be appropriate to save the surpluses generated in the trust funds by using the revenues to reduce the debt, rather than cutting taxes or increasing spending. Indeed, for similar reasons, many states already separate their pension reserves from funds available for tax cuts and other spending.
iv. Will Medicare be allowed to go bankrupt?
Medicare’s long-term financial problems are more dire than social security’s. CBO projected last summer that the Medicare trust fund would be insolvent by 2012. A reasonable estimate is that the recent improvement in overall budget status pushed the date of insolvency back a few years. Nevertheless, the $800 billion on-budget surplus that is forecast for 2000-9 is predicated on the notion that nothing will be done to address Medicare’s problems. Clearly, any diversion of general revenues to Medicare would reduce the amount available for tax cuts.
v. Projected revenues
Federal revenues were 20.5 percent of GDP in 1998, the highest level since 1944, when they were 20.9 percent. They are projected to rise slightly relative to GDP in 1999 and then to decline by about 0.5 percentage points. From 2003-9, revenues are projected to be 20.2 percent, a larger share than in any year from 1945 to 1997. Revenues from corporate income taxes, payroll taxes, and other taxes are each expected to decline by about 0.2 percentage points of GDP (Table 6).
Income tax revenues are forecast to grow at 4.3 percent per year, roughly the same as the 4.4 percent growth of GDP from 1998 to 2009. After the explosive income tax growth of the past five years, the revenue forecast may seem relatively benign. But the forecast may be less conservative than it appears, because it seems to assumes that most of the recent surge of revenue relative to GDP will be permanent.
Estimating the proportion of the revenue surge that is assumed to be permanent is difficult to do in a precise way. In Table 7, we provide some rough measures of this proportion. For example, most of the revenue surge occurred in the individual income tax, which rose from 7.9 percent of GDP in 1994 (after OBRA 1993 had taken effect) to a projected 9.8 percent in 1999, only to fall to 9.6 percent in 2003-7, before rising to 9.7 percent in 2009. Using the estimated low of 9.6 percent of GDP suggests that 85 percent of the rise in income tax revenues relative to GDP from 1994 to 1999 is implicitly assumed to be permanent. Using all federal revenues suggests that 76 percent of the surge is assumed to be permanent in the forecast. The table shows that, depending on the tax measure and year used, the forecast assumes that somewhere between 72 and 95 percent of the revenue surge is assumed to be permanent.
Whether this assumption is reasonable depends on whether the sources of the gain are considered likely to continue. As noted above, about one-third of the revenue surge is due to higher capital gains realizations, which are in turn due to the surging stock market in the last few years. If half of the surge in realized capital gains continues in the future (as gains accrued in recent years are gradually realized) and all of the other components of the surge continue to hold, then roughly 83 percent of the revenue surge will prove permanent. However, if less than half of the capital gains surge continues and if any of the remaining two-thirds of the surge proves temporary, the permanent component of the revenue surge could fall well below 80 percent.
Small changes in the proportion that is assumed permanent can have large changes in the 10-year budget estimates. If the implicit assumption overstates the actual share of the revenue surge that is permanent by 10 percentage points, then future revenues would be lower by about 0.2-0.3 percent of GDP. Including the costs of added debt service, this would reduce the surplus by $300-450 billion over the 2000-9 period.
There are three generic reasons why surplus projections are difficult. First, the surplus is a residual, the difference between revenues and outlays. Roughly speaking, the surpluses over the next 10 years are projected to be about 10 percent of revenues or of outlays. Thus, relatively small changes in the economy, or in revenues or spending relative to the economy, can have large impacts on the surplus.
Second, changes in government’s fiscal position at one point tend to build on themselves over time. That is, short-run mis-estimates are typically amplified as the forecast horizon lengthens. For example, an increase in revenues reduces current deficits, but it also reduces interest costs, which reduces future debt and deficits.
Third, the economy is difficult to predict. CBO (1999, p. 85) reports the 10-year growth of real wages, salaries and corporate profits per member of the potential labor force, which is the labor force adjusted for cyclical variations in the economy. This figure was as high as 45 percent in the late 1960s, but then fell to negative 15 percent by 1982, and has since risen to over 10 percent. CBO (1999, p. 82) also reports that, in their 5-year forecasts from 1988 to 1998, the first projection of the surplus in a particular year was off by an average (absolute value) of 13 percent of ou