July 9, 2007

Honorable Judd Gregg
Ranking Member
Committee on the Budget
United States Senate
Washington, DC 20510

Dear Senator:

Under current law, rising health care costs and the aging of the population will cause federal spending on Medicare, Medicaid, and Social Security to rise substantially as a share of the economy. If tax revenues as a share of gross domestic product (GDP) remain at current levels, that additional spending will eventually cause future budget deficits to become unsustainable. To prevent those deficits from growing to levels that could impose substantial costs on the economy, the choices are limited: Revenues must rise as a share of GDP, projected spending must fall, or both.

In response to your letter of June 18, 2007, the Congressional Budget Office (CBO) has prepared the attached analysis of the potential economic effects of using higher tax rates alone to finance the projected increases in spending over the next several decades. That analysis also examines the effect of such tax changes on various illustrative taxpayers. In keeping with CBO's mandate to provide objective, impartial analysis, this report makes no policy recommendations.

CBO would be pleased to address any further questions you have. I can be reached at (202) 226-2700. The staff contact for the macroeconomic analysis is Doug Hamilton, who can be reached at (202) 226-2770; the contact for the tax analysis is David Weiner, who can be reached at (202) 226-2689.

Sincerely,

Peter R. Orszag
Director

Attachment

cc:	Honorable Kent Conrad Chairman

Financing Projected Spending in the Long Run

July 9, 2007

Under current law, rising health care costs and the aging of the population will cause federal spending on Medicare, Medicaid, and Social Security to rise substantially as a share of the economy. If tax revenues as a share of gross domestic product (GDP) remain at current levels, that additional spending will eventually cause future budget deficits to become unsustainable. To prevent those deficits from growing to levels that could impose substantial costs on the economy, the choices are limited: Revenues must rise as a share of GDP, projected spending must fall, or both.

The Congressional Budget Office (CBO) has analyzed the potential economic effects of using higher taxes alone to finance the projected increases in spending over the next several decades. CBO's analysis reaches the following conclusions:

• To finance the projected increases in spending, revenues could be raised in a variety of possible ways, and those alternative tax policies would have different effects on the economy and on the distribution of tax payments. In response to a request from Senator Gregg, this letter attachment examines the implications of raising marginal income tax rates; in general, other mechanisms for raising revenues would impose lower macroeconomic costs but may also be less progressive.
  
  
• Differences in the economic effects of alternative policies to achieve a sustainable budget in the long run are generally modest in comparison to the costs of allowing deficits to grow to unsustainable levels. In particular, the difference between acting to address projected deficits (by either reducing spending or raising revenues) and failing to do so is generally much larger than the implications of taking one approach to reducing the deficit compared with another.
  
  
• Health care cost growth is the most important factor affecting the long-term projections of spending—and thus the amount of revenues needed to finance that spending. That factor is significantly more important than other commonly cited factors, such as aging of the population.
  
  
• If health care costs per beneficiary grew 1 percentage point faster than per capita GDP on average each year in the long run and the resultant spending path was financed entirely through higher income tax rates, real (inflation-adjusted) growth of GDP would be reduced somewhat relative to a scenario in which spending on Medicare, Medicaid, and Social Security was reduced. The impact on the macroeconomy could be attenuated by not relying exclusively on higher income tax rates to raise revenues.
  
  
• However, if health care costs per beneficiary grew an average of 2.5 percentage points faster than GDP per capita each year, as they have over the past four decades, and the spending was financed solely with a proportional increase in income tax rates, the economic costs would be significant and the circumstance probably impossible to sustain through 2050. Again, the economic effects could be mitigated by using alternative tax policies that did not raise marginal income tax rates as much.

CBO's Long-Term Projections

Significant uncertainty surrounds the outlook for spending in the long term, but almost all observers agree that the budget is on an unsustainable path. Unless changes are made to current budget policy, rising health care costs and the aging of the population will put increasing pressure on the budget via three significant federal programs: Medicare, Medicaid, and Social Security. Rising health care costs, in particular, will cause federal spending to grow rapidly. Over the past four decades, Medicare's and Medicaid's costs per beneficiary have increased, on average, about 2.5 percentage points faster per year than has per capita GDP.¹ If those costs continued growing at the same relative rate over the next four decades, federal spending on those two programs alone would rise from 4.5 percent of GDP today to about 20 percent in 2050.

In December 2005, CBO examined the budgetary impact of a variety of alternative assumptions about the future course of spending and taxes. The key variable affecting the long-term fiscal balance was the rate at which health care costs grow relative to income. Indeed, it exerts a significantly greater influence on the budget over the long term than other commonly cited factors, such as the aging of the population.² Under one scenario in CBO's December 2005 report, health care costs per beneficiary were assumed to grow 1 percentage point faster than per capita GDP each year on average in the long run. In that scenario, total federal spending (excluding interest on the debt) increased from 19 percent of GDP in 2005 to 25 percent in 2050.³ If, however, such excess health care cost growth averaged 2.5 percentage points per year in the long run, as it has for the past 40 years, total federal spending (excluding interest) would climb sharply from 19 percent of GDP in 2005 to 34 percent of GDP in 2050.

Those projected increases in spending would produce ever-rising budget deficits that would eventually become unsustainable if revenues remained at their current share of GDP. Thus, at some point, policymakers will need to adjust the paths for spending and/or revenues to prevent budget deficits from seriously damaging the economy. The sooner such changes are made, the smaller the economic damage will be.⁴

Approach for Analyzing Economic Effects of Taxes

What would happen to the economy if the increases in spending were financed entirely through higher taxes? Because different taxes affect the economy in different ways, the answer would depend in part on how revenues were raised, and in this analysis, CBO examined two alternative tax policies to illustrate how different policies can affect the economy (as described in the next section).

The effects of those scenarios on the economy also depend on many other factors. A major one is the mix of spending that the taxes finance. Different types of spending have different effects on the economy. Because the growth of health care costs plays such an important role in the long-run outlook for the budget, the macroeconomic effects of taxes will become increasingly dependent on how households view the costs of government-financed health care. Do they value each dollar of health spending paid by the government as an additional dollar of cash income? Or is it worth less to them than a dollar of cash income? That issue is important because the people who receive benefits from the Medicare and Medicaid programs at any point in time tend to be older than the people who pay for those programs, and those redistributions of income across age cohorts can affect macroeconomic outcomes. Because people's perceptions about additional health spending are uncertain, CBO employed polar assumptions about how much the additional health spending would be valued by households. One assumption was that households would value a dollar of health spending paid by the government as equivalent to a dollar increase in cash income. The other assumption was that households would not view the increase in government paid health care as a gain at all.

Uncertainties also exist about the degree to which households would respond to changes in marginal tax rates and after-tax income and how those responses in work and saving decisions would play out in the economy at large. CBO used two different models of economic behavior to reflect the range of opinion among economists about how people respond to taxes. Both models take account of the dynamic effects of higher tax rates on the economy and how those changes in the economy would in turn affect revenues. However, both models are simplified representations of the economy and thus provide only a rough guide to the potential effects of the tax scenarios on the economy. To simplify the analysis further, CBO generally restricted its analysis to examining the long-run implications of alternative tax and spending policies and did not focus much attention on the transition to those long-run outcomes. Such transitions, however, are often extended over decades and thus represent crucially important components of the real-world effects of policy changes (especially because policy interventions intended to alter the path of the transition to the long run can alter the long-term effects of the overall policy in substantial ways).

One model is an open-economy version of the textbook growth model that CBO uses in its annual Analysis of the President's Budgetary Proposals. Economic output in the model depends on the number of hours supplied by workers, the size of the capital stock, and total factor productivity (the state of technological know-how). The model accounts for international capital flows in a simple way: It assumes that wage rates and interest rates are fixed at base levels and that capital flows across borders until that condition holds. The labor supply response is determined by CBO's tax microsimulation model, which uses a sample of taxpayers and provides a detailed accounting of the individual income tax system.⁵ Households in the model are not forward looking; instead, members of the households base their decisions about working and saving entirely on current economic conditions.

The second model is the stochastic overlapping generations model with incomplete markets that CBO also uses in its analysis of the President's budget.⁶ In that model, households are forward looking and their members decide how much to work and save in order to make themselves as well off as possible over their lifetime. They face uncertainty about future wages and the length of their life and may be subject to borrowing constraints. The model makes two polar assumptions about the openness of the economy to international capital flows. One assumption is that the economy is completely closed—no capital can flow in or out; the other is that the economy is completely open and cannot affect wage rates or interest rates (the same assumption that the open-economy textbook growth model uses).

Although CBO's analysis focuses on the effects of tax policies on GDP, that is only a measure of economic activity in the marketplace; it is not a measure of consumers' well-being. Changes in tax policy can affect behavior and well-being in ways that are not fully reflected in GDP. For example, higher tax rates may create incentives for a range of behaviors, such as substituting fringe benefits for cash compensation, making use of more tax-deductible items, and engaging in other methods to avoid taxes. Those actions have efficiency costs to the economy that are not fully captured in GDP. The income tax system can also reduce the variability of after-tax income relative to before-tax income and thus can help to smooth out shocks to before-tax income.⁷ GDP, however, does not directly measure the value of such income smoothing through the tax system. Finally, changes in tax policy can also affect the distribution of after-tax income, and the two tax scenarios analyzed here have very different distributional implications (as discussed in the last section).

Two Alternative Tax Scenarios

The first tax scenario CBO examined assumes that current tax law is unchanged for the individual income tax through 2050. All other taxes are assumed to remain constant as a share of GDP. Although statutory tax rates are not changed, income growth raises effective tax rates. Real income growth can push taxpayers into higher marginal tax brackets and reduce the value of various parameters of the tax system, such as the personal exemption. Nominal income growth can further raise effective tax rates through provisions of the tax code that are not indexed for inflation—for example making taxpayers subject to the alternative minimum tax (AMT) or reducing the value of the child tax credit. As a result, the amount of revenues collected as a share of the economy rises significantly between 2006 and 2050. By CBO's calculations, this scenario would raise roughly the amount of revenues needed to finance the spending path in which excess health care cost growth is 1 percentage point per year in the future. However, it would not be sufficient to finance the spending path in which such cost growth is 2.5 percentage points per year.

The second tax scenario assumes that the entire tax code is indexed for both growth in real income and inflation and that the increase in spending is financed each year by an across-the-board, proportional increase in marginal income tax rates.⁸ Under this scenario, tax rates can be set to finance different spending paths, although the resulting economic effects will differ.

To highlight the economic effects of rising revenues and spending, those two tax scenarios are compared against a base scenario in which revenues and spending are held at the shares of GDP that they are in 2006, which implicitly assumes significant spending reductions from the levels that would occur under current law. That base scenario also implicitly assumes that changes in the tax code would be needed to keep the revenue share from rising, as it would under current law.

Maintain Current Law for the Individual Income Tax

According to simulations by the models, the scenario of maintaining current law for the individual income tax would reduce real GDP in 2050 by between 1 percent and 6 percent from what it would have been if revenues and spending were kept at their 2006 shares of GDP.⁹ Another perspective on the same conclusion is that real GDP would be between 140 and 152 percent higher than in 2006 under the tax scenario, rather than 155 percent higher under a scenario in which revenues and spending remained at their 2006 shares of GDP.

The losses in GDP are larger the more that the government-paid health spending is valued by households. To the extent that health spending is valued, the additional health resources reduce people's incentives to work and save. Also, the GDP losses are smaller the more the economy is open to capital flows from abroad. An influx of foreign capital could offset any slowdown that the higher tax rates caused in the accumulation of capital domestically. The inflows of capital from foreign investors are not free, however. A larger share of the economy's smaller output would have to be used to service the debts owed to foreign lenders, which implies that less income would be available to U.S. residents. In other words, capital inflows attenuate the impact on GDP but not on national income (or gross national product).

Under all of those assumptions, as noted above, the reductions in GDP estimated by CBO are relatively small in comparison with how much the economy could grow if the budget was put on a sustainable path. If fiscal sustainability was not achieved, however, budget deficits would continually mount and eventually cause a persistent decline in economic growth and the standards of living in the United States. Thus, the costs of failing to put the budget on a sustainable path are potentially very large: Failing to address the fiscal gap ultimately puts at risk the nation's long-term economic growth itself, whereas the differences among various approaches for eliminating that gap typically represent only a modest share of such growth.

Index the Tax Code and Raise Income Tax Rates

The second scenario would index the 2006 individual income tax code for both inflation and growth in real income and then finance the rise in projected spending with higher individual and corporate income tax rates. CBO examined the implications of aiming to finance two alternative spending paths: The first has health care costs per beneficiary in the long run growing 1 percentage point faster per year than per capita GDP does; the second, 2.5 percentage points faster per year.

Excess Health Care Cost Growth of 1 Percentage Point. Under the scenario in which, on average, health care costs per beneficiary rise 1 percentage point faster than per capita GDP and in which raising marginal income tax rates is the only mechanism used to balance the budget, individual income tax rates would have to rise by at least 70 percent to finance the increase in spending. Before any economic feedbacks or changes in behavior are taken into account, the lowest tax rate on individual incomes would have to be raised from 10 percent to 17 percent; the tax rate on incomes in the current 25 percent bracket would have to be increased to 43 percent; and the highest statutory rate on individual incomes would have to be increased from 35 percent to 60 percent. The top corporate income tax rate would also have to increase from 35 percent to 60 percent. Those estimates of tax rate changes are meant to be illustrative; official estimates of tax rate and revenue changes for any specific proposal would be carried out by the Joint Committee on Taxation.

Under this scenario, the level of real GDP in 2050 could be reduced by between 3 percent and 16 percent from what it would have been if revenues and spending, as shares of GDP, were kept at 2006 levels. In other words, real GDP in 2050 would be between 114 percent and 147 percent higher than in 2006, compared with roughly 155 percent higher under a scenario in which revenues and spending were maintained at their 2006 shares of GDP. For the reasons discussed in the prior section, the open-economy textbook growth model produces smaller effects than the stochastic overlapping generations model does. The impacts on GDP are also larger the more that health spending is viewed as cash income and smaller the more that the economy is open to capital flows from abroad.

The two tax scenarios—maintaining the policy that exists under current law and indexing the tax code and raising rates—illustrate the importance of the details of the tax policy. The first scenario does not raise effective marginal tax rates as much because real economic growth substantially reduces the proportion of individual income that is exempted from taxes because of the personal exemption and credits. In other words, maintaining current law embodies a form of base broadening. The contrast between the two scenarios highlights the fact that raising taxes can take a wide variety of forms that significantly affect the economic outcome.

Excess Health Care Cost Growth of 2.5 Percentage Points. How would the economy be affected if tax rates were used to finance the spending path in which excess cost growth is 2.5 percentage points per year? Answering that question is difficult because the economic models that economists have developed so far would have to be pushed well outside the range for which they were initially developed. Any numerical estimate would be very speculative and heavily dependent on the model producing it.

Nonetheless, CBO's calculations indicate that tax rates would have to be raised by substantial amounts to finance this scenario. Before any economic feedbacks are taken into account, and again assuming that raising marginal tax rates was the only mechanism used to balance the budget, the tax rate in the lowest tax bracket would have to be increased from 10 percent to 26 percent; the tax rate on incomes in the current 25 percent bracket would have to be increased to 66 percent; and the tax rate in the highest bracket would have to be raised from 35 percent to 92 percent. The top corporate income tax rate would also increase from 35 percent to 92 percent. Such tax rates would significantly reduce economic activity and would create serious problems with tax avoidance and tax evasion. Revenues could fall significantly short of the amount needed to finance the growth of spending, and thus tax rates at this level may not be economically feasible.

Other Tax Policies

Alternative tax policies could be employed that would mitigate the economic effects delineated above. In particular, policies that relied less on marginal income tax rates could have substantially smaller effects on the economy. Indeed, tax policies that reduced the income of households but did not affect the marginal incentives to work and save would have similar effects on the economy as reductions in government benefit payments to households to the extent that those benefit payments and the tax payments were made to and from the same households.

Effects of Two Tax Scenarios on Illustrative Taxpayers

The two tax scenarios have significantly different implications for marginal tax rates and consequent effects on the economy, illustrating the importance of designing tax policy to raise revenues efficiently. In addition, the different policy approaches have different implications for the distribution of tax liabilities.

Those differences can be illustrated by comparing changes in individual income and payroll tax liabilities for taxpayers in 2005 (the most recent year for which data on median income are available) with the burdens under the two scenarios in 2050. For example, in 2005 a single person with the median income of about $27,000 paid about $6,500, or 24.0 percent, of his or her income in personal income and payroll taxes (see the table). Under the current-law scenario, in 2050 a taxpayer at the median income would owe income and payroll taxes equal to 31.8 percent of his or her income.¹⁰ Alternatively, under the scenario in which the tax code is indexed and marginal tax rates are the exclusive mechanism used to balance the budget—and excess health care cost growth is 1 percent—a single taxpayer with median income would pay 30.0 percent of his or her income in income and payroll taxes in 2050.¹¹ For a married couple with two children and income of about $66,000, the median in 2005, the share of income owed would increase from 20.0 percent that year to 37.6 percent in 2050 under the first scenario and to 25.5 percent in 2050 under the second scenario.