Abstract
This article examines the fiscal outlook and tax reform options in the United States. The major conclusions include the following: the United States faces a substantial fiscal shortfall in the medium- and long term; both spending cuts and tax increases should contribute to the solution; tax increases need not do significant harm to economic growth; and there are sensible ways to both reform tax structure and raise revenues, including the redesign of tax expenditures, the creation of a value-added tax or a carbon tax, or an increase in the gasoline tax.
The Great Recession and its aftermath have left the United States with a difficult fiscal situation: a weak economy that would benefit from short-term stimulus and the prospect of medium- and long-term deficits that indicate the need for fiscal consolidation after the economy recovers. Any future deficit reduction will likely require a combination of spending cuts and revenue increases. While tax reform would be a laudable goal even in the absence of a fiscal problem, the need to build a better tax structure—one that is both more efficient and more equitable—grows as revenue requirements rise.
This article considers how the federal tax system could be designed to simultaneously address three goals: promoting economic growth, improving equity, and raising revenue. The second section summarizes the medium and long-term fiscal outlook and reviews the arguments for higher revenues as part of a fiscal solution. In the third section, we discuss how to broaden the income tax base by reducing and reforming tax expenditures. Although public attention often focuses on raising revenues by raising income tax rates, this could create significant avoidance and prove economically damaging given the current narrow income tax base (Altshuler, Lim, and Williams 2010). Broadening the income tax base would be more conducive to economic growth and would reduce the distortions and inefficiencies created by the tax system. It would also be fairer and simpler to treat different types of income and expenditures similarly, and it could raise substantial net revenue even if some of the proceeds were used to finance lower rates.
In the fourth section, we explore how a value-added tax (VAT) could also be part of the fiscal solution. A VAT could raise significant revenue and would raise national saving if the proceeds were used for deficit reduction. Moreover, the regressive aspects of a VAT could be offset by other federal policies.
In the fifth section, we discuss how a carbon tax and/or a higher tax rate on gasoline could reduce externalities, make markets more efficient, and raise significant revenues. As with a VAT, the distributional issues can be offset via other policies. The sixth section is a short conclusion.
Fiscal Outlook and Its Implications
Fiscal Outlook
The United States faces large federal fiscal deficits in the immediate future, the next decade, and the longer term. Although the near-term deficits—generally resulting from the automatic and discretionary fiscal policy responses to the “Great Recession”—are generally thought to be assisting a return to full employment, the medium and long-term horizons are more disconcerting.
Using a plausible “business as usual” baseline, Auerbach and Gale (2013) show that federal deficits will exceed US$8.4 trillion, or 4.0 percent of gross domestic product (GDP), between 2014 and 2023. The deficit will fall from a high of 10.1 percent of GDP in 2009 to a low of 3.3 percent of GDP in 2016, before rising to 4.4 percent in 2023 (figure 1). After 2023, deficits are poised to rise further. The debt-to-GDP ratio will pass its 1946 high of 108.6 percent in the early 2030s if Congress continues “business as usual” (figure 2).
Alternative deficit projections, 2013–2023.
Alternative debt projections, 2013–2023.
Unlike the aftermath of World War II, however, the debt-to-GDP ratio will continue to rise after surpassing the previous peak. Long-term models show expenditures rising significantly as the aging of the populace and growth of per capita health care expenditures cause Medicare and Medicaid outlays to grow rapidly (figure 3). Current estimates place the fiscal gap—the immediate and permanent increase in taxes or reduction in spending that would keep the long-term debt-to-GDP ratio at its current level—between 3.5 and 5.7 percent of GDP through 2089 and 4.0 and 7.8 percent on a permanent basis. And if deficit reduction were delayed given the lackluster state of the economy, the long-term fiscal gap would grow larger and require deeper spending cuts or more revenue. For example, if the adjustments were delayed until 2018, the year after Congressional Budget Office (CBO) projects that a return to full employment, the fiscal gap would increase by up to 0.4 percentage points of GDP.
Alternative projections of revenue and noninterest outlays, 2013–2090.
The Need for Spending Cuts and Revenue Increases
For several decades, federal health care programs have been increasing as a share of GDP, and they are projected to continue rising, due to both population changes and increasing relative price of medical care. Since spending is projected to rise faster than GDP in the indefinite future, it is clear that spending cuts must be part of the solution.
However, there are several reasons to consider tax increases, as well as spending cuts, as part of the fiscal solution. First, the sheer magnitude of the fiscal gap suggests that a spending-only solution would impose very substantial reductions on spending that might not be seen as equitable. At 5 to 7 percent of GDP, the fiscal gap is several times larger than the savings that were generated in budget deals in the past. The 1983 Social Security Reform reduced deficits by about 1.0 percent of GDP in the four years after passage while the 1990 and 1993 budget deals reduced deficits by about 1.4 percent of GDP and 1.2 percent of GDP, respectively, over the five years after passage. 1 In addition, Americans seem particularly reluctant to cut government spending on Social Security and Medicare, two of the key drivers of long-term spending, than on other forms of spending. For instance, a 2011 Gallup poll showed that over 60 percent of Americans were unwilling to cut social security and/or Medicare, which was true across the political spectrum.
Second, as a matter of political equilibrium, it seems likely that a sustainable budget deal would draw from both sides of the ledger. Indeed, prior major deals have included both tax increases and spending cuts. Budget discipline has been successful only when imposed on both sides of the fiscal ledger. Congress reduced spending and raised taxes in the 1983 Social Security reforms and the 1990 and 1993 budget deals. For example, in the 1990 budget deal, 49 percent of the reductions came from higher tax receipts, 34 percent from reduced defense spending, and 17 percent from other cuts in spending (Steuerle 2004).
The spending cuts in enacted in 2011 as part of the deal to raise the debt limit were a spending-only adjustment. But they were followed by a tax-only adjustment in the American Taxpayer Relief Act of 2012. As a result, legislation in 2011 and 2012, on net, cut from both sides of the ledger (relative to a current policy baseline). The bills, however, only solve a small part of the long-run imbalance, reducing estimates of the extended policy permanent fiscal gap from 6 to 9 percent of GDP (Auerbach and Gale 2012) to 5 to 7 percent of GDP (Auerbach and Gale 2013). 2
Third, as a matter of equity, a tax increase is the primary way to have high-income households share significantly in the burden of fixing the deficit; spending cuts typically do not have a large impact on high-income households. 3
Fourth, it is more effective to control spending by requiring that it be paid for with current taxes than to allow deficits to grow. Although the “starve the beast” hypothesis argues that keeping revenues down is an effective approach to curtailing spending, the hypothesis is not consistent with recent experience. 4 Romer and Romer (2009), for example, find that tax cuts designed to spur long-run growth do not lead to lower government spending; if anything, they find that tax cuts lead to higher spending. This finding is consistent with Gale and Orszag (2004a), who argue that the evidence of the last thirty years supports a “coordinated fiscal discipline” view, in which tax cuts were coupled with increased spending (as in the 1980s and 2000s) and tax increases were contemporaneous with spending reductions (as in the 1990s).
Long-term Growth Effects of Tax-financed Deficit Reductions
An increase in taxes will not necessarily slow long-term economic growth. Tax changes have two broad sets of long-term effects on the economy. 5 The first set operates through income and substitution effects. Direct changes in relative prices, incentives, and after-tax income affect the degree to which households are willing to work and save and to which firms invest and hire.
The second broad effect is on national saving. A reduction in the deficit raises public saving, which typically results in higher national saving, the sum of household, corporate, and government saving. This important effect is often ignored in discussions of tax policy and economic growth. Sustained deficits usually have deleterious long-term effects as they reduce future national income through higher interest rates, lower national savings, and increased indebtedness to foreign investors. Gale and Orszag (2004b) estimate that a 1 percent of GDP increase in the deficit will raise interest rates by twenty-five to thirty-five basis points and reduce national saving by 0.5 to 0.8 percentage points. Engen and Hubbard (2004) obtain similar results with respect to interest rates. Thus, relative to a balanced budget, a deficit equal to 6 percent of GDP would raise interest rates by at least 150 basis points and reduce the national saving rate by at least 3 percent of GDP. The International Monetary Fund (IMF; 2010) estimates that, in advanced economies, an increase of 10 percentage points in the initial debt-to-GDP ratio reduces future GDP growth rates by 0.15 percentage points. Hence, if this result is extrapolated linearly (albeit, cautiously the relationship may be nonlinear), an increase in the debt-to-GDP ratio from 36 percent in 2007 to 83 percent by 2023 (Auerbach and Gale 2013) could reduce the growth rate of GDP by about 0.7 percentage points. Thus, tax increases dedicated to deficit reduction could help spur economic growth relative to continuing policy as normal.
The net long-term effect of a tax change is the result of the two effects outlined earlier, which are sometimes offsetting and sometimes mutually reinforcing. Stokey and Rebelo (1995), for example, show that even the very large tax increases associated with World War II—on the order of 10 percent of GDP—apparently had no discernible impact on the long-term economic growth rate. Gale and Potter (2002), taking a very different approach than Stokey and Rebelo, find that negative effect on national saving of the 2001 tax cuts outweighed its positive impact on incentives; the net effect on growth was negative. This suggests that fully eliminating the 2001 tax cuts would have raised long-term economic growth prospects.
Tax Expenditures
In formal terms, tax expenditures are “revenue losses attributable to provisions of the Federal tax laws which allow a special exclusion, exemption, or deduction from gross income or which allow a special credit, preferential rate of tax or a deferral of liability” (The Congressional Budget and Impoundment Control Act of 1974 ([Public Law 93-344]).
Background
For many purposes, it is useful to categorize tax expenditures by how they mechanically affect income tax liability. Table 1 lists the largest tax expenditures by revenue cost and the type of tax expenditure each represents.
Top Ten Total Income Tax Expenditures.
Source: Office of Budget and Management. Analytical Perspectives, Budget of the U.S. Government, Fiscal Year 2014.
Note: Includes individual and corporate portions.
Exclusions exempt some forms of income from taxation. For example, employer-sponsored health insurance is the economic equivalent of income for workers, but it is exempt from taxation. Like wage payments, employer payments for employee health insurance are tax deductible for the business; unlike wages, however, employer-sponsored health insurance benefits are not taxed in the individual income tax.
Deductions reduce taxable income dollar for dollar. So-called above-the-line deductions are applicable to all taxpayers who meet the requirements. Itemized deductions—including deductions for mortgage interest deduction, state and local taxes, and charitable contributions—are typically only used by those whose sum of such deductions exceeds the standard deduction. A dollar’s worth of an exclusion or a deduction reduces taxable income by a dollar, and so saves more taxes for a household in a higher marginal tax bracket.
In contrast, credits reduce income tax liability dollar-for-dollar. Whereas the first two categories alter the income subject to taxation, credits directly reduce the tax owed to the federal government. Non-refundable credits can reduce one’s tax liability to US$0. Refundable credits, once tax liability is negated, can provide a net payment to the taxpayer, effectively creating negative tax liability. For example, the earned income credit is refundable, and the child credit is refundable under certain circumstances. 6
Preferential rates reduce the tax rate on some types of income relative to others. For example, long-term capital gains and dividends are taxed at a maximum rate of 15 percent, whereas other forms of income are currently taxed at rates as high as 35 percent. The taxpayer benefits from the difference between the marginal tax rate on regular income and the tax rate on the preferred form of income.
Deferrals postpone the taxation of some forms of income. For example, when employees contribute to 401(k)s, their contributions are deductible, the asset grows tax-free, and tax is only applied when funds are withdrawn. This postponement reduces the net effective tax rate on the income in a manner proportional to the tax rate on the alternative use of the funds.
As these examples make clear, the definition of a tax expenditure is somewhat subjective. It requires application of the word “special” in the definition, and it requires an implicit assumption about what the “normal” or reference tax system looks like. In broadest terms, the notion of a normal system underlying the current tax expenditure definitions is a progressive income tax, where all forms of income are taxed at the same rate, but unrealized capital gains are not subject to taxation. If the normal tax system, however, were thought to be a consumption tax rather than an income, the deferrals applicable to 401(k)s and other retirement saving would not be considered tax expenditures, since the current tax treatment would represent the normal treatment of saving under a consumption tax. On a similar note, personal and dependent exemptions and the standard deduction are not generally considered tax expenditures, since everyone may choose to take such options.
The sheer diversity of tax expenditures should also be noted. Although table 1 lists the largest items, the Office of Management and Budget (2013) identifies 169 total tax expenditures: 145 of which affect the individual income tax and 85 of which affect the corporate income tax (some affect both taxes). Meanwhile, the Joint Committee on Taxation (2013) lists over 200 tax expenditures. The items range from specialized subsidies for particular industries to broadly used initiatives—for example, mortgage interest, charitable contributions—that many consider core elements of social policy in the United States and that have been present since almost the inception of the income tax code in 1913. The diversity of tax expenditures sometimes makes it difficult to generalize the merits of tax expenditure reform, as many (and often the largest) provisions are not just “loopholes” designed to benefit a select few but to pursue broad public purposes. That is, simply because a tax expenditure is a deviation from a normal tax code does not mean that it is undesirable or should be eliminated. Nevertheless, tax expenditures are a fruitful place to begin thinking about tax reform, since the alternative ways to raise revenue—higher tax rates and new taxes—create obstacles.
Tax Reform and Tax Expenditures
The canonical focus for income tax reform is to create a system with a broad base that taxes all sources and uses of income at the same rate so as to generate lower statutory rates. Tax expenditure reform would be essential to achieving these goals. Broadening the base entails restricting the use of exclusions and deductions. Taxing all sources and uses of income, at the same effective rate, entails restricting the use of preferential rates, credits, and deferrals. In short, tax expenditure reform has the potential to raise revenue and to make the tax system fairer, more efficient, and simpler. But the diversity of tax expenditures and the situations they address requires attention to the particular details of each case.
Likewise, if a guiding focus of budget reform is to reduce government spending, it should be noted that many major tax expenditures act essentially as government spending programs that happen to be embedded in the tax code rather than in outlays (Batchelder and Toder 2010; Marron 2011; Marron and Toder 2012). Thus, tax expenditure reform in many cases can be thought of as reducing effective government spending.
The potential benefits of reforming tax expenditures as a partial solution to current tax and fiscal problems is well known. Both the President’s National Commission on Fiscal Responsibility and Reform (2010; the Bowles–Simpson Commission) and the Bipartisan Policy Center’s Debt Reduction Task Force (2010; the Domenici–Rivlin Commission) recommended massive scaling back of tax expenditures in order to use the revenue to reduce the deficit and decrease statutory tax rates. President Bush’s 2005 tax reform commission similarly recommended eliminating and limiting many tax expenditures, although to a lesser extent than did the deficit reduction panels.
Revenue
There are two key points regarding tax expenditures and revenue potential. First, the revenue value of most tax expenditures rises with the marginal tax rate. For example, a deduction or exclusion that reduces taxable income by US$1 would save fifteen cents in tax liability for someone in the 15 percent tax bracket and thirty-five cents for someone in the 35 percent bracket. Similarly, the value of deferral options and preferential tax rates is related to the ordinary income tax rate. As a result, reducing tax rates will mechanically reduce the value of tax expenditures but is unlikely to generate net new revenues and should be distinguished from closing or curtailing tax expenditures, which typically will raise revenue.
Second, tax expenditure reform has the potential to raise significant amounts of revenue. Although precise estimates are difficult to compute, illustrative calculations indicate the potential for revenue raising. Assuming the tax code reverts to its pre-2001 rates, a summation of tax expenditures listed in the FY2014 Budget reveals that tax expenditures would reduce revenues by US$1.32 trillion in the 2015 fiscal year, with US$1.12 trillion coming from lower individual income receipts and US$160 billion from lower corporate income receipts (Marron 2012; Office of Management and Budget 2013). Since refundable credits have outlay effects (US$100 billion) and the elimination of exclusions and other provisions would increase payroll receipts and excise taxes (US$120 billion), eliminating all tax expenditures could theoretically raise US$1.54 trillion in 2015 (Marron 2012). This estimate still may understate the potential revenue of base broadening because of interaction effects; for example, Toder and Baneman (2012) found that the combined revenue loss from all nonbusiness individual income tax expenditures would have been 9.6 percent greater than the summed individual revenue loss in 2011. 7
At the same time, the potential revenue gains should be adjusted downward for a couple reasons. Behavioral effects may significantly alter the revenues that could be raised from tax expenditure reform; Poterba and Sinai (2008) found that portfolio adjustments would offset 14 percent of static revenue increase if the mortgage interest deduction were eliminated. In addition, the potential revenue from reform may be limited by administrative feasibility, which Nguyen et al. (2012) show account for about 17 percent of tax expenditures and political constraints. All of these items make precise estimates of the revenue potential from restricting tax expenditures difficult.
Distributional Effects
Tax expenditure reform not only has the potential to raise revenues but to do so in a progressive manner. High-income households tend to be able to use more of the existing tax expenditures than lower-income households, and—as noted earlier—the tax benefit per dollar of tax expenditure used (except for credits) rises with a household’s marginal tax rate. Toder and Baneman (2012) found that if all nonbusiness individual income tax expenditures had been eliminated for 2011, the outcome would have been broadly progressive, with the top 1 percent of the income distribution experiencing a decline of 19.8 percent in after-tax income, whereas the bottom quintile would have seen only a 7.5 percent decrease (figure 4).
Decrease in after-tax income from eliminating all individual income tax expenditures.
Yet, distributional estimates of tax expenditures should be interpreted with caution (Burman, Toder, and Geissler 2008; Toder, Harris, and Lim 2011; Toder and Baneman). First, changes in tax expenditures may induce change in behavior, as noted earlier. Second, the use of the revenues would affect the overall distributional impact (Burman, Toder, and Geissler 2008).
Moreover, the aggregate changes, noted above, obscure the fact that different tax expenditures have widely varying effects on the distribution of income (Hungerford 2006; Altshuler and Dietz 2008; Burman, Toder, and Geissler 2008; Toder and Baneman 2012). Itemized deductions, exclusions and in particular the preferential rates for capital gains and dividends barely benefit the lower-income quintiles but heavily benefit the upper-income quintiles. Above-the-line deductions and non-refundable credits, on average, reduce after-tax income by a similar proportion among the income quintiles. Refundable credits, like the Earned Income Tax Credit (EITC) and the Child Tax Credit, deliver substantial income assistance to low-income households.
Efficiency
Given the diversity of tax expenditures, it is difficult to generalize about their efficiency impact. In general, treating sources and uses of income the same for tax purposes is thought to be the right benchmark from which to begin consideration of efficiency within an income tax system. A reduction in the tax rate for one activity generally requires higher rates on other activities to raise the same amount of revenue. This was, for example, the guiding notion behind the Tax Reform Act of 1986, even if it was not fully achieved then, as well as the Bowles–Simpson and Domenici–Rivlin proposals.
There are, of course, important reasons why this rule is not a perfect guide. Different sources or uses of income may have different elasticities, which may dictate different optimal tax rates, according to the Ramsey rule. Likewise, the presence of externalities may also indicate that special treatment of particular activities is appropriate. If a tax expenditure corrects a market failure, it can increase economic efficiency.
The efficiency effects of reductions in tax expenditures are often contrasted to increases in income tax rates. The frequent claim that reducing tax expenditures does not raise effective marginal tax rates, though, can be overstated. For example, consider someone in a 35 percent marginal tax bracket who is able to shelter, through tax expenditures, 20 percent of his income. His effective marginal tax rate is 28 percent, since he pays 35 percent tax on 80 percent of his income. If the tax expenditures were eliminated and his statutory tax rate reduced to 28 percent, his effective marginal tax rate would remain at 28 percent. That is, although the reduction in statutory marginal tax rates reduces the effective tax rate, the reduction in tax expenditures raises the effective marginal tax rate. In this particular case, those two effects happen to exactly offset.
In addition to the direct efficiency loss of higher tax rates, the political economy of tax expenditures can lead to indirect efficiency losses. Burman and Phaup (2011) suggest that taxpayers’ perception of a lighter tax burden through tax expenditures may encourage them to demand more government services via tax expenditures than they would through outlays, thus causing an inefficient provision of government services. Furthermore, discretionary spending (but not entitlement spending) goes through periodic review, but tax expenditures do not. Provision of government services through tax expenditures, therefore, may be both less effective and less efficient than an equal amount of outlay spending.
Moreover, most tax expenditures do not correct market failures and instead promote an inefficient allocation of resources. The itemized deduction for mortgage interest, for example, does not correct a failure of the housing market. It actually appears ineffective at promoting homeownership by instead encouraging households to acquire bigger mortgages and larger houses (Gale, Gruber, and Stephens-Davidowitz 2007; Toder et al. 2010). And it may even reduce homeownership, as the subsidy is capitalized into home prices, reducing the demand among young workers (Bourassa and Yin 2007).
The employer-sponsored insurance (ESI) exclusion subsidizes the cost of health insurance and increases access to group insurance. Gruber and Poterba (1996), however, found that it also leads to overconsumption of high-cost, inefficient health insurance plans, which increases overall health care expenditures. The recent health care legislation limited the value of health insurance premiums that are excluded from taxation by introducing a tax on “Cadillac” health plans, effective in 2018. This may still enable access to insurance more generally while mitigating the inefficient consumption of health care.
It is also unclear whether the deferral of tax on retirement saving in pensions, 401(k) plans, and individual retirement accounts are effective at increasing private saving: those who take advantage of retirement saving often would have saved using alternative means, meaning the tax expenditure tends to alter the type of private saving much more than the level of private saving. The deferral of retirement saving also imposes significant revenue losses that reduce public saving (Engen, Gale, and Scholz 1996; Poterba, Venti, and Wise 1996; Gale 1998; Benjamin 2003; Gelber 2011; Chetty et al. 2012).
This does not mean that important policy goals cannot be achieved through tax expenditures. The EITC has been effective at promoting work among low-income single mothers. Eissa and Liebman (1996) found that the EITC, which functions as a wage subsidy that is provided through the tax code, increased labor force participation of low-income single mothers by 1.9 to 2.8 percentage points, and others have found similar effects (Meyer and Rosenbaum 2001; Meyer 2002; Eissa and Hoynes 2005).
Policy Options
The most radical option would be to repeal all tax expenditures. This would raise substantial revenue and increase the progressivity of the tax code. Nevertheless, it seems unlikely, for political and administrative reasons, and it may not be advisable on economic grounds either. Just as the public discussion of government spending does not typically involve eliminating major programs like Social Security or Medicare, eliminating a whole stream of core social and economic policies merely because they exist in the tax code rather than on the outlay side would not necessarily make sense.
A more practical, but still sweeping, option would be to convert itemized deductions to flat 15 percent credits (Gale 1997; Batchelder, Goldberg, and Orszag 2006). This would raise about US$1.2 to US$1.3 trillion over the 2013–2022 projection window. 8 It would improve efficiency by limiting the inefficient allocation of resources. It would improve horizontal equity by equalizing the tax benefits of tax-preferred activities across taxpayers. And it would improve vertical equity by eliminating the regressive nature of current itemized deductions. For example, converting the mortgage interest deduction to a 15 percent non-refundable credit would nearly double the number of tax units with cash income less than US$50,000 that receive a tax benefit from having a mortgage on their home (Baneman et al. 2011). If the credits were made refundable, the progressivity of the change would increase, but the revenue effects would fall.
A less extreme version of the idea of converting deductions to flat credits is the Obama administration’s proposal to cap the benefits of itemized deductions at 28 percent. This effectively leaves the deductions as they are for people in tax brackets of 28 percent or less and converts the itemized deduction to a 28 percent credit for people in higher tax brackets.
An alternative proposal would be to place an overall cap on the value of tax expenditures. Feldstein, Feenberg, and MacGuineas (2011) propose to cap the tax reduction value of tax expenditures at 2 percent of adjusted gross income (AGI). For example, a taxpayer with AGI of US$50,000 could reduce tax liability by up to US$1,000 under such a proposal. If he was single and faced a 25 percent marginal tax rate, he would be able to use up to US$4,000 worth of deductions and exemptions. The proposal has obvious political attractions for our elected leaders, in that they would not have to curtail specific tax expenditures. Feldstein et al. estimate that the 2 percent cap could have raised US$278 billion in 2011. 9
Yet, the proposal raises several issues. In general, it is important to distinguish the idea of having a cap from the particular cap that Feldstein et al. proposed. One issue is what should be included under a cap? In their proposal, Feldstein et al. include itemized deductions, health insurance, and the child credit but leave out capital gains and retirement saving. Other choices would generate a progressive change.
A second concern is how the cap adjusts with respect to income. Feldstein et al. make the tax value a uniform share of income and set that share at 2 percent. This formulation could affect lower- and middle-income households in unexpected ways. For example, a household with median income of US$50,054, an average employer-sponsored health insurance policy worth US$15,745, and a marginal tax rate of 15 percent would not be able to deduct the entire tax value of the plan from its income, and the cap would prevent using any other deductions. 10 A high-income household, however, would be able to deduct the full tax value of its ESI, in addition to its deductions. A more progressive option could, for example, make the cap a decreasing function of income.
The implicit order of deductible payments under a cap is also relevant. Since households must pay their state and local taxes, many would end up using their entire cap on this tax preference. Only if there is value remaining in the cap after state and local taxes are paid will one receive a tax benefit for other deductions or exclusions.
Another option for tax expenditure reform would be to tailor solutions that recognize the distinct differences in each of the tax expenditures. This is perhaps the most appealing from an economist’s perspective, as each item could be evaluated separately but may be more difficult politically. The system that emerged from such evaluations would, in our view, share the following characteristics. Capital gains would be taxed as ordinary income (Burman 1999, 2012). Dividends would be taxed once and only once at either the corporate or individual level. The mortgage interest deduction would be converted to a US$10,000 first-time homebuyers tax credit (Gale, Gruber, and Stephens-Davidowitz 2007). The charitable deduction would continue in full as an above-the-line deduction as President Bush’s Advisory Panel on Federal Tax Reform recommended in 2005. Retirement-saving deductions would be converted to 15 percent matching credits (Gale 2011). Employer-provided health care would continue to be exempted at the firm level, but would be taxed as regular income at the individual level. The resulting increase in tax liability would be either completely or partially offset by a uniform refundable credit, similar to the proposal by Senator McCain in 2008, that incentivizes taxpayers to demand lower-cost health plans (Furman 2008).
The VAT
Under a VAT, businesses pay taxes on the difference between their total sales to other businesses and households and their purchases of inputs from other businesses. That difference represents the value added by the firm to the product or service in question. 11 The sum of value added at each stage of production is the retail sales price, so the VAT simply replicates the tax patterns created by a retail sales tax and is like other taxes on aggregate consumption. The key distinction is that VATs are collected at each stage of production, whereas retail sales taxes are collected only at point of final sale. Furthermore, the VAT is easier to enforce and is widely regarded as having a superior administrative structure to a retail sales tax.
Although it would be new to the United States, the VAT is in place in about 150 countries worldwide and in every Organisation for Economic Co-operation and Development (OECD) country other than the United States. Experience suggests that the VAT can raise substantial revenue, is administrable, and minimally harmful to economic growth. Additionally, the VAT has at least one other potential advantage worth highlighting: a properly designed VAT might help the states deal with their own fiscal issues. This section discusses these issues and addresses several concerns that have been raised about the VAT.
Revenue
Among non-US OECD members in 2006, the VAT raised almost 7 percent of GDP in revenue, and accounted for almost 19 percent of revenue raised at all levels of government. As with any tax, revenue from a VAT depends on the rate structure and the base. The standard VAT rate, the rate charged on most goods and services, has remained relatively steady in recent years in non-US OECD countries. In 2007, it ranged from a low of 5 percent in Japan to a high of 25 percent in Denmark, Iceland, Norway, and Sweden. The average rate was 18 percent (OECD 2008).
The VAT “yield ratio” measures VAT revenues as a share of GDP divided by the standard VAT rate. A ratio of 0.3, for example, implies that a 10 percent VAT raises 3 percent of GDP in revenues. 12 Note that the yield ratio does not include the net costs of policies intended to compensate low-income households for VAT payments, nor do they include the offsetting effects that the VAT may have on other revenue sources. The yield ratio simply measures how much revenue is actually gained from the VAT itself.
In 2006, in non-US OECD countries, the yield ratio ranged from a low of 0.28 in Mexico to a high of 0.69 in New Zealand. Most countries fell within a range of 0.3 and 0.4 (OECD 2008). The yield ratio depends critically on the extent to which the VAT tax base is kept broad, rather than narrowed by preferential rates or exemptions on certain goods or services. In practice, most OECD countries apply preferential rates to some items. Of the twenty-nine OECD countries with a VAT in 2007, seventeen countries “zero rated” certain goods (meaning that VAT is not charged on the retail sale of the good, but credits are awarded on the VAT paid on the inputs), and twenty-one applied at least one nonzero reduced rate to a subsector of goods. Only Japan and the Slovak Republic have no preferential rates (OECD 2008).
Toder and Rosenberg (2010) estimate that the United States could raise gross revenue of US$355 billion in 2012 through a 5 percent VAT applied to all consumption except for spending on education, Medicaid and Medicare, charitable organizations, and state and local government. This would represent about 2.3 percent of GDP and produce a yield ratio of 0.45 (table 2).
Revenue Effects in 2012 of a 5 percent VAT.
Source: Toder and Rosenberg (2010).
Note: GDP = gross domestic product; VAT = value-added tax.
However, as discussed further below, governments often provide either subsidies or exemptions in the VAT. One way to do so is to exclude some preferred items. For example, exempting rent, new home purchases, food consumed at home, and private health expenditures from the VAT in the United States would reduce revenue by 38 percent, cutting the yield ratio to 0.28.
A different way to provide subsidies is to give each household a cash payment. Using the broad base, the provision of a cash payment of US$437 per adult and US$218 per child would, according to Toder and Rosenberg (2010), cost US$97.7 billion. Note that, under this option, the official revenue collected by the VAT would remain at US$355.5 billion and the measure of the yield ratio—given by VAT revenues and the standard rate of 5 percent—would remain at 0.45. But what might be called the effective revenue—that is, the revenue gain from the VAT net of the costs of making the compensatory cash payments—would fall to US$257.8 billion, or 1.64 percent of GDP, giving an “effective” yield ratio of 0.33.
Imposing the VAT would reduce net business income, which would in turn reduce other revenues. Toder and Rosenberg estimate that declines in other tax receipts would offset about 27 percent of gross VAT revenues. This would reduce “effective” revenues—after netting out the costs of cash payments and the loss in other revenues—of 1.02 percent of GDP for either base, resulting in an “effective” yield ratio of 0.2.
These figures imply, after allowing for offsetting adjustments in other taxes and the costs of either cash payments or narrowing the base as described earlier, that a 10 percent VAT would raise just over 2 percent of GDP in revenues.
Efficiency
A broad-based VAT that is levied uniformly on all goods and services would not distort relative prices among consumption goods. Similarly, a VAT with a constant tax rate over time would not distort household-saving choices, nor would it distort business’s choices regarding new investments, financing instruments, or organizational form. 13 Relative to higher income tax rates—which would distort all of the choices noted above—the VAT has much to offer in the way of incentives. Like the income or payroll tax, however, the VAT would distort household choices between work and leisure. The VAT is border adjustable; it would exempt exports and tax imports. While this is sometimes touted as providing economic benefits, it is actually a neutral treatment of these items.
A substantial literature, based on economic theory and simulation models, documents the potential efficiency gains from substituting a broad-based consumption tax for an income tax (Auerbach 1996; Fullerton and Rogers 1996; Altig et al. 2001). These gains arise from a combination of broadening the tax base, eliminating distortions in saving behavior, and imposing a onetime tax on existing wealth.
The tax on existing wealth merits additional discussion. As a tax on consumption, the VAT can be regarded as a tax on the wealth and income that households use to finance current and future consumption: wealth that exists at the time of the transition to the VAT, future wages, and extra-normal returns to capital (Hubbard and Gentry 1997). 14 The tax on existing wealth is a lump-sum tax, since the wealth has been already accumulated. Lump-sum taxes are preferable to other forms of taxation on efficiency grounds, since they do not distort economic choices. In fact, the lump-sum tax on existing wealth is a major component of the efficiency gains due to the creation of a consumption tax. 15
The efficiency and growth effects due to an add-on VAT would include both losses from the increased distortion of work/leisure choices and substantial gains noted earlier from the onetime tax on existing wealth and substantial gains from deficit reduction, discussed previously.
Distributional Effects and Offsetting Policies
In theory, the distributional burden of the VAT depends crucially on how household resources are measured. Typical distributional analyses are made with respect to current income. The VAT is regressive if households are classified by, and the tax burden is measured as a share of current income. Because the VAT is a proportional tax on consumption, and because lower-income households tend to spend a larger proportion of their income than higher-income households, the VAT imposes higher burdens—as a share of current income—on lower-income households.
However, several other perspectives are possible. The VAT is a proportional tax if households are classified by current consumption since all households are taxed at the same rate on the amount they consume. Likewise, to the extent that current consumption mirrors average lifetime income, the VAT is also proportional with respect to lifetime income. Empirical research broadly confirms these notions (Caspersen and Metcalf 1994; Metcalf 1994; Toder and Rosenberg 2010). However, empirical analysis is complicated by the fact that alternative methods of distributing the burden of a consumption tax, such as distributing the burden to consumption versus wages and capital less investment, can produce drastically different estimates of progressivity, even though they are equivalent in theory (Burman, Gravelle, and Rohaly 2005).
As mentioned earlier, the VAT imposes a onetime tax on existing wealth, a feature that is desirable on efficiency grounds but is more controversial with regard to fairness. We believe a onetime tax on wealth would be fair, and in fact would be quite progressive. There is concern that imposing a VAT would hurt the elderly, a group that has high consumption relative to its income. However, it is the case that Social Security and Medicare are the principal sources of income for a substantial proportion of low-income elderly households. Since those benefits are effectively indexed for inflation, low-income elderly households would be insulated from any VAT-induced increases in the price of consumer goods or health care services. 16 High-income elderly households, who receive much lower shares of their income in the form of indexed government benefits, would need to pay more in taxes but could afford to do so.
Concerns about the regressivity of the VAT are complex, but they should not obstruct the creation of a VAT for two reasons. First, while we accept the validity of distributional considerations, what matters is the progressivity of the overall tax and transfer system, not the distribution of any individual component of that system. Clearly, the VAT can be one component of a progressive system.
Second, it is straightforward to introduce policies that can offset the impact of the VAT on low-income households. The most efficient way to do this is to simply provide households either refundable income tax credits or outright payments. For example, if the VAT rate were 10 percent, a US$3,000 demogrant would equal VAT paid on the first US$30,000 of a household’s consumption. Households that spent exactly US$30,000 on consumption would pay no net tax. Those that spent less on consumption would receive a net subsidy. Those that spent more on consumption would, on net, pay a 10 percent VAT only on their purchases above US$30,000. Toder and Rosenberg (2010) estimate that a VAT coupled with a fixed payment to families is generally progressive, even with respect to current income.
In contrast, many OECD governments and state government offer preferential or zero rates on certain items like health care or food to increase progressivity. This approach is largely ineffective because the products in question are consumed in greater quantities by middle-income and wealthy taxpayers than by low-income households. 17 Furthermore, this approach creates complexity and invites tax avoidance as consumers try to substitute between tax-preferred and fully taxable goods and policy makers struggle to characterize goods. For example, if clothing were exempt from the VAT, Halloween costumes classified as clothing would be exempt, while costumes classified as toys would not.
Carbon Taxes
Throughout this article, we use the phrase “carbon tax” to refer to a tax on carbon dioxide and possibly other greenhouse gas emissions. Although a carbon tax would be a new policy for the federal government, the tax has been implemented in several other countries. Finland, Norway, Sweden, and Denmark instituted carbon taxes in the early 1990s, followed by the Netherlands and Germany in the latter part of the 1990s. The United Kingdom followed suit in 2001. Australia introduced a carbon tax in 2011. North American jurisdictions have also implemented carbon taxes. The town of Boulder, Colorado, adopted a carbon tax in 2006, and Montgomery County, Maryland, did so in 2010. The Canadian provinces of Alberta and Quebec adopted carbon taxes in 2007, followed by British Columbia in 2008.
Revenue
Carbon taxes can raise significant amounts of revenue. For instance, in 2007, the tax raised revenue equivalent to about 0.3 percent of GDP in Finland and Denmark and 0.8 percent in Sweden. A well-designed tax in the United States could similar amounts. A number of studies have estimated the revenue effects of carbon taxes, with estimates ranging from 0.6 percent of GDP for a US$20 per ton tax (CBO 2010; Rausch and Reilly 2012) to 1.6 percent of GDP for a US$41 per ton tax (Paltsev et al. 2007) and several intermediate estimates (Metcalf 2008; Shapiro, Pham, and Malik 2008). Estimates in McKibbin, Morris, and Wilcoxen (2009) and Metcalf (2010) suggest that a US$30 per ton tax would raise about 1 percent of GDP in revenue. In terms of gauging how large a tax that would be, Bauman (2010) has estimated that a tax of US$25 per ton would raise gasoline prices by 25 cents a gallon.
The net effects of a carbon tax on the deficit will depend, of course, not only on the magnitude of the tax and the behavioral response by consumers and firms, as the studies above consider, but also on the offsets provided for distributional or transition purposes, and the uses of the funds. In many instances to date, carbon tax revenues have not been used for deficit reduction. Norway and Sweden do include carbon tax revenue as part of general government receipts, which suggests a possible effect on deficit reduction. But carbon tax revenue in Denmark is returned to industry and directed toward environmental subsidies. Several nations have used carbon tax revenue to reduce other taxes (Sumner, Bird, and Smith 2009). Australia coupled its carbon tax with a substantial increase in the tax-free level of income (and other tax changes). The Netherlands and Sweden have exempted a large portion of the industrial sector from the tax, as well as helping low-income households offset the burden of the tax (the latter measure was also implemented by Germany). Quebec deposits carbon tax revenues into a fund devoted to public transportation and environmental initiatives, while British Columbia makes its carbon tax revenue neutral by reducing corporate and personal income tax rates and providing an annual credit of US$100 per adult and US$30 per child to lower-income citizens.
Efficiency
In principle, carbon taxation receives high marks on efficiency criteria. Indeed, the basic rationale for a carbon tax is that it makes good economic sense: unlike most taxes, carbon taxation can improve the efficient allocation of resources by accounting for externalities in the market price. Externalities can be severe. Stavins (2007) notes that the efficiency benefits of a carbon tax are often understated since the largest efficiency gains come in the form of internationally shared reduced greenhouse gas emissions. While the United States is the largest per capita emitter of carbon dioxide, China is the largest overall emitter, and the European Union makes a significant contribution as well. Therefore, enacting a program that would lead to better cooperation with other countries and reduce emissions across the world would be better suited to deal with the well-known problems brought about by global warming, such as rising sea levels, more frequency in extreme temperatures, among others.
Taxes on carbon can address these externalities. Not surprisingly, most analyses find that a carbon tax could significantly reduce emissions. For example, Metcalf (2008) estimates that a US$15 per ton tax on carbon emissions would reduce greenhouse gas emissions by 14.0 percent. Sumner, Bird, and Smith (2009) estimate that the European countries’ carbon taxes have had a significant effect on emissions reductions, attributing reductions of up to 15 percent to the carbon tax. Andersen (2010) estimates that the six European countries that introduced carbon taxes saw reduction in fuel demand of 2.6 percent in the 1994–2003 period compared to a baseline with no such taxes in place. Furthermore, the University of Ottawa (2012) found that the carbon tax implemented in British Columbia led to a 9.9 percent reduction in greenhouse gas emissions in the province, compared to just 4.6 percent for the rest of Canada.
In addition to reducing emissions, a carbon tax could improve other economic incentives by reducing other tax rates or paying down the deficit (Parry and Williams 2011). A carbon tax could have other benefits too. It would reduce the US economy’s dependence on foreign sources of energy, and would create better market incentives for energy conservation, the use of renewable energy sources, and the production of energy-efficient goods. The permanent change in price signals from enacting a carbon tax would stimulate new private sector research and innovation in developing new ways of harnessing renewable energy and energy-saving technologies.
The implementation of a carbon tax could also be used a mechanism to phase out the panoply of targeted subsidies for biofuel production. For example, since the 1970s, the United States has subsidized the production of ethanol and other biofuels as a means of reducing greenhouse gas emissions, reducing US dependence on foreign energy, and stimulating production in the agricultural sector. These subsidies are costly to administer, both in terms of distortions in market behavior and lost revenue. Because they reduce the price of gasoline, some research suggests the subsidies may increase instead of reducing greenhouse gas emissions because the increased emissions from increased gasoline consumption may more than offset the reduced emissions from substituting ethanol for gasoline. 18 For example, in 2009, the exemption of biofuels from highway motor fuel taxes reduced excise tax receipts by US$6 billion (CBO 2010). 19 A carbon tax that accurately reflected the price of these negative externalities would be a more efficient mechanism for achieving reductions in greenhouse gas emissions and improved energy security.
Finally, a well-designed carbon tax could improve the efficiency of tax collection. Metcalf and Weisbach (2009) note that taxing a few upstream producers is more efficient than taxing all downstream consumers. This would not change the incidence of the tax, but it could reduce the costs of compliance, enforcement, and administration.
Distribution
Distributional concerns over carbon taxes stem from the observation that low-income households devote a higher proportion of their income to consumption and will thus bear a higher burden of the tax relative to high-income households. The distributional effects of carbon taxation have been well studied (Bull, Hassett, and Metcalf 1994; Metcalf 1999, 2007; Hassett, Mathur, and Metcalf 2009). The regressivity finding is consistent across studies, but varies in magnitude. Metcalf (2008) analyzes the distributional effects of a carbon tax and finds that it would reduce the after-tax income of taxpayers in the first decile by 3.7 percent, compared to just an 0.8 percent reduction for the wealthiest decile. Findings are dependent on whether incidence is measured on a current income versus lifetime basis, with the tax being more regressive when measured on a current income basis relative to lifetime income basis. For example, Hassett, Mathur, and Metcalf (2009) find that the indirect component of a carbon tax (i.e., higher prices due to higher costs of production) is significantly more progressive, whereas the direct component, which focuses on the changes in the cost of gas and electricity, is regressive. Finally, the incidence varies with timing: the carbon tax can either fall forward in the form of higher consumer prices or backward in the form of lower returns to factor inputs. Bovenberg and Goulder (2001) and Paltsev et al. (2007) find that the short- and medium-term incidence falls primary on consumer prices.
Importantly, the regressive impact of a carbon tax could be offset in any of a number of ways, similar to offsets for distributional effects of the VAT, discussed in the previous section. Most prominent among these options would be refundable income tax credits. Thus, while the regressivity of a carbon tax should be addressed, it should not be considered an obstacle to implementation of carbon taxes.
Gasoline Taxes
Raising taxes on gasoline is another option. While a modest excise tax on gasoline sales already exists in the United States, it is substantially lower than in other industrialized nations.
In the United States, federal excise taxes on gasoline amount to 18.4 cents per gallon, with local tax rates typically taxing gasoline at additional 20–30 cents per gallon in 2010. The OECD average for gasoline excise taxes is approximately US$3.39 per gallon, about seven times the rate of the US tax. 20 OECD taxation of gasoline ranged from US$0.34 per gallon (Mexico) to US$5.14 per gallon (Turkey); the United States has the second lowest rate of gasoline taxation among OECD countries (OECD 2011). In addition, per-mile fuel taxes in the United States are low by historical standards, falling by 40 percent in real terms since 1960 (Parry, Walls, and Harrington 2007).
American levels of gasoline taxation are also well below the cost of the externalities of gasoline consumption. Parry, Walls, and Harrington (2007) estimate the per-gallon externality cost of gasoline at US$2.38 per gallon; about half the externality cost is due to congestion and the rest is from accidents, pollution (including greenhouse gases), and oil dependency.
A higher tax on gasoline could raise significant amounts of revenue. CBO (2009) estimates that a 50 cent increase in the excise tax would raise about US$600 billion over ten years—0.3 percent of GDP.
A gasoline tax is less efficient than a carbon tax, since the former covers a narrower range of externality-producing goods, but it could still result in significant emission reductions. Davis and Kilian (2009) find that a 10 cent per gallon increase in the US gasoline excise tax would reduce total carbon emissions by 0.5 percent overall and by 1.5 percent from vehicles. Sterner (2007) similarly estimates that fuel demand in Europe would be twice as high if European governments had implemented a gasoline excise tax schedule similar to that in the United States.
Like carbon taxes, gasoline taxes will fall disproportionately on low-income households, especially in the short run when households have difficulty adjusting their behavior to avoid the tax (Poterba 1989, 1991).
Some analysts have proposed a variable tax on gasoline to help stabilize the price of energy, reduce economic volatility, and provide a minimum price signal that should encourage production of alternative energy sources, all while still raising revenue. For example, Westin (2010) proposes an oil price stabilization tax, where an excise tax would serve as a mechanism for creating a floor on the price of domestically consumed gasoline.
Conclusion
Given the sheer magnitude of the long-term fiscal gap, revenue increases will need to be an important part of any resolution to reduce the deficit. However, this presents both challenges and opportunities. The most difficult challenge is political: enacting tax increases when a vast majority of Republican members of Congress have signed a “No New Taxes” pledge.
Yet, the need for revenues also presents the opportunity to reform the tax system in ways that can raise significant amounts of revenue, enhance long-term growth, and maintain or improve progressivity. Implementing the reforms in this article—broadening the income tax base, establishing a consumption tax, and/or implementing a carbon tax—would raise revenue, and still allow for lower statutory marginal income tax rates and credits or other offsetting mechanisms to alleviate any negative distributional consequences. Moreover, such reforms could yield significant improvements in the efficiency of economic decisions and tax administration while also promoting other goals such as reducing greenhouse gases and raising investment and national saving.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.