The late economist David Bradford was fond of a joke that he constructed to illustrate the intersection of tax law and the budget process. He would propose a marvelous new way to cut taxes without affecting government services: Instead of wasting tax revenues on military equipment purchases, Congress could implement a “Weapons Supply Tax Credit,” under which arms manufacturers would receive a tax credit for delivering to the U.S. Government weapons meeting certain specifications. The amount of the credit would equal what Congress might formerly have spent on purchasing those weapons. Then Congress could announce that, through this “targeted tax relief,” taxes had been slashed without jeopardizing our security or increasing the deficit.

The joke, of course, is that nothing at all would have changed; the federal government still would obtain the same weapons and incur the same economic cost to do so. Our accounting for the transactions, however, would differ. Instead of recording government revenues from taxes collected and government expenditures for national defense, we would just report net lower taxes collected. Before, the government took in $1,000 and spent $100 on fighter planes. After, the government would record just $900 in revenues, and some “free” planes would arrive at the Air Force’s doorstep. On paper, the government had gotten smaller; in reality, it would be as large as ever.

Bradford’s joke is meant as a gentle parody to illustrate the empty formalism of our concepts of government revenues and expenditures. When the government subsidizes people or businesses by writing them checks, we all recognize that intervention as government spending. When the government subsidizes the same people or businesses to the same extent by giving them a targeted tax break, that action often is mischaracterized as “lower taxes” or “smaller government.”

Specialists term these synthetic government spending programs “tax expenditures.” Tax expenditures are really spending programs, not tax rollbacks, because the missing tax revenues must be financed by more taxes on somebody else. Like any other form of deficit spending, a targeted tax break without a revenue offset simply means more deficits (and ultimately more taxes); a targeted tax break coupled with a specific revenue “payfor” means that one group of Americans is required to pay (in the form of higher taxes) for a subsidy to be delivered to others through the mechanism of the tax system.

Like all parody, Bradford relies on hyperbole to make his point. How, then, would he have reacted to an Internal Revenue Service news release from April 6th, 2009, announcing the publication of Notices 2009–23 and 2009–24? The first of those notices explained to taxpayers how to apply for $250 million in tax credits to be allocated by the IRS and the Department of Energy (surely a bureaucratic odd couple) for delivery of certain “Phase II [qualifying] gasification projects.” The second announced $1.25 billion in tax credits to be awarded, again by the same two agencies, for certain advanced coal facilities. In effect, the notices announced the transfer of hundreds of millions of dollars from the federal government to selected energy companies.

Bradford’s joke has lost its punch line. We now find ourselves living inside the parody and thinking it normal.

Tax expenditures have grown in importance to the point where they are now the dominant instruments for implementing new discretionary spending policies. We spend more than twice as much through tax expenditures as we do through old-fashioned explicit non-defense spending programs. Tax expenditures dissolve the boundaries between government revenues and government spending. They reduce both the coherence of the tax law and our ability to conceptualize the very size and activities of our government.

Tax expenditures not only distort tax policy and obfuscate our understanding of government operations; they also adversely affect the workings of Congress. Contemporary tax expenditure practices have elevated the tax-writing committees to a special status: a Congress within the Congress. These committees now fill both fundamental functions of a legislature: they raise revenues and they spend those revenues themselves through the tax subsidies that they marry to tax increases in shaping “revenue neutral” legislation. These three consequences of our over-reliance on tax expenditures — the growing incoherence of tax policy, the obfuscation of the size and activities of our government, and the elevation of the tax-writing committees to a special status — are all profoundly corrosive phenomena.

A Tax Expenditure Fable

How exactly do tax expenditures erode our ability to conceptualize the size and activities of government? And how can tax expenditure analysis improve political discourse in the real world? To answer those questions, I have constructed a little example involving the small but self-reliant country of Freedonia. Its economy is comprised of 10 fruit and vegetable growers, each earning $1,000 pre-tax, for a total gross domestic product of $10,000. Each grower pays income tax to support the Freedonian army at a flat rate of 15 percent, for total tax revenues of $1,500.

Freedonia’s sole kumquat producer is particularly resourceful. Armed with scientific reports showing the many health benefits of kumquat consumption, he convinces the Freedonian legislature that kumquat production deserves tax incentives, to bring kumquats within the reach of every Freedonian family. The legislature responds by effectively exempting kumquat production from its income tax through an innovative kumquat production tax credit.

But Freedonia is not a profligate state, and it believes in fiscal discipline in the form of pay-as-you-go budget rules. Therefore, to keep the kumquat credit revenue-neutral, the legislature pairs the new preference with an 11.1 percent tax hike on the other producers, to maintain tax revenues at $1,500. (Freedonian tax policy allows for rounding error.) That means that the other fruit and vegetable farmers will each pay $167 (instead of $150) in tax on their $1,000 of income.

In a world without tax expenditure analysis, the Freedonian legislature can argue that nothing has changed: government revenues are constant, and there is no increase in government spending or borrowing. But this is plainly wrong; things have changed, in both the private and public sectors.

First, the tax incentive increases kumquat production and consumption. The equilibrium price and quantities sold of kumquats will be different relative to other fruits and vegetables after the tax incentive. Economists believe that, in the absence of some identifiable market failure, markets set prices better than legislatures do, but the kumquat credit alters the quantity of kumquats sold relative to the case in which the tax burden of all fruit and vegetable growers was equal. Unless the overall health of Freedonians really is improved by the kumquat credit (perhaps because of prior rampant borderline scurvy among the population), the result will be a less efficient allocation of our collective resources.

Second, the introduction of the kumquat credit in an apparently virtuous “revenue neutral” fashion has another profound economic effect: tax rate increases on the incomes of all the fruit and vegetable producers who do not receive targeted tax relief. All taxes, no matter how beautifully implemented, impose “deadweight losses.” That is, some transactions that are rational in a world without taxes become too expensive in a world with those taxes and do not take place. And deadweight loss increases faster than the tax rate: in standard presentations, in fact, at the square of the tax rate.

What all this means is that, by virtue of granting “revenue neutral targeted tax relief,” the Freedonian government may raise the same aggregate revenues as it did previously, but impose more deadweight loss on the remaining taxable Freedonian private sector. This result is one of the great ironies of many tax expenditures, particularly those that fall into the category of business incentives — once the incentive’s impact on tax burdens for others is considered, it impoverishes the country even more than it enriches the beneficiaries of the legislative largesse.

Third, by virtue of its new kumquat credit, the Freedonian government just got bigger, even though aggregate nominal tax revenues remain constant. The best way to see this effect is to employ Louis Kaplow’s favorite mode of analysis and analogize the new kumquat credit to a uniform 11.1 percent tax hike on all of Freedonia’s fruit and vegetable producers, followed by a $167 kumquat crop farm subsidy payment to the kumquat producer. By recasting the tax expenditure in this way, as a constant tax burden and a separate transfer payment, the two different functions of government are restored to their customary formal presentation, and the words “revenue” and “spending” can be applied consistently to economically identical (but formally different) modes of implementation. As so recast, it is easy to see that Freedonia’s economic handprint on the private sector is no longer $1,500 in tax revenues, but rather $1,667 in economic terms. The government is bigger in every meaningful sense of the word.

Fourth, deadweight loss cannot be avoided by electing “targeted tax relief” without revenue offsets. This point may seem obvious to many readers, but I am confident that, without explicit discussion of the point, at least some policymakers would conclude that the only problem with tax expenditures is trying to pay for them.

The simple fact is that, as the third lesson sought to demonstrate, the kumquat credit is a form of government spending. The government has four choices as to how to finance its spending. First, it can pay for its incremental spending on a current basis by raising taxes on someone else. Second, the government can borrow money today and repay it with future taxes. Third, the government can borrow today and inflate its way out of the problem — but inflation is just another kind of a tax imposed haphazardly and often cruelly on capital owners or claimants to fixed revenue streams like pensioners. Fourth, the government can borrow today and default on its debt in the future. No other options are available.

If you agree with me that the fourth way is probably a bad idea, then you are left with the realization that each new tax expenditure necessarily implies a tax increase. I can think of no more important principle of public finance for policymakers to absorb, and tax expenditure analysis can help to focus the mind on this inescapable truth.

The Measure of Tax Expenditures

Let us now turn from little Freedonia to the fractious political realities of the United States. By any measure, tax expenditures represent an enormous part of the U.S. Government’s operations. As of October 2008, the Congressional Research Service reported 247 tax expenditures worth $1.2 trillion. Ninety percent of that money represented tax expenditures for individuals, while 10 percent ($118 billion) represented a reduction of corporate income tax.

How can I put $1.2 trillion of 2008 tax expenditures into context? That sum is greater than the entire amount raised by the individual income tax in 2008, or for that matter all federal discretionary spending in that year (in each case, about $1.1 trillion). Indeed, it is more than twice as much as all non-defense discretionary spending in 2008 ($528 billion).

Tax expenditures have grown rapidly in number over the years. The Joint Committee on Taxation (JCT) staff’s first tax expenditure list in 1972 counted 60 such items. In 2008, the committee staff counted 247 items. If $118 billion in 2008 corporate tax expenditures sounds small in relation to the amount spent on individual tax subsidies, the sum represents roughly 39 percent of 2008 corporate tax receipts ($304 billion) or, if you prefer a somewhat less aberrational year, 32 percent of 2007 corporate tax receipts ($370 billion).

Bradford’s joke uses military spending as its example, but very few tax expenditures relate to national defense. It is instructive to compare tax expenditures to explicit non-defense discretionary outlays. To do so, consider the regularly published overview of the functional components of the federal budget, prepared by the Center on Budget and Policy Priorities (CBPP). The document separates the budget into six high-level categories: defense, Social Security, Medicare/​Medicaid/​CHIP, “safety net programs,” “interest on the national debt,” and “everything else.” Both the CBPP and the Congressional Research Service Compendium of Tax Expenditures follow the standard federal budget presentation to construct their categories, except that the CBPP puts payments to federal government retirees and veterans in the “everything else” category. Eliminating that item, about 13 percent of the federal budget ($390 billion) was spent in 2008 on “everything else” in its more ordinary sense.

To compare explicit outlays to tax expenditures in the areas of government services to which most tax expenditures actually are directed, I subtracted from the $1.222 trillion in total 2008 tax expenditures those items that the Congressional Research Service categorized as “social services” expenditures (other than the charitable contribution deduction), “income security” expenditures (excluding employer-provided life or disability insurance and tax expenditures relating to private pensions, as they are above and beyond the “safety net”), and veterans affairs expenditures. The result ($1.074 trillion) can then usefully be compared to the magnitude of explicit outlays in the CBPP’s “everything else” category, excluding payments to federal retirees and veterans benefits ($390 billion). By doing so, one discovers that our non-defense, non–safety net annual spending through tax subsidies is about 275 percent of the amount of explicit government outlays in those areas. In other words, when looking at education, transportation, scientific research, and every other activity by which the federal government touches the day-to-day lives of middle class and affluent Americans under the age of 65, our official scorekeeping captures only 27 percent of the government’s actual spending.

So tax expenditures are enormous in absolute terms, are larger than explicit government spending in comparable areas, and have grown rapidly. As a share of GDP, tax expenditures are now at a much higher level than in 1974, when federal accounting for tax expenditures was first officially adopted. At the same time, explicit discretionary spending actually has declined substantially as a percentage of GDP from levels around 10 percent of GDP in the early 1980s to less than 8 percent today.

In 1974, for example, the simple sum of all tax expenditures amounted to 5.7 percent of GDP. Tax expenditures climbed from that level to an all-time high in the mid-1980s of 9.7 percent of GDP and then fell because of the base broadening and rate reductions of the Tax Reform Act of 1986. Tax expenditures reached a modern low of 5.3 percent of GDP in 1991.The rate stayed in the neighborhood of 6 percent of GDP during most of the 1990s, but then began a steep climb.

By my calculations, the simple sum of all tax expenditures in fiscal year 2008 totaled an extraordinary 8.6 percent of GDP, the highest percentage since the mid-1980s. To put this number in context, if tax expenditures today were the same percentage of GDP as was the case in 1974, the simple sum of 2008 tax expenditures would have been some $412 billion lower than the actual estimates! Contrast that number with Congress’s fiscal year 2010 budget resolution, which calls on the tax-writing committees to scrounge up a total of $97 billion in tax “loophole closers” over the next five years — less than $20 billion per annum.

The 2008 figure is very close to the situation in 1985, when tax expenditures amounted to 8.7 percent of GDP. Aggregate tax expenditures as a percentage of income taxes were also very close in the two years (87 percent in 1985 and 84 percent in 2008). This fact is particularly telling because most tax expenditures are expressed as deductions or exclusions, and their value fluctuates with tax rates: in lower-rate environments, tax (deduction) expenditures have lower value. In general, 2008 was a much lower tax rate environment than was the case in 1985. The fact that tax expenditures today are running at roughly the same percentage of GDP and income tax revenues as in 1985 confirms that tax expenditures have multiplied in degree as well as in number.

Official tax expenditure estimates by the Treasury Department and the JCT staff include only income tax expenditures. This estimate understates the importance of tax subsidies in some important areas, particularly energy policy. A recent Department of Energy study, for example, found that tax expenditures for energy production amounted to $10.4 billion in 2007; of this amount, nearly $3 billion was attributable to excise tax credits for ethanol production alone and absent from official JCT staff and Treasury Department lists of income tax expenditures. Consistent with my larger theme, that Department of Energy study also found that, in constant 2007 dollars, tax expenditures for energy production and conservation more than tripled over eight years, from $3.2 billion in 1999 to $10.4 billion in 2007, while total federal government financial support (including tax expenditures) only doubled.

The Negative Effects of Tax Expenditures

Tax expenditures deprive the committees of Congress with substance-matter expertise of the ability to determine how tax expenditures are designed or spent. They do not track the efficacy of the tax programs, they do not necessarily coordinate that spending with their own explicitly appropriated spending, and they have lost the ability to argue that their priorities should be preferred over those reflected in the tax legislation.

The tax-writing committees increase spending on policies of their choosing by decreasing the salience of those benefits to most observers (but not, of course, to the beneficiaries) and decreasing the salience of the tax costs incurred to finance those spending policies. Lower tax salience is associated with bigger government (that is, a larger tax base). The result is a classic example of fiscal illusion in which, arguably, both taxpayers and many members of Congress underestimate the tax increases implicit in “revenue neutral” legislation.

The growth of government through tax expenditures is facilitated by both their lack of presentation in the formal budget and the blurring of expenditure authority within the Congress. We cannot determine by inspection of our budget how much support the federal government provides to the energy sector, for example, nor do we know the nature of those programs. Because the facts are not presented in a straightforward manner, we cannot debate fairly the efficiency costs of a system in which spending and revenues are disguised from both citizens and most legislators. Tax expenditures augment fiscal illusion, and fiscal illusion, in turn, drives poor policy.

The budgetary imperative to spend through the tax system also interferes with the internal workings of Congress. Petitioners for federal largesse can and do file claims with multiple committees: if the Farm Bill created through the normal authorization process does not contain what you want, just ask the Senate Finance Committee for a tax credit. The result is appropriated spending and tax credits that duplicate, overlap, or conflict with one another.

This phenomenon has been widely studied with respect to social services, but it also applies to energy policy and most other instances of discretionary spending (other than military procurement where, perhaps mindful of stirring up the ghost of David Bradford, we have not yet implemented the Weapons Supply Tax Credit). Moreover, no government agency is charged with presenting to the public and to Congress as a whole an annual comprehensive picture of the total costs of all the discretionary outlay budget functions, including tax subsidies as well as direct outlays, although the component data are published and studies of individual areas are prepared from time to time.

In the same vein, “permanent” tax subsidies are not subject to any sort of review or oversight by the congressional committees charged with substance matter expertise, and no comprehensive congressional review of tax expenditures exists. So tax expenditures, once implemented, are essentially unmonitored by any arm of Congress. Instead, they simply disappear below the surface into the mainstream of baseline revenues.

Containing the Growth of Tax Expenditures

Why have the other committees of Congress allowed their own relevance to be diluted in the manner that I have described? Why do the committees with substantive responsibility for energy policy, for example, tolerate a situation in which 63 percent of all federal energy subsidies and support are routed through the tax system?

Here, I can only offer three tentative observations. First, tax expenditures have electorally useful characteristics. They include the permanent authorization (or at worst semi-automatic renewal) of funds for beneficiaries; the certification of eligibility by fund recipients rather than agencies; and the ability to spend money invisibly compared to items with explicit budget line items. Second, the phenomenon of logrolling — actually a technical term of political science, not an aspersion — may be at play. Third, the role of congressional committees is in the process of rapid evolution. As Richard E. Cohen wrote in a National Journal article last year,

Since the 1970s and ’80s, the once-mighty committee system controlled by autocratic chairmen has crumbled. Instead, power has been centralized in recent decades in the leadership, with major legislation often written ad hoc by party leaders working with a few key chairmen or other members behind closed doors, outside of the committees.

What can be done about the irresistible political attraction of tax expenditures? The most important step would be to require that all tax expenditures be recorded as spending for all budget purposes. That is, tax subsidies not only would be presented as spending in future budgets, but would go through the same legislative processes as do explicit spending proposals, including referral to the appropriate authorization committee. This proposal essentially would explicitly recognize that taxes in the 2008 budget were about $3.7 trillion rather than the $2.5 trillion listed in the actual budget. For the same reason, 2008 spending was about $4 trillion rather than the $3 trillion officially listed.

By making tax expenditures explicitly on-budget spending programs, the size of government would no longer be hidden. In addition, all uses of government funds would compete on an open, level playing field.

In contrast, current practice, comprising “revenue neutral” tax legislation filled with new tax subsidies paired with tax increases, fulfills neither of these objectives. The process is expressly designed to avoid any visible imprint on the budget, and the programs so favored have not been forced to compete with other spending programs for scarce government resources in other committees or among the members as a whole. The “revenue neutral” package is analyzed as simply summing to a net zero impact and is accepted by the floor as a unit, rather than having its constituent parts examined with the care that would be devoted to debating a package of explicit spending proposals offset by explicit tax increases.

A variation of my proposal that is more sensitive to political realities would cap invisible tax expenditures at a percentage of the previous year’s GDP. All tax expenditures above the cap would be expressed as explicit spending. For example, we could cap unarticulated tax expenditures at their 1974 level of 5.7 percent of GDP. If this were applied to Fiscal Year 2008, it would bring on budget over $400 billion of tax expenditures.

I would go one step further and also require that the tax-writing committees not be permitted to take up a tax subsidy measure without referral from the relevant authorization committee with substance-matter jurisdiction. The substantive committee, in turn, would be required periodically to recertify that the subsidy should be renewed. By introducing this measure, substantive oversight and policy design would be centralized in the committee charged with maintaining substance-matter expertise in the area. The larger agenda is to elevate the role of the authorization committees (those with substance-matter responsibility) to serve as the clearing house for all proposals touching on their area of competence. My hope would be that by doing so they might serve as a better counterweight to the budget, appropriations, and tax-writing committees.

Conclusion

Tax subsidies have grown in importance to a size unimaginable 40 years ago, when tax expenditures first were identified as a budget problem. The current congressional appetite for tax expenditures has distorted not just tax policy, but also the budget process and congressional oversight of the money it spends. The tax-writing committees have employed tax expenditures to extend the committees’ reach and to become, in fact, a Congress within the Congress. They present new spending policies embedded in “revenue neutral” tax legislation in ways that are largely invisible in standard budget presentations. It is time to redirect tax expenditures to those cases where they represent the most appropriate technical delivery mechanism for government spending.

Readings

  • “Rethinking Tax Expenditures and Fiscal Language,” by Daniel Shaviro. Tax Law Review, Vol. 57, No. 2 (2004).
  • “Tax Expenditure Framework Legislation,” by Edward D. Kleinbard. National Tax Journal, Vol. 63 (June 2010).
  • “The Congress within the Congress: How Tax Expenditures Distort Our Budget and Our Political Processes,” by Edward D. Kleinbard. Ohio Northern University Law Review, Vol. 36, No. 1 (2010).
  • “The Integration of Tax and Spending Programs,” by David A. Weisbach and Jacob Nussim. Yale Law Journal, Vol. 113 (2004).