Beruflich Dokumente
Kultur Dokumente
Daniel N. Shaviro
September 2003
Daniel N. Shaviro*
Preliminary Draft
I. INTRODUCTION
Tax expenditure analysis is like a hardy plant with shallow roots that spreads
widely, resisting the occasional effort to extirpate it, and yet has little if any effect on the
soils in which it sprouts. At least sixteen countries release tax expenditure data,
pertaining not just to income taxes but also to sales, value-added, payroll, inheritance,
excise, property, motor vehicle, and betting-and-lottery taxes.1 However, the aim of
original American proponent Stanley S. Surrey, that tax expenditures once identified be
mostly converted into direct expenditures or repealed altogether,2 has made little if any
headway.3 This reflects not just political support for particular provisions and their
character as tax rules, but broader rejection of key tenets of tax expenditure analysis.4
*
Professor of Law, NYU Law School.
1
Organization for Economic Co-operation and Development, TAX EXPENDITURES:
RECENT EXPERIENCES 7, 65 (1996) (henceforth, “OECD Report”). The countries
included in this list are Australia, Austria, Belgium, Canada, Finland, France, Germany,
Ireland, Italy, Japan, Luxembourg, the Netherlands, Portugal, Spain, the United
Kingdom, and the United States.
2
Stanley S. Surrey, PATHWAYS TO TAX REFORM 179-180 (1973). Tax expenditure
analysis apparently sprung up more than a decade earlier in Germany, without the same
normative underpinnings and apparently without being noticed in the United States. See
Harry A. Shannon, The Tax Expenditure Concept in the United States and Germany: A
Comparison, 33 Tax Notes 201 (1986).
3
See Victor Thuronyi, Tax Expenditures: A Reassessment, 1988 Duke L.J. 1155, 1170-
1181.
4
See id. at 1178; Bruce Bartlett, The End of Tax Expenditures As We Know Them?, 92
Tax Notes 413, 419 (2001).
2
Such rejection, while encouraged by canards,5 reflects as well powerful criticisms of the
analysis that emerged almost immediately in the American academic literature6 and have
persisted since.7
To complain that tax expenditure analysis has not been politically more effective,
in the precise way that Surrey envisioned, might be asking not just too much but the
wrong thing. Even if one accepts his normative goals, one may rightly ask to what extent
a mere fiscal language innovation either can or should categorically shape political
conversation that the concept has generated in the tax policy community, too narrowly
focused on the defensibility of using of a canonical “reference tax base” in identifying tax
expenditures.
This article aims to put the tax expenditure concept on a more fundamental
footing, grounded in a fuller appreciation of fiscal language issues generally. Among its
1) The basic claim of tax expenditure analysis, that certain tax rules are “really”
spending, is not quite correct, because “taxes” and “spending” are not coherent categories
to begin with. However, the underlying idea can be restated in terms of Richard
Musgrave’s famous distinction between the allocative and distributional branches of the
5
I have in mind here the claim that tax expenditure analysis reflects the “sinister premise
…. [that one should] think of all income as virtual state property, and forbearance to tax
away every last penny of it as itself a tax expenditure.” Charles Fried, “Whose Money Is
It?,” Washington Post, January 1, 1995, page C-7. I discuss this view of tax expenditure
analysis infra at __.
6
See Boris I. Bittker, Accounting for Federal “Tax Subsidies” in the National Budget, 22
Nat’l Tax J. 244 (1969).
7
See, e.g., Douglas A. Kahn and Jeffrey S. Lehman, Tax Expenditure Budgets: A Critical
Review, 54 Tax Notes 1661 (1992).
3
fiscal system.8 Allocation involves affecting the amount, use, and character of all assets
in our society, while distribution involves affecting who has what. Thus, paying police
and building roads are allocative activities, while using income measures to determine tax
for its familiarity) are best defined as allocative rules that, as a formal matter, are found
degree and, in some cases more than others, of reasonably contested opinion. Tax
expenditure analysis ought to be more flexible and varied in its groupings than in the
Surrey tradition, where each rule is canonically classified as tax expenditure or not
relative to a specified reference tax base that itself reflects both contestable distributional
made for tax penalties or negative tax expenditures, arising in cases where a tax rule is (at
least arguably) harsher than a purely distributional policy might be thought to suggest.
3) One key reason for the value of tax expenditure analysis as an exercise, even in
the flawed manner in which it is currently done, is that it addresses the confusion in
public policy debate that may occur when proponents of placing particular allocative
rules in the tax system exploit the common tendency to define “taxes” and “spending”
entirely formally, and yet to treat the categories as genuinely meaningful. When an
allocative program uses tax benefits, such as special deductions or credits, in lieu of
direct appropriations, “taxes” and “spending” for the fiscal system as a whole may appear
8
See Richard Musgrave, THE THEORY OF PUBLIC FINANCE v (1959).
4
expenditure analysis, undermined this clarifying function by also enlisting the analysis as
a weapon in battles concerning what the government’s distribution policy should look
like – in particular, his support for progressivity and comprehensive income taxation.
Even those who share his distributional views ought to recognize by now that tax
expenditure analysis cannot be politically effective – even leaving aside the intellectual
language issues that underlie classifying particular government rules as taxes or spending.
Section III discusses how tax expenditures could be and have been defined. Section IV
addresses the important (but little-noticed) discussion of tax expenditure analysis by the
U.S. Treasury Department that was included in the official Budget of the United States
Government for 2003 (henceforth, the “Treasury Discussion”),9 and offers suggestions
for the future use of tax expenditure analysis. Section V offers a brief conclusion.
9
See Analytical Perspectives, Budget of the United States Government for Fiscal Year
2003, pages 95-127.
5
Political Weapon
The government, through its agents’ actions and the rules it imposes, does a great
vouchers), we may describe it as operating a fiscal system. The fiscal character of its
actions is a matter of degree, and can be emphasized or not as one likes. Thus, as I have
discussed elsewhere, one might think of minimum wage laws either as workplace
off-budget transfer to those same workers.10 The rationale for distinguishing between the
fiscal system and everything else the government does is simply ease of measurement.
For example, one can more easily place a dollar value on an individual’s retirement
pension, or even the subsidized healthcare services that she receives, than on her benefit
Even once one has determined which rules to treat, for a given purpose, as within
the fiscal system, one needs further organizing terminology. Otherwise, one simply has a
welter of rules with no clear organizing principle to help in thinking about them. Among
the more prominent fiscal language terms in current or recent American usage are taxes,
spending, fines, user fees, budget deficits or surpluses, trust funds for programs such as
Social Security, and lockboxes. These are not Platonic categories, but rather clusters of
ideas, often with associated connotations. Thus, consider taxes and spending. The
10
See Daniel Shaviro, The Minimum Wage, the Earned Income Tax Credit, and Optimal
Subsidy Policy, 64 U. Chi. L. Rev. 405 (1997).
6
former might be defined as payments to the government and the latter as payments from
the government, with additional refinements to distinguish them from the likes of
contract payments, fines, and user fees. This does not, however, exhaust their widely
accepted meaning. At least in American usage, there is a bit more baggage to consider.
involving high taxes and spending. All else equal, therefore, political actors will want to
categorize their opponents’ proposals as involving high taxes and spending, and their
own proposals as involving low taxes and spending, even if the only real difference in
purportedly objective descriptive tool and a weapon of political combat. However, its
use as a political weapon is parasitic on its claim to offer objective description. For
rather than reflecting something significant about the rule’s substance, then any inference
we are being invited to draw from the label, such as that it is an example of “big
Unfortunately, the characteristics that underlie our fiscal language categories are
often purely formal, rather than meaningful. Consider the distinction between taxes and
spending, based on whether cash is traveling to the government or from it. Anyone who
ever shops or has a job, is likely to make some payments to the government. Indeed,
nearly everyone ends up making net lifetime payments to the government, even if she
gets good value in her overall dealings, since the receipt of public goods is excluded from
7
the calculus. Against this background, it makes no economic difference whether, at any
given moment, one gets a dollar back (“spending”) or is allowed to pay a dollar less
The distinction between taxes and spending thus depends on pure form. For
example, “ spending increases” can be converted into “tax cuts” if they are netted against
tax payments before made in a manner that looks sufficiently as if the tax system is the
responsible agent. And we seemingly can have “smaller government” in lieu of “bigger
government,” even without any policy change, simply by netting things out so that the
What ought we really to mean by “big government”? Or, leaving aside any
implications that it is good or bad, what should interest us, from a consequential
standpoint, about a given government program? Presumably, what we really care about
is its effects on allocation or distribution – that is, on what we have as a society and who
among us has it. A bigger government, therefore, is one with greater rather than lesser
Our fiscal language would actually be dealing in substance if it thus described the
counter-factual state baseline might pose quandaries. (No government? But also no
Hobbesian state of nature? Current policy as the baseline for the effects of proposed
changes?)11 Yet even if this left some scope for political warfare over fiscal language –
since baselines may look normative even if chosen just for convenience – at least the
11
For further discussion of these issues, see Daniel N. Shaviro, The Bush II Tax Cuts:
Steps Towards Bigger Government?
8
incidence and allocative effects of various programs are scarcely matters of consensus.
The use of cruder and more formal measures is therefore unsurprising – although more
ought to be done in developing the substantive measures, and although the ones we now
examples of the problems with common usage of “taxes’ and “spending” – all drawn
from the United States due to my greater familiarity with its fiscal system, but (I hope)
having broader application. The theme that should emerge from them is that prevailing
fiscal language is too manipulable in some respects yet too rigid in others. The first and
last examples show how it can be manipulated, subject to rules of the game that are
arbitrary even if at times effective. The middle two are more sinister, because they show
how fiscal language can mislead even sincere advocates of a given policy or, through its
rigidity, rule out what might be reasonable policy options on bogus grounds.
12
For an initial effort at describing how the allocative and distributional measures might
be constructed, see David Bradford, Reforming Budgetary Language, CESifo Working
Paper No. 619 (December 2001).
9
Benefits
In 1993, President Clinton sought to reduce the federal budget deficit through a
package that combined spending cuts with tax increases. Mindful of years of Republican
attacks on Democrats’ claimed proclivity to “tax and spend,” his Administration started
out by promising a 3-1 ratio of spending cuts to tax increases. However, this gradually
As an aside, the fact that the ratio dropped, notwithstanding the evident political
advantages of “cutting spending,” shows that the relevant fiscal language was not
that furnished particular goods and services, while changing distribution so that high-
income individuals would bear a greater burden. The former programs were mainly
classified as spending, while the latter aim in large part had to be done through the tax
system. So in practice the spending and tax measures were not wholly unrelated to
[C]ritics redid the numbers in many ways …. [and] the Senate Budget
Committee minority (Republican) staff concluded that the version of the
Clinton program that passed the House incorporated $6.35 in taxes for each
$1 in spending cuts. The argument involved in part matters of defining
baselines and of netting …. [One big dispute, however, concerned] the
Clinton Administration’s proposal to increase the portion of Social Security
retirement benefits that are subject to income taxation (a proposal that was
subsequently enacted). The Administration described this proposal as a
10
reduction in spending of $21.4 billion over five years. It did not count the
change in its table of “revenue provisions,” where tax changes are
customarily summarized. Critics cried foul and they found support in the
analysis by the even-handed Congressional Budget Office that placed the
provision in the category of “revenue proposals.”13
was not alone in thinking that it was. He also captured third place in the 1993
this maneuver plus his insistence on “using the word ‘investment’ as a substitute for the
word ‘spending’ in his rhetoric on economic policy.”14 Apparently, however, the Reagan
Administration had gotten away with classifying income taxation of Social Security
benefits as a benefit cut15 - naturally, without protest from Clinton’s subsequent critics.
The Clinton Administration argued that its proposal really was a spending cut
because its effect was to reduce seniors’ net Social Security retirement benefits. It was
“wrong,” apparently, because the Social Security checks that seniors got in the mail
would not be affected. Instead, the change affected their computations of taxable income
and thus the income tax payments they sent to the federal government. So apparently it
Suppose, however, that there had been a politically acceptable way to give the
Social Security Administration information from seniors’ tax returns that it could use to
reduce all benefit checks by exactly the amount of the tax increase under the actual
13
Id. a 4-6.
14
Jan Ackerman, Forked Tongues Prevail on High; Pentagon Gets Annual Doublespeak
Award from Teachers Group, Pittsburgh Post-Gazette, November 22, 1993, page B-1.
15
Steven Greenhouse, Clinton’s Economic Plan: Seeing Figures, 2 Sides Calculate
Clinton’s Math, New York Times, February 22, 1993, page A-14.
11
proposal and enactment. Then, despite identical economic effects, this presumably would
have been accepted as a spending change rather than a tax change. Under this scenario,
however, a brand-new concern might have arisen. The United States would have been
means-testing Social Security benefits – a gross violation, in some people’s view, of the
sacred social compact under which Social Security is a universal program, and yet
evidently not a concern under the actual (and substantively identical) 1993 enactment.
In his proposed federal budget for 2003, President Bush proposed tax cuts that the
Congressional Budget Office estimated would cost $1.5 trillion over ten years despite
being backloaded through gradual phase-in.16 Congress then passed tax cuts that were
expected to cost $318 billion over ten years, although the reduction largely reflected
sleazy accounting tricks such as “sunsetting” new provisions that the proponents actually
planned and expected to retain. The tax cuts were enacted notwithstanding enormous
Cutting taxes at a time when United States fiscal policy already fell so far short of
being on a sustainable path raised questions of what (if anything) the Administration and
its Congressional allies had in mind for the long term. One widely bruited theory among
observers held that the unstated underlying goal must be to “starve the beast,” i.e., reduce
16
[Cite CBO]
17
Estimates of the long-term fiscal gap, or the present value of the excess of expected
spending over expected taxes under a reasonable projection of current policy, ranged
from $44 trillion to $74 trillion. [Cite Gokhale & Smetters; Shaviro in World Econ. J.]
12
politically unfeasible, would be inevitable down the road.18 Given existing budget
priorities, this scenario would have to involve enormous future cuts in Social Security
and Medicare benefits. Reducing outlays, say, on highways, farm subsidies, and welfare
However deceptive one might find the Administration’s methodology under this
scenario (which offered the only way to make sense of its long-term policy), it could at
Administration correct, however, in thinking (if it did) that the government really would
be smaller over time if taxes were cut now and entitlements cut (even more than would
This is far from clear. The belief that it is clear may rest on spending illusion, or
confusing the size of the nominal dollar flows between individuals and the government
with the actual size of government. Again, we should think of the size of government in
terms of its allocative and distributional effects relative to some state of affairs where it
was doing less. Reciprocal cash flows, however, such as the case where you and the
government give each other a dollar, do not involve such effects. Social Security and
Medicare have enough in common with mere dollar swaps that even the allocative and
distributional effects which they concededly have do not come anywhere near to being
18
If so, however, then it was curious that the Administration also was proposing spending
increases estimated at $750 billion over ten years, including a costly expansion of
Medicare to include a new prescription drug benefit. [Cite CBO].
13
that is actuarially fair, in that people’s retirement benefits equal the value of their
allocative effects if people are far-sighted and can borrow and lend at the interest rate
used to equate contributions and benefits. Such a program would therefore have no
effects whatsoever, despite its involving billions of dollars of observed “taxes” and
Security in the United States, are not nearly so vacuous. They may involve vast
redistribution – in practice, chiefly from younger to older age cohorts. In addition, they
easily dissave in advance the accruing value of their retirement benefits. They therefore,
Even so, however, these effects are not well-measured by the observed cash flows from
the programs’ taxes and spending.20 Surely the government would be much bigger, in
real terms, if more of the gross amounts were being redistributed and if the outlays took
the form of specific government-produced goods and services that the benefiting seniors
of the Bush Administration policy of enacting enormous tax cuts in 2003, significantly to
19
Cite MAKING SENSE OF SOCIAL SECURITY REFORM and Nicholas Barr.
20
The case of Medicare is slightly more complicated, because seniors are required to
consume the retirement benefits in the form of healthcare. It is likely, however, that the
benefits are worth a high percentage of their cash cost to the recipients, reflecting the
relative low compensated elasticity of healthcare expenditure. Cite WHO SHOULD PAY
FOR MEDICARE.
14
the benefit of the wealthier (and thus generally older) members of current generations,
and to be made up down the road through reductions in Social Security and Medicare
benefits? In answering this question, one should keep in mind that current seniors have
done much better under these programs on a lifetime basis than younger generations were
expected to do even before the 2003 tax cuts. Largely for this reason, lifetime net tax
rates (lifetime taxes paid minus transfers received, divided by lifetime income) are
expected to be much lower for current and older generations than for future and younger
generations.21 In consequence, the 2003 tax cuts arguably made the government bigger.
They lowered one already-low set of lifetime net tax rates in exchange for raising another
already-high set of such rates. They thereby may have increased total redistribution
through the U.S. fiscal system (just as defaulting on government bonds might have this
effect). The 2003 tax cuts also may have increased the total distortionary effects of the
fiscal system, by likely requiring higher marginal tax rates in the future, thus violating the
principle of “tax smoothing” under which rates should remain relatively constant in order
ideological goals (rather than just short-term political goals) in its 2003 budget policy, it
may have been deceived by the fiscal language of nominal taxes and spending into doing
the opposite of what its presumed anti-government ideology might dictate. Yet the news
gets worse still. If there is any good reason at all to favor current over future generations,
it is that the latter may end up, for technological reasons, being wealthier (transfers
21
Cite generational accounting estimates.
22
See Robert J. Barro, On the Determination of the Public Debt, 87 J. Pol. Econ. 940
(1979). Note also the point about distortion rising with the square of the rate.
15
through the fiscal system aside) than we are.23 So the Bush Administration, contrary to
People on the cusp of escaping poverty, such as by shifting from part-time to full-
time work, often face astonishingly high effective marginal tax rates. Several years ago, I
estimated that an American single earner with two dependent children who lived in a
state with relatively generous welfare benefits would potentially be better-off earning
$10,000 per year than $25,000 per year.24 The benefit of earning an extra $15,000 per
year before taxes would be more than offset by the combination of various positive tax
Needy Families, Food Stamps, the earned income tax credit, Medicaid, and rent subsidies
percent, I was concededly rejecting the distinction that fiscal language conventions draw
between taxes and spending. I was equating income-related spending phase-outs with
income-related taxes. The rationale I offered was that a dollar is a dollar: income-related
spending reductions have the same incentive and distributional effects at the margin as
income-related tax increases.26 For example, if earning a dollar leaves me only eighty-
five cents ahead of where I was previously, it should not matter whether this reflects a
23
Cite Do Deficits Matter. Note the contrary argument from Medicare that we may want
to transfer resources to richer people who can do more with a dollar due to technological
advances.
24
Cite my Tax Notes piece on MTRs.
25
Note Kotlikoff on whether it pays to work.
26
Cite the MTR article.
16
fifteen-cent income tax liability or the loss of fifteen cents worth of cash-equivalent Food
Stamps.
The structural reason for imposing high marginal tax rates on people as they try to
poor. Middle-income people, for example, are thought not to need them. If a benefit is
to be limited to the poor, then it must be phased out (or eliminated in a sudden “notch”)
somewhere within the range where one’s income, earnings, or assets become great
enough to indicate that one no longer is poor. Accordingly, despite the widely
reasoning suggests that an economic penalty equivalent to that from imposing a high rate
There was actually a relatively simple structural approach to easing this dilemma,
known to policymakers as the negative income tax or demogrant approach. Under this
approach – really just a change in fiscal language – basic welfare grants would be
nominally universal, but financed by an income-related levy, with the consequence that
they could match the desired generosity (be it high or low) of any conventional welfare
scheme without necessarily imposing high effective marginal tax rates on those escaping
poverty.
The basic idea can be illustrated as follows. Suppose we think of $10,000 as the
poverty line, and $7,500 as the minimum amount that we can tolerate any member of the
“deserving” poor ending up with. (For present purposes, it does not matter how we
27
The dilemma can be avoided by simply not offering large benefits to the poor. The
time limits on welfare benefits that the United States imposed in connection with its 1996
welfare reform are an example of this approach. Cutting benefits is not generally the
right solution, however, if one actually wants to aid the poor.
17
define the deserving poor, so long as deservingness is defined other than in terms of
income.) Under a conventional welfare scheme, deserving people with zero income
might get a $7,500 grant, ratably phased down to zero as their incomes rose to $10,000.
Within this income range, therefore, under this approach we would have imposed a 75
percent effective marginal tax rate, disregarding any other taxes or means-related
transfers.
In order to pay for this program, something else would have to be done. Suppose
we previously had an income tax with a zero bracket for up to $10,000 of income and a
flat 25 percent rate above that, and that we proposed to pay for the program solely by
raising the flat rate to 30 percent. We would now effectively have two marginal tax rates:
Under the demogrant approach to describing the welfare system, all deserving
people as defined without respect to income – even Bill Gates if he otherwise qualified –
would nominally get the $7,500 demogrant. (It might, however, be netted against income
tax liabilities rather than literally being paid.) We still would have to pay for the
income tax rate on up to $10,000 of income and a 30 percent rate on income above that.
We now would have a set of programs that was identical to those involving conventional
welfare. People’s net receipt or tax at all income levels would be exactly the same. This
choice seems politically unlikely, however, since most people believe that income tax
rates should be flat or graduated. Thus, adoption of the demogrant approach – or more
precisely, the demogrant description of the welfare system – seems likely to lead to a
different outcome, with lower marginal rates for the first $10,000 of income and thus
18
necessarily (all else equal, and assuming a current-year breakeven budget constraint) of
Would the likely outcome of saying we have demogrants be better than the likely
outcome of saying we have a conventional welfare system? That depends on how one
evaluates the substantive policy issue of designing rate structures. Clearly, however,
marginal rate decisions would be more transparent under the demogrant approach.
Moreover, that approach would offer a wider range of choices if the conventional welfare
approach, in practice, virtually compels a given marginal rate structure once we have
determined (perhaps on very different grounds) the minimum transfer and maximum
the welfare system would face severe obstacles, rooted in the conventional terminology
(b) By giving demogrants to everyone (even Bill Gates if, income aside, he counts
as “deserving”), it gives money to people who do not need the aid because they are not
(c) It requires marginal tax rates to be higher than under conventional welfare.
But this is only true if we decide on a bottom-tier marginal rate of less than 75 percent.
Even in that case, marginal rates are merely different (higher in some ranges and lower
19
than others), rather than higher. Conventional welfare seems to have lower marginal
In sum, adoption of the demogrant approach would itself be only a fiscal language
change, but one that (if it were considered persuasive) might plausibly encourage the
adoption of real policy change. Acceptance of it is impeded, however, by its tension with
prevailing fiscal language norms that people evidently do find persuasive. People might
resist assimilating the demogrant to the income tax in their thinking even if it was
presented as a refundable tax credit on income tax returns. The nub of the problem
would lie in the cases where people got a net payment, ineluctably converting the
demogrant into “spending” rather than a “tax” notwithstanding the economic principle
that a dollar is a dollar and that transfers, therefore, are merely negative taxes.28
“Spending Cuts”
Bradford describes his pretended “secret plan” to eliminate a budget deficit by formally
cutting spending rather than taxes. In Step 1, all defense spending on weapons
procurement would be eliminated. In Step 2, a new “weapons supply tax credit” (WSTC)
would be enacted. “To qualify for the WSTC, manufacturers will sign appropriate
28
See, e.g., Musgrave, supra n. __, at 272.
20
characteristics. The WSTC, which may be transferred to other taxpayers without limit,
may only be used in payment of income tax. Step 2 is, apparently obviously, a tax cut.”29
Step 3 (changing the Bradford plan slightly) would be an income tax rate increase,
precisely making up the revenues that had been lost through enactment of the WSTC. In
form, therefore, tax revenues would be constant, while spending would have declined
steeply due to Step 1. As a result – putting President Clinton to shame – one could claim
that one had solely used spending cuts to accomplish deficit reduction, without raising
overall taxes at all. In substance, however, all that one would really have done is raise
Bradford concedes that “the chances are pretty good some astute politician or
journalist would notice what is going on in this case.” If so, then “the policy process
would [have] see[n] through to the economic substance. But maybe not; permit me to
doubt.” 30 For what it is worth, my own take on this little story is that the WSTC, because
it is transferable, violates the accepted definition of a proper “tax” no less than it would
if, like the demogrant, it were made explicitly refundable to the extent in excess of the
through the failed political experiment of safe harbor leasing, at least in the United States
“tax” benefits cannot properly be sold to taxpayers with positive liability to offset, any
proposal to work, because otherwise weapons manufacturers might not have sufficient
29
Bradford, supra n. __, at 8.
30
Bradford, supra n. __, at 8.
31
[Describe safe harbor leasing]
21
tax liability to use it in full. (Suppose, for example, that most of their income derives
from these very contracts.) Apart from this design problem, however, it is hard to see
why the Bradford plan could not work – at least, as to new weapons programs, so that its
Suppose that someone objected that no normative income tax would include a
WSTC, and that therefore it must be a “tax expenditure.” The following dialogue might
then ensue.
“Because I have consulted a number of experts, and not one of them would
Now we could lower the boom, by quoting a recent criticism of tax expenditure
analysis by Douglas Kahn and Jeffrey Lehman. Does not this appeal to the experts
falsely “presume[] that some of us should be deemed to know the answers [to defining a
normative tax base] better than others?”32 Accordingly, Kahn and Lehman ask, how is
this any different from a hypothetical case in which someone tries to answer the question:
“Should the National Zoo house panda bears?” by stating his opinion and that of other
self-proclaimed experts that “normative zoos” should only house bears and that pandas
Two things about their critique should be obvious. The first is that they have a
point. How exactly do we identify the “normative income tax,” and why should we care
32
Kahn and Lehman, supra n. __, at 1665.
33
Id.
22
about it even if we could? The second is that their argument nonetheless misses
something important.34 In terms of the discussion above, while it cannot be quite right
that the WSTC is “really” spending if the distinction between “taxes” and “spending” is
vacuous to begin with, still, there is an evident distinction between offering the WSTC
and, say, allowing a shopkeeper to deduct from the income tax the salaries that he pays
his clerks. Where the tax expenditure debate went off the rails, reasonably drawing
Kahn’s and Lehman’s ire yet throwing them off-course as well, was not in its aim of
identifying “special” provisions such as the WSTC, but in its means of doing so, through
Can we distinguish between the WSTC and ordinary business expense deductions
without becoming self-appointed zoo experts who are opining about panda bears? How
could we better expose the fiscal language scam that underlies the WSTC without
perpetrating our own fiscal language fallacies, and without courting all of the controversy
and lack of acceptance that has dogged tax expenditure analysis for the last thirty years?
Is tax expenditure analysis fatally flawed by its failing to use the more fundamental
language of allocative effects and distributional effects, or is there still a useful role that it
could play? I address these issues in the next section, after describing the origins of tax
expenditure analysis.
34
In fairness, they do describe their criticism of tax expenditure budgets as “pragmatic,
not nihilistic. We do not believe that all arguments are equally good, or equally
persuasive …. We find it valuable to point out [as tax expenditure analysts mean to do]
those provisions of the code that depart from what one would expect to find if one’s sole
concern were measuring” some measure, such as income, of relative wellbeing. Id.
23
early as 1954, writers had noticed the “equivalence between special tax deductions,
credits, and other allowances and government subsidies.”35 They apparently did not view
this as a criticism. Rather, “the German literature generally affirms that the tax system
furnishes a useful instrument for implementing economic and social policy and
acknowledges that it is often used for such ‘nonfiscal’ purposes.”36 By 1959, the German
government had begun reporting on subsidies in the federal budget, including those
supplied through the tax system, with an eye to improving budgetary control. The stated
motivation was that, if subsidies generally were to be reduced, these “invisible” and
indirect subsidies (measured by the foregone revenue) should be placed on a par with the
subsidies that were provided more overtly through direct spending.37 Budgetary reports
on tax subsidies that were classified as indirect spending were being made regularly by
1967, although it is unclear how widely this was noticed outside Germany.
rules in the tax code, was introduced in the United States with considerably more fanfare
Marketeers, a New York financial group, Stanley S. Surrey, the Assistant Secretary of the
Treasury for Tax Policy, called for a “tax expenditure budget” that would report the
35
Shannon, supra n. __, at 203.
36
Id. at 204.
37
See id.
24
revenue cost of “deliberate departures from accepted concepts of net income … [through
which] our tax system does operate to affect the private economy in ways that are usually
tax language.”38
Surrey subsequently supplied a convenient creation myth for U.S. tax expenditure
analysis. In September 1967, or two months before the speech, while sitting in a hearing
of the Ways and Means Committee of the House of Representatives, he ostensibly had
experienced a sudden “illumination.”39 The hearing had been called to assess President
Johnson’s proposal that a 10 percent income tax surcharge be enacted to help pay for the
Vietnam War. Many committee members, however, wanted to cut spending rather than
just raising taxes. They spent several days examining how this could be done, and
For the moment, the committee, in its desire to see expenditures controlled
and thus make a tax increase more palatable if it must be voted, became an
Appropriations Committee. But in its scrutiny of the expenditures listed in
the budget, the committee had forgotten what it knew as a tax committee.
Never once in its examination of the direct expenditures listed in the budget
did the committee pause to consider the dollars involved in the tax incentives
and tax subsidies contained in the Internal Revenue Code.
It was not for lack of knowledge. The committee members were completely
aware that through tax benefits the income tax law provided financial
assistance to this or that business [as well as to state and local governments
and to such groups as the aged, the sick, and the blind] …. But the
committee kept the financial assistance furnished by these tax provisions
38
Quoted in Surrey, PATHWAYS TO TAX REFORM, supra n. __, at 3.
39
Id. at 3.
25
completely separated and isolated in its mind from the task at hand. Indeed,
the connection with that task simply did not occur to the members.40
The experience of sitting through this, Surrey tells us, “suddenly illuminated
many questions.”41 Could this gap in the members’ understanding be addressed? What if
he were to have the Treasury Department staff prepare a tax expenditure budget, parallel
to the official budget listings for direct expenditures? This would aid the policy process,
the next time spending cut issues arose, both by creating a ready list of special tax
provisions that really were spending, and by giving policymakers the numbers that they
would need to put these provisions on a par with direct spending in the budget process.
In actual fact, Surrey seems to have had something similar in mind as early as
1953, when he wrote a paper (apparently never published) criticizing “technical escape
routes [from the income tax] for favored groups” that Congress had deliberately enacted,
and quantifying at length the revenue lost due to these provisions.42 Moreover, for the
past year, the U.S. Treasury department had been devoting “substantial resources to
outlining the subsidy elements of the tax code. With Surrey’s approval, Gabriel G.
Rudney, a Treasury economist, had spent a year at the Brookings Institution43 working on
the question of how a tax expenditure budget would be presented. Rudney had even
40
Surrey, supra n. __, at 1-2.
41
Id. at 3.
42
See Stanley S. Surrey, “Our Schizophrenic Income Tax,” 1953, p. 21, The Stanley S.
Surrey Papers, Harvard Law School Library, Modern Manuscript Division. Cambridge,
MA, Box 23, Folder 7. I am grateful to Dennis Ventry for bringing this paper to my
attention.
43
Jonathan Barry Forman, Origins of the Tax Expenditure Budget, 30 Tax Notes 537,
538 (1986).
44
See id at 538 n. 10.
26
The reason for the creation myth is easy to discern. Surrey’s stated motivation of
simply improving the budget process (the same motivation as that operating in Germany),
while undoubtedly sincere, was also somewhat anodyne. He had a more controversial
motivation as well. Tax reform, defined as broadening the base of the income tax so that
high-income taxpayers would pay more, had long been a personal cause of his. The tax
expenditure budget thus served for him as a tool of tax policy, not just budget policy. It
was meant to serve as a hit list, identifying provisions that should be repealed from the
tax code and either disappear altogether or else reappear as direct spending, and not just
placed on a par with direct spending whenever budgetary balance was being evaluated.
Back in private life (at Harvard Law School) once President Nixon had
succeeded President Johnson, Surrey soon made explicit his arguments against tax
expenditures. They generally were “upside-down subsidies,” unfairly aiding the rich
more than the poor, because their value depended on one’s marginal tax rate.45
and in the budget process.”46 A further concern, not always emphasized by him in this
setting, went to tax politics, which (even before his Treasury stay) he had observed was
From the start, therefore, tax expenditure analysis in the United States was both a
45
Surrey, supra n. __, at 134-138.
46
Id. at 141.
47
See Stanley S. Surrey, The Congress and the Tax Lobbyists—How Special Tax
Provisions Get Enacted, 70 Harv. L. Rev. 1145 (1957).
27
expressly made certain political calculations, such as keeping off the list, for now, items
such as gifts and imputed rent whose inclusion would unduly have “puzzled” the public
because expert understanding that they were really income remained too “novel.”48 The
fact that he had a well-known political agenda – and, in a town like Washington, might
have been assumed to have one even if he didn’t – inevitably colored reception of the
idea. From the start it attracted controversy that, in Germany, where it was more of a
straight budgetary tool without the same perceived tax policy implications, it did not.
Under Surrey’s direction, the Treasury Department published its first tax
expenditure budget in 1968, although he was unable to get it included in the President’s
official budget. The Nixon Administration, being “cool to the tax expenditure concept”49
but not completely opposed, continued both to prepare estimates like those done under
Surrey and to deny them official status in the President’s budget. However, the
Congressional Budget and Impoundment Act of 1974, a response both to deficit concerns
and to turf wars between a Congress controlled by the Democrats and a weakened
President Nixon, made mandatory the inclusion of tax expenditure estimates both in the
provisions of the federal tax laws which allow a special exclusion, exemption, or
deduction from gross income or which provide a special credit, a preferential rate of tax,
48
Surrey, PATHWAYS TO TAX REFORM, supra at 18. [But note no tax penalties, e.g., no
inflation adjustment, and double corporate tax subject to realization.]
49
Forman, supra n. __, at 541.
50
See id. at 544-545.
51
Public Law No. 93-344, 88 Stat. 297 (July 12, 1974).
28
less political freight, was not significantly different, apart from being terser. A 1970 law
defining “tax concessions” that must be estimated for purposes of comparison with direct
spending specified that they are “special exceptions to the general tax norm, which result
In the end, then, aided by the fortuity of the Watergate crisis, tax expenditure
analysis won official recognition in the United States notwithstanding the controversiality
that it derived from its association with a particular tax reform agenda. This did not,
however, mean that it had avoided the taint of being viewed in many quarters as a
tendentious exercise in furtherance of that agenda. Boris Bittker – the tax law’s great
“fox” who saw many things, and scourge of all the “hedgehogs” who saw one big thing –
had dealt the heaviest blow back in 1969, when he pounced on Surrey’s call for a “full
accounting” of expenditure items in the income tax law.53 The accounting was far from
full, Bittker noted, and the decisions regarding what to include and exclude were
arbitrary. Worse still, many rules’ status as tax expenditures or genuine tax rules simply
could not be ascertained, due to the lack of an “agreed conceptual model” for the
In fact, Bittker’s article is more of a “Yes, but” than a “No,” because he agreed at
the end that “a more limited accounting” could be useful, “provid[ing] information that
would be helpful in applying our political, economic, and ethical criteria in making
policy judgments about the income tax system.”55 However, it was a “Yes, but” in which
the “but” came first and accounted for 16-3/4 of the article’s 18 pages. Bittker may
52
See Shannon, supra n. __, at 205.
53
Bittker, supra n. __.
54
Id. a 258.
55
Id. at 260-261.
29
therefore have played an important role in conveying the message that experts were split
about the cogency of tax expenditure analysis, and that the analysis was more partisan
weapon than objective descriptive tool. Then again, given the role Surrey unmistakably
wanted the analysis to play in prompting a certain kind of tax reform, perhaps this
reception was inevitable among those who did not share his vision.
The other big objection to tax expenditure analysis that colored its acceptance was
its supposed implication that “our money belongs to the government, and that the
government is doing us a favor by not taxing it.”56 This is not a correct criticism unless
the “normative tax” that is being used to define “spending” applies at a 100 percent rate
to all of our money. Tax expenditure analysis identifies differences between the tax
treatment of different items, and assumes no taxation in excess of that which the
treatment of higher-taxed items, and has not traditionally been used to measure instances
Still, once one sheds spending illusion, tax expenditure analysis can aid an anti-
such as the “targeted tax cuts” of which President Clinton was so fond. The government
is presumably a lot bigger allocatively, for example, if it imposes high tax rates that are
subject to numerous exceptions, than if it raises the same revenue, with the same overall
progressivity, through lower rates that are subject to fewer exceptions and thus do far less
56
Thuronyi, supra n. __ at 1178 (noting that “[p]oliticians commonly attack the tax
expenditure concept” on this ground).
30
Is spending illusion the only reason conservatives have often been hostile to tax
expenditure analysis? Certainly it is a big part; the illusion is so tenacious that even
Nobel Prize-winning economists, such as Gary Becker and James Buchanan, succumb to
its siren lure.57 Conservatives may also have been responding, however, to the liberal
political agenda of many liberal supporters of tax expenditure analysis. Stanley Surrey,
for example, “felt very strongly that the tax system should be sharply progressive, and he
regarded ‘all the Mickey Mouse stuff in the Code as attenuating the progressivity of the
rate structure.’”58 Tax expenditure analysis may therefore have been viewed as a stalking
horse for greater progressivity (rather than for distributionally neutral base-broadening),
The end result of all these cross-currents has been a state of affairs in which tax
expenditure analysis lacks the real acceptance one might have expected from its
has been seriously undermined by the Bittker line of criticism, which is frequently
repeated and has not been effectively answered. Moreover, Bittker did not even make
one of the strongest criticisms of all, if we interpret the analysis as resting on a distinction
between “real” taxes and “real” spending that even he (at least as to the latter) did not
think to challenge. If taxes and spending are themselves arbitrary categories, then the
57
See Gary S. Becker and Casey B. Mulligan, Deadweight Costs and the Size of
Government, National Bureau of Economic Research Working Paper 6789 (1998)
(arguing that countries with less efficient tax systems have smaller governments because
they spend less); Shaviro, DO DEFICITS MATTER, supra, at 103 (Buchanan and kindred
theorists “show their misunderstanding when they treat the imposition of excess burden
through taxation as an alternative to feeding Leviathan rather than as en example of
Leviathan at work”).
58
Forman, supra n. __, at 538.
31
critique that a given tax rule is really spending does not quite state correctly the point that
widespread acceptance of their structure and implicit norms in order to do their work.
Thus, when President Clinton spoke of “saving Social Security” (defined as keeping the
official budget surplus at least as great as the Social Security surplus), neither the goal
nor his definition of what it implied were effectively challenged until he ceased to be on
the scene demanding it.59 Tax expenditure analysis seems instead to be viewed in many
Things have simply remained stuck in this posture for the last thirty years, and with the
the use of tax expenditure analysis elsewhere. In other countries, the analysis has been
Surrey-style tax reformers nor anti-tax and anti-government ideologies play as large a
role as they have in the United States. Lesser controversy, however, seems to have been
accompanied by less extensive inquiry into the underlying definitional problems60 and
59
[This notwithstanding silliness of the definition, as per MAKING SENSE OF SOCIAL
SECURITY REFORM].
60
See Shannon, supra n. __ at 204-205 (noting that, “because tax expenditures are
viewed as legitimate means of implementing social and economic policy in Germany ….
definitional issues have received much less attention” there than in the United States);
OECD Report at 11-14 (noting countries’ widely differing approaches to defining tax
expenditures).
32
The time is ripe, therefore, for reexamining the premises of tax expenditure
analysis, with an eye to its fiscal language prongs both as a tool for attempted objective
description and as (inevitably) a weapon in political combat. I begin with the former
prong, and with a more abstract approach than that which generally has been used in
practice.
really a spending rule, and should thus be restated as hypothetical Tax Rule B plus
Spending Rule C, which in combination are equivalent to it. If the rule at issue is
something like the WSTC, which one has determined ought not to be in the tax system to
begin with, the process of re-description is relatively simple. Tax Rule B is simply the
absence of any such rule, and its entire revenue consequences are attributed to its re-
described form as Spending Rule C. However, if the tax rule is “wrong” yet there ought
to be some tax rule – as in the case of accelerated depreciation, if one determines that it
exceeds “correct” tax depreciation (such as economic depreciation) – then the process is
more cumbersome. One must do more work in specifying hypothetical Tax Rule B
before we can attribute its net revenue loss, relative to actual Tax Rule A, to hypothetical
61
Estimates of the cost of tax expenditures generally are done on an ex post rather than
an ex ante basis, to measure the amount by which revenues actually were reduced by the
provisions given the investments that taxpayers made. See OECD Report at 14. They
thus differ from, and will often exceed, the revenue gain that would result from repealing
33
So long as hypothetical Rules B plus C are indeed equivalent to actual Tax Rule
restatement needs to be motivated. After all, we could just as easily decompose Tax Rule
A into the even more favorable Tax Rule D (say, triple WSTCs) plus Negative-Spending
Rule E (requiring the taxpayer to refund two-thirds of the triple WSTCs). Tax Rule A
could then be described as a tax penalty relative to D, as measured by E. One thus needs
to explain why a particular counter-factual should be chosen from among the infinite
possibilities as capturing the “true” character of the actual observed Tax Rule A.
In practice, the claim that actual Tax Rule A is actually hypothetical Tax Rule B
plus Spending Rule C, rather than some other pair of hypothetical rules that would add up
the same, rests on the further claim that, if the legislature had only meant to raise revenue
under its general policy for doing so, B is what it would have enacted. This claim is easy
enough to credit when A is the WSTC and B is its absence. Even for cases as easy as
In the effort to make sense of it, we can take either of two approaches: a broader
one that reinterprets what we might mean by taxes and spending, and a narrower one that
discerns a general character to particular sets of tax rules, such as the “income tax” or the
“motor vehicle tax.” Each approach has both advantages and limitations, and they should
the provisions, with resulting reallocation of investment away from items that are no
longer being taxed as favorably.
34
Under a more broadly theoretical approach than that which is typically used, the
tax system might be defined in terms of function, rather than of the direction of cash flow
between taxpayers and the government. It provides an important mechanism for dividing
equitably the costs of government, and more broadly “put[ting] into practice a conception
of economic or distributive justice.”62 We might have an income tax, for example, not
income as evidencing ability to pay, which has been selected as the proper criterion for
Richard Musgrave, in his famous tripartite conceptual division of the public sector into its
(negative taxes in Musgrave’s parlance), it does the work of the distribution branch.
Indeed, that is all it deliberately does to the extent that its allocative effects, such as on
Identifying the “tax system,” as the term is used in tax expenditure analysis, with
taxes, such as a pollution tax meant to give polluters socially optimal incentives, clearly
involve the allocation branch, yet nonetheless are “taxes” in common usage and could
62
Liam Murphy and Thomas Nagel, THE MYTH OF OWNERSHIP: TAXES AND JUSTICE 3
(2002).
63
Musgrave, supra n. __, at v.
35
principle be integral to decisions by the distribution branch. Suppose, for example, that
the legislature very slightly preferred income taxation to consumption taxation from a
purely distributional standpoint, yet enacted a consumption tax due to concern about the
effect of the income tax on saving. Would we not then say that the “tax system” used a
consumption tax, rather than an income tax accompanied by spending rules to reverse its
discouraging effect on saving? Nonetheless, identifying the tax system with Musgrave’s
distribution function goes a long way to help explain the tax expenditure concept. We
can readily grasp that the WSTC is allocative rather than distributional, as is accelerated
economic income would be taxed but for the aim of subsidizing a particular set of
industries.
Calling the WSTC an allocative provision, and thus not part of the “tax system”
“distribution branch.” Suppose it were established that the weapons being purchased by
the government through the WSTC were worthless, and that the reason for the enactment
was simply to benefit the companies that were getting the credits. Suppose even that this
motivation were overt (say, to rescue troubled companies), rather than involving
deception of the public by legislators and lobbyists. We presumably would not reclassify
the WSTC as being part of the tax system after all, even though its motivation would then
be distributional. After all, so reclassifying it would imply that, when interest groups
strongly influence the enactment of a preferential tax rule, the case for treating the rule as
a tax expenditure is weakened. Yet this presumably is the opposite of what advocates of
tax expenditure analysis have in mind, and no one would suggest reclassifying spending
36
as “really” taxation when interest groups use it to line their pockets. Accordingly, the
distribution branch, insofar as it is being used to define the “tax system” for tax
Examples from the tax expenditure debate in the United States help to show both
the importance and the limits of this broad approach to defining tax expenditures. An
early controversy concerned whether medical deductions were properly classified as tax
expenditures in the official accounts. William Andrews argued that the deductions
should not be so classified, on the ground that they were an appropriate adjustment in
measuring ability to pay. In form, he was making an argument about the treatment of
his argument depended, not on the dictionary meaning of income as such, but rather on
the broad distributional relevance of incurring medical expenses.64 Whether one agrees
or not with Andrews’ analysis, he was raising the right type of question, and one that
On the other hand, consider the United States rules under which interest on
municipal bonds, or those issued by state and local governments, is generally tax-exempt,
whereas other interest income is generally taxable. One cannot say which of these two
approaches to taxing interest income is correct, from the standpoint of the Musgravean
distributional branch, unless one is prepared to wade into the decades-old controversy
concerning the choice between income and consumption taxation. One can, however,
easily see that the contrast between the tax treatment of municipal bond interest and other
64
William Andrews, Personal Deductions in an Ideal Income Tax, 86 Harv. L. Rev. 309
(1972).
37
hand, so the disparity could only be defended on allocative grounds, presumably as a way
investment. So tax expenditure analysis ought to apply here somehow, but one needs
more than just the Musgravean benchmark to apply it, unless one is prepared to assert –
more controversially than the point about inconsistency seems to require – that either the
income tax or the consumption tax approach to interest income is correct from a broad
distributional standpoint.
Several further examples may help in showing how the Musgravean distributional
household characteristics in the United States income tax include (1) distinct filing
statuses for married individuals, single individuals, and unmarried heads of households
(with implications for such features as the width of lower rate brackets), (2) deductions
called personal exemptions for each qualifying member of the household, and (3) the
allowance of child tax credits. As it happens, only the last of these is officially treated as
a tax expenditure, although a view that personal exemptions ought to be on the list might
be inferred from their being disallowed under the alternative minimum tax, which applies
a lower rate to a broader base, ostensibly to prevent tax preferences from unduly reducing
one’s tax liability relative to one’s economic income. Bittker noted that all adjustments
38
for household characteristics are potentially up for grabs in defining the tax expenditure
budget.65
From a broad Musgravean perspective, two points should be clear. The first is
that household characteristics, like earnings and (under some views) medical expenses,
plainly are relevant to the operations of the distribution branch. For example, in
assessing a given individual’s material wellbeing, all household resources, not just those
that he or she personally owns, are potentially relevant. Mrs. Bill Gates is unlikely to
need public assistance even if she earns nothing. Moreover, it is certainly plausible
(whatever view one ultimately takes) that, as between two individuals or couples with the
same income, a distributional adjustment might be appropriate if one had seven children
and the other none.66 Thus, in basing tax liability on household characteristics, we are in
the realm of reasonable disagreement about the distribution function.67 Thus, all of these
rules – even more clearly than medical deductions, and unlike the WSTC – can be
differences in the means used to adjust for household characteristics (for example, the use
of a credit versus a separate filing status that affects the width of rate brackets) have no
65
Bittker, supra n. __, at 253.
66
Moreover, a household’s rules for allocating resources among its members may affect
the incidence of a tax or transfer. Suppose, for example, that we wanted to start taxing
male chauvinist husbands while aiding their wives. Any attempt to do this would have to
take account of the possibility that the chauvinists would end up with the same share of
household resources as previously. A further point supporting the relevance of household
characteristics to distribution is that the aggregate wellbeing of two or more individuals
in a household might differ from that of the same number of individuals with the same
total earnings who are not together in a household. Consider, for example, the issue of
whether cohabiting couples enjoy economies of scale that might be distributionally
relevant.
67
However, a particular adjustment for household characteristics might reasonably be
viewed as going beyond the realm of reasonable disagreement. Suppose, for example,
that married couples with children were wholly exempted from all taxes.
39
direct bearing on this conclusion. It is hard to see, for example, why the distribution
branch would be confined to using rate bracket adjustments rather than credits. Thus, it
is hard to see why their form should dictate which (if any) of them are treated as tax
expenditures.
such as the United States, with an at least partly “classical” corporate income tax system
impose tax on certain corporate earnings twice: first at the corporate level, and then again
at the shareholder-level when after-tax corporate earnings are distributed. Is this properly
attributed to the Musgravean distribution branch? That depends on how one interprets
the question.
On the one hand, the distribution branch can only distribute tax burdens among
municipal bond interest and other interest. From this perspective, one might conclude
that double corporate taxation should be classified as a negative tax expenditure or tax
penalty.
On the other hand, the double tax may reflect voters’ genuine though mistaken
belief that corporations, like flesh-and-blood individuals, actually can bear tax burdens.
It also may reflect some policymakers’ belief that double corporate taxation is a
politically convenient way of increasing the progressivity of the fiscal system. And it
may reflect administrative considerations in operating the distribution branch, if not only
40
collecting tax at the corporate level but denying relief at the shareholder level aids
moment when he supposedly first thought of tax expenditure analysis, of whether the tax
rule reflected deliberative allocative policy of a sort that typically is left to appropriations
committees, we might be inclined to say no. From these perspectives, one might
conclude that double corporate taxation should not be classified as a negative tax
Given these conflicting perspectives, the answer to how double corporate taxation
should be classified depends on the questions that one is trying to illuminate through tax
expenditure analysis. Depending on one’s purposes, one could define the Musgravean
distribution branch in various ways. For example, is one interested in what a relatively
confusion, political calculation, and administrative problems? Is one concerned about tax
rules that might, as a practical matter, alternatively involve appropriations, or all those
that have allocative effects other than on work and saving (or indeed any allocative
effects whatever)?
economic income would reach unrealized fluctuations in asset value. It also would adjust
comprehensively for price level changes, so that nominal inflationary gain would not be
treated as having made one better-off. An income tax that makes neither of these
adjustments arguably includes a tax expenditure with respect to unrealized gain, and a tax
penalty with respect to unrealized loss and the lack of an inflation adjustment. On the
other hand, these features of most actual income taxes might be rationalized as reflecting
41
progressivity of the fiscal system or to the tax burden on saving that distinguishes an
income tax from a consumption tax. Certainly it is hard to imagine Surrey’s tax
committees imposing an inflation-adjusted tax that reached unrealized gain and loss as
part of a mere reallocation of functions between the tax and appropriations committees of
the legislature (with the latter being expected to respond with suitable adjustments to
corporate taxation in requiring further specification of the purposes that tax expenditure
* * * * *
In sum, tax expenditure analysis can be rescued from the vacuity of the distinction
between taxes and spending if we reinterpret it to identify provisions in the tax laws that
Musgrave’s distributional branch would be unlikely to employ when acting on the basis
reasonable to classify the WSTC, but not the allowance of ordinary business deductions,
deductions, is reasonably debatable, while for others, such as interest income, the proper
result is reasonably debatable but certain disparities in the law seem clearly to reflect a
“spending” (i.e., allocative or narrowly distributional) motivation. One also needs greater
68
In the case of a Pigovian tax, one would have to rely instead on the broad allocative
purpose of the tax. Thus, if pollution were generally taxed but with an exception for
cement factories that had nothing to do with distinctions between their pollutants and
those emitted by other factories, we might call this a tax expenditure whether the aim was
to benefit cement factory owners (a narrow distributional goal) or to increase cement
production (an allocative goal distinct from that of the tax).
42
specification of the purposes of tax expenditure analysis in order to classify tax rules that
are clearly “wrong” distributionally but that might reflect tax administrative concerns.
means conclusive. It helps to show, moreover, that Bittker’s critique of tax expenditure
analysis was overstated, unless one confines the critique (as Bittker perhaps intended) to
rebutting overstated claims that one particular rendering of the “true” distribution system
distribution branch should do, we do not lack widely shared approaches to how we
should think about what it is trying to do. Bittker was too swayed by the American
does, it might sound too woolly-headed to be used officially. Not surprisingly, then, tax
identifying the “non-tax” rules in a tax system. This generally involves identifying the
“normal” or “normative” features of a given set of tax rules, based on the “concept of a
normative tax of the type under consideration.”69 Features that vary from the ideal but
typically are built into the “normative tax.” Classifications may also be influenced by
69
Stanley S. Surrey and Paul R. McDaniel, TAX EXPENDITURES 3 (1985).
43
class of taxpayer,”70 and whether it could as a practical matter be moved outside the tax
rather than on broadly gauged distributional concerns, simply hides the ball. For
example, it may cause the double corporate tax, realization requirement, and lack of an
“normative tax of the type under consideration,” as distinct from the tax we actually
observe (which by definition is what it is, and thus would have no “tax expenditures” if
used as its own reference standard), we must to some extent ask the very same questions
as under the Musgravean analysis. Consider the tax expenditure budgets that thirty-three
American states (plus the District of Columbia) publish with respect to their sales taxes.
Several states classify exemptions from the sales tax for inter-business transactions as tax
expenditures.71 To explain why this is wrong, one must look to the broader question of
what the distribution branch might reasonably use a sales tax to do – i.e., “produce a
cannot entirely avoid the broader questions by appealing to the contours of a particular
70
OECD Report, supra n. __, at 9.
71
See John L. Mikesell, The Normal State Sales Tax: The Vision Revealed in State Tax
Expenditure Budgets, 2003 State Tax Today 66-4.
72
Id.
44
have certain advantages where the features of the “type” have attracted sufficient
consensus. Suppose, for example, that a sales tax exempts a particular consumer good, in
addition to (as a sales tax) generally exempting returns to saving such as interest income.
consumer good and returns to saving in addition to everything in the existing sales tax
base, might nonetheless find it useful to distinguish, through the tax expenditure measure,
between the two types of exclusion. The exclusion of the consumer good might support
different inferences about the likely political process (perhaps involving interest group
politics) than the fact that the tax was a species of consumption tax rather than an income
tax. In addition, even if one believed (contrary to most observers) that consumption taxes
are generally more distortionary than income taxes, one might find it convenient to
distinguish between the economic distortions that one attributed to choosing the wrong
base and those that one attributed to taxing consumption unequally. (For example,
people who disagreed about the former claim of distortion might nonetheless agree with
the latter claim.) Accordingly, a measure that identified the exclusion of the consumer
good while ignoring the divergence from income taxation might be useful and
informative. The battle (in broader Musgravean terms) between the two rival tax bases
The narrower approach also has problems, however, which have been starkly
exposed in the American income tax setting. It is a truism that the United States income
45
tax is in fact a “hybrid income-consumption tax”73 (while also having features that are
consistent with neither). In a whole host of examples, a preference from the income tax
correct from the consumption tax perspective. It also is far from clear that the system as
question of which system is better is continually in play politically. Even when the
does – the relative merits underlie political battles about matters ranging from tax-free
savings accounts to depreciation and other cost recovery rules to corporate integration.
Against this background, Surrey successfully argued early on that the official tax
expenditure estimates he was seeking should be conducted solely from an income tax
perspective. He presented this argument in ostensibly neutral terms, and wholly without
Each tax has its own appropriate structure and each has its advantages and
disadvantages. But the scope of each such tax in its actual application must
be tested by its concepts, which concepts led to its choice in the first instance.
The structure of a normative income tax is not to be tested by the values or
concepts used by those who prefer that a consumption tax be chosen instead,
and vice versa. A tax expenditure budget for an income tax, to be useful in
seeing what objectives that tax has been asked to carry in addition to taxing
net income, is to be framed by using a normative definition of “income.”74
73
See, e.g., Henry J. Aaron, Harvey Galper, and Joseph A. Pechman, (eds.), UNEASY
COMPROMISE: PROBLEMS OF A HYBRID INCOME-CONSUMPTION TAX (1988).
74
Surrey, PATHWAYS TO TAX REFORM, supra n. __, at 21.
46
In other words, we actually have an income tax – that is what we have enacted – so
income tax principles must be used in the tax expenditure budget whether they are right
Formally speaking, the argument here is identical to the one I outlined above
concerning the sales tax under which the exemption of a particular consumption good,
but not the exemption of returns to saving, might reasonably be treated as a tax
expenditure. The surrounding circumstances are different, however. Purely from the
standpoint of internal logic, it is not entirely clear that what we have is best classified as
an income tax. Moreover, any inference that departures from it must reflect allocative or
explained as reflecting a consumption tax approach that is known to have wide appeal.
Even if one nonetheless accepts Surrey’s argument, the brute fact is that, in a
crucial sense, it did not prevail. His success with regard to official reporting of tax
expenditures was not accompanied by the kind of widespread acceptance of the analysis
that he sought. This is no surprise. Consumption tax advocates, who frequently have
been hostile to tax expenditure analysis75 even though they share Surrey’s goal of taxing
alternative kinds of consumption and saving more neutrally, could plausibly interpret his
arguments as amounting to the following: “Please forget for the moment that I am an
income tax advocate. I want you to accept as normative the type of tax that we officially
have, which I know you do not like, because we have it, and even though in fact we do
not have it. I will of course do the same for you, either in comparatively trivial settings
75
See, e.g., Institute for Research on the Economics of Taxation, IRET Congressional
Advisory No. 95, “What’s Wrong With the Tax Expenditure List/How to Fix It (January
6, 2000); Bartlett, supra n. __; Norman B. Ture, “Unreleased Testimony on Tax
Expenditures,” Tax Notes, December 21, 1981, page [1535].
47
under present law (such as state sales taxes) or if you win outright at the federal level,
which we both know is highly unlikely. Please do not be swayed by the fact that, by
accepting the income tax as normative for these purposes, you may help me to win our
ongoing disputes concerning whether the existing system should be made more income
This is hardly an offer one cannot refuse. The suspicion that tax expenditure
analysis served unacknowledged and controversial political agendas may further have
been heightened by several others of its key features, as practiced in the United States
income tax setting. One was its one-way ratchet in measuring tax expenditures but not
tax penalties. Another was its treating double corporate taxation as the norm, on the
ground that it was a merely structural feature of the U.S. system, when this was not even
consistent with neutral income taxation. Here was another feature of the existing tax
system that Surrey, among others who shared his backing for greater progressivity, just
Haig-Simons77 income tax were also built into the “normal income tax structure” – for
Surrey had even suggested that such items might be added to the official list once the
In the meantime, Surrey had left himself open to such critiques as the following:
“[H]aving abandoned the purity of Haig-Simons, [he] is adrift in a sea of value judgments
76
[Cite for Surrey’s support for the double corporate tax.]
77
See Henry C. Simons, PERSONAL INCOME TAXATION 50 (1938) (defining “income” as
one’s consumption plus change in net worth during the relevant accounting period).
78
Surrey, PATHWAYS TO TAX REFORM, supra n. __, at 18.
48
and his is no better than any other expert’s. Thus, it is presumptuous for him to label his
definition (i.e., the Treasury’s) as the one correct definition, any deviations from which
will be labeled tax expenditures.”79 Such critiques could not be fully rebutted by arguing
that, say, the realization requirement really is “structural” to the income tax in the sense
of serving administrative aims and not being readily transmutable into a direct spending
program. Nor would it have helped, even if Americans were generally more interested in
what other countries do, to point out that similar approaches to tax expenditure analysis
are followed elsewhere without arousing similar controversy. The setting was simply too
politically loaded for a creature as unlovely (and unloved) as the “normal income tax
structure” to win requisite acceptance as the one and only baseline for measuring
In short, the problems with using a “normative tax of the type under
consideration” came home to roost because it combined being innately too controversial
with having too many compromises built into the particular implementation that was
being used. Not sufficiently respected as an objective descriptive tool, it could not
too far from center stage, on the insufficiently accepted ground that they have been
wholly resolved for purposes of the particular tax at issue, U.S. proponents of tax
expenditure analysis unduly limited its capacity both to provide interesting information
and to attract the acceptance that it would have needed to work as hoped.
79
Bartlett, supra n. __, at [415?]. [Cite also Bittker & the Ture quote in Bartlett.]]
49
Tax expenditure analysis is too inherently flexible a tool to have only one or a
single set of narrowly defined purposes. Restating actual Tax Rule A as hypothetical Tax
Rule B plus Spending Rule C need only be interesting and informative in order to be
uses of the tax expenditure concept (on the ground that some other term should be used if
cases where it seems that tax rules are deliberately being used to have allocative or
narrowly distributional effects of a kind that often are done through direct spending. Or
we might interpret the juxtaposition more objectively, as identifying such effects even
where the tax rule might be strongly influenced by administrative considerations. The
proper choice of emphasis may depend on where one’s interests lie, as between budget
From a purely budgetary standpoint, like that underlying Surrey’s creation myth
and the actual German history of tax expenditure analysis, one might mainly be interested
alternatively be done through “spending.” For example, if one is trying to reduce budget
deficits, one might want to place narrowly crafted special tax benefits on a par with
appropriations, while not being greatly interested in the gap between realized income and
50
economic income. Even from such a perspective, however, one might want to be
apprised of the differences between provisions such as the WSTC that clearly have no
rationalized, whether because they reflect a consumption tax approach or otherwise (like
distributional issues.
If we add a tax policy perspective to our underlying set of interests, then all
arguable or clear departures from the “ideal” Musgravean system are of interest, whether
or not they could be converted into direct spending. Thus, one might want to compare
the existing U.S. “income” tax to a pure Haig-Simons income tax, and also to a broad-
based consumption tax.80 Under a tax policy perspective, moreover – seemingly the
dominant one in American uses of tax expenditure analysis – the aim of offering more
information, rather than less, would be advanced by making reasonable distinctions in the
accounts. For example, even if one includes medical deductions on the ground that they
are possibly or in some views a tax expenditure, one might want (no less than under the
budgetary perspective) to distinguish between them and provisions that (after the fashion
of the WSTC) are unambiguously “spending.” One also might want to distinguish clearly
adjustment), as well as those that are arguably structural (such as the classical corporate
tax), from those that are more like the WSTC. So the main difference between the
80
One could go deeper still. For example, if one thinks that we tax income or
consumption only because they are signals of some underlying attribute, such as “ability”
or “endowment,” that cannot be directly observed,80 then one might be interested in a
revenue estimate for the cost of failing to tax ability that is unexercised by reason of the
tax disincentive to work.
51
budgetary and tax policy perspectives is simply that, under the latter, somewhat more
The argument I am making here for more varied and informative tax expenditure
analysis would approach banality (who wants to argue for less information?) if the United
States history of the analysis were different. Indeed, various countries provide alternative
“spending” is relatively ambiguous.81 In the U.S. income tax setting, however, the
analysis has been sufficiently waylaid by the aftershocks from Surrey’s initial decision to
present it as a purely budgetary tool, rather than as a tool in both budgetary and tax
policy, that such a proposed revision verges on the radical. Proponents of Surrey-style
income tax reform may be reluctant to surrender the preeminence of their view within the
official reports, however doubtful the advantages that they gain therefrom. And some
opponents of the Surrey view may be too hostile to the analysis in all forms – whether out
of reflexive distaste based on its history, or spending illusion, or because they actually
indeed expansion. Perhaps there are grounds for optimism, however, in light of the
makes tax expenditure analysis potentially so valuable. Its mode of description may help
to counter the undue political advantage, in some settings, of doing allocative policy or
narrow distributional policy through the tax code. Surrey’s story about sitting in the 1967
Ways and Means hearing, the details of which are entirely believable even if it was not
81
See OECD Report at 10.
52
really the moment when he conceived of tax expenditure analysis, describes only a trivial
(because easily corrected) manifestation of the problem, in the form of political actors not
The deeper problem, especially after thirty years of tax expenditure analysis, is
not that policymakers fail to recognize the practical interchangeability of tax rules and
spending rules. Nor is it that they fail to give tax rules significant scrutiny, even though
once enacted tax rules generally do not have to be reenacted annually in the manner of
discretionary spending. It is well documented that tax expenditures often get extensive
ongoing scrutiny.82 The main problem that tax expenditure analysis may help to address
lies, rather, in the speciousness of the otherwise prevailing fiscal language distinction
between “taxes” and “spending.” Supposed tax cuts are perceptually different than
supposed spending,83 even if the two are substantively identical, for reasons that include
but are not limited to anti-government ideology as applied through the lens of spending
illusion. A deeper cause is the heuristic bias that decision researchers call the endowment
distinction between money that is never paid in to the Treasury and money that is first
82
See, e.g., John F. Witte, THE POLITICS AND DEVELOPMENT OF THE FEDERAL INCOME
TAX 334 (1985).
83
See, e.g., Martin A. Sullivan, Tax Expenditure Budgets, Now More Than Ever, 86 Tax
Notes 1187, 1188 (“During the 1990s, President Clinton has perfected a political tactic
that has done wonders for the Democratic party, but at the same time it has complicated
the tax code. Tax-and-spend liberalism has been replaced with "tax expend" liberalism.
Rather than directly funding new government programs, the president knows that
politically it is far easier to implement social programs through the tax code.”).
84
Shaviro, WHEN RULES CHANGE, supra n. __, at 87.
53
The endowment effect might not work politically in favor of using the tax code if
the beneficiaries from tax breaks were as subject to it as the general public. They might
then think that they were getting less from a tax break, just as the public gets the
impression that it is giving less. Heuristic biases, however, work more powerfully when
one’s attention to an issue is relatively casual. The direct beneficiaries of tax benefits can
therefore readily grasp that an extra dollar in their pockets is a dollar either way. The
target, when tax rules are used to exploit the endowment effect, is the general public,
which pays far less attention but is always a looming presence on the scene, capable of
inducing rejection of transfers that break through the haze to attract hostile scrutiny.85
the big contribution that tax expenditure analysis can potentially make. And while this
falls in the “weapon” category of fiscal language uses, since it aims to affect political
of a sort that can withstand scrutiny. Not only is it intellectually defensible unless one
sees no difference between the WSTC and measuring income for broad distributional
purposes, but it aims to improve understanding, rather than seeking to dictate a given
eliminated, only that we should think about them differently than the usual “tax” and
“spending” categories imply. Thus, while those who want more use of the tax code,
rather than less, to camouflage policies that otherwise might involve direct spending, may
85
See Daniel Shaviro, Beyond Public Choice and Public Interest: A Study of the
Legislative Process as Illustrated by Tax Legislation in the 1980s, 139 U. Pa. L.Rev. 1,
40-42 (1990).
54
have reason to oppose tax expenditure analysis, the reason is not one that they can
forthrightly acknowledge.
advancing this purpose by making tax expenditure analysis more broadly acceptable (as
well as more genuinely informative) is well worth the sacrifice of its currently tighter link
to a Haig-Simons tax reform agenda. Even if we assume that such an agenda is the
proper one in tax policy, its advancement in this manner has simply proved too question-
begging. It seems clear, in retrospect, that Surrey over-reached, but retrenchment is still
possible.
55
document in its own right, and to flesh out this paper’s suggested approach to tax
expenditure analysis. While I try to avoid excessive detail in the critique, it delves a bit
more intimately into details of the United States income tax than some readers may want
to go. Any such readers may prefer to skip ahead to subsection C, which offers a
The Treasury Discussion starts by accepting the basic idea behind tax expenditure
analysis, that some existing tax provisions seem to reflect a very different enterprise than
equitably raising revenue, and by noting the arbitrariness of the “normative income tax”
baseline that official estimates have always used. It then proposes three main changes:
redefining the baseline income tax to come closer to the Haig-Simons ideal, identifying
negative tax expenditures as well as positive ones, and preparing estimates under a
consumption tax baseline as well as an income tax baseline. A fourth change, evident in
the analysis thought not listed at the front, is greater appreciation of ambiguity, through
the creation of categories for items that are “possibly” or “probably not” tax expenditures
The use of a consumption tax baseline may ensure a hostile reception of the
Treasury Discussion in some circles. The problem is not just that income tax advocates
are reluctant to give ground, but that the Bush Administration’s general tax policy stance
has raised partisan hackles. Beyond advocating massive tax cuts that can only be
56
shift to consumption taxation.86 This shift, moreover, is viewed as of a piece with its
other tax policy aims, notwithstanding that in principle (even if not in any Bush
tax.87 There will be a tendency, therefore, to apply principles of guilt by association and
dismiss the Treasury Discussion as just another exercise in radically reshaping tax policy,
to be opposed by anyone who does not share the Administration’s political goals.
Applying that view here would be a mistake, however. As I argued in section III,
income tax baseline in tax expenditure analysis, since both systems are plausible and both
are reflected to some extent in the system we have. The question is whether we should
have more information that is reasonably meaningful rather than less, and whether tax
expenditure analysis can be rescued from its current status as the perceived tool of big
government liberals. Moreover, serious work by the Treasury’s professional staff should
not be rejected simply out of dislike for agendas emanating from the top, even if we can
see that those agendas provide an additional motivation for wanting tax expenditure
analysis to be diversified. Anyone who doubts the fair-mindedness of the exercise should
keep in mind that the Treasury Discussion suggests a comprehensive income tax baseline
that is far closer than the existing one to Haig-Simons income, in addition to outlining a
86
See, e.g., Robert J. Wells, Wage Slaves Beware – You May Soon Be Financing the
Federal Government, 98 Tax Notes 1029 (February 17, 2003).
87
See Daniel N. Shaviro, Replacing the Income Tax With a Progressive Consumption
Tax (2003).
57
The Treasury Discussion groups tax expenditures under existing practice, along
with those that might newly be estimated, into a number of different categories. In the
next two subsections, I first review the Treasury Discussion’s categories, with comments
on some of the more significant items, and then offer suggestions for modifying the
Treasury Discussion notes that a number of items that are treated as tax expenditures
Representative examples include the exclusion for municipal bond interest, the deferral of
income from controlled foreign corporations, and the difference between accelerated
depreciation and an assumed measure of economic depreciation.88 The list, being limited
interest, however, is the fact that the items in this category account for less than half of
the estimated revenue loss for 2004 for all of the items that are currently listed as tax
expenditures.
But With Some Qualifications – The more noteworthy items in this category include the
following:
such as income taxes – The Treasury Discussion notes that, in a comprehensive income
88
The Treasury Discussion notes that some of these items, such as accelerated
depreciation, are measured differently under its comprehensive income tax baseline than
under the existing “normal income tax” baseline. [Cite.]
58
measure, the imputed value of state and local government benefits would be included in
income, but state and local taxes would be deducted. Calling allowance of the deductions
a tax expenditure therefore rests on the view that disallowing them would be a good
enough proxy for directly including the value of the benefits. As we will see further
below, this proxy disallowance issue arises repeatedly and ought perhaps to be handled
somewhat differently.
under such a system, “all gains would be taxed as accrued, so there would be no deferred
unrealized gains on assets held at death.”89 This is true enough, but the response of
local income taxes, which would be properly deductible from a comprehensive income
tax standpoint if there were absolutely no correlation between taxes paid and benefits
received, the point here is simply that the gain should have been taxed even sooner. If
the Treasury Discussion distinguished structural from other departures from the
comprehensive income tax baseline, as I suggest below, this item could be moved to the
clear tax expenditure category with a cross-reference to the “structural” section, noting
that it would disappear if current accrual taxation for all assets were adopted.
circumstances, with respect to one’s dependent children is possibly a tax expenditure, the
Treasury Discussion states, “[t]o the extent that the personal and dependent care
exemptions and the standard deduction [which are not listed as tax expenditures] properly
89
Treasury Discussion at 131.
59
remove from taxable income all expenditures that do not yield consumption value.”90
The status of the child credits is not clearcut, however, because, under an opposing view,
they are “appropriate modifications that account for different taxpaying capacity,”
The analysis here strikes me as somewhat off the mark. Are the costs of raising a
child, which personal exemptions and the standard deduction may very minimally help to
defray, “expenditures that do not yield consumption value”? I can personally testify to
and I am strongly under the impression that they share this preference.91 It also is pure
formalism to distinguish between the above items and child credits on this ground, when
none of the items bears any particular relationship to the expenditures supposedly not
constituting consumption.
The supposedly countervailing point, that child tax credits might be appropriate
issue in fact lies. While plainly it is reasonable to adjust for household status in some
way, not everyone would agree with this adjustment – or, indeed, with any of the others
in the current U.S. income tax. My suggestion, as I discuss further below, is to do either
wage subsidy under certain circumstances that is administered through the income tax via
90
Id. at 132.
91
Perhaps the underlying view is that “necessities” do not constitute consumption. This
plainly is mistaken, if we think of consumption as an input to measuring wellbeing,
unless we think that people who fail to get the necessities are no worse off than those
who do get them. The real point, presumably, is that we want to assure that people can
afford necessities before we start taxing them. This is better viewed as a rate bracket
point than a point about what constitutes consumption.
60
a refundable credit.92 The Treasury Discussion mystifyingly groups this with the child
credit, perhaps on the ground that it is household-related since taxpayers with up to two
dependent children get significantly larger wage subsidies. I would argue that the earned
income tax credit does not belong in the tax expenditure tables to begin with, since it is
simply a part of the overall rate structure, although its household-related aspects could be
from those that are “possibly” tax expenditures on the ground that their status is
personally views medical expenses and charitable contributions (and I find arguments
that the latter are not part of consumption to be extremely weak),93 their contested status
supports keeping them out of the “clear” category, on the ground that it is unwise to try to
dictate here.
Tax – The two main items here are the deductibility of mortgage interest and of state and
local property taxes on owner-occupied homes. The Treasury Discussion notes that,
under a comprehensive income tax, these items would be deducted and the implicit rental
92
Note how EITC relates to marital status and # of children.
93
The Treasury Discussion, supra at 132, notes that “charitable contributions could
represent a transfer of purchasing power from the giver to the receiver,” thus reducing the
giver’s net worth. This misses the point that a voluntary giver evidently preferred them
to outlays (also reducing net worth) that everyone agrees are consumption. This is not to
say that charitable deductions are inappropriate, but only that they cannot reasonably be
viewed as leaving the giver worse-off. Medical expenses are more ambiguous because,
even if they improve wellbeing given one’s underlying medical condition, they may be
evidence of worse health status that is relevant to wellbeing. Cite Kaplow on the
insurance issues.
61
income earned on the housing included.94 This is identical to the point made with respect
to other state and local nonbusiness taxes, a point that the Treasury Discussion does not
directly acknowledge. It states, however, that they are “extremely crude proxies for the
implicit rental income earned on owner-occupied housing. The interest deduction proxy,
for example, ignores implicit rental income earned on a house that is unencumbered by
any mortgage.”95
This fails to establish that the proxies for implicit rental income are cruder overall
than taxes as a proxy for state and local government benefits. The claim that the proxy
here is too crude may relate to the Treasury’s revenue estimates, which estimate the 2004
2004 revenue loss of only $20.5 billion for excluding the implicit rent. With all due
respect to the Treasury’s revenue estimators, however, these estimates are almost certain
arbitrages on their homes? Suppose we even restrict the comparison to that between the
$20.5 billion estimate for excluding implicit rent and the estimated $68.4 billion revenue
impact for home mortgage interest deductions. Even if all homes are 100 percent
leveraged (suggesting that the issue of homes unencumbered by debt is too trivial to
worry about), this implies that people are paying interest that is more than 330 percent of
the implicit return they are earning on the asset. This could only happen in the very short
run, if home prices suddenly plummeted and the market had not yet adjusted, but even in
that scenario it would leave a wake of distress (such as defaults and foreclosures) beyond
94
Treasury Discussion at 133.
95
Id.
62
anything that has recently been observed in the U.S. housing and credit markets.
Accordingly, both these estimates and the Treasury’s conclusion that the mortgage
interest and real property tax deductions are probably not tax expenditures need
reconsideration. As I further discuss below, these items could be grouped with a broader
category of what I call “proxy” disallowance rules in the current tax system.
of these items is relatively speculative and preliminary, but the main items mentioned
earnings that have been taxed at the corporate level.”96 The estimated revenue gain from
the negative expenditure, based on assuming the continuation of pre-2003 law, was $25.3
notes that various loss disallowance rules in the U.S. income tax, such as those for
passive losses (pertaining to suspected tax shelters) and net capital losses, depart from
accurate Haig-Simons measurement. It fails to make the point, however, that the loss
requirement. They therefore raise proxy disallowance issues and should not be treated as
96
Treasury Discussion at 139.
63
approach of income taxation, under which interest expense from dissaving should
this can be viewed as a negative tax expenditure to the extent that it would increase asset
basis and depreciation deductions while reducing interest inclusions, albeit a positive tax
the Treasury Discussion notes, various education expenses should be amortized and
allowed over the period when they are generating income, rather than being disallowed
(on the premise that they are entirely consumption expenses) as they are, for the most
part, under present U.S. law. This is correct as far as it goes, although one would have to
try to distinguish, for purposes of the estimates, between the investment and consumption
preferential. When people forego current work in order to increase their future earnings
potential through education, one could argue for including the imputed return in current
income and then amortizing it along with educational outlays. To be sure, treating this as
a positive tax expenditure would raise the question of just how broadly tax expenditure
analysis should apply to foregone earnings. A comprehensive income tax baseline does
little to illuminate this question, since it offers no clear basis for specifying when we
64
However, the absence of a positive tax expenditure for this item could raise proxy
taxation issues about the negative tax expenditure with regard to educational outlays.
only new items for which the Treasury Discussion includes estimates are homeowners’
imputed rent and corporate double taxation. Among the various items that the Discussion
says might be added in the future are the deferral of unrealized gain and loss, the
exclusion of gifts received, and the allowance of foreign tax credits rather than just
deductions. The rationale for the gift and foreign tax items is that the suggested baseline
treatment might be more consistent with a comprehensive income tax approach, even if
one were to conclude that the Musgravean distribution branch should or might follow the
existing rules.
Discussion notes, full use of this baseline would add to the need for a set of negative tax
expenditure estimates, since all provisions in current law that follow an income tax
leaving various issues unresolved, it does not discuss how the nuanced and varied
categories described above could be integrated into the existing structure of official tax
expenditure reporting. Among the aspects of that structure that make such integration
97
See Daniel Shaviro, Endowment and Inequality, in Joseph J. Thorndike and Dennis J.
Ventry (eds.), TAX JUSTICE: THE ONGOING DEBATE 135 (2002).
65
more difficult are the structure’s being (1) organized by budgetary function, such as
“Transportation” and “Community and regional development,” and (2) extremely lengthy
due to its including a host of relatively small items (such as numerous separate listings
for different uses of tax-exempt municipal bonds). This structure is better suited for the
use of tax expenditure analysis as a tool of budgetary policy than of tax policy.
One way the Treasury could proceed in the future would be to separate out the
budgetary and tax policy functions. For budgetary reporting purposes, the analysis could
largely keep its present form, including the exclusion of structural and negative tax
expenditure items. Change could be limited to clearly identifying (perhaps through the
use of distinct lists) the items that (1) are not tax expenditures from a consumption tax
baseline, and (2) arguably are not tax expenditures even from an income tax baseline.98
For tax policy reporting purposes, items such as all municipal bonds could be
aggregated, and those below a threshold size (say, $1 billion per year) kept off the list.
The remaining items could then be organized into a manageable set of categories,
including in all cases both positive and negative adjustments. While there is no single
right way of organizing the tax policy list, one possibility, which I offer here in the hope
include imputed rent, the double corporate tax, the realization requirement, and the lack
of inflation adjustments.
98
Household adjustments could either be eliminated from the list or else expanded (with
clear identification of their contestable character) to include additional items, such as
personal exemptions.
66
Items listed here might include medical deductions and exclusions for employer-provided
health insurance, charitable deductions, and the exclusion of gifts received. If household
adjustments were included here, rather than being left out altogether, additional items
adjustments could also be compared to such alternative baselines as (1) a pure separate
returns system for couples and (2) a pure income-splitting system for couples.
D. Proxy departures from the baseline that may offset other departures –
Items listed here might include state and local taxes, home mortgage interest deductions,
C. Proxy departures from the baseline that may offset other departures
99
A structural departures list like that for the comprehensive income tax might not be
needed here. Consumption taxes typically do not use realization rules or apply double
corporate taxation. Moreover, they do not present the problem of inflation adjustment if
they use expensing for capital outlays and do not require measurement of interest income
and expense.
67
V. CONCLUSION
Tax expenditure analysis is too potentially useful to be rejected on the ground that
no single conceptual model for a tax system has won universal acceptance. Nor does the
ultimate emptiness of the distinction between “taxes” and “spending,” which the analysis
attempts to fix, prevent it from improving information and reducing the tendentiousness
of our fiscal language. At least in the United States, efforts to make tax expenditure
analysis do too much – by serving as an instrument for one particular vision of tax reform
– have unnecessarily undermined its acceptance. By adapting it to use more flexible and
varied measures that clarify its relationship to underlying distributional aims and that take
account of reasonable disagreements as to those aims, we can hope to improve both its
informational content and its general background influence on budgetary and tax policy
debate.
important opportunity to advance actual U.S. practice in this area. Unfortunately, the
Treasury analysis is at risk of being engulfed in the bitter partisan wrangling that
increasingly surrounds the American tax policy process. Income tax advocates may be
too suspicious of the Bush Administration’s underlying motivations to give the analysis
the Treasury Discussion helps point the way can nonetheless win broader acceptance, and
even be permanently incorporated into official reporting practice, then at least there will
be one small bright spot as ever-rising deficits and a radically unstable tax law lead