Is the Tax Expenditure Concept Still Relevant? - … the Tax Expenditure Concept Still Relevant?Published in: National Tax Journal · 2003Authors: Leonard E BurmanAffiliation: Georgetown

Is the Tax Expenditure Concept Still Relevant?

613

National Tax JournalVol. LVI, No. 3September 2003

Abstract - The term “tax expenditure” refers to departures fromthe normal tax structure designed to favor a particular industry,activity, or class of persons. Most budget experts view the tax ex-penditure budget as a useful tool in managing the size and scope ofthe federal government, but a growing contingent of conservativecritics has raised questions about the concept. The paper examinestax expenditure measurement issues, the debate about their relevanceand new measures tentatively explored by the BushAdministration’s Budget, the way tax expenditure estimates areused in the U.S., and how they might be made more useful.

INTRODUCTION

The term “tax expenditure” is attributed to Stanley S. Sur-rey who, as Assistant Secretary of the U.S. Treasury for

Tax Policy, instructed his staff to compile a list of preferencesand concessions in the income tax that had the nature of ex-penditure programs. His goal was straightforward: to drawattention to these items in hopes of building momentum fortax reform, which would redirect the tax system toward itscore function of raising revenues.

Stanley Surrey and coauthor, Paul R. McDaniel, definedthe concept thus in their 1985 treatise on the subject:

The tax expenditure concept posits that an income tax is com-posed of two distinct elements. The first element consists ofstructural provisions necessary to implement a normal incometax, such as the definition of net income, the specification ofaccounting rules, the determination of the entities subject totax, the determination of the rate schedule and exemption lev-els, and the application of the tax to international transactions.The second element consists of the special preferences foundin every income tax. These provisions, often called tax incen-tives or tax subsidies, are departures from the normal tax struc-ture and are designed to favor a particular industry, activity, orclass or persons. They take many forms, such as permanentexclusions from income, deductions, deferrals of tax liabilities,credits against tax, or special rates. Whatever their form, thesedepartures from the normative tax structure represent govern-ment spending for favored activities or groups, effected throughthe tax system rather than through direct grants, loans, or otherforms of government assistance (p. 3).

Is the Tax ExpenditureConcept Still Relevant?

Leonard E. BurmanThe Urban Institute,Washington, DC 20037

Georgetown University,Washington, DC 20007

NATIONAL TAX JOURNAL

614

Seven years after Treasury first pub-lished a list of tax expenditures in 1967,the Congressional Budget Act of 1974 re-quired the Administration to publish a listof tax expenditures as part of its annualbudget submission. The concept alsogained widespread acceptance outside ofthe United States. Both Canada and theUnited Kingdom started publishing listsof tax expenditures in the late 1970s, andmany other OECD countries had eitheradopted formal tax expenditure budgetsor conducted preliminary studies by 1985(Surrey and McDaniel, 1985).

The Budget includes tax expenditures,defined as deviations from the “normal”individual and corporate income taxbases, along with its estimates of directexpenditures. Until 2002, it also includeda list of tax expenditures against a trans-fer tax (estate and gift taxes) baseline, butthose items were excluded from the FY2003 budget because “ . . . there is no gen-erally accepted normal baseline for trans-fer taxes and . . . [the tax was] . . . repealedunder the Economic Growth and Tax Re-lief Reconciliation Act of 2001(EGTRRA)”1 (U.S. Office of Managementand Budget, 2002, p. 95). In principal, taxexpenditures could also be defined withrespect to other taxes, such as excise taxes,but it has not been done on a systematicbasis (Davie, 1994).

Most budget experts view the tax ex-penditure budget as a useful tool in man-aging the size and scope of the federalbudget, but a growing contingent of con-servative critics has raised questions aboutthe concept. That critique was elevated tonew heights in 2002 when PresidentBush’s budget introduced the tax expen-diture section with a warning that “. . .the Administration believes the meaning-fulness of tax expenditure estimates is un-

certain . . .” and promised a new moremeaningful presentation in future years(U.S. Office of Management and Budget,2002, p. 95).

The paper discusses the measurementof tax expenditures as implemented in theUnited States, the debate about the rel-evance of the tax expenditure concept ashighlighted by the Bush Administration,and the way tax expenditure estimates areused in the United States.

MEASURING TAX EXPENDITURES

Every year, the Treasury Departmentcompiles a list of income tax expenditures,which is included in the Administration’sBudget released in early February. TheJoint Committee on Taxation prepares asimilar list for Congress. The purpose ofthe tax expenditure estimates is to subjectspending programs administered throughthe tax code to the same Congressionalscrutiny and control as direct expendi-tures.

It is not clear whether this process hasbeen effective, because it is hard to tellwhat the level of tax expenditures wouldhave been absent annual revelation. Sur-rey and McDaniel (1985) calculated thattax expenditures grew relative to GDP andmuch faster than direct outlays in the first15 years that tax expenditure estimateswere produced, between 1967 and 1982.Toder (1998) reported that tax expendi-tures grew only slightly faster than GDPfrom 1980 to 1999.

As the authors of those studies ac-knowledge, their measures are rough ba-rometers at best of the cost of tax expen-ditures. A tax expenditure estimate reflectsthe amount by which tax liability is re-duced due to a particular tax provision,but it does not measure the revenue that

1 Note that repeal is not effective until 2010, and only for one year. Also, the Administration produces taxexpenditure estimates for other expiring provisions, such as work opportunity tax credits, so the repeal argu-ment for excluding the estate tax is a bit disingenuous. The more likely reason is that the Administration doesnot believe that the estate tax should exist. The Joint Committee on Taxation has never published estate taxexpenditures.


615

would be gained by eliminating that pro-vision for two reasons. First, the estimatedoes not include any behavioral response,which would be incorporated in a revenueestimate. Thus, for example, if the Hopescholarship tax credit—a tax credit for thefirst two years of post–secondary educa-tion—were eliminated, many taxpayerswho would have used that credit wouldinstead opt for the slightly less generouslifetime learning tax credit or other taxsubsidies aimed at higher education. Inconsequence, the revenue savings to theTreasury would be only a fraction of theamount of Hope tax credits allowed.2

In addition, government estimates oftax expenditures do not account for lossesin tax revenues from other revenuesources, most notably payroll taxes, whichare used to finance retiree health care andannuities in the U.S. For example, Sheilsand Hogan (1999) estimate that the lossin payroll tax revenues from the tax ex-clusion for employer contributions tohealth insurance—the largest tax expen-diture in 2004—is more than half of theincome tax revenue loss alone.3 Thus, thetotal income plus payroll tax expenditurecould be at least $150 billion comparedwith the estimate for the income taxexpenditure alone of $100 billion (SeeTable 1).

Another problem in attempting togauge the importance of tax expendituresover time is that they may not simply be

summed to come up with a tally of thecost of the web of tax expenditures as awhole because there are potentially sig-nificant interactions among the differenttax provisions. For example, the cost of thededuction for home mortgage interest—the second largest tax expenditure onTable 1—and the deduction for state andlocal taxes—the sixth largest—is less thanthe sum of the two estimates. If either taxpreference were eliminated, many fewertaxpayers would itemize deductions,making the value of the second tax pref-erence significantly smaller.4

The bottom line is that the sum of taxexpenditures provides very little informa-tion about the cost of tax expenditures asa whole to the federal government. How-ever, comparing that sum over time mayprovide some information about trends intax expenditures, especially if the compo-sition of tax expenditures does not changemarkedly. Since 1980, however, there havebeen eight major tax changes, so there islikely to be considerable noise in the trendestimates.5 In consequence, we know verylittle about how tax expenditures havechanged over time as a share of the fed-eral budget or in their relative importancecompared with direct spending programs.

The Century Foundation WorkingGroup on Tax Expenditures, a bipartisangroup convened to evaluate current taxexpenditures and make recommendationsabout how to improve monitoring and

2 The same problem could occur in the measurement of the effect of spending programs, but is limited by thefact that almost all outlay programs in the U.S. are subject to binding spending limits. In contrast, most taxexpenditures are unlimited—available to all eligible individuals. Indeed, the more likely avenue for shiftingcosts from eliminating direct spending programs is that tax expenditures might increase. For example, if thegovernment were to eliminate the guaranteed student loan program, which implicitly subsidizes interestrates for higher education loans, tax deductions for student loan interest would increase.

3 Pensions would be the largest tax expenditure in 2004 if the tax exclusion on employer and employee contri-butions to pension plans were grouped together as they are in the Joint Committee on Taxation’s analysis.The Budget presentation separates these two items.

4 Taxpayers can take advantage of itemized deductions only to the extent that the total of all those deductionsexceeds a standard deduction, which varies by filing status. About two–thirds of tax filers in 2001 could notbenefit from the deduction for charity, for example, because their deductible expenses were less than thestandard deduction (Balkovic, 2002–2003).

5 Significant changes to U.S. tax laws were enacted in 1981, 1982, 1984, 1986, 1990, 1993, 1997, and 2001. Mostanalysts expect a sizable tax cut to be enacted this year, so the trend of frequent major changes is unlikely toabate any time soon.


616

123456789

101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657

Exclusion of employer contributions for medical insurance premiums and medical careDeductibility of mortgage interest on owner–occupied homesNet exclusion of pension contributions and earnings: Employer plansNet exclusion of pension contributions and earnings: 401(k) plansCapital gains (except agriculture, timber, iron ore, and coal) (normal tax method)Deductibility of nonbusiness state and local taxes other than on owner–occupied homesDeductibility of charitable contributions, other than education and healthAccelerated depreciation of machinery and equipment (normal tax method)Step–up basis of capital gains at deathExclusion of interest on public purpose State and local bondsNet exclusion of pension contributions and earnings: Individual Retirement AccountsDeductibility of State and local property tax on owner–occupied homesChild creditCapital gains exclusion on home salesExclusion of interest on life insurance savingsSocial Security benefits for retired workersDeferral of income from controlled foreign corporations (normal tax method)Net exclusion of pension contributions and earnings: Keough PlansExclusion of workers’ compensation benefitsDeductibility of medical expensesWorkers’ compensation insurance premiumsGraduated corporation income tax rate (normal tax method)Extraterritorial income exclusionEarned income tax creditCredit for increasing research activitiesException from passive loss rules for $25,000 of rental lossDeductibility of charitable contributions (health)Deductibility of charitable contributions (education)Social Security benefits for dependents and survivorsSelf–employed medical insurance premiumsCredit for low–income housing investmentsSocial Security benefits for disabledExclusion of veterans death benefits and disability compensationCredit for child and dependent care expensesParental personal exemption for students age 19 or overLifetime Learning tax creditHOPE tax creditDeduction for higher education expensesExpensing of research and experimentation expenditures (normal tax method)Exclusion of income earned abroad by U.S. citizensExclusion of reimbursed employee parking expensesExclusion of benefits and allowances to armed forces personnelTax credit for corporations receiving income from doing business in U.S. possessionsDeferred taxes for financial firms on certain income earned overseasAdditional deduction for the elderlyLow and moderate income savers creditPremiums on group term life insuranceSpecial ESOP rulesInventory property sales source rules exceptionExclusion of interest on hospital construction bondsExclusion of scholarship and fellowship income (normal tax method)Empowerment zones, Enterprise communities, and Renewal communitiesExemption of credit union incomeCapital gains treatment of certain incomeDeferral of income from post 1987 installment salesAccelerated depreciation on rental housing (normal tax method)Exclusion of interest on owner–occupied mortgage subsidy bonds

120,16068,44067,87055,29053,93050,91033,99031,11028,50027,31023,13022,16021,31020,86020,74018,9307,9007,6166,4606,3406,1905,7005,5105,0904,9904,9204,5804,3504,1403,6903,6403,5703,4003,2303,2302,9802,8802,8802,7602,6802,2902,2402,2402,1302,0501,8601,8301,7901,6201,4401,2601,1701,1601,1201,1001,0801,050

724,520375,910340,550307,700259,560212,430187,000 –31,570152,380138,730108,25086,310

115,670110,770122,260100,79044,99041,28436,48033,80035,33030,67031,62027,1509,830

21,62025,26024,01022,83021,35019,97021,14018,94012,71011,08014,11014,2309,440

18,79014,28012,73011,4305,7407,540

10,4905,8909,4509,9608,9708,2506,8307,1906,6406,2405,710

–4,5706,030

TABLE 1INCOME TAX EXPENDITURES RANKED BY SIZE IN 2004

(In Millions of Dollars)

Rank Provision 2004 2004–2008

Source: U.S. Office of Management and Budget (2003), Tables 6–3 and 6–4.


617

reporting, recommended a number ofchanges in the way tax expenditure infor-mation is reported (Toder, Wasow, andEttlinger, 2002). These include: annual es-timates of the cost of all tax expenditurestogether and grouped by budget category;historical estimates of total tax expendi-tures based on a consistent methodology;and detailed information about the dis-tribution of tax benefits as well as an as-sessment of how well the tax expenditureswork.

IS TAX EXPENDITURE A“MEANINGFUL CONCEPT?”

Measuring tax expenditures raises anumber of thorny issues, some of whichunderlie the current Administration’s con-cerns about the tax expenditure budget.The most serious issue is how to definethe “normal income tax.” Surrey andMcDaniel (1985) argue that it should be acomprehensive Haig–Simons measure ofincome with adjustments to reflect prob-lems of administration. As vague as thatguideline is, the actual choice is even morenebulous. Surrey and McDaniel are agnos-tic about fundamental issues, such aswhether the normal income tax should beindexed for inflation. The basic view isthat a fixed relatively comprehensivebaseline should be chosen and it will serveto be a useful measuring rod againstwhich to gauge progress or lack thereofin improving the tax system.

There are some peculiar consequencesof this approach. Expensing and acceler-ated depreciation of investments aretreated as tax expenditures, whereas thetaxation of capital gains on a realizationbasis, rather than as they accrue, is treatedas part of the normal tax.6 All of those ex-

amples convey tax benefits through asimilar mechanism—taking advantage ofthe time value of money. Accelerating de-ductions and deferring income are twosides of the same tax minimization strat-egy, but only the deduction is accountedas a tax expenditure. An even more starkcontrast is with the treatment of savingsbonds, on which the interest income isdeferred until the bond is cashed in, whichis basically identical to the treatment of capi-tal gains. But the former is considered atax expenditure, because accrual taxationof bond interest is straightforward (andthe norm for most bonds), whereas accrualtaxation of gains is not.

The normal income tax contains a clas-sical corporate income tax with no offsetfor double taxation. This is a source ofconsternation among Administrationeconomists. They recently proposed leg-islation that would eliminate the doubletaxation of corporate income. Against thenormal income tax baseline, that proposalwould be considered a tax expenditure,although it is likely that the normal taxwould be modified to incorporate the newnorm if it were enacted.

In any event, the FY 2003 Budget prom-ised to consider measuring tax expendi-tures against a more comprehensive mea-sure of income and to consider reportingnegative tax expenditures for the firsttime. Against such a baseline, the taxationof dividends (as well as part of the tax oncapital gains) may be represented as anegative tax expenditure, and repeal ofdouble taxation would result in neutraltaxation with no tax expenditure.

This year’s Budget delivered on thepromises. Negative tax expenditures werereported for some items, such as the ex-cess of economic over tax depreciation for

6 Haig–Simons income would include accrued capital gains, but the normal tax measures capital gains on arealization basis. Three factors drove this decision: historical precedent (gains have always been taxed formost individuals on a realization basis); the widely held belief that accrued but unrealized gains are notincome; and the administrative difficulty of taxing gains when the sale price is not observable. Surrey andMcDaniel (1985) seem ambivalent on this choice, deeming it as appropriate as of 1985, but one that should bereexamined over time.


618

non–residential structures. An appendixto the tax expenditure chapter in the FY2004 Budget also examined two alterna-tive baselines—a comprehensive incometax and a consumption tax. That analysisincluded an estimate of the tax expendi-ture against the comprehensive incometax baseline from double–taxing corpora-tions: –$25.4 billion in 2004.

But, interestingly, the list of items thatare tax expenditures under current law,but would not be under a comprehensiveincome tax baseline is very short. Severalof the items, such as deductibility formortgage interest and property taxes, maybe viewed as proxies for tax expendituresunder the comprehensive baseline (suchas the nontaxation of imputed rent).7 Thissuggests that, despite its lack of theoreti-cal rigor, the current tax expenditure listprovides a useful perspective on the ex-tent of distortions and deviations of ourincome tax from the economic ideal.

The most contentious issue is whetherto define the normal tax as an income taxor a consumption tax, and this seems tobe the Administration’s main concern. Ifthe income tax is considered the norm,then savings tax incentives—such as tax–exemption for individual retirement ac-counts and pensions, and preferential taxrates for capital gains—are considered taxexpenditures. The FY2003 Budget arguedthat the growing prevalence of tax–freesavings vehicles might suggest a changein norm. “. . . [T]he growing presence oftax–deferred savings vehicles in the taxcode suggests that these may today bepart of the ‘normal’ income tax circa 2002”(U.S. Office of Management and Budget,2002, p. 96).

If a consumption tax is taken as thenorm, then the taxation of interest anddividends are negative tax expenditures—that is, taxation in excess of the norm—and tax–exempt pensions and individual

retirement account are part of the normaltax and thus not worthy of note. Againstthis baseline, preferential tax rates oncapital gains constitute a negative taxexpenditure because they exceed thebenchmark rate of zero. In contrast,against the income tax baseline, the fail-ure to tax realized capital gains at full ratesis the third largest tax expenditure (SeeTable 1).

The FY 2004 Budget categorizes the 30largest tax expenditure items in terms ofwhether they would appear in a con-sumption tax expenditure list (See Table2). Just over half of the items would “prob-ably” be tax expenditures against bothbaselines. The 13 items that are considerednot tax expenditures include savings taxpreferences, accelerated depreciation, taxdeferral on foreign income, and limita-tions on deductibility of losses.

As noted in the Administration’s Bud-get, the choice of consumption as a baseleaves many issues unresolved, becausedifferent variants of consumption taxwould include very different items as taxexpenditures. For example, Senate Fi-nance Committee Chairman WilliamGrassley recently proposed to repeal theexclusion for a limited amount of earnedincome of Americans living abroad (Sec-tion 911 of the Code). That exclusion isclearly a tax expenditure measuredagainst an income tax and would also bea tax expenditure measured against a con-sumed income tax, but would probablynot be treated as a tax expenditure againsta value added tax. The first variant of con-sumption tax would tax earnings (fromwhich the 911 exclusion would be an ex-ception) but exempt the normal return tocapital, whereas the latter variant wouldtax only domestic consumption (and in-come from any source would only betaxed to the extent that it is spent in theU.S.).

7 The Budget analysis also lists a number of items that might not be tax expenditures under the comprehensiveincome tax baseline, because those items tend to be justifiable as an approximation to a tax expenditure againstthe theoretical benchmark.


619

There is clearly an ideological elementto the debate about tax bases. People whofavor an income tax also tend to favor thecurrent method of measuring and display-ing tax expenditures. Those who wouldprefer heavier reliance on consumptiontaxes would favor defining the normal taxas a broad–based consumption tax.8 Giventhat the actual income tax is a hybrid sys-tem containing many elements of incomeand consumption taxation, there is noobjective way to resolve this dispute. Dis-

playing tax expenditures against multiplebaselines may be helpful, but it may alsocontribute to overall confusion.

One aspect that will be helpful is indemonstrating the extent of overlap be-tween the two lists. Many tax expendi-tures would exist against any baseline(e.g., charitable deduction, child taxcredit), but it is interesting to note thatsocial tax expenditures are the commonground, whereas business tax expendi-tures and savings tax breaks do not ap-

8 Indeed, conservative economist, Bruce Bartlett (2001) argues that the current methodology creates a “ . . . biasin favor of liberal tax policy.”

TABLE 2CATEGORIZATION OF CURRENT TAX EXPENDITURES UNDER A COMPREHENSIVE

CONSUMPTION TAX BASELINE1

Revenue EffectDescription (2004)

A. Tax Expenditure Under a Consumption BaseExclusion of Social Security benefits for retired workersExclusion of workers’ compensation benefitsCredit for increasing research activitiesExclusion of Social Security benefits of dependents and survivors

B. Probably a Tax Expenditure Under a Consumption BaseExclusion of employer contributions for medical insurance premiums and medical careDeductibility of mortgage interest on owner–occupied homesDeductibility of nonbusiness state and local taxes other than on owner–occupied homesDeductibility of charitable contributions, other than education and healthDeductibility of State and local property tax on owner–occupied homesChild creditDeductibility of medical expensesExtraterritorial income exclusionEarned income tax creditDeductibility of charitable contributions (health)Deductibility of charitable contributions (education)Deductibility of self–employed medical insurance premiums

C. Probably Not a Tax Expenditure Under a Consumption BaseWorkers’ compensation insurance premiums

D. Not a Tax Expenditure Under a Consumption BaseNet exclusion of pension contributions and earnings: Employer plansNet exclusion of pension contributions and earnings: 401(k) plansCapital gains (except agriculture, timber, iron ore, and coal) (normal tax method)Step–up basis of capital gains at deathExclusion of interest on public purpose State and local bondsNet exclusion of pension contributions and earnings: Individual Retirement AccountsCapital gains exclusion on home salesExclusion of interest on life insurance savingsAccelerated depreciation of machinery and equipment (normal tax method)Deferral of income from controlled foreign corporations (normal tax method)Net exclusion of pension contributions and earnings: Keogh plansGraduated corporation income tax rate (normal tax method)Exception from passive loss rules for $25,000 of rental loss

18,9306,4604,9904,140

120,16068,44050,91033,99022,16021,310

6,3405,5105,0904,5804,3503,690

6,190

67,87055,29053,93028,50027,31023,13020,86020,74016,663

7,9007,6165,7004,920

1The measurement of certain tax expenditures under a consumption tax baseline may differ from the officialbudget estimate even when the provision would be a tax expenditure under both baselines.Source: U.S. Office of Management and Budget (2003), Appendix Table 2.


620

pear as tax expenditures against con-sumption tax base. Those are also the taxexpenditures that are worth most tohigher income taxpayers.

But even those who favor a consumptiontax could find useful information in thecurrent tax expenditure budget. A hybridincome–consumption tax, as we have in theUnited States, may actually do more toimpair national savings than a pure incometax, because of the nonneutralities amongdifferent kinds of saving and investment.The tax expenditure list provides at least acrude measure of these nonneutralities, in-sofar as it shows that particular industriesbenefit far more than others.

Indeed, the Budget presentation ac-knowledges this point:

The hybrid character of the existing taxsystem leads to many provisions thatmight make good sense in the context ofa consumption tax, but that generate in-efficiencies because of the problem of the“uneven playing field” when evaluatedwithin the context of the existing tax rules.It is not clear how these should be classi-fied. For example, many saving incentivesare targeted to specific tax–favoredsources of capital income, and so poten-tially distort economic choices in waysthat would not occur under a broad–based consumption tax (U.S. Office ofManagement and Budget, 2003, p. 135).

The tax expenditure list, however, doesnot present this information in the mostuseful form. A better measure would beestimates of the effective tax rate by in-dustry and type of investment. This in-formation would be useful as a comple-ment to the tax expenditure budget.Against any baseline, a tax system thatproduces the same revenue with relativelycomparable effective tax rates (or usercosts) will be more efficient than one thatproduces widely varying effective taxrates (Gravelle, 1994).

The most peculiar argument against thetax expenditure concept is the notion thatit assumes that all income belongs to thegovernment unless government deigns torefund it in the form of tax breaks. Inter-estingly, neither this argument nor the con-cept of tax expenditure is a new one.Brooks (1986) reports that in 1863, WilliamGladstone, then a Tory member of the Brit-ish parliament, railed against the exemp-tion from income tax of charitable contri-butions. He complained that the charitablededuction would make no sense as a di-rect expenditure, conflicting as it wouldwith efforts to bring “ . . . the whole ex-penditure of the State . . . within the con-trol, and under the eye, of the House ofCommons. If this money is to be laid outupon what are called charities, why is thatportion of the State expenditure to be al-together withdrawn from view . . . and tobe so contrived that we shall know noth-ing of it, and have no control over it . . . ?”9

The rebuttal from Sir StraffordNorthcote could be lifted from the mod-ern ultraconservative’s critique of tax ex-penditures: “‘The right hon. Gentleman,if he took £5 out of the pocket of a manwith £100, put the case as if he gave theman £95 . . . ’” (Brooks, 1986, p. 684). Morethan a century later, the Republican ViceChairman of Congress’s Joint EconomicCommittee, Jim Saxton, complained that“[t]he tax expenditure concept reliesheavily on a normative notion that shield-ing certain taxpayer income from taxationdeprives government of its rightful rev-enues” (Saxton, 1999).

The irony of this aspect of the debate isthat conservatives in other contexts objectto runaway growth of government spend-ing. Presumably it is relevant in evaluat-ing spending on housing programs, forexample, to note that the largest new con-struction program is implemented not asa direct expenditure managed by the De-partment of Housing and Urban Develop-

9 Cited in Brooks (1986, pp. 683–84).


621

ment, but the low–income housing credit.The largest cash assistance program forlow–income families is the earned incometax credit. And so on. All of these programscould be implemented as virtually identi-cal direct expenditure programs, and pre-sumably conservatives in Congress wouldwant to monitor their cost and effective-ness if they were thus transformed.

More generally, periodically evaluatingthe size and effectiveness of tax expendi-tures is a necessary (although not suffi-cient) requirement for good government.Even those who believe that lower taxesand smaller government are always agood thing should care about tax expen-ditures, because the revenues lost to suchprograms could be used for other pur-poses—such as lowering tax rates. Indeed,the Tax Reform Act of 1986, enacted twodecades after the implementation of taxexpenditure estimates, illustrates thistrade–off. Top marginal tax rates for indi-viduals were cut from 50 percent to 28percent, and from 46 percent to 34 percentfor corporations in a package that wasdesigned to be revenue neutral (Birnbaumand Murray, 1987). The dramatic rate re-ductions were financed entirely by elimi-nating or curtailing tax expenditures.10

Millions of taxpayers were also removedfrom the tax rolls.

While some might argue that certain taxexpenditures that were eliminated wouldhave been worth slightly higher tax rates,that is exactly the kind of debatepolicymakers should have. It cannot hap-pen without a full assessment of the taxexpenditures in the code.

USE OF TAX EXPENDITURES IN THEUNITED STATES

Both the Executive and Legislativebranches of the U.S. government and

occasionally private researchers measureand evaluate tax expenditures. The U.S.Budget groups tax expenditures togetherby budget category, which allows analyststo examine immediately how much sup-port is provided to various governmentactivities.

To illustrate the analysis that might beapplied to tax expenditures, take the firstitem in the Budget tables. It might surprisesome to learn that a small share of mili-tary compensation is provided in the formof an exclusion from tax of certain ben-efits and housing allowances for militarypersonnel. This allowance is worth $11.4billion over five years—a tiny fraction ofdirect spending on defense—but not aninsignificant sum. Presenting the estimatein the budget highlights the existence ofthe subsidy (a narrative section describeseach provision) and allows policymakersand others to examine whether it is an ef-fective way to spend governmental re-sources.

Although few in the U.S. question theneed for a strong national defense, somemight wonder if the $11 billion might bebetter spent directly on military compen-sation or perks. The exclusion is worthmost to highly compensated military of-ficers and those with substantial othersources of income, either because theirspouses work or they have unearned in-come. At a time when press reports indi-cate that some military families are livingin poverty, it is unlikely that a direct boostto compensation would be targeted at thebest off members of the military. On theother hand, some would argue that on–base military housing is difficult to value.Adopting the most administrable solutionof making housing allowances tax–free ona military base, but taxable off the base(where value of the subsidy is easy to de-termine), would be inequitable. The other

10 There was also a shift of tax liability from individuals to corporations that some viewed as anti–growth, butthe point is that eliminating tax expenditures made possible substantial tax rate cuts that most economistswould view as efficiency–enhancing.


622

military exclusions are related to particu-lar circumstances where filing tax returnsmight be a particular hardship.

This is not the place to resolve these is-sues. Suffice it to say that the existence ofthe tax expenditure list and budget esti-mates invites this kind of scrutiny. If thegovernment did not measure exclusionssuch as the one for military pay and perks,it is unlikely that legislators or analystswould ever submit any but the largest taxsubsidies to scrutiny.

Other elements of the tax–expenditurebudget routinely invite public debate. Themortgage interest deduction is the larg-est housing subsidy by far and dwarfs thesize of not only other tax expenditures butalso other direct expenditures for hous-ing in the federal budget. The subsidy hasbeen criticized as an upside down subsidythat provides the greatest benefit to up-per middle class and upper classhomeowners, while the greatest needs areamong lower–income families strugglingto pay rent or afford a home (Howard,1997). In 2001, $65 billion of tax revenueswere diverted to subsidize owner–occu-pied housing, compared with $14 billionon rental vouchers, $6 billion for publichousing, and $3 billion for the low–in-come housing tax credit—the largest pro-grams aimed at helping low–income rent-ers. (Another $22 billion in assistance forhomeowners was conveyed via the de-duction for property taxes.)

One issue in evaluating tax expendi-tures is that they are difficult to directlycompare with outlay programs becausethe tax expenditure is net of any offset-ting tax receipts that they may generate(because the income received is taxable).In contrast, outlay estimates for spendingprograms simply reflect the amount ofmoney that the agency will spend with-out an explicit adjustment for offsettingtax receipts. For this reason, the Treasury

also produces a set of tables showing taxexpenditures in “outlay equivalent”terms—in other words, the amount thatwould have to be budgeted for an equiva-lent direct spending program.11

For example, if the excluded compen-sation and benefits provided to militarypersonnel were instead provided in theform of a wage supplement, the outlayequivalent would increase from the $11.4billion tax expenditures to a $13.3 billionoutlay. The increase is the difference be-tween after–tax and pre–tax income. Themortgage interest deduction would in-crease from $6.0 billion over five years to$8.7 billion if implemented as a directmortgage subsidy, and the low–incomehousing credit would increase from a $3.3billion per year (after–tax) credit to a $4.5billion per year (pre–tax) cash outlay. Thedifference arises because, if the moneywere paid directly to investors as cash, theincome would be taxed at an average rateof 26 percent.

The Administration’s Budget presenta-tion shows the data in several differentways. Table 1 shows the 57 largest incometax expenditures (those costing more than$1 million—about 0.01 percent of GDP) inthe United States in Fiscal Year 2004.12 Thetable shows that the largest tax expendi-tures are quite large. Nontaxable fringebenefits are among the largest tax expen-ditures. The exclusion of employer–provided health insurance from incomeis expected to cost $120 billion in losttax revenues in 2004, nearly 15 percentof income tax revenues. The exclusionof defined contribution 401(k) plansand defined benefit employer pensionplans each costs over $55 billion, or 6 per-cent of income tax revenues. Clearly mar-ginal income tax rates could be substan-tially reduced if these major sources ofcompensation were included in taxableincome.

11 See U.S. Office of Management and Budget (2003, pp. 112–15).12 The budget ranks all tax expenditures. The smaller ones are excluded from Table 1, but not from the Budget.


623

For some tax preferences that involvedeferral of income tax liability, the Bud-get includes estimates of the presentvalue (PV) of tax benefits with respect toactivity undertaken in 2002 (See Table 1).These estimates provide an indication ofthe cost of an investment tax subsidy overthe life of the investment. They are alsouseful in showing the tax expendituresolely with respect to activity undertakenover a fixed time period—in this case,2001. A significant portion of saving orinvestment tax subsidies such as the ex-clusion of pension contributions and earn-ings reflects revenue lost on saving madein prior years. For example, the tax expen-diture for individual retirement ac-counts—number 11 on Table 1—largelyreflects forgone tax revenues on amountsdeposited in IRAs one or more years ago.As a result, the PV of annual contributions

(10.6 billion) is less than half of the totalannual cost of the IRA exclusion ($23.1billion).

The Joint Committee on Taxation (JCT)presents a similar tabulation of tax expen-ditures annually. There is not 100 percentoverlap between the items on the two lists,largely because the JCT opts for a broadermeasure of normal income. There are alsodifferences in the way different provisionsare combined together. Nonetheless, theestimates are largely consistent, and theJCT document explains the significant dif-ferences that exist.

The JCT makes an important contribu-tion by tabulating the distribution ofbenefits from selected tax expendituresby income class. A subset of JCT’s 2002analysis is shown in Table 3. The tax ex-penditures at the top of the table—all de-ductions from income—primarily benefit

TABLE 3DISTRIBUTION BY INCOME CLASS OF SELECTED INDIVIDUAL TAX EXPENDITURE ITEMS, 2001

(Money Amounts in Millions of Dollars, Returns in Thousands)

Charitable ContributionDeduction

Returns Amount

$13239817

16462,930

10,70414,07021,94514,570

$66,934

501183

2,4353,1973,7129,7237,0077,3782,247

36,933

671076

1,9382,7593,2338,8796,6666,9762,110

33,704

2$66241584

11764,8716,744

12,36922,841

$48,894

511166

2,1652,9003,3779,2526,9837,2552,068

35,217

$2$117$341$597$947

$3,519$4,275$6,654$4,786

$21,238

591257

2,4983,1713,6459,7057,2197,8902,592

38,035

$2122365695

1,0664,0585,2518,913

19,959

$40,428

Income Class($thousands) Returns Amount Returns Amount Returns Amount

Below $10$10 to $20$20 to $30$30 to $40$40 to $50$50 to $75$75 to $100$100 to $200$200 and over

Total

Income Class Returns Amount Returns Amount Returns Amount

Below $10$10 to $20$20 to $30$30 to $40$40 to $50$50 to $75$75 to $100$100 to $200$200 and over

Total

State and Local TaxDeduction

Real Estate TaxDeduction

Mortgage InterestDeduction

Earned Income TaxCredit Child Care CreditChild Tax Credit

523,2564,1583,6713,2477,1494,9734,204

1

30,709

$221,2033,1473,3993,1537,2165,1783,857

(1)

$27,176

5,3705,8684,4342,520

35216000

18,560

$6,76015,0539,2542,703

21221000

$34,002

168

429583638

1,5081,1631,500

227

6,117

($4)20

222348364745604806128

$3,236

Notes: Estimates for child tax credit and earned income tax credit include refundable portionSource: Joint Committee on Taxation, Estimates of Federal Tax Expenditures for Fiscal Years 2003–2007 (JCS–5–02).Washington, DC: Government Printing Office, 2002.


624

upper–middle and upper–income house-holds. Only about 30 percent of tax filersitemize deductions, and most of thosehave higher incomes. In consequence,two of these tax expenditures (deductionfor charitable contributions and for stateand local taxes) provide just under half oftheir benefits to families with incomesover $200,000. All of them provide overhalf of benefits to the approximately 5percent of tax filers with incomes over$100,000.

In contrast the child tax credit is dis-tinctly a middle income tax subsidy (seebottom panel of Table 3). Indeed, it is notavailable to families with very high in-comes (it starts to phase out at $110,000 ofincome). In consequence, almost half ofits benefits accrue to families with incomesunder $50,000. The earned income taxcredit is explicitly targeted at low–incomefamilies. Almost two–thirds of its benefitsaccrue to families earning under $20,000per year. The child and dependent care taxcredit is primarily a middle–class tax ben-efit. Lower–income families often cannotafford to pay for care, even with a taxcredit, and very low–income families can-not benefit from the credit because it is notrefundable.

The most systematic and comprehen-sive evaluation of tax expenditures is doneby the Congressional Research Service ofthe Library of Congress every two yearsfor the Senate Budget Committee. In 2002,the report evaluated 128 tax expenditures(Congressional Research Service, 2002).The evaluation includes a description ofthe provision, its cost, impact (includingdistribution of tax benefits by incomewhen available), and rationale. It con-cludes with a summary assessment ofthe arguments for and against the provi-sion. There are also references to other

relevant research on the specific tax ex-penditure.

Beyond the summary descriptions andestimates, there is no regular and system-atic evaluation of tax expenditures con-ducted by the executive branch of the fed-eral government. The Government Perfor-mance Results Act of 1993 (GPRA), whichwas intended to “provide for the estab-lishment of strategic planning and perfor-mance measurement in the Federal Gov-ernment,” requires annual evaluations oftax expenditures by the Office of Manage-ment and Budget in consultation with theDepartment of the Treasury.”13 Clinton’sTreasury Department, of which I was apart from 1998 to 2000, was unenthusias-tic about performing these evaluations,reasoning that a comprehensive evalua-tion of tax expenditures would necessar-ily raise serious objections to measuresenthusiastically advanced by the Admin-istration. The result would either be awaste of staff time, as a credible analysiswould never be published, or a white-wash that would damage the credibilityof the Treasury staff. Although the menuof favorite tax expenditures changedwhen President Bush took office, the Of-fice of Management and Budget has notpublished any new tax expenditure analy-ses as part of GPRA, suggesting that thesame concerns still hold sway.

Other government agencies analyze taxexpenditures on an ad hoc basis. The Gen-eral Accounting Office has evaluated nu-merous tax subsidies at the request ofCongress. A 1994 study concluded that taxexpenditures had been growing fasterthan direct expenditures and recom-mended that tax–writing committees ofCongress look for ways to draw more at-tention to tax expenditures as part of thebudget process, although it is unclear

13 According to the Committee Report that accompanied the legislation, “[t]he Committee expects that annualperformance reports would . . . be used to report on . . . tax expenditure assessments. These assessments shouldconsider the relationship and interactions between spending programs and related tax expenditures. The Com-mittee hopes that such reports will foster a greater sense of responsibility for tax expenditures with a directbearing on substantial missions and goals” (Senate Committee on Government Affairs, 1993, Section 4).


625

what result that recommendation had, ifany (General Accounting Office, 1994).The Congressional Budget Office (CBO)analyzes a set of tax and budget options,including repealing or modifying a num-ber of tax expenditures, in its annual Bud-get Options volume (Congressional Bud-get Office, 2003). The CBO also occasion-ally writes reports about particular tax ex-penditures (see, e.g., Burman, 1992). TheCongressional Research Service of the Li-brary of Congress produces similar re-ports at the request of Congress. In addi-tion, all of these agencies and the JCT andTreasury are sometimes requested to ana-lyze existing and proposed tax expendi-tures at Congressional hearings.

The public estimates and analysis re-quire a good deal of staff resources, data,and computer services. Treasury’s Officeof Tax Analysis has a staff of about 50,many of whom work on aspects of the taxexpenditure budget. Congress’s JointCommittee on Taxation has a smaller staff,but many of them are also devoted to rev-enue estimating and tax expenditure esti-mates. Estimates are based on a large(>100,000) micro data file for individualsand a corporate data file that contains tax

returns for all of the largest corporations.Data are brought to bear from othersources, such as the Survey of ConsumerFinances, a comprehensive survey of in-dividual wealth holding produced by theFederal Reserve Board.

Outsiders often use these data to evalu-ate tax expenditures. The Century Foun-dation recently used tax expenditure datato conduct a thorough review of tax ex-penditures in the U.S., as discussed ear-lier (Toder, Wasow, and Ettlinger, 2002).Eric Toder has published a number ofstudies of the changing nature of tax ex-penditures. He documented a markedshift away from business tax expendi-tures—savings incentives, investment taxcredits, accelerated depreciation, specialprovisions for favored industries, etc.—in favor of social tax expenditures—pro-grams that look like federal assistanceprograms for individuals. For example,the value of the earned income taxcredit—an income support program forworking poor families—has increaseddramatically over the past two decades.Numerous tax credits have been imple-mented to subsidize low–income housingconstruction, higher education, child care,

Figure 1. Trends in Tax Expenditures 1980–2000 (as percent of GDP)


626

and child rearing. His chart documentingthis shift, extended by Elaine Maagthrough 2000, is shown in Figure 1. Al-though the problems of aggregating taxexpenditures makes the point estimatesunreliable, there is little doubt that a seachange in the nature of tax expenditureshas taken place. Toder (1998) discusses thefactors that have contributed to this shiftand its consequences.

Howard (1997) has documented thesame phenomenon, writing from a politi-cal scientist’s rather than an economist’sperspective. He calls this shift in relianceon tax expenditures “the hidden welfarestate.”

CONCLUSIONS

Early in the Bush Administration, con-servative ideologues questioned the use-fulness of the tax expenditure concept andmany feared that this very useful toolmight be banished from government bud-get presentations. Instead, the Adminis-tration has so far continued to produceits comprehensive summary and esti-mates of income tax expenditures withonly minor political filtering (the deletionof estate tax expenditures most notably).In addition, this year’s Budget includes adiscussion of the issues that would beraised by measuring tax expendituresagainst alternate baselines—a compre-hensive income tax and a comprehensiveconsumption tax. That analysis shedslight on the robustness of tax expenditureestimates.

Under any baseline, the tax expenditureconcept is a useful way of showing howgovernment affects the allocation of re-sources both directly—by financing pub-lic activities via tax concession—and in-directly, by altering after–tax prices andthus distorting the allocation of resources.One suspects, though, that theAdministration’s preference to shift thefocus of analysis from an income taxbaseline to a consumption tax baseline is

part of a larger strategy to sneak a con-sumption tax in through the back door.As the 2004 Budget points out, however,savings subsidies in the context of a hy-brid income–consumption tax system cancreate substantial resource misallocations.

It would thus be much better to havethe debate about income tax versus con-sumption tax in the open and, oncesettled, attempt to make the tax system asneutral as possible. Measures and analy-sis of tax expenditures against the appro-priate baseline will continue to be helpfulin advancing that goal.

Acknowledgments

I am grateful for helpful comments fromBruce Davie, Jane Gravelle, and partici-pants at the National Tax AssociationSpring Symposium on May 29, 2003, andthe World Bank seminar on “Tax Expen-ditures, Fiscal Transparency, and CountryPractices,” January 27, 2003. DeborahKobes provided invaluable research assis-tance. Views expressed or implied are theauthor’s alone and should not be attrib-uted to any institution with which he isaffiliated. Contact: [email protected].

REFERENCES

Balkovic, Brian.“Individual Income Tax Returns, Prelimi-nary Data, 2001.” SOI Bulletin 22 No. 3 (Win-ter, 2002–2003): 136–46.

Bartlett, Bruce.“The Flawed Concept of Tax Expenditures.”National Center for Policy Analysis. http:// w w w. n c p a . o r g / o p e d / b a r t l e t t /apr3001.html, 2001.

Birnbaum, Jeffrey H., and Alan S. Murray.Showdown at Gucci Gulch: Lawmakers, Lobby-ists, and the Unlikely Triumph of Tax Reform.New York: Random House, 1987.

Brooks, Neil.Review of Surrey and McDaniel (1985). Ca-nadian Tax Journal 34 No. 3 (May–June, 1986):681–94.


627

Burman, Leonard E.“The Cost–Effectiveness of the Low–IncomeHousing Tax Credit Compared With Hous-ing Vouchers.” CBO Staff Memorandum. Con-gressional Budget Office, April 1992 (re-printed in Tax Notes, July 27, 1992).

Congressional Budget Office.Budget Options. Washington, D.C.: Govern-ment Printing Office, March 2003.

Congressional Research Service.Tax Expenditures: Compendium of BackgroundMaterial on Individual Provisions, Committeeon the Budget, United States Senate, 107th Con-gress, 2nd Session. Senate report 107–80, De-cember 2002.

Davie, Bruce.“Tax Expenditures in the Federal Excise TaxSystem.” National Tax Journal 47 No. 1(March, 1994): 39–62.

General Accounting Office.“Tax Policy: Tax Expenditures Deserve MoreScrutiny.” GAO Report GAO/GGID/AIMD–94–122, June 1994.

Gravelle, Jane G.The Economic Effects of Taxing Capital Income.Cambridge, MA: The MIT Press, 1994.

Howard, Christopher.The Hidden Welfare State: Tax Expenditures andSocial Policy in the United States. Princeton:Princeton University Press, 1997.

Joint Committee on Taxation.Estimates of Federal Tax Expenditures for Fis-cal Years 2003–2007 (JCS–5–02). Washington,D.C.: Government Printing Office, 2002.

Saxton, Jim.“Tax Expenditures: A Review and Analysis.”Joint Economic Committee, http://w w w. h o u s e . g o v / j e c / f i s c a l / t a x /expend.pdf, August 1999.

Senate Committee on Governmental Affairs.“Government Performance and Results Actof 1993, Report of the Committee on Gov-ernmental Affairs, United States Senate, toAccompany S. 20.” Report 103–58, June 1993.

Sheils, John, and Paul Hogan.“Cost of Tax–Exempt Health Benefits in1998.” Health Affairs 18 No. 2 (March/April,1999): 176–81.

Surrey, Stanley S., and Paul R. McDaniel.Tax Expenditures. Cambridge, MA: HarvardUniversity Press, 1985.

Toder, Eric.“The Changing Role of Tax Expenditures:1980–99.” In Proceedings of the Ninety-FirstAnnual Conference on Taxation. Washington,D.C.: National Tax Association, 1998.

Toder, Eric.“The Changing Composition of Tax Incen-tives 1980–99.” The Urban Institute. Wash-ington, D.C., 1999.

Toder, Eric, Bernard Wasow, and Michael P.Ettlinger.

Bad Breaks All Around: The Report of The Cen-tury Foundation Working Group on Tax Expen-ditures. New York: The Century FoundationPress, 2002.

U.S. Office of Management and Budget.Analytical Perspectives, Budget of the UnitedStates Government Fiscal Year 2001. Washing-ton, D.C.: Government Printing Office, 2000.

U.S. Office of Management and Budget.Analytical Perspectives, Budget of the UnitedStates Government, Fiscal Year 2003. Washing-ton, D.C.: Government Printing Office, 2002.

U.S. Office of Management and Budget.Analytical Perspectives, Budget of the UnitedStates Government, Fiscal Year 2004. Washing-ton, D.C.: Government Printing Office, 2003.

Documents

Is the Tax Expenditure Concept Still Relevant? - … the Tax Expenditure Concept Still Relevant?Published in: National Tax Journal · 2003Authors: Leonard E BurmanAffiliation: Georgetown