Special Studies, May 27, 2008

Robert D. Dietz, Ph.D.

Report available to the public as a courtesy of HousingEconomics.com

Table of Contents
Print the Full Article
See previous Special Studies

Each year, the Congress and the Treasury Department report estimates of tax expenditures as part of the annual budget process. A tax expenditure is a provision of the tax code that reduces tax burdens for some taxpayers based on certain qualifying criteria. [1] In general, tax expenditures are targeted to a certain group of taxpayers, and thus are viewed negatively by some tax policy analysts. For others, tax expenditures are economically or otherwise justified policies that are governed by the tax code.

The original intent of tax expenditures was to provide an equivalent measure for tax provisions relative to appropriations in the government’s budget—that is, tax expenditures were intended to measure revenue losses associated with tax rules in the same way appropriations measure outlays associated with spending programs. However, as this paper demonstrates, tax expenditures do not measure how much revenue the government can expect to gain by repealing particular provisions of the tax code.

With respect to housing policy, the Congress reported the following tax expenditures for tax year 2008:

• Home mortgage interest deduction: $79.8 billion

• Real estate tax deduction: $14.3 billion

At times, the housing industry and homeowners view these numbers with apprehension, fearing that opponents of the property tax and the mortgage interest deductions may cite them as evidence that housing receives excessively favorable treatment under the tax code. In addition to arguing that there are economic, historical, administrative and other policy justifications for housing’s current treatment in the tax code (the subject of a future Housing Economics Online article), there are several caveats to keep in mind when considering tax expenditure analysis with respect to owner-occupied housing:

• Tax expenditures are not equivalent to revenue estimates of repealing the policy.

• Tax expenditures are difficult to estimate and require subjective judgments; estimates from Congress, the Treasury, and private sources vary, sometimes greatly.

• Individual tax expenditures interact, so it is not valid to generate a “total housing tax expenditure” by simply summing individual estimates.

• An alternative tax expenditure (for net imputed rent) that views housing as an investment produces a significantly smaller tax expenditure estimate.

In fact, the Treasury Department has recently begun to report a tax expenditure estimate for net imputed rent. However, because the Treasury Department continues to report tax expenditures for the mortgage interest and property tax deductions, it is important to note that it is incorrect to add these two measures to the tax expenditure estimate for net-imputed rent, as this would result in double-counting.  [2] 

Tax Expenditure Reporting by Government and Other Institutions

The government sources for the identification and estimation of tax expenditures are the Congressional Joint Committee on Taxation (JCT) and Treasury Department’s Office of Tax Analysis (OTA). JCT and OTA are the official “scorekeepers” for tax legislation. The staff economists of these agencies estimate and publish the official numbers concerning both revenue estimates (the expected change in tax revenue from changes in tax law) and tax expenditures (the impacts of present-law tax policy on tax receipts).

To generate their estimates, both the JCT and OTA maintain tax microsimulation models, known as Individual Tax Models (ITM). An ITM is basically a large turbo-tax calculator that generates hundreds of thousands of statistically-generated individual tax returns from sampled, confidential IRS tax return data. Using these simulated taxpayers, the models calculate the effects on federal tax revenue from hypothetical changes to tax law. For baseline calculations, the models typically assume that present tax law is held constant, including eliminating those provisions of the tax code that are scheduled to expire. To develop an ITM’s forecast, certain assumptions and forecasts concerning demographics, macroeconomics, and other economic characteristics must be included in the model. It is not unusual for differing assumptions to be used by JCT and OTA, which can lead to divergences in their estimates.

Furthermore, there are other institutions, typically think tanks, which maintain ITMs to produce tax estimates. Two of the best known are the Tax Policy Center model and the National Bureau of Economic Research (NBER) TAXSIM model. These ITMs rely on the less detailed public-use IRS tax data, but operate in a similar manner to the JCT and OTA models. This paper uses both the publicly-available estimates from JCT and OTA, as well as estimates from the NBER model.

Mortgage Interest Deduction

The tax code allows mortgage interest payments to be deducted for itemizing taxpayers for interest allocable to a primary or secondary residence for mortgage principal amounts of up to $1 million. Interest allocable to home equity loans of up to $100,000 may also be deducted. These amounts are not indexed to inflation. Only itemizing taxpayers may deduct qualified mortgage interest or real estate taxes. As with other itemized deductions, complications with respect to the Alternative Minimum Tax (AMT) are possible. In general, interest allocable to a home equity loan and property tax payments may not be deducted against AMT taxable income, although mortgage interest classified as acquisition indebtedness, including debt used to finance home repair and improvement, may be deducted.

Figure 1 presents the history and forecast of the tax expenditure for the mortgage interest deduction, as reported by the JCT, OTA, and the NBER model. As noted earlier and seen in the Figure, there is significant variation with respect to the estimates of this tax expenditure. [3] 

For 2008, JCT reports that the mortgage interest deduction totals $79.9 billion, while OTA reports $94.8 billion, and the NBER model estimates $104 billion. Regardless, these estimates indicate that the mortgage interest deduction is one of the largest tax expenditures, behind only the exclusion for 401(k)and other retirement plan contributions ($114.1 billion according to JCT), the exclusion for employer-provided health insurance ($116.5 billion according to JCT), and the preferred tax rates on capital gains and dividends ($127.9 billion according to JCT).

It is clear from these estimates that the size of the tax expenditure for the mortgage interest deduction is increasing and is expected to increase, in part due to the boom years in the housing market. However, the NBER estimates indicate that as a percentage of forecasted individual tax revenues, the tax expenditure for the mortgage interest deduction is expected to remain constant at approximately 9%. In 2011, when income tax rates increase, it will fall to approximately 8% of individual tax revenues.

Because many observers use tax expenditures as revenue estimates, it is important to note that a tax expenditure estimate is not a revenue estimate. Hence, these estimates do not indicate what revenue would be gained by the Treasury if Congress repealed the mortgage interest deduction. Consistent with most tax expenditures, the revenue estimate for this proposal would be considerably smaller due to expected changes in behavior, which tax expenditure estimates do not include. In particular, if the mortgage interest deduction was repealed or significantly scaled back, many homeowners would reallocate their holdings of investments in order to reduce their mortgage debt, or more dramatically, delay or defer homeownership. Combined with the decline in housing prices that would accompany such a change, the resulting revenue estimate for such a proposal would be significantly smaller than the corresponding tax expenditure estimate.

Real Estate Tax Deduction

Figure 2 shows the tax expenditure for real estate taxes paid for owner-occupied homes. For 2008, the tax expenditure is expected to total $14.3 billion by JCT, $16.4 billion by OTA, and $15.9 billion by the NBER model. Figure 2 also reveals an important aspect of tax expenditure estimate: the degree to which tax expenditures are dependent on other provisions in the tax code. In this case, the inability of taxpayers to deduct real estate taxes when paying AMT “recaptures” part of the tax expenditure. No AMT patch is in present-law for years 2008 and thereafter, so larger numbers of taxpayers pay AMT. This reduces the use of the real estate tax deduction and lowers its expected tax expenditure. Furthermore, the 2001 and 2003 tax cuts expire at the end of 2010, which reduces the AMT effect, thereby increasing the tax expenditure in 2011. Once again, this analysis demonstrates that placing a single objective number on tax expenditures is a difficult exercise. For example, with no AMT in place, the 2008 NBER tax expenditure estimate would increase from $15.9 billion to $34.9 billion.

Aggregating Tax Expenditures

Figure 3 demonstrates an important rule concerning tax expenditure estimate reporting that is repeatedly violated by policymakers and tax analysts. Unlike government spending accounts, tax expenditures cannot be summed to generate tax expenditures for groups of provisions because of how they are calculated. [4] Tax expenditure estimation involves multiple interactions with other provisions of the tax code; hence, summing individual estimates typically overstates the aggregate tax expenditure due to double counting. Primarily, this effect is due to the standard deduction and the itemization decision. ITMs allow taxpayers to choose to itemize or not, a limited exception to the general rule that tax expenditure estimates do not allow for changes in behavior.

For example, if Congress repeals the mortgage interest deduction, many taxpayers will not itemize and therefore claim the standard deduction. This effect reduces the hypothetical revenue loss, and lowers the tax expenditure estimate for the mortgage interest deduction or any other tax expenditure that would likely cause larger numbers of taxpayers to cease itemizing deductions. Moreover, for a tax expenditure that repealed the mortgage interest and real estate tax deductions, an even larger set of taxpayers would claim the standard deduction, thereby producing a smaller combined tax expenditure. These interactions mean that summing individual estimates will produce inaccurate results.

Using the NBER TAXSIM model, this concept can be illustrated for the housing tax expenditures. As Figure 4 demonstrates, the sum of the two individual estimates for the mortgage interest deduction and the real estate deduction overstates the true tax expenditure for both provisions. NBER’s TAXSIM reports a 2008 tax expenditure of $104 billion for the mortgage interest deduction and $15.9 billion for the real estate tax deduction, yielding a sum of $119.9 billion. The simultaneous TAXSIM estimate, the correct measurement, is $115.4 billion. While only a difference of about 4% in 2008, in prior years the difference has been larger due to higher marginal income tax rates. For example, in 2003, summing the tax expenditures overstated the total housing tax expenditure by 9.93%. Moreover, the larger the bundle – that is, the total number of provision being combined – the larger the overstatement will be.

Net Imputed Rent

While the analysis above shows that care must be taken when trying to establish a single tax expenditure for owner-occupied housing by adding individual components together, a more fundamental issue is at stake—whether to treat housing as consumption or investment. Tax law and tax economics theory indicate that deductions for interest on investment should be permitted. Deductions for state and local taxes should be permitted, because they are a proxy for public investment. For businesses, deductions for interest and tax expenses are considered “normal” and are not listed as tax expenditures. Why, then, are they on the tax expenditure list for individual taxpayers?

The reason for the contrast between business taxpayers and individuals lies in the differing views with respect to the economic role of owner-occupied housing of tax lawyers, who established and maintain the present system of tax expenditure reporting, and some economists. From the tax law perspective, ownership of a home is personal consumption, no different from owning any other durable good like a television or refrigerator. For most economists, a home is a source of a flow of services that generates “income” (i.e. imputed rent) to the homeowner. This income is paid and received by and to the homeowner but is not a market-based transaction, so its value must be imputed from government data. It is calculated and reported in the Commerce’s Department’s Bureau of Economic Analysis national income product accounts, alongside income for rental housing operated by businesses in order to estimate housing’s contribution to GDP.

From a tax policy perspective, income is taxed only for market transactions. Hence, net imputed rent (imputed rent minus appropriate business deductions for taxes, depreciation and interest expense) is not taxed for administrative and compliance purposes. For example, housework done by members of household is not taxed, while housework paid for and completed by a cleaning service is taxed because it involves a market transaction.

Nonetheless, the Treasury Department in 2006 began reporting a tax expenditure for net imputed rent. For 2008, the tax expenditure is $5.4 billion. The Treasury estimate varies greatly year-to-year, with an estimate of $28.8 billion in 2006 and climbing from the 2008 value to $28 billion in 2013 due to expected changes in the non-market realized imputed rent measure.

Note however that, if users do not recognize that the Treasury Department’s list of tax expenditures includes two alternate ways of measuring tax expenditures for owner-occupied housing, they could be guilty of double-counting. It is inappropriate under any circumstances to add a tax expenditure for net-imputed rent to the tax expenditures for the mortgage interest and real estate tax deductions. Owner occupied housing needs to be treated consistently as either consumption or investment, not both. If housing is considered an investment, so that its net imputed rent income should be taxed as synthetic business income, then the absence of such a tax is a legitimate tax expenditure. However, in this case deductions for interest and taxes are equivalent to normal business expenses and should not count as tax expenditures. Alternatively, if housing is considered a consumption good, then tax expenditures for the interest and tax deductions may be reported, but a tax expenditure for net imputed rent is not.

This analysis begs the question: what is the correct approach for measuring tax expenditures for owner-occupied housing?  If one views owner-occupied housing as a form of investment for which services flow to the homeowner, then the correct measure is the tax exclusion for net-imputed rent. On the other hand, if you view owner-occupied housing as a consumption good, then the ability to claim deductions for interest and taxes, which is not allowed for other forms of personal consumption, is the appropriate tax expenditure.

All things considered, given the treatment of owner-occupied housing for GDP measurement purposes and the required tax policy treatment of other parts of the capital stock, the net imputed rent measure is the most appropriate measure of the tax expenditure for owner-occupied housing. This implies that the best estimate we have at present for tax expenditures for owner-occupied housing in 2008 is the Treasury Department’s “net imputed rent” number of $5.4 billion in 2008 (increasing to $28 billion in 2013), rather than any of the larger numbers that could be produced from the estimates of tax expenditures for the mortgage interest and real estate tax deductions. [5] 

______________________

Appendix – Tax Expenditure Definition

The technical definition of a tax expenditure comes from the Congressional Budget and Impoundment Act of 1974 (“the Budget Act”). The Budget Act created the modern Congressional budget-making process, including the establishment of the Congressional Budget Office, the tax legislative scoring and analysis roles of the Joint Committee on Taxation, and a requirement for annual reporting of tax expenditures. The Act defines tax expenditures as:

“Revenue losses attributable to provisions of Federal income tax laws which allow a special exclusion, exemption, or deduction from gross income or which provide a special credit, a preferential rate of tax, or a deferral of tax liability.”

This definition of tax expenditures is open to considerable interpretation. In particular, the legislative history for the Budget Act indicates that tax expenditures are to be determined in reference to “normal income tax law.”  However, determining what is the “normal” income tax is inherently a subjective exercise based on one’s view of how the income tax should operate and what public policy objectives it should incorporate. The more a given observer believes the tax code should only raise revenue for the government with a minimum of market interference, the more likely they are to classify particular tax rules as tax expenditures.

For example, the first tax expenditure lists, published in 1968 by Assistant Secretary of the Treasury for Tax Policy Stanley Surrey were part of an effort to report synthetic government spending programs that were governed by the tax code. Surrey’s objective was to improve the budget process and increase support for base-broadening tax reform. However, Surrey was also an enthusiastic supporter of the progressive system of income tax rates that tax higher income levels at higher marginal and average tax rates. At that time, the progressive system of rates was not classified as a tax expenditure by Surrey’s staff, so clearly some “preferential” rates were more equal than others when it came to tax expenditure classification. This subjective nature of tax expenditure classification and measurement has rightly limited tax expenditure analysis’s influence on public policy, as compared to the more objective reporting of government spending.

To measure a tax expenditure, an economist at JCT or OTA must first specify the normal income tax. All deviations from that hypothetical, ideal income tax system – justified or not – are correspondingly classified as tax expenditures. To estimate the size of a given tax expenditure, the economist hypothetically repeals provision, and measures the amount of revenue that would be gained by the Treasury. It is important to note that for tax expenditure analysis, the economist assumes that there is no change in behavior relative to present law. Hence, if the tax expenditure encourages, for example, college enrollment, then for estimation purposes the economist assumes that there is no change in college attendance relative to present law. This estimation process is different than revenue estimation (i.e. “scoring” tax legislation) for which governments economists do incorporate behavior; although no macroeconomic feedbacks are included, which would constitute “dynamic” scoring.

____________________

Footnotes:

[1] A more technical definition of tax expenditure can be found at the end of this paper.

[2] This article references findings from the 2008 working paper, “Reconsidering Tax Expenditure Estimation: Challenges and Reforms,” by Rosanne Altshuler and Robert Dietz, which was completed as part of the NBER Conference “Incentive and Distribution Consequences of Tax Expenditures.”  For this paper, the author used estimates from MIT NBER TAXSIM model. I would like to thank NBER for making this resource available. There has been a recent resurgence in interest in tax expenditure estimation, and this paper is part of that debate. For example, the Chief of Staff of the Joint Committee of Taxation on May 1, 2008 indicated in a recent speech “Rethinking Tax Expenditures” (http://www.house.gov/jct/Rethinking_Tax_Expenditures.pdf) that the JCT will soon be reforming how they report tax expenditures.

[3] The historical estimates in this paper from OTA and JCT come from the institutions’ forecast estimates of a given year, as published in the year prior to the estimate. Hence, the estimates for 2007 are those forecasts reported in 2006. The 2006 estimates come from 2005. This method of reporting is used because neither institution publishes historical series of tax expenditures. In contrast, the historical estimates from the NBER model are based on historical data and were estimated in 2008, but these estimates rely on the inferior public-use SOI data. The estimates for 2008 and thereafter are from the most recent forecast from each model.

[4] Despite this, papers summing tax expenditures, and reporting incorrect estimates, appear routinely. Recent examples include: Carraso, Steuerle, and Bell. 2005. “The Trend in Federal Housing Tax Expenditures.” Urban Institute. U.S. General Accountability Office. 2005. “Tax Expenditures Represent a Substantial Commitment and Need to be Reexamined.”  GAO-05-690.

[5] In addition to the numbers reported here, another relatively large tax expenditure for owner-occupied housing is the principal residence gain exclusion, which allows a homeowner to exclude up to $500,000 in capital gain ($250,000 for a single taxpayer) for the sale of a primary residence, which totals $29 billion for 2008 according to JCT. This exclusion is not examined in this paper. Also, it is important to note that taxpayers face significant restrictions in their ability to deduct a capital loss on the sale of a residence, unlike other capital assets. This “negative tax expenditure” exists but is not currently reported by either JCT or OTA due to their general convention of not reporting negative tax expenditure. More information can be found in Altshuler and Dietz (2008).