Archved - Government of Canada
Tax Expenditures and Evaluations - 2000: 3

Archived information

Archived information is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please contact us to request a format other than those available.

- Table of Contents - Previous - Next -

Part 2
Tax Evaluations and Research Reports

Defining Tax Expenditures

I. Introduction

The principal goal of a tax system is to raise revenue in order to finance government operations and programs. As the 1997 federal budget explained: "In raising revenue, it is essential for governments to use a tax system that is fair…." [1] Governments also use a variety of tax instruments - credits, exemptions, rate reductions, deductions, deferrals, rebates and carry-overs - to achieve an array of social and economic objectives.

These deviations from a benchmark tax system, or tax concessions, are often described as tax expenditures, implying that the tax system is being used to deliver assistance that could have been provided through spending. This raises two issues. First, identifying tax concessions is not a simple exercise. Second, this approach labels all tax concessions as tax expenditures and hence as a substitute for program spending.

A simple example highlights the complexities involved in identifying tax concessions. Consider two countries, both having a statutory tax rate of 30 per cent. Suppose Country A has an additional, lower, 15-per-cent tax rate on incomes that are less than $30,000 dollars. Now suppose that Country B, instead of the second 15-per-cent tax rate, allows a variety of tax credits, deductions and exemptions, targeted at those with incomes below $30,000. As a result of these tax concessions, assume that Country B effectively has the same tax burden for all citizens as Country A.

Despite the two tax structures having identical taxation outcomes, the list of tax concessions for Country A will differ greatly from that of Country B. Because common practice allows the tax rate structure to be taken as part of the benchmark, Country A will have no tax concessions to report, while Country B will have a long list. This is obviously not reasonable, and becomes even more problematic if all tax concessions are simply assumed to be substitutes for program spending.

The example suggests that determining tax expenditures should be a two-step, rather than a one-step, process. The first step should determine tax concessions starting with a benchmark tax system. The second step should determine, using an appropriate set of criteria, whether the tax concession is deliverable on the spending side. If it is, then it is a tax expenditure; if it is not, then it is effectively a tax reduction. This two-step approach implies that tax expenditures are a subset of tax concessions: hence, all tax expenditures are tax concessions; not all tax concessions are tax expenditures; and tax concessions that are not tax expenditures are effectively tax reductions.

Recognition that not all tax concessions are tax expenditures is evident in the tax expenditure reports of Germany, the United States and the United Kingdom. [ 2 ] They refer to tax expenditures as "subsidies" or "special exceptions in the tax code that serve programmatic functions" [3] in their efforts to identify tax expenditures within tax concessions. Further, each of these three countries uses a form of the two-step approach to identify tax expenditures. This distinction between tax concessions and tax expenditures is pursued in this paper using an approach similar to that of Germany, the United States and the United Kingdom. It is worth emphasizing that this paper does not seek to limit the number of tax concessions that are reported, but rather to provide guidance with respect to their interpretation. Accordingly, the goals of the paper are to:

In section II of this paper, various theoretical approaches to defining the benchmark are discussed. Section III identifies elements of the tax structure that are absolutely necessary and, as such, are considered components of the benchmark. Section IV examines the issue of specific types of tax concessions necessary to the overall structure of the tax system, and highlights those that are in place for this reason. Section V discusses the experiences of other countries in defining and reporting tax expenditures. Section VI develops possible criteria for determining which tax concessions are tax expenditures. Section VII presents a classification system for tax expenditures based on these criteria. Some concluding remarks complete the paper.

II. The Tax Concession Debate: What Is the Benchmark?

A tax concession represents a deviation from a benchmark tax system. However, there is a lack of consensus in determining the benchmark tax system. This lack of consensus has been attributed to the fact that "in a many-consumer economy, there is an indefinitely large number of efficient tax structures, no one of which has prior claim to being the best." [4] As a result, a variety of possible benchmarks have been developed. These benchmarks vary in their treatment of certain fundamental components such as the tax unit, tax base and tax period. Each benchmark is unlike the others in how it defines one or more of these components and, therefore, each benchmark produces a different set of tax concessions.

The following paragraphs discuss the possible approaches to defining the benchmarks for each of the personal income tax, the corporate income tax, and the Goods and Services Tax/Harmonized Sales Tax (GST/HST). They will also illustrate how the list of tax concessions changes with each benchmark.

Personal Income Taxation
Haig-Simons Comprehensive Income Base

When the discussion of the tax expenditure concept began in the 1960s, there was some consensus that the appropriate norm for the personal income tax was the Haig-Simons (or Schanz-Haig-Simons) comprehensive income base.  [5] The Haig-Simons Comprehensive Income Base requires the taxation of real current additions to purchasing power, or real increases in wealth. Under a Haig-Simons base, all income must be included as taxable income. This means that income from all sources, including labour income, rents, dividends, interest, transfers, accrued capital gains, imputed rent, value of household services, gifts and inheritances, must be taxed uniformly. In addition, Haig-Simons allows for the deduction of expenses to earn income and requires indexation for inflation in order to measure real, not nominal, additions to wealth.

The problem with using the Haig-Simons ideal is that it is difficult to implement. Even though theory dictates that no forms of income should be tax-exempt, in practice there may exist measurement, compliance or collection issues preventing full and comprehensive income taxation. As a result, a long list of tax concessions would result. For example, measurement problems are likely to be encountered when attempting to tax the imputed value of household services. Not only would it be difficult to place values on certain types of household services, but it would also be virtually impossible to accurately measure the amount of time each individual spent doing these activities. Under the Haig-Simons base, non-taxation of household services would be considered a tax concession, the value of which would be unknown.

Even when measurement is not the issue, compliance problems may lead to an unduly long list of tax concessions. For instance, the Haig-Simons base requires the taxation of all gifts received and allows for the deduction of all gifts given. However, it is highly unlikely that gifts under a certain value will actually be reported and included in taxable income. Furthermore, even if the non-taxation of gifts were reported as a tax concession, it is unlikely that the calculated tax concessions would be very accurate as it is impossible to know the true value of lost tax revenue resulting from unreported gifts.

Another practical problem with using the Haig-Simons base as the benchmark involves tax collection. For example, Haig-Simons taxation demands that capital gains be taxed on accrual, not on realization. Lost tax revenue from taxation upon realization is considered a tax concession. However, tax collection becomes a problem when taxing capital gains on accrual. The tax payable from accrued capital gains is usually measurable (with the possible exception of works of art), but taxpayers may encounter problems raising funds to pay the tax owing, as even though gains have accrued, taxpayers do not realize the gains until they sell the assets. This means that, in order to pay the tax, they may be forced to sell assets they would otherwise retain. In this case the tax would alter economic behaviour and result in an inefficient allocation of economic resources. For this reason, despite the Haig-Simons comprehensive income mandate, taxation of accrued income actually leads to problems of economic distortions. Auerbach suggests a solution for this tax collection problem whereby the tax liability could be realized only upon sale of the asset, but interest could be charged for the deferral of tax from the date of accrual of the gain. [6] Under these conditions, tax collection is less of a problem, but the cost of tax administration - tracking gains and losses and calculating interest - is substantially higher.

Labour Income Base

An alternative income base would involve considering only labour income. Under a labour income base, unlike a comprehensive income base, income from investment and savings is exempt from taxation. Investments are made from after-tax labour income, but the returns generated by the investment are not subject to tax. This means that, under the labour income base, the present discounted values of both current and future consumption are the same. As a result, the appeal of the labour income base is that it does not distort decision making between present and future consumption.

Unlike some other approaches, a labour income base treats human capital in the same manner as other forms of capital: it considers human capital as a non-registered asset. That is, investment in human capital is made from after-tax dollars, but no tax is imposed on future gains, such as increased earnings, that result from the investment. [7] Thus, it does not allow for the deduction of investment expenses in human capital, such as tuition fees and other educational expenses. Furthermore, it requires the taxation of potential earnings that are foregone in order to increase the value of human capital, by going to school, for example. In addition, increases in earnings attributed to the accumulation of human capital are not subject to taxation. [8] Deviations from these rules are regarded as tax concessions. In the real world, both the taxation of foregone earnings and the non-taxation of increased earnings due to human capital investment are infeasible and impractical. Accordingly, the resulting list of tax concessions due to the treatment of human capital as a registered, rather than non-registered, asset would be of limited use for comparison and policy purposes as the required treatment of human capital under the labour income base is impractical.

Consumption Base

Instead of using an income base for taxation, some theorists advocate the use of a consumption or expenditure base. Under a consumption base, people are taxed on what they spend rather than on what they earn, where spending is measured as the difference between income and saving. The benefit of a consumption tax base is that it does not discriminate between taxpayers that receive and spend income at different times. As a result, some argue that the consumption base may better reflect a taxpayer's ability to pay than does the Haig-Simons base. [9]

Under a consumption tax base, all assets, including owner-occupied housing, are given registered treatment. Using the example of owner-occupied housing, deductions for down payments, mortgage payments, and spending on repairs, maintenance and improvements would be allowed, and proceeds from the sale of the owner-occupied housing would be subject to taxation. Imputed rent, however, would be taxed, as it is part of consumption. In current practice, the reverse is considered the norm: no deductions are allowed and proceeds from the sale of owner-occupied housing are not subject to tax. As a benchmark, the consumption tax base may be problematic in certain cases as it has the potential to raise measurement and compliance issues when determining imputed rent and spending on maintenance and repairs.

Lifetime Consumption Base

A lifetime consumption base differs from an annual consumption base in that it allows income averaging and taxation based on lifetime consumption. The lifetime consumption base does not require taxation of imputed household services (as in the case of Haig-Simons) or registered treatment of human capital (as in the cases of both the labour income base and the Haig-Simons base). Nevertheless, several problems arise when comparing the existing tax structure to the lifetime consumption base. This base requires the taxation of gifts and inheritances, as they are deemed part of consumption. The problems this raises are similar to those found under the Haig-Simons base. Secondly, despite its theoretical appeal, in reality it is virtually impossible to impose taxes on total lifetime consumption. As a result, if the benchmark were taken to be lifetime consumption, then tax concessions would be calculated as deviations from the lifetime consumption benchmark. These calculated tax concessions would have little meaning and would not be usable for the purposes of comparison with direct spending. Compounding this problem is the difficulty involved in annualizing taxes for the purpose of comparison with actual tax collections. For these reasons, adoption of the pure lifetime consumption base may not be ideal.

Corporate Income Taxation

Determining the base of taxation for corporate income leads to a smaller range of options and views. There are two principal trains of thought regarding corporate income taxation.


Some theorists believe that corporate taxes may be superfluous and that only the individual should be taxed. In this case, the ideal tax base would require taxation of all sources of income at the individual level. The alternative approach is set out in the 1997 Canadian budget, which provides three reasons for business taxation. [10] First, businesses benefit from public goods and services, and so businesses should pay taxes in order to compensate for "the cost of a public good or service that confers a benefit on the firms." [11] Second, without a corporate tax, individuals would be able to defer paying taxes on capital gains or income by investing in a corporation and having the income or capital gains accrue within the corporation. Third, by taxing corporations, income that is earned in Canada and accrues to foreigners and foreign corporations operating in Canada can be taxed.

Haig-Simons Comprehensive Income Base

Those who favour taxing the income of the corporate sector often refer to the Haig-Simons income base as the appropriate corporate benchmark tax structure. The Haig-Simons base is the same for corporate income as for personal income in that it requires the taxation of comprehensive income with adjustments for inflation. The corporate income tax can be viewed as a tax on economic rents and on the return on equity capital. This translates into a withholding tax on shareholders' Haig-Simons income.

There are two main problems with adopting the Haig-Simons base as the ideal corporate tax base. First, Haig-Simons requires indexation for inflation so that only real gains in wealth are subject to taxation. Although this may be a desired feature of the tax base, there are serious problems implementing it because it is difficult to accurately measure real gains as opposed to nominal gains. Second, it makes no provision for the cyclical nature of business. Taxes are assessed based on real accrued gains in wealth. The Haig-Simons base indicates that losses should be subtracted from gains so that only real increases in wealth are subject to taxation. However, Haig-Simons does not address what happens if losses exceed gains and there is a net decrease in wealth. The logic of Haig-Simons implies that if losses exceeded gains, a credit would be granted for the negative income tax due. However, there is no clear indication as to whether Haig-Simons would require this. Without a provision for the cyclical nature of business, carry-overs allowed in the tax system may be considered tax concessions when compared with the Haig-Simons benchmark, even though they are in place to deal fairly with corporate losses and to recognize the existence of the business cycle.

Goods and Services Tax/Harmonized Sales Tax
Consumption Base

The GST/HST is a value-added tax, broadly based on consumer goods and services. The most logical benchmark is, therefore, a consumption base. Under a consumption base, all goods and services consumed are subject to a uniform tax rate. If any good or service is either excluded from the tax or subject to a tax rate different than the statutory rate, there will be a tax concession. This benchmark allows no provisions for horizontal or vertical equity as every individual is subject to the same tax rate on all goods and services.


The examination of the theoretical arguments for the use of various tax bases and their impact on the determination of tax concessions has revealed that there is no single correct benchmark. With the exception of the GST/HST, where the use of the consumption base is clearly the most appropriate benchmark, there are plausible arguments for and against defining the benchmark in each of the aforementioned ways. As a result, none of these proposed benchmarks is ideal in its pure form. The use of any of these bases in their pure form would create tax concessions that would be of limited use for the intended function of tax expenditure reports - comparative analysis and policy examination.

The difficulty of selecting a benchmark tax structure is integral to the difficulty in defining a tax concession, since a change in the benchmark results in a change in what is classified as a tax concession. While it is generally accepted that a tax system should be efficient, equitable, simple to administer and easy to comply with, theory provides no indication as to which requirement should take precedence and no rule as to how to balance these criteria. The focus of this paper, however, is not on selecting the benchmark. The discussion of the difficulties involved in selecting a benchmark tax structure was provided to demonstrate that the intricacies of defining tax expenditures begin before the notion of a tax expenditure is even discussed. For the purposes of this paper, the benchmark tax structure is taken from Tax Expenditures: Notes to the Estimates/Projections. [12]

III. Components of the Benchmark

This paper does not propose criteria for defining tax concessions, but rather takes the tax concessions as they have been defined in Tax Expenditures: Notes to the Estimates/Projections. It recognizes that there are certain elements that make up every tax system, which must be defined and understood as integral to the structure of the system. These components do not constitute tax concessions. The Canadian tax system defines these components - the tax unit, the taxation period, the tax rate structure and the treatment of inflation - as follows.

Tax Unit

While it is accepted that there must exist a definition of the tax unit, the "ideal" measurement of the tax unit is not dictated by theory. In the Canadian personal income tax system, the benchmark tax unit is defined to be the individual. An alternative to this might be to define the tax unit as the family, as is possible in the United States. The existing Canadian tax system levies taxes based on the income of an individual, even though some tax concessions depend on family income. However, if the benchmark tax unit were defined to be something other than the individual, such as the family, but in practice the unit of taxation remained the individual, then the calculated tax concessions would be of very limited use for policy purposes.

In the case of the corporate income tax system, the benchmark tax unit is the corporation. This definition of the corporate tax unit is most highly correlated to the treatment of the tax unit in the existing tax system. Even though there are occasional measures to recognize inter-corporate ties, such as the deferral of capital gains through rollover provisions, in general, taxes and tax provisions are based on the single, legal, corporate entity. If anything but the single corporation were used as the benchmark tax unit to calculate and report tax expenditures, the amounts would be of limited policy relevance and would be difficult to use for comparison purposes because the unit of taxation in practice would remain the individual corporation.

Under the GST/HST, the ideal tax unit is taken to be the consumer of goods or services in Canada, whether the consumer is an individual person, business or corporation. One of the related tax provisions, the GST/HST credit, depends on family income. In this respect it more closely resembles an expenditure program, for which qualification is generally based on family income. In the existing system, however, all other provisions deal with the consumer as defined above.

Taxation Period

The tax period is another integral part of the benchmark tax structure. In Canada, the benchmark tax period for the personal income tax and for the GST/HST is the calendar year. For the corporate income tax structure, the taxation period is the fiscal period ending in the calendar year. This is more of a convention than a rule, established for ease of accounting. It is, however, consistent with the Haig-Simons benchmark for income taxation. As noted earlier, proponents of the lifetime consumption benchmark argue that annual income does not accurately define ability to pay, as income and consumption patterns change throughout the lifetime of an individual. In their view, the taxation period should be considerably longer than a year in order to allow for income averaging.

Tax Rate Structure

It is also necessary for the benchmark to define the tax rate structure. Deviations from the benchmark rate structure would then be considered tax concessions. For the Canadian corporate income tax, the benchmark rate is defined to be the same as the current standard rate of corporate income tax, including the surtax and the provincial abatement. Any preferential rates, such as the low tax rate for manufacturing and processing and the low rate for small businesses, represent a departure from the benchmark and are considered to be tax concessions. The GST/HST also has a flat benchmark tax rate. Any deviation from the statutory rate of 7 per cent is a tax concession.

In the Canadian personal income tax system, the existing progressive tax structure, including surtaxes, is considered the benchmark tax structure. Although, in order to be consistent with the corporate income tax and the GST/HST, it would be best to use one single rate as the benchmark, if this were followed in the personal income tax, problems regarding the measurement and interpretation of other tax concessions would arise. If the benchmark rate were set at a flat rate, as it is in the corporate income tax and the GST/HST, then the progressive structure of the personal income tax system would lead to large tax concessions. Furthermore, if the benchmark tax rate diverged greatly from the existing tax rate structure, then any tax concessions would not be accurately comparable to direct spending. Inclusion of the progressive structure in the benchmark rate means that vertical equity is recognized as a key component of the personal income tax benchmark.

Treatment of Inflation

For the purposes of identifying tax concessions, the treatment of inflation plays a minimal role. In contrast, when estimating the value of tax concessions, how the benchmark treats inflation is important. For example, as Canada moves to a fully indexed personal income tax system in 2000, the value of the tax concessions associated with credits and personal amounts will automatically increase in nominal terms. In addition, a measure in the tax system requiring only two-thirds of capital gains to be included in taxable income is designed partly to help avoid taxation of purely inflationary gains. But unless the true difference between nominal and real gains is one-third of total capital gains, there will be either a positive or negative tax concession. The exact difference between nominal and real gains, however, is a measurement issue and can be dealt with when estimating the value of tax concessions.

The indexed system that is in place as of 2000 is close to the Haig-Simons benchmark, which posits that only real additions to wealth should be taxed. Thus, under that benchmark, lack of indexation would be considered a negative tax concession. In contrast, strict adherence to proportional taxation on a consumption base does not require any indexation.

IV. Other Tax Features Included in the Benchmark

In addition to the basic components of the benchmark, there are certain other features that are elements of a fair tax system. Measures to include these features should be considered part of the benchmark and should not be considered tax concessions.

Measures to Reduce or Eliminate Double Taxation of Income

It is obvious that the same earned income should not be taxed more than once. This may, however, effectively arise if that income is taxable under two tax systems. In order to determine whether or not a tax measure helps to reduce double taxation, it is essential to examine the integration of the various tax systems. For example, the dividend gross-up and credit allow individuals to gross up dividends received from taxable Canadian corporations, include this amount in income and then claim a tax credit for a portion of the grossed-up amount. This provision recognizes that some of the taxes on dividend income have already been paid at the corporate level and that this income should not be taxed again under the personal income tax. Thus, the gross-up and credit reduce double taxation of income and increase the neutrality of the overall tax structure; therefore, they should not be considered as tax concessions. Despite attempts to reduce double taxation, under-integration may still exist. As a result, any remaining instances of double taxation of income should be considered negative tax concessions. In contrast, if the dividend gross-up and tax credit were calculated to be greater than the tax on dividends paid by the corporation, then a situation of over-integration would exist and this would be considered a tax concession.

Similar reasoning lies behind the foreign tax credit. It recognizes that income earned abroad has already borne tax and so provides a tax credit to avoid this same income being taxed again in Canada. However, if the credit were larger than the real income tax paid, then the difference would be considered a tax concession. In contrast, if the credit were smaller than the actual income tax paid, then a situation of under-integration of the tax systems would exist. In this case, the difference between the tax credit and the actual income tax paid would be considered a negative tax concession.

Loss Carry-Overs

Loss carry-overs recognize the cyclical nature of business and acknowledge that losses suffered in one period may be offset by gains in another. Income in one year may not be an accurate indicator of ability to pay taxes due to substantial losses in a previous year. Consequently, carry-overs allow losses to be applied to past or future income in order for individuals, businesses and corporations to smooth income to a certain degree over the business cycle. In Canada, tax provisions that recognize the impact of business cycles on income include the carry-over of non-capital losses, the carry-over of net capital losses, and the carry-over of farm losses and restricted farm losses.

Deduction of Expenses to Earn Income

In general, businesses act to maximize profits, where profits are defined to be revenue less the costs of generating that revenue. It is generally accepted that taxes should be assessed on income net of expenses incurred to earn income, rather than gross income.

Under both the personal and corporate income tax systems, legitimate expenses incurred to earn income are deductible from gross income in order to determine the appropriate base of income taxation. Nevertheless, the definition of legitimate expenses in the personal income tax is subjective and, in some cases, controversial. Certain expenses such as child care expenses, attendant care expenses, and meals and entertainment expenses are viewed by some as legitimate costs of doing business, but are seen by others as spending resulting from personal choices. Consequently, the classification of certain expenses as tax provisions essential to the benchmark tax structure is not absolute and indisputable.

V. Tax Expenditures: Country Experiences

An examination of tax expenditure reports across countries shows that their structure, costing and contents differ widely, revealing the absence of a universally accepted definition of what constitutes a tax expenditure. In Canada, the current reporting of tax expenditures is very broad, covering any tax concession that is not considered one of "the most fundamental structural elements of each tax system" and implying a very wide definition of tax expenditures.  [13] In addition, if there is any doubt as to whether or not a tax concession fits this definition, revenue losses resulting from the reduction are still reported, but under a category entitled "memorandum items." Not every country produces as comprehensive a list of tax concessions as Canada does. In order to determine the definition of a tax expenditure in Canada, it may be useful to consider the definitions of tax expenditures currently in use in other countries.

The United Kingdom attempts to divide tax reliefs into two categories. Those reliefs that are alternatives to, and have similar consequences as, public spending are referred to as tax expenditures. Those forms of tax relief that are either an integral part of the tax structure or simplify administration or compliance are called structural reliefs. Structural reliefs include measures such as the personal allowance and double taxation relief. Tax expenditures include measures such as the exemption of capital gains on the sale of a principal residence and the exemption of the first £8000 of reimbursed relocation packages provided by employers. However, the government acknowledges that "the distinction between structural reliefs and tax expenditures is not always straightforward," [14] and include a third category of tax reliefs, which consists of tax concessions that combine elements of both the structural and expenditure categories. Into this somewhat open-ended category, they put tax concessions such as age-related allowances and the tax exemptions for child benefits and disability living allowances.

France defines a tax expenditure as a tax measure that results in a loss of tax revenue for the state and a corresponding easing of the tax burden on the taxpayer that would not have occurred under the application of general tax law. Nevertheless, additional criteria including duration, general applicability and neutrality are also used to define tax expenditures in cases of uncertainty. Cases of uncertainty, however, are rare.

Unlike France, Germany does not have a formal definition of tax expenditures. It examines every tax concession within the context of the tax system as a whole to determine whether or not it is a subsidy. Consideration is also given to the scope of the tax measure: "The larger the circle of taxpayers which benefits from a tax concession, the less it tends to be considered a subsidy in the report." [15] In addition, Germany uses consistency as a criterion for identifying tax expenditures. The issue here is whether a tax measure is applied consistently or in an ad hoc manner. Those provisions that are considered ad hoc are classified as tax expenditures.

With consistency as the only criterion, however, even if a tax concession brings the tax system closer to the benchmark tax structure, it could still be considered a tax expenditure. Bruce [16] cites the example of the (now defunct) Canadian investment income deduction as an ad hoc response to a flaw in the existing tax structure in that, without the deduction, increases in investment income due only to inflation would have been taxed. In this case, an ad hoc provision actually moves the system closer to the Haig-Simons ideal, but under the German classification system it would be considered a tax expenditure. Bruce therefore concludes that "consistency is not a sufficient condition" for determining whether or not a tax concession is a tax expenditure; the tax concession must also accord with the norm. [17]

Germany also states the official reason for the tax concession. After determining the list of tax expenditures, it classifies them by sector and then further classifies them by the aim of the tax measure. By denoting the purpose of the tax measure as one of support, adjustment aid, productivity aid or other purpose, Germany attempts to ascertain the amount of government financial support given to each objective.

Although according to the United States Senate Committee on the Budget, tax expenditures "may, in effect, be viewed as spending programs channelled through the tax system," [18] the U.S. definition of tax expenditures is somewhat unique among Organisation for Economic Co-operation and Development (OECD) countries. Unlike other OECD countries, the United States uses two different tax structures to report two sets of tax expenditures: the normal tax structure and reference law. Tax expenditures under the normal tax structure are similar, but not identical, to those reported in Canada. They include any deviation from the basic structure of the tax system, where the basic structure includes both a standard deduction and deductions for expenses to earn income, and is not limited to a specific rate structure or unit of taxation. The normal tax base, however, strays from a Haig-Simons comprehensive income base as it allows for the following: income being taxable when realized rather than on an accrual basis, corporate income being taxed separately from individual income, and the tax system being based on nominal values, i.e., there is no adjustment for inflation.

Tax expenditures under reference law are more limited. They more closely reflect the existing tax laws and consist only of "special exceptions in the tax code that serve programmatic functions." [19] In order for a tax concession to be classified as a tax expenditure under reference law, two conditions must be satisfied. First, "absent the special provision, the tax laws provided general rules to enable a taxpayer to determine his income tax due and payable," where the general rules constitute the reference law. [20] That is, the tax concession must not be necessary for the calculation of tax payable. Second, the tax concession must apply "to a sufficiently narrow class of transactions or transactors" such that its objectives could be realized through program spending using government funds. [21]

The reference law baseline differs from the normal baseline in several ways. Reference law recognizes as part of the baseline the varying tax rates within the corporate tax structure. Thus, the maximum statutory tax rate is not recognized as the benchmark, as it is under the normal baseline. As a result, preferential tax rates applied both to the first $10 million of corporate income and to capital gains are not considered tax expenditures under reference law. Reference law also does not include in its benchmark the taxation of transfer payments from the government to individuals. Exclusion of these cash transfers from income, therefore, does not result in tax expenditures under reference law. In general, tax expenditures under the reference law baseline are also tax expenditures under the normal tax base line, but the reverse is not always true. Examples of tax concessions considered tax expenditures under the normal tax base line but not under the reference law baseline include accelerated depreciation on rental housing, buildings other than rental housing, and equipment and machinery; and exclusion of scholarship and fellowship income.

The United Kingdom, France, Germany and the United States all maintain the common goal of establishing a more accurate definition of a tax expenditure. The actual definitions developed by each country vary considerably. The United Kingdom and Germany, however, introduce an interesting method of determining tax expenditures: examining each tax concession to ascertain whether it is an alternative to public spending or a subsidy. This idea of defining a tax expenditure as an alternative to public spending or as a subsidy indicates that, in order for a tax concession to be a tax expenditure, the concept of deliverability must be examined.

VI. Developing the Criteria for Defining Tax Expenditures

In determining the criteria for defining tax expenditures, both theory and practice provide useful insights. In Surrey's definition, tax expenditures are "governmental financial assistance programs…carried out through special tax provisions rather than direct Government expenditures." [22] Together with the United Kingdom definition of tax expenditures as alternatives to public spending and the German definition of tax expenditures as subsidies, this means that tax expenditures are direct spending programs delivered through the tax system. [23] A reasonable criterion is that a tax concession should be considered a tax expenditure if it can be delivered equivalently through program spending. "Delivered equivalently" means that a tax concession be delivered outside the tax system without altering program or administrative costs or modifying the distribution of benefits. If a tax concession cannot be delivered equivalently it is not a tax expenditure, but rather a tax reduction.

Distribution of Benefits

The most difficult aspect of "equivalent deliverability outside the tax system" ("deliverability" for short) to satisfy is retention of the same distribution of benefits. Conceptually, the distribution of benefits can be exactly the same only in the case of refundable tax credits. Under both the corporate and personal income tax systems, in theory, non-refundable tax credits may be deliverable outside the tax system because they do not affect the calculation of taxable income. For example, if the same people who claimed a non-refundable credit under the tax system were instead to receive the benefit under a spending program, then the distribution of benefits would be the same under both systems. Because the credit is non-refundable, however, it is only valuable to those with positive taxable income. Consequently, if an individual's total non-refundable tax credits exceed total taxes payable, then some credits will not be used up. An individual who qualifies for a credit, but cannot claim the full credit due to insufficient taxable income, does not receive the benefit.

For example, suppose Individual A claims the basic personal credit in the amount of $1,229.27 (17 per cent of $7,231). However, Individual A's calculated income tax owing is only $1,000. Because the credit is non-refundable, Individual A can claim only $1,000 of the credit. The remaining $229.27 cannot be claimed due to insufficient levels of taxable income and income tax payable. As a result, Individual A will not receive the full benefit of the tax credit.

If this basic personal tax credit were replaced by a direct spending program that retained the same distribution of benefits, Individual A would still only be able to receive partial benefits. However, a different taxpayer, Individual B, with income tax payable in the amount of $2,000 would be able to claim the credit in its entirety. To match this distribution using direct spending would involve designing a program that provided low or zero benefits to lower-income recipients and higher benefits to higher-income recipients. This type of program is unlikely to be acceptable in practice for vertical equity reasons.

If, instead, the spending program granted Individual A full benefits of $1,229.27, then clearly the distribution of benefits would change. Not only would Individual A, who received only partial benefits under the tax benefit, now receive full benefits under the spending program, but so would other individuals who previously received zero benefits under the tax concession. Adopting this approach to direct delivery would result in a program whose benefits would be like those of a refundable tax credit. That is, more people would qualify for this benefit. Consequently, for some tax concessions like   non-refundable credits and deductions, the distribution of benefits may change, making those tax concessions only potentially deliverable outside the tax system.

Program Cost

For many tax concessions there are conceivable replacement spending programs which provide the same total benefit to the overall population. This allows program costs to remain constant. However, program costs are highly dependent on the distribution of benefits. If the primary goal is to retain the same program costs, then it is likely that the distribution of benefits will change and, consequently, the original objectives of the tax concession will not be achieved. Only if the same individuals receive the same benefits will the objectives be achieved at the same program cost. As shown above, if the distribution of benefits changes, it is most likely that more people will qualify for the benefit under the spending program than through the tax concession. Thus, if the distribution of benefits changes, then the original objectives of the tax concession will only be achieved at a higher program cost.

Some tax credits are deliverable outside the tax system as the program costs are almost identical and the distribution of benefits is likely to be the same. For instance, the tuition fee credit, accompanied by its infinite carry-forward, can be viewed as almost equivalent to program spending. Although it is non-refundable, the credit can be carried forward indefinitely to offset future taxable income. Only if the qualifying individual never has taxable income will the benefit from the credit not be realized.

Despite the carry-forward, the tuition credit is not fully, but "almost fully" deliverable outside the tax system. If the tuition credit were delivered outside the tax system, the benefits would be realized in the same year that the tuition was paid. However, for an individual that did not use up the tuition credit in the tax system due to insufficient income, but carried it forward one or more years, the value of the credit would not be identical to the value of a credit claimed in the year that tuition costs were incurred. The value of the credit claimed in future years would have to be discounted to the present value and would be less than the value of a credit claimed in the year the costs were incurred. In this case, the program costs of delivering a tuition benefit would be slightly higher than the costs of the tuition credit in the tax system. As the difference in benefit levels and program costs between the two delivery methods is very small, the tuition credit combined with the carry-forward provision can be considered deliverable outside the tax system. This argument applies to all non-refundable credits with a carry-over provision.

However, delivery of a tax concession through direct spending cannot always be achieved at the same or even similar program cost. Direct delivery of concessions such as the spousal credit, equivalent-to-spouse credit, dependant credit, age credit and basic personal credit would likely result in increased program costs or, if the distribution of benefits is considered, a change in the distribution of benefits. Because these tax concessions are all non-refundable credits without a carry-over provision, they are wasteable credits. That is, if the credit is not claimed in full in the year it is earned, then the full benefit from the credit will never be realized. In order for program costs to remain the same under direct delivery, it is likely that the administration costs of determining the exact benefit for each individual would be prohibitive.

Administration Cost

For a tax concession to be fully deliverable outside the tax system, it needs to have similar administration costs. In some cases, such as the refundable Canada Child Tax Benefit, the administration costs will be similar, as the tax concession could be converted into a program, and delivered as it is currently, in the form of a cheque.

However, some tax concessions could undergo very substantial increases in administration costs if they were removed from the tax system and delivered directly. For example, a spending program cannot deliver the provision for non-taxation of capital gains on principal residences without very significant increases in administration costs. The recipients of the benefit change from year to year and the amount of the tax benefit depends on the marginal tax rates faced by the taxpayer. Because the value of the benefit varies from taxpayer to taxpayer depending on their position in the tax rate structure, large expenses would have to be incurred in order to determine both the qualifying individuals and the value of the benefit to each individual.

In general, in the personal income tax system, the administration costs of delivering deductions and exemptions as direct spending would be very substantial. The purpose of both deductions and exemptions in the tax system is to "arrive at an accurate assessment of income, i.e., to determine the proper tax base for a taxpayer."  [24] Deductions and exemptions reduce gross income to taxable income and, therefore, apply and have value only to those with positive gross income. However, due to the progressive nature of the personal income tax system, the value of a deduction or exemption varies according to the individual's marginal tax bracket. The value of a deduction or exemption is highest for those in the top marginal tax bracket and falls to zero for those who do not pay any taxes.

For example, the provision for non-taxation of employer-paid insurance benefits for group private health and dental plans exempts the employee from paying tax on these benefits. The exemption is considered undeliverable because of the very fact that it is used to determine taxable income, and hence, the value would vary across individual employees according to their marginal tax brackets. As a result, it would be virtually impossible to duplicate the benefits from this tax concession in a spending program.

Consequently, deductions and exemptions in the personal income tax system are considered infeasible for delivery outside the tax system because of overwhelming administration costs.

Unlike the personal income tax system, the corporate income tax system and the GST do not contain progressive tax structures. As a result, deductions and exemptions have the same value to all who qualify for them and the administration costs of delivering them outside the tax system would not be prohibitive.

VII. Classification System

Each tax concession can be assessed based on the above criteria to determine how deliverable it is outside the tax system. Because of the variability in types of tax concessions and the limited ability of theory to direct their classification, it is not possible to consider each tax concession as simply deliverable or non-deliverable. Consequently, it will not be possible to classify each tax concession into just two categories, those being tax expenditures and other tax concessions. Due to varying levels of deliverability, there are some uncertain cases that require individual determination. As a result, the system of classification that emerges from this set of deliverability criteria consists of four levels, but with a degree of uncertainty about the appropriate classification of some provisions.

Type I Tax Expenditures

The first classification level comprises those tax concessions that are closest to direct expenditures. They are broadly deliverable outside the tax system as their delivery through direct spending leads to no significant alteration in the distribution of benefits and no substantial changes in program and administration costs. In general, Type I tax expenditures consist of refundable tax credits and rebates, as qualification for these tax provisions and their delivery through direct spending would remain the same as their delivery through the tax system.

Type II Tax Expenditures

Type II tax expenditures are tax concessions that are deliverable outside the tax system with a similar distribution of benefits and program costs. Direct delivery of these tax concessions will, however, result in measurable increases in administration costs. In general, Type II tax expenditures comprise non-refundable tax credits with a carry-over and/or transferability. In addition, because of the single tax rate in the corporate tax system, Type II tax expenditures also include many exemptions and deductions in the corporate tax structure. Under the GST/HST, Type II tax expenditures comprise most zero-rated goods and services.

Type III Tax Expenditures

These tax concessions, if delivered through direct spending rather than the tax system, would significantly increase both administration costs and program costs and would substantially change the distribution of benefits. In general, Type III tax expenditures include non-refundable tax credits that have no provision for transferability or carry-over, and therefore might not be claimed in full. In the case of the GST/HST, type III tax expenditures consist of the tax-exempt goods and services which, if delivered directly, would alter the distribution of benefits and increase program costs.

Tax Reductions

Tax reductions are those tax concessions that simply cannot be delivered outside the tax system without incurring substantial and prohibitive increases in administrative costs as well as changes in the distribution of benefits and program costs. Not only would the distribution of benefits and program costs change, as in the case of Type III tax expenditures, but the value of these tax reductions would vary depending on the marginal tax bracket faced by the individual. In general, the tax reduction category comprises deductions, deferrals and exemptions in the personal income tax system. In the corporate income tax system, tax reductions consist of deferrals. Deferrals often have variable lengths, and thus the benefit of the deferral may vary from corporation to corporation.

Part of the Benchmark

This category consists of those tax concessions which were identified earlier in this paper as essential components of the tax system, because they reduced double taxation, ensured the correct measurement of income or dealt with business losses. Because these reductions are essential elements of the tax system, the deliverability criteria do not apply to them.

VIII. Concluding Remarks

This paper highlights the fact that the definition of a tax expenditure is not absolutely precise and that a universally accepted definition does not exist. For this reason, the tax expenditure report provides as comprehensive a set of information as possible. This avoids getting into a controversy about whether an item is a tax expenditure or not. Analysts have found this approach useful.

While some analysts define a tax expenditure as being any deviation from a benchmark tax structure, this paper shows that a deviation from a benchmark can either be a tax reduction or program spending delivered through the tax system. This suggests the need for a two-step approach to defining tax expenditures. The first step requires the identification of tax concessions – deviations from the benchmark tax structure. The second step involves determining which tax concessions are tax expenditures

The analysis clarifies an approach to classifying tax expenditures with the goal of helping readers to analyze the information currently provided in the tax expenditure report. This approach could also be used to change how information on tax expenditures is presented. Comments are invited as to whether such a reclassification would be useful, bearing in mind that the same amount of information on deviations from the benchmark will continue to be provided.


  1. Department of Finance Canada, Budget 1997: Budget Plan, (Ottawa: Public Works and Government Services Canada, 1997), p. 145. [return]

  2. Tax expenditure reports are important publications as they permit comparisons between tax expenditures and direct spending measures. The extent and nature of the uses of tax expenditure information varies among countries. Most countries use it for debating spending and taxation issues in parliamentary committees. Germany also uses the tax expenditure report to examine sectoral subsidies, while the United Kingdom uses it to reconsider the form and method of tax relief and its cost. In the United States it contributes to tax reforms. And in Belgium tax expenditure information is even an element in labour negotiations. [return]
  3. Budget of the United States Government, Analytical Perspectives: Fiscal Year 1998 (Washington: U.S. Government Printing Office, 1997), p. 84. [return]

  4. Robin Boadway and Frank Flatters, "Tax Expenditures and Alternatives for Evaluating Government Activities Conducted Through the Tax System," Tax Expenditures and Government Policy, ed. Neil Bruce (Kingston: John Deutsch Institute for the Study of Economic Policy, 1988), p. 80. [return]

  5. David E. Wildasin, "Tax Expenditures: The Personal Standard," Tax Expenditures and Government Policy, ed. Neil Bruce (Kingston: John Deutsch Institute for the Study of Economic Policy, 1988), p. 137. [return]

  6. A. J. Auerbach, "Retrospective capital gains taxation," American Economic Review 81 (1991), pp. 167-178. [return]

  7. Non-registered treatment of assets is in contrast to registered treatment of assets, where a tax deduction is allowed for the initial investment but, when the asset is sold, both the initial investment and accrued gains are subject to taxation as long as they are not transferred into other registered assets. [return]

  8. James B. Davies and France St. Hilaire, Reforming Capital Income Taxation in Canada: Efficiency and Distributional Effects of Alternative Options (Ottawa: Economic Council of Canada, 1987) p. 36. [return]

  9. Davies and St. Hilaire, Reforming Capital Income Taxation in Canada, p. 7. [return]

  10. Department of Finance Canada, Budget 1997: Tax Fairness (Ottawa: Public Works and Government Services Canada, 1997), p. 16. [return]

  11. Vijay Jog and Jack M. Mintz, "Business Tax Expenditure Accounts: Their Purpose and Measurement," Tax Expenditures and Government Policy, ed. Neil Bruce (Kingston: John Deutsch Institute for the Study of Economic Policy, 1988), p. 191. [return]

  12. Department of Finance Canada, Tax Expenditures: Notes to the Estimates/Projections (Ottawa: Public Works and Government Services Canada, 2000). [return]

  13. Department of Finance Canada, Tax Expenditures, 1999, p. 33. [return]

  14. OECD, Tax Expenditures: Recent Experiences (Paris: Head Publication Services OECD, 1996), p. 98. [return]

  15. OECD, Tax Expenditures: Recent Experiences, p. 66. [return]

  16. Neil Bruce, "Pathways to Tax Expenditures: A Survey of Conceptual Issues and Controversies," Tax Expenditures and Government Policy, ed. Neil Bruce (Kingston: John Deutsch Institute for the Study of Economic Policy, 1988), p. 29. [return]

  17. Ibid. [return]

  18. Congressional Research Service, Tax Expenditures: Compendium of Background Material on Individual Provisions, prepared for the Committee on the Budget, United States Senate (Washington: U.S. Government Printing Office, 1997), p. 2. [return]

  19. Budget of the United States Government, Analytical Perspectives: Fiscal Year 1998 (Washington: U.S. Government Printing Office, 1997), p. 84. [return]

  20. OECD, Tax Expenditures: Recent Experiences, p. 108. [return]

  21. OECD, Tax Expenditures: Recent Experiences, p. 108. [return]

  22. Stanley Surrey, Pathways to Tax Reform (Cambridge, Mass.: Harvard University Press, 1973), p. 6. [return]

  23. Budget of the United States Government, Analytical Perspectives, p. 84. [return]

  24. Wildasin, "Tax Expenditures: The Personal Standard," p. 155. [return]

- Table of Contents - Previous - Next -