MARGINAL TAX RATES WIDELY MISUNDERSTOOD, SAYS CRS STUDY.

                       CRS REPORT FOR CONGRESS

Despite this prominence, however, marginal tax rates are widely misunderstood, both in terms of their meaning and their significance. Although a relatively simple concept, determining the actual, as opposed to the statutory, marginal tax rate faced by taxpayers can be an extremely complex task. The calculation requires consideration of many interacting income tax provisions and, in some cases, other laws and programs.

Some of this complexity is inevitable, but some could be avoided. For instance, the results of many of the tax benefit phaseouts could be accomplished more directly by simply changing the statutory marginal rate structure. Simplification of the system would make actual marginal tax rates easier to determine, improve taxpayer understanding of the tax system, and facilitate the decisions taxpayers face every day -- regarding work, saving, and investment -- in which the marginal tax rate is a relevant consideration.

Because of the substantial complexity in calculating marginal tax rates, some of the research on the behavioral effects of the tax system may be in error. While most of the studies take account of some of the complexities involved in calculating marginal tax rates, most do not take full account of all of the pertinent factors. Moreover, recent research does not take account of taxpayer confusion about marginal tax rates. The tax rate that actually governs behavior is the tax rate the taxpayer thinks he is subject to. Concentrating on estimated rates rather than the perceived rates that actually govern behavior may lead to erroneous conclusions about the effects of marginal tax rates on behavior.

While much of the attention on marginal tax rates focuses on effective tax rates which are higher than the statutory rates, most taxpayers are in fact subject to effective marginal tax rates that are lower than statutory rates. Many taxpayers facing effective marginal tax rates which greatly exceed the statutory rates are found in the lowest income brackets where phaseouts of the earned income tax credit and welfare programs produce exceedingly high effective marginal tax rates.

Based on the available evidence, however, the effect of marginal tax rates on real economic behavior is debatable. While changes in marginal tax rates probably affect the timing of transactions, investment portfolios, and labor force participation of secondary earners, they appear to have relatively little effect on work effort, saving, and investment. This does not imply that marginal tax rates are unimportant. Marginal tax rates distort economic decisions and reduce welfare; higher tax rates produce larger welfare losses. Hence, low marginal tax rates and a broad tax base are goals that maximize public welfare.

CONTENTS

I. What Are Marginal Tax Rates and Why Are They Important?

Marginal tax rates have been the focus of much policy attention over the past dozen years. Reducing marginal tax rates was one of the principal campaign themes of presidential candidate Ronald Reagan in 1980. The Economic Recovery Tax Act of 1981 (ERTA), one of the major pieces of tax legislation during the 1980s, delivered on that promise by reducing rates approximately 23 percent across the board. The Tax Reform Act of 1986, probably the single most significant piece of tax legislation during a decade when many important tax revisions were enacted, further reduced marginal tax rates in exchange for substantial income tax base broadening. Marginal tax rates were further revised in the Omnibus Budget Reconciliation Act of 1990. And during the 1992 presidential campaign, the level of marginal tax rates affecting upper-income taxpayers was one of the main points of contention between the two candidates.

While marginal tax rates have been much discussed and legislated, there is still confusion and uncertainty regarding what they are and what they mean. A marginal tax rate is a simple concept, but application of the concept can be quite complex. The calculation of marginal tax rates for different circumstances can be complicated, requiring consideration of effects of many interacting tax provisions and, in some cases, other laws and programs. A basic understanding of economic theory and research on how markets react to tax changes is also required. And understanding what marginal tax rates mean in terms of their effects on the economy and the importance of those effects involves interpretation of a vast, diverse, and sometimes difficult literature, the implications of which are vigorously debated.

This report explains marginal tax rates and reviews their significance. Section I defines marginal tax rates and discusses their importance. Section II explores the difficulties in calculating marginal tax rates. Section III examines the marginal tax rate effects of a number of provisions of the Federal individual income tax, including phaseouts, special treatment of certain income, and the special treatment for some uses of income. Section IV examines overall marginal tax rates taking into account not only the Federal income tax, but also other Federal taxes and non-tax programs. Section V provides a brief review of and introduction to the economic research on the effects of marginal tax rates on behavior and economic welfare. Finally, Section VI discusses the implications of the analysis in this report.

I. WHAT ARE MARGINAL TAX RATES AND WHY ARE THEY IMPORTANT?

A. Average Versus Marginal Tax Rates

In defining marginal tax rates it is useful to distinguish between an average tax rate and a marginal tax rate because these two concepts are sometimes confused. The two are different conceptually, and in most tax systems they are also different numerically.

An AVERAGE TAX RATE is simply the tax liability calculated as a percentage of income. For example, if a taxpayer's income is $40,000 and he pays $2,000 in tax, then his average tax rate would be 5 percent ($2,000/$40,000). Average tax rates are useful in gauging the level of the tax burden, comparing the level of one tax with another, and comparing taxes paid by different taxpayers or on different activities or products. Average tax rates are frequently used, for example, to discern the distributional effects of a tax by comparing the relative level of the tax burden at various points on the income scale. Such comparisons can show whether a tax system is progressive (average tax rates rise as income increases), regressive (average tax rates fall as income increases), or proportional (average tax rates remain constant along the income scale). The Federal individual income tax, for example, is a progressive tax under which average tax rates rise as income increases.

A MARGINAL TAX RATE, on the other hand, is the rate of tax applicable on an increment of income. For instance, if a taxpayer is in the 31 percent income tax rate bracket and earns an additional dollar of taxable income, he would owe $0.31 in additional income tax ($1 of taxable income times the 31 percent marginal income tax rate). Part of the progressivity of the Federal individual income tax is achieved through a system of graduated statutory marginal income tax rates.

Under the current marginal income tax rate schedule the first $34,000 of taxable income on a joint return is taxed at a 15 percent rate. The next $48,150 of taxable income is taxed at 28 percent. Any taxable income in excess of $82,150 ($34,000 plus $48,150) is taxed at a 31 percent rate. Generally, because of the structure of the individual income tax, a taxpayer's marginal tax rate will be higher than his average tax rate.

B. Statutory Versus Effective Marginal Tax Rates

Statutory marginal tax rates, that is, the tax rates written into the tax law (such as the 15 percent, 28 percent, and 31 percent rates referred to in the previous paragraph), are not necessarily accurate indicators of the actual marginal tax rates faced by taxpayers. Many tax provisions have the practical effect of changing marginal tax rates from those stated in the tax law. For this reason, it is useful to define an EFFECTIVE MARGINAL TAX RATE as the actual rate of tax applicable to an increment of income taking into account all relevant tax and non-tax considerations.

The Federal individual income tax has many provisions which cause the effective tax rate to differ from the statutory rate. As an example, consider the limitation on itemized deductions for high- income taxpayers. Itemized deductions which would otherwise be allowed must be reduced by 3 percent of the amount by which a taxpayer's adjusted gross income exceeds a threshold ($108,450 in 1993). Hence, for taxpayers with income above the threshold who itemize deductions, an extra dollar of income is taxed as if it were $1.03, since, in addition to the extra dollar of income, the taxpayer loses $0.03 worth of itemized deductions. The provision thereby raises the effective marginal tax rate faced by affected taxpayers by 0.93 percentage points. ¹

Provisions in the income tax are not the only factors affecting marginal tax rates. Other taxes and government programs may directly, indirectly, or implicitly raise or lower effective marginal tax rates. In section III, the marginal tax rate effects of a number of income tax provisions as well as other taxes and other Federal programs are discussed.

C. Why Are Marginal Tax Rates Important?

Marginal tax rates are important primarily because of their influence on economic behavior and resource allocation. These influences occur because taxes change relative prices from those that would exist in private markets in the absence of taxation or if all government revenue were raised through lump-sum or head taxes. The relative price changes occur because taxes are not applied uniformly; some things receive favorable tax treatment and other things are entirely exempt.

Under an income tax, for instance, a taxpayer's choice regarding how much time to devote to work versus how much to take in leisure is affected because the wages from work are taxed but the pleasure from leisure is not. An income tax also affects savings decisions. Savings can be viewed as a trade-off between current and future consumption since savings is an amount that could be consumed currently but, instead, is set aside for consumption in the future. An income tax affects this trade-off because the interest earned on savings is taxable, thereby raising the price of future consumption relative to the price of current consumption.

By changing relative prices, taxes affect the decisions people make and the allocation of resources. Economic theory suggests that, except for certain well known market failures (such as monopoly or pure public goods or externalities), ² the resource allocation achieved by the private market without government intervention is the most efficient allocation. This implies that unless the resource allocation effects of a tax offset market failures, they reduce economic efficiency. In general, the higher the marginal tax rates, the larger the relative price distortions and the greater the loss in economic efficiency. Therefore analysis of the level of effective marginal tax rates and differences in these rates across assets or market activities can provide indications of how efficient or inefficient the tax system is with regard to resource allocation.

II. DIFFICULTIES IN DETERMINING MARGINAL TAX RATE

While the concept of a marginal tax rate is relatively clear, in some cases the application of the concept is not straightforward. This section discusses three difficulties in determining marginal tax rates.

A. What is Pre-Tax Income and What is the Tax?

In some cases, the amount of income and/or the tax that should be taken into account in computing the marginal tax rate may not be immediately obvious. Calculation of the marginal social security tax rate provides an illustration of this problem.

An individual who earns an extra $100 of wages will have to pay the social security tax on the income (assuming his wages are not above the social security wage base, which is $55,500 in 1992). The social security tax rate is 7.65 percent (this is the combined rate for old-age, survivors, disability insurance and hospital insurance (OASDI)). Hence, $7.65 will be withheld from the individual's $100 of wages for the social security tax. Based on these facts, the marginal social security tax rate would seem to be 7.65 percent.

But the story is actually more complicated. The social security tax is imposed on both the employee and the employer. The individual's employer also must pay $7.65 on the $100 of wages, so the government collects a total of $15.30 in social security tax.

Despite the employer's payment of half of the tax, most economists would argue that the employer does not, in fact, bear the burden of the tax. That is, the portion of the tax paid by the employer is shifted, either to employees through lower wages, or to customers through higher prices. Assume for purposes of illustration that the entire tax is shifted to employees through lower wages. In this case, it is appropriate to include the entire social security tax -- both the employee and employer portions -- in calculating the marginal tax rate on the employee. This might suggest that the marginal tax rate is 15.3 percent. But this answer is also incorrect.

It is inconsistent to include the employer paid portion of the tax without also including it in pre-tax income. Given that the employer portion of the tax is shifted to employees, then if the social security tax did not exist (or the employer portion of the tax did not exist), the worker would be paid $7.65 more for the work for which he is now paid $100. His $100 wage, therefore, actually represents $107.65 of PRE-TAX INCOME, and this is the figure that should be used in the marginal tax rate calculation. Under these assumptions the marginal social security tax rate is, therefore, 14.2 percent (15.3/107.65). ³ (Even this figure is not the final answer when all considerations are taken into account, as discussed in the next subsection.)

In this example, neither the amount of tax nor the amount of income used in the marginal tax rate calculation are the figures that, at first, might seem to be appropriate. The example illustrates the care required in calculating and using marginal tax rates.

B. Marginal Tax Rates for Taxes With Offsetting Effects

A second complication in calculating marginal tax rates is that some taxes or tax provisions are directly linked to offsetting effects or benefits. The social security tax can also be used to illustrate this point. Payment of the social security tax entitles the taxpayer to future social security benefits. In many instances, these offsetting effects should be taken into account in determining the marginal tax rate.

By definition, a tax is a required payment that does not purchase any directly related benefit. Payments in exchange for something of equal or greater value are not taxes; they are fees or charges. Examples are entrance fees at national parks and local government charges for water and sewerage services.

A taxpayer receives something of value as a result of paying the social security tax: future social security benefits. The transaction is not exactly like a purchase in the private market because the value of the benefits may not equal or exceed the amount of the tax (although in many cases it does). But only the net tax (the tax minus the benefits) should be treated as a tax; the portion of the tax that is offset by benefits is more appropriately regarded as a fee.

A recent article by Feldstein and Samwick takes this approach to measuring marginal social security tax rates. ⁴ Focusing on OASI (old age and survivors' insurance), for one group of lower to middle- income taxpayers (all of whom are subject to the same statutory tax rate) the authors find effective net marginal social security tax rates ranging from 11.2 percent to -1.59 percent. ⁵ The 11.2 percent rate is for female dependent spouses who receive no social security benefits from their marginal wage income because they will receive benefits as a result of their husbands' social security covered earnings. The -1.59 percent marginal tax rate is for a 60 year old male with a dependent spouse; in this case the benefits received exceed the amount of the marginal tax, so the marginal effective tax rate is actually negative.

There are other situations in which offsetting effects influence the marginal tax rate. If a taxpayer's marginal income is used to make a deductible individual retirement account (IRA) deposit, for example, his apparent marginal tax rate would be 0 since the IRA deduction fully offsets the tax on the income. But depositing the income in the IRA brings with it a future tax liability when the funds are withdrawn. It also provides a benefit to the taxpayer in the form of deferral of the tax liability on both the deposit and the investment income in the IRA. Depending on the circumstances, the IRA deposit could be subject to an effective net marginal tax rate ranging from slightly lower than the statutory tax rate to a rate that is substantially negative. ⁶

C. What is the Relevant Margin?

An additional complication in calculating marginal tax rates is determining the relevant margin. It is useful to distinguish between a POINT marginal tax rate and an INTERVAL marginal tax rate.

The point marginal tax rate is the most commonly used concept. It is the marginal tax rate at a specified income level. In simple terms and for applied work, it is frequently described as the tax on the next (or last) dollar of income. ⁷ If earning an additional dollar of income would raise an individual's tax liability by 28 cents, for example, then his point marginal tax rate is 28 percent.

But in many situations, such as considering a new job, the taxpayer may be contemplating large changes in income rather than small changes. In these cases, the point marginal tax rate may not be relevant. Instead, a marginal tax rate calculated over the income increment being considered would be more appropriate. This is referred to as an INTERVAL marginal tax rate.

Consider, for example, a person thinking about a new job that would increase his annual income by $20,000. If he is in the 28 percent tax bracket, his point marginal tax rate is 28 percent. But the higher income may push him into the 31 percent tax bracket. The higher income could increase his tax by $6,000, for example, representing a 30 percent effective marginal tax rate over the $20,000 INCOME INTERVAL. In this case, using the point marginal tax rate in estimating the financial effects of taking the new job would provide misleading results.

Since it is the more commonly used concept, in this report "marginal tax rate" refers to the POINT marginal tax rate unless it is stated otherwise.

III. FACTORS AFFECTING MARGINAL TAX RATES

A variety of factors influence marginal tax rates. First, of course, is the statutory tax rate. As emphasized in the discussion above, however, the statutory tax rate is only one among many elements determining the effective marginal tax rate. The effective marginal rate may also be affected by phaseouts or thresholds. Phaseouts are generally used to reduce and eventually eliminate special benefits such as the earned income tax credit or the individual retirement account deduction. These two phaseouts are discussed further in the subsection below. Thresholds are also frequently used to reduce or limit special benefits for those above specified income levels. For example, the 3-percent reduction in itemized deductions for taxpayers with incomes above a specified threshold was described above (section I B).

The income tax also provides special treatment for some types of income and some uses of income. Phaseouts and thresholds in the income tax generally are used to increase the marginal tax rate, whereas special treatment of a type of income or use of income usually reduces the marginal tax rate. Finally, marginal tax rates can be affected by interactions among several taxes or between taxes and spending programs. Some low-income families, for example, are affected by the phaseout of the earned income tax credit under the income tax and also by the phaseout of welfare benefits, both of which raise their effective marginal tax rate, in some cases to very high levels.

These effects are discussed and illustrated further in the following subsections.

A. Two Examples of Phaseout Provisions

Two provisions have been chosen to illustrate the effects of phaseouts on marginal income tax rates. The first is the phaseout of the earned income tax credit, which affects lower-income taxpayers; the second is the phaseout of deductible contributions to individual retirement accounts (IRAs), which affects middle-income taxpayers.

Phaseout of the Earned Income Tax Credits (EITC)

The earned income tax credit (EITC) is a refundable tax credit designed to offset the social security tax liabilities of low-income working individuals with children and to ensure that families with incomes below the poverty threshold do not have positive income tax liabilities. It consists of three parts; a basic earned income credit; a supplemental young child credit; and a supplemental health insurance credit. In 1993, the basic credit is 18.5 percent for families with one child and 19.5 percent for families with two or more children and is applied against the first $7,750 of earned income. The maximum basic credit for a one-child family would be $1,433.75; the maximum basic credit would be $1,511.25 for families with two or more children. Since the credit is refundable, families with earned income below $7,750 have a negative effective marginal tax rate.

The basic EITC is phased out starting at an income of $12,200 and is reduced to zero when income reaches $23,053. Because the maximum basic credit is phased out over an income range of $10,853, each dollar of income over the phaseout threshold reduces the maximum basic credit by $ 0.1321 for families with one child and by $ 0.1393 for families with two or more children. For example, a family with one dependent child and earned income of $17,000 would have a basic EITC of $799.67. The $799.67 basic credit was calculated by subtracting 13.21 percent of the difference between the family's earned income ($17,000) and the lower phaseout threshold ($12,200) from the maximum basic credit ($1,433.75).

When the phaseout of the basic EITC overlaps the bottom tax bracket it produces an effective marginal tax rate that is substantially higher than the statutory marginal tax rate of 15 percent. For instance, consider the result when a family with one child earns an additional dollar of taxable income. The additional dollar of income would increase tax liability by $ 0.15. At the same time, their EITC would be reduced by $ 0.1321. Hence, earning $1 of additional income costs the family $ 0.2821 in net taxes. Although the statutory marginal tax rate is only 15 percent, the overlap of the EITC phaseout with the tax bracket results in an effective marginal tax rate of 28.21 percent.

The effect of phasing out the EITC on marginal tax rates is somewhat more pronounced for families with two or more children since the phaseout occurs at a slightly more rapid pace than for one-child families. In addition, when the supplemental young child credit and the supplemental health insurance credit are taken into account, the effects on marginal income tax rates are still greater.

Phaseout of Deductible Individual Retirement Account (IRA) Contributions

Deductible contributions to IRAs are phased out over specific income ranges for individuals who are covered by employer-sponsored retirement plans. (Individuals not covered by employer-sponsored plans can make deductible IRA contributions regardless of their income levels). For joint returns, the phaseout of deductible contributions occurs between AGI levels of $40,000 and $50,000. For single returns, the phaseout occurs between AGI levels of $25,000 and $35,000. (Individuals covered by employer-sponsored retirement plans whose incomes exceed these AGI levels can make non-deductible contributions to their IRAs.)

The phaseout of deductible IRA contributions can have a significant effect on marginal tax rates. As noted above (section II B), depending on the circumstances an IRA deposit can be subject to an effective marginal tax rate ranging from slightly lower than the statutory tax rate to a rate that is substantially negative; the reduced effective rate is due to deferral of the tax on the IRA deposit and on the IRA investment earnings. Consider the case of a two-earner married couple whose AGI is $40,000 and who make a $4,000 IRA contribution ($2,000 to each of their accounts) bearing an effective tax rate approximately equal to zero.

Since their AGI is at the bottom phaseout threshold, their $4,000 IRA contribution is fully deductible. If they receive one extra dollar of income, however, their AGI would exceed the phaseout threshold and part of their IRA contribution would no longer be deductible. Since, for joint returns, IRA deductibility is phased out over a range of $10,000 (between AGIs of $40,000 to $50,000), an extra dollar of income means that $ 0.40 ($4,000/$10,000) of their IRA contribution would no longer be deductible.

Hence, an extra dollar of income would be taxed as if it were actually $1.40 ($1 of income plus $ 0.40 in reduced deductions). If the couple were in the 15 percent statutory marginal tax rate bracket, the phaseout would be equivalent to taxing the extra dollar of income at 21 percent ($1.40 times the 15 percent statutory marginal tax rate). If they were in the 28 percent statutory marginal tax rate bracket, their effective marginal tax rate on the extra dollar of income would be 39.2 percent.

It is worth noting that this example shows nearly the maximum effect of phasing out deductible IRA contributions on the marginal tax rate. ⁸ Single individuals who can contribute only $2,000 to their IRAs would experience smaller increases in their effective marginal tax rates as a result of the phaseout of deductible contributions. And all taxpayers whose IRA deductions bear a marginal tax rate higher than zero would also experience smaller increases. The IRA deduction of someone who is close to retirement, for example, may have an effective marginal tax rate only 4 or 5 percentage points below the statutory rate. In this case, eliminating the deductibility raises the effective marginal tax rate only by the 4 or 5 percentage points.

B. Other Phaseouts and Thresholds Affecting Marginal Income Tax Rates

The preceding examples are but two of the many phaseouts and thresholds in the individual income tax that produce effective marginal tax rates differing from the statutory marginal tax rate schedule. Others include:

o phaseout of the child and dependent care credit, which reduces the credit for employment-related child and dependent care expenses from 30 percent to 20 percent over an AGI range of $10,000 to $28,000;

o phaseout of the 15 percent credit for the elderly or the permanently and totally disabled, which reduces the amount to which the credit applies based on the amount of certain pension, annuity, or disability benefits received and the amount by which AGI exceeds certain thresholds ($10,000 on a joint return);

o phaseout of the personal exemption for high-income taxpayers by 2 percent for each $2,500 (or fraction of $2,500) by which AGI exceeds certain threshold amounts ($162,700 on joint returns for 1993);

o phaseout of the rental realty loss allowance, which reduces the $25,000 of passive losses attributable to rental real estate activities that can be used to offset income from nonpassive sources by one half the amount by which AGI exceeds $100,000;

o phaseout of the alternative minimum tax exemption (on joint returns, the basic exemption of $40,000 is phased out over an alternative minimum taxable income range of $150,000 to $310,000);

o the requirement that certain social security recipients include a portion of their social security benefits in taxable income if their benefits and other income exceed certain thresholds ($32,000 on a joint return); and

o the reduction of otherwise allowable itemized deductions for high-income taxpayers by 3 percent of the amount by which AGI exceeds a threshold ($108,450 for 1993 for all tax returns except married filing separately).

This is not an exhaustive list; others could be added. For example, the itemized deductions for medical and dental expenses, gifts to charity, casualty and theft losses, and miscellaneous deductions are all subject to separate AGI-related thresholds in addition to the 3-percent overall reduction of itemized deductions for high-income taxpayers. But this list includes the most significant of the phaseouts and thresholds. ⁹

As stated above, most of these provisions have the effect of raising effective marginal tax rates. The special treatments for some types of income and some uses of income considered in the next two subsections, on the other hand, decrease effective marginal tax rates.

C. Special Treatment for Some Types of Income

Some types of income receive special treatment under the Federal income tax. In most cases, this treatment is exemption; that is, the income is not subject to tax. In a few cases, a special tax rate applies to certain types of income.

The form of income that first comes to mind when tax-exempt income is mentioned is probably interest on State and local government bonds. But there are many other forms of tax-exempt income, some large, some small. In fact, the tax expenditure budget lists 33 types of excluded income other than interest on State and local bonds. ¹⁰ The largest single tax expenditure, for example, is the exclusion of pension contributions and earnings ($56.6 billion in 1993). And trailing not far behind is the exclusion of employer contributions for medical insurance premiums and medical care ($46.4 billion in 1993). Exclusion of social security benefits (for most taxpayers) is another large tax expenditure ($24.5 billion in 1993). But the exclusions range from the exclusion of income earned abroad by U.S. citizens to the exclusion of scholarship and fellowship income. To the extent that an individual's marginal income includes any of these favored types of income, his effective marginal tax rate is reduced.

At different times in the history of the income tax, different types of income have been subject to different tax rates. From 1972 through 1980, for example, wage and salary income was subject to a maximum marginal tax rate of 50 percent, but "unearned" income was subject to marginal tax rates that ranged up to 70 percent. And capital gain income has frequently had special treatment. Currently, long-term capital gain is subject to a maximum statutory marginal tax rate of 28 percent, although the effective marginal tax rate can be higher if the income is affected by any of the phaseouts or thresholds. ¹¹

D. Special Treatment for Some Uses of Income

Some uses of income also receive special treatment under the Federal income tax. If a taxpayer's marginal income is used for deductible expenses, for example, his effective marginal tax rate will be lower than the statutory rate.

In general, higher-income taxpayers have higher itemized deductions. Not all taxpayers, of course, itemize deductions, but for those who do this factor reduces their marginal tax rate. For most taxpayers, for example, higher income results in higher State income tax payments, which are deductible under the Federal income tax (for those who itemize). Higher income may also result in higher charitable contributions, which also are deductible. Higher-income people also tend to own more expensive automobiles (resulting in higher personal property taxes), live in more expensive homes (resulting in higher real property taxes), and have larger mortgages (with larger interest payments) which also generate higher itemized deductions. These relationships are not, of course, associated with a single marginal dollar of income for a specific taxpayer. But they are characteristic of the aggregate data and are relevant to a taxpayer considering a sizeable income increment such as may be associated with a new job, for example.

As mentioned above, some itemized deductions are subject to a floor that is related to AGI. Certain specified miscellaneous itemized deductions, for example, may be deducted only to the extent that they exceed 2 percent of AGI. Medical care and dental care expenses in excess of 7.5 percent of AGI are deductible. And, high- income taxpayers must reduce itemized deductions by 3 percent of the amount by which AGI exceeds a threshold amount ($108,450 for 1993). At the margin, this limitation has the same effect as an AGI-related floor on deductions. The effective marginal tax rate, taking into account the effects of deductible expenses and AGI-related floors, is the following:

                          m = r(1 + a - b)

Most discussions of marginal tax rates do not take into account the potential tax effect of use of the marginal income. As the example illustrates, however, it is an important consideration.

IV. OVERALL EFFECTIVE MARGINAL TAX RATES

Taking into account the interactions among all the various factors affecting marginal tax rates, what marginal rates do taxpayers actually face? This is a difficult question to answer comprehensively, but some evidence can be provided. The discussion first focuses on rates under the Federal individual income tax. Next, marginal rates resulting from the combined influence of all Federal taxes are considered. Finally, the effects of non-tax programs are also briefly discussed.

A. The Federal Individual Income Tax

A recent paper by Lerman and Cilke computed effective marginal tax rates under the Federal individual income tax. ¹² While they provided example calculations showing cases in which effective marginal rates could differ significantly from the statutory rates, in fact, these are not typical. They used the U.S. Treasury Department microsimulation model to calculate effective marginal tax rates for a large sample of individual income tax returns. ¹³ Figure 1, reproduced from their paper, shows one relevant set of results.

FIGURE 1

[figure omitted]

The lines in the graph show cumulative distributions of tax returns (vertical axis) arrayed by effective marginal tax rates on wage income (horizontal axis). The effective marginal tax rates were calculated by increasing wage income on each tax return and comparing the tax at the higher wage with the actual tax on the return. The results include the effects of the deductibility of a higher State income tax, but assume that other deductions (such as charitable contributions) do not change with income.

Results are shown separately for tax returns falling into each of the statutory marginal tax rate brackets. The solid line in the figure, for example, shows the distribution for returns in the 0 marginal tax rate bracket. About 85 percent of these returns have an effective marginal tax rate of 0. As the graph shows, about 10 percent of these returns have negative effective marginal tax rates; these negative rates are attributable to the effects of the earned income tax credit. About 5 percent of the returns have effective marginal tax rates greater than 0. These rates result from the phaseout of the earned income tax credit and perhaps other provisions such as the child and dependent care credit and the alternative minimum tax.

The data plotted in the figure reveal several interesting facts. First, for most taxpayers the effective and statutory marginal tax rates are the same. As stated above, this is so for 85 percent of the returns in the 0 tax rate bracket. For returns in the 15 percent, 28 percent, and 31 percent brackets these percentages are 71 percent, 56 percent, and 12 percent, respectively. Second, for most taxpayers whose effective marginal tax rate differs from their statutory rate, the difference is not large. Even in the 31-percent bracket -- where the rates differ in 88 percent of the cases -- most taxpayers have effective rates within 3 or 4 percentage points of the statutory rate.

Third, for most taxpayers whose effective and statutory marginal tax rates differ, the effective rate is LOWER than the statutory rate. The cases that usually draw attention in policy discussions are ones in which the effective rate exceeds the statutory rate, but these are not in the majority. Furthermore, the proportion of returns with effective rates lower than the statutory rate would be even larger if the calculations took into account likely increases in deductions associated with higher incomes. Finally, there are some taxpayers, albeit few, who face effective marginal tax rates significantly higher than the statutory rates. Returns with the highest marginal rates -- those at 45 percent and above -- are probably affected by several phaseouts and thresholds, some of which can have large effects. Phaseout of the rental realty loss allowance, for example, raises the effective marginal tax rate by 15.5 percentage points for a taxpayer in the 31 percent tax bracket.

An alternative way of examining effective marginal tax rates taking into account all of the factors determining them is to derive estimates based on average tax rates. Effective marginal tax rates are indicated by the slope of the average tax rate line.

The solid line in Figure 2 graphs average Federal individual income tax liability against average income for population income deciles for 1990. The graph shows data for the first nine income deciles and two portions of the top decile. The two highest-income points on the line are for the 90th to 95th percentiles and the 95th to 99th percentiles, respectively. The point for the top 1 percent of income earners is not shown in the figure.

FIGURE 2

[figure omitted]

This graph and figures 3 and 4, are based on data provided by the Congressional Budget Office (CBO). The income definition on which the data are based is CBO's family income measure, which begins with adjusted gross income as defined in the tax code but also adds cash transfers and two business tax imputations. The cash transfers included in family income are social security benefits, unemployment insurance benefits, veterans' benefits, workers' compensation, AFDC, SSI, and other cash welfare benefits. The business tax imputations are for the employer share of the social security tax and for the Federal corporate income tax. ¹⁴

FIGURE 3 AND 4

[figures omitted]

The points for the first two income deciles on the solid line in the figure -- at average incomes of $4,695 and $10,154 -- show tax liabilities slightly below 0. At these income levels the average tax liability is negative due the effects of the earned income tax credit. The remaining points rise upward toward an average tax liability of $20,500 at an average income of $125,800 for taxpayers in the 95th to 99th percentiles. For the top 1 percent of income earners (not shown in the figure), average income was $548,970 and average tax liability was $118,029.

The slope of this line is the average effective marginal tax rate since it indicates the average increase in tax liability as income increases from one point to the next on the line. The slope of each line segment of the average tax liability line is shown as a point on the average effective marginal tax rate line in the figure. ¹⁵ Effective marginal tax rates begin at virtually 0, rise rapidly between the 2nd and 5th deciles, and then rise more gradually. The highest effective marginal tax rate shown in the figure is 22.1 percent for taxpayers in the 95th to 99th percentiles. If data for the top 1 percent of taxpayers were shown in the figure, the additional effective marginal tax rate point would be only 1 percentage point higher, at 23 percent.

Hence, while statutory marginal tax rates rise to 31 percent, average effective marginal tax rates calculated for population income deciles remain about one-fourth below that level. There are factors which can both raise and lower the effective marginal rate from the statutory rate; in practice, on average the factors lowering marginal rates predominate.

To trace recent trends, Figure 3 plots average effective marginal tax rate lines for four years: 1977, 1980, 1985, and 1990. The horizontal axis represents deciles rather than income so marginal tax rates can be compared at the same point in the income distribution each year. Between 1977 and 1980 effective marginal tax rates rose for all but the highest-income taxpayers as high inflation rates pushed taxpayers into higher statutory tax rate brackets. The rate decline at the top of the income scale was probably due to the capital gains tax cut in the Revenue Act of 1978. The effects of the Economic Recovery Tax Act of 1981 (ERTA) -- which included a 23- percent across-the-board cut in marginal tax rates -- are clearly visible in the drop in effective marginal tax rates from 1980 to 1985 in the graph.

The major tax policy change between 1985 and 1990 was the Tax Reform Act of 1986. This legislation substantially revised the tax code, lowering statutory marginal tax rates and broadening the tax base. But it was designed to be "distributionally neutral," that is, to leave average tax rates in each income bracket nearly unchanged. This goal implies that effective marginal tax rates also would remain nearly unchanged. The graph shows that this objective was largely achieved.

The data plotted in Figure 3 are evidence that the statutory marginal tax rate is not a reliable indicator of the effective marginal tax rate actually experienced by taxpayers. While there are significant differences in the effective marginal rates shown in the graph, they are small compared to the differences in statutory marginal rates between these years. In 1977 and 1980, the top statutory marginal tax rate was 70 percent, although earned income was subject to a top rate of 50 percent. In 1985 the top statutory marginal rate was 50 percent. In 1990 the top statutory rate was 28 percent, although the 5 percent "bubble" raised the actual rate to 33 percent. Despite these high statutory marginal tax rates, the highest effective marginal rate was 29.7 percent in 1977. In 1990 it was 23 percent.

B. All Federal Taxes

The methodology used in Figures 2 and 3 can be used to develop estimates of effective marginal tax rates for the entire Federal tax structure. Such estimates are shown in Figure 4, which is comparable to Figure 2. The lines for average income tax liability and average effective marginal income tax rates are the same as graphed in Figure 2. The heavy solid line plots average total tax liability, including the individual income tax, the social security tax (both the employee and employer portions), the corporate income tax (allocated to households), and excise taxes. The top dotted line in the graph shows average effective marginal tax rates for the total tax system. These rates are the slopes of the line segments of the average total tax liability line.

For the total tax system, the average effective marginal tax rate begins at 11.7 percent and quickly jumps to 23 percent. From there, however, the marginal tax rate rises gradually to about 30 percent between the 8th and 9th income deciles, and then declines somewhat to about 27.5 percent at the highest income levels (including the top 1 percent of income earners, not shown in the figure). Thus, effective marginal tax rates for the total tax system are significantly higher than just for the individual income tax.

But it is interesting to compare the effective marginal rates for the total tax system with the statutory rates for the individual income tax. The total tax system imposes effective marginal tax rates on lower-income and middle-income taxpayers that are substantially higher than the statutory rates these taxpayers face under the income tax; this relationship is not true, however, for high-income taxpayers.

C. The Effects of Non-Tax Programs

The discussion so far has focused on the tax system, which is natural given that the subject is marginal tax rates. But some other government programs have effects similar to imposition of a tax. The combined effects of several programs operating jointly can yield very high effective marginal "tax" rates. When these effects are taken into account, the highest effective marginal tax rates are found in the low-income ranges where the benefits of welfare programs phase out.

To illustrate these effects, Table 1 shows example calculations of welfare benefits, taxes, disposable income, and the effective marginal tax rate for a mother with two children. ¹⁶ Welfare benefits and State income tax liabilities vary from State to State; those shown in the table are for the State of Pennsylvania, which is used for the illustration. The general conclusions regarding marginal tax rate effects that will be drawn from the example, however, are not specific to a single State.

In this example, benefits under the Aid to Families with Dependent Children (AFDC) program phase out as earnings rise to a level of $8,000. Food stamp benefits are relatively constant up to $8,000, but phase out above that income level. The earnings are subject to the social security tax (only the employee portion of the tax is taken into account in the example). The Federal individual income tax liability is negative up to beyond an income level of $20,000 due to the earned income tax credit. And the State income tax begins having an effect in the example when earnings reach $9,000.

The right column in the table shows the combined effective marginal tax rate. It is derived by computing the increase in disposable income as a fraction of the income increment and subtracting the result from 1. For example, as the mother's earnings increase from $4,000 to $5,000, her disposable income rises from $9,957 to $10,324, an increase of $367. Hence, disposable income rose by 36.7 percent of the earnings increase, implying an effective marginal tax rate of 63.3 percent (1 - .367 = .633).

For most of the income increments up to earnings of $20,000, the effective marginal tax rate exceeds 60 percent. The marginal tax rates can go even higher under certain circumstances. For instance, at the point where food stamp benefits are lost because income passes the eligibility limit, the loss of benefits can be far greater than the increase in income. Rather than being phased out as income increases, food stamp benefits are reduced to zero at certain income limits producing a notch effect. (Under the Food Stamp program, income eligibility limits are set at 130 percent of the Federal poverty guidelines.). Consider a family earning $14,400 who would be eligible for $1,344 in food stamps. If the family's income increased by $85, then its income would exceed the food stamp income limit of $14,484, and food stamp benefits would be reduced to zero. This is equivalent to a marginal tax rate on the $85 increment of income of over 1,500 percent.

       EARNINGS AND BENEFITS FOR A MOTHER WITH TWO CHILDREN WITH

V. EFFECTS OF MARGINAL TAX RATES ON BEHAVIOR AND ECONOMIC WELFARE

Up to this point, the analysis in this report has focused on the myriad factors affecting marginal tax rates and the difficulties in determining them. From this discussion it should be evident that, in many instances, the calculation of effective marginal tax rates is a complicated task. Moreover, recent legislative actions which have expanded the use of phase-out provisions have made it even more complex.

In this section, the focus is on the effects of marginal tax rates. But the complexity in determining the rates remains a relevant consideration. There are many decisions of taxpayers that should be affected by their marginal tax rates: how much to work, how much to save, what kind of investments to make, what to consume. But the tax rate that will affect these decisions is the taxpayer's PERCEIVED marginal tax rate, which may or may not equal his ACTUAL marginal tax rate because of the complexity involved in its determination.

There is very little direct evidence regarding taxpayer perceptions of their marginal tax rates. Some surveys have found that taxpayers do not have a good understanding of the tax system in general or of their average tax rate or tax liability. ¹⁷ If taxpayers do not have accurate perceptions of their average tax rate, they are unlikely to know their marginal tax rate, which is a more complex concept. The only known surveys focusing on taxpayer awareness of marginal tax rates were done in the United Kingdom and Italy; both found relatively few taxpayers who could approximate their marginal tax rates. ¹⁸ If taxpayers typically are confused about their marginal tax rates, this suggests that tax rates may not have much systematic effect on taxpayer behavior. Certain easily understood taxes or tax rate changes might have noticeable effects, but the tax system as a whole might have relatively small effects.

Even if marginal tax rates were known precisely, economic theory provides little guidance regarding their effects on behavior. Changes in marginal tax rates produce both substitution and income effects, frequently in opposite directions. The substitution effect is the shift toward greater use of the relatively lower priced item in response to a change in prices brought about by a change in marginal tax rates. Most tax rate changes also produce an income effect; that is, they leave the taxpayer richer or poorer. This income effect may also affect behavior, possibly countervailing the substitution effect.

To fully appreciate the nature of income and substitution effects on behavior, consider the response to a decrease in the marginal tax rate on saving. Because the lower tax rate increases the after-tax rate of return on saving -- which reduces the "price" of future consumption (in terms of foregone current consumption) -- taxpayers might substitute some future consumption (saving) for present consumption. This substitution effect, in isolation, should cause aggregate saving to rise. An increase in the after-tax return to saving, however, means an individual could actually save less (consume more currently) and still achieve the same level of future consumption. That is, in real terms the individual is richer. This income effect could induce the individual to consume more in the present, thus reducing savings. As in this case, the income and substitution effects often offset each other, making it difficult to determine theoretically how a change in marginal tax rates will affect aggregate behavior. Hence, knowledge of the effects of marginal tax rates on behavior must be gained from observation.

The empirical evidence about the effects of changes in marginal tax rates on behavior is also inconclusive. While some studies have shown strong relationships between changes in marginal tax rates and changes in behavior, others have shown little or no correlation. The degree of correlation between changes in tax rates and behavior also seems to vary depending on the type of behavior under consideration, for example, whether it involves charitable contributions, labor supply, or saving decisions.

Studies consistently have shown that tax rates are one of the most important factors affecting the level and distribution of charitable giving. ¹⁹ Yet even in this area the link between marginal tax rates and behavior is not always strong. For instance, despite the increase in the after-tax cost of charitable contributions that occurred as a result of the marginal tax rate reductions in the Tax Reform Act of 1986, the predicted decrease in charitable contributions did not occur. (The 1986 Act also eliminated the non-itemizer deduction for charitable contributions and subjected the untaxed appreciation of donated gifts to the alternative minimum tax.)

In fact, charitable contributions increased after 1986. As a recent paper by Clotfelter made clear, however, although the aggregate level of donations increased, the distribution and types of gifts changed considerably. ²⁰ After 1986, donations by higher- income individuals and gifts of appreciated property declined, but the difference was made up in the aggregate by donations from lower- income individuals. Hence, the substitution effect appears to have been operative in the upper-income ranges. In the case of the Tax Reform Act, there should have been virtually no income effect because the overall tax changes were nearly revenue neutral in all of the income brackets. Hence, the substitution effect was weak or was overwhelmed by other factors in the lower-income brackets.

The supply of labor and hours worked is another area where behavior has shown a responsiveness to changes in marginal tax rates. Many studies have demonstrated that although primary earners are rather insensitive to marginal tax rate changes, secondary earners, especially married women, are highly responsive to changes in the after-tax wage rate. ²¹ Once again, however, changes in marginal tax rates are not the only variable affecting labor supply. As Bosworth and Burtless point out, much of the gain in the labor supply during the 1980s should not be attributed to the marginal tax rate reductions that occurred during the period, since much of the labor supply response emanated from lower-income households who were not affected by the tax changes. ²²

Probably the most controversial area concerning the relationship between behavior and tax changes involves private saving and investment. Two topics in particular have received extensive examination; the effects of individual retirement accounts (IRAs) on saving and the effects of the capital gains tax rate on capital gains realizations. The empirical evidence presented in the literature to date has been inconclusive. Indeed, those studies that claim to have found a strong linkage between marginal tax rate changes and the level of saving or investment have been subject to criticism on a number of grounds. ²³

Slemrod has suggested that there seems to be a hierarchy of behavioral responses with respect to tax rate changes. ²⁴ At the top of the hierarchy is the timing of economic transactions. The evidence for this is fairly straightforward; in anticipation of capital gains tax rate increases, realizations rise. IRA contributions increase in the face of new restrictions on their deductibility. And most recently, investments in equipment fall in anticipation of enactment of a new investment tax credit, and corporate executives exercise stock options, and investors buy tax- exempt bonds in anticipation of a tax rate increase. All of these actions involve the timing of transactions in order to maximize tax savings in the short-run.

In the second level of hierarchical responses, Slemrod places the re-shuffling of portfolios and accounting procedures. These responses include things like the reformulation of personal debt from non-deductible loans to home-equity type loans which retained their deductibility after the 1986 Act.

At the bottom of the hierarchy, Slemrod places the real decisions of individuals. These would include work versus leisure choices and present versus future consumption (saving) decisions. This is the area where there appears to be the least evidence of responsiveness to tax rate changes.

Even if behavioral responses to tax rate changes are small, however, it does not imply that the efficiency consequences associated with marginal tax rates are necessarily insignificant. Economic theory is clear: taxes affect economic welfare negatively because they distort relative prices and therefore distort resource allocation. Even in cases in which behavior does not change or changes very little in response to a tax rate change, a substitution effect still occurs. If behavior does not change, this implies that the substitution effect is offset by the income effect or other factors. But the substitution effect still exists and results in diminished welfare. ²⁵ Moreover, the welfare loss associated with a tax increases exponentially as marginal tax rates increase. Hence, a basic tenant of public finance theory is that to minimize the welfare loss accompanying a tax raising a given amount of revenue, the tax base should be defined as broadly as possible so that marginal tax rates can be kept as low as possible and tax rate differentials are minimized.

VI. IMPLICATIONS

Marginal tax rates have been a central issue in Federal tax legislation for the past dozen years. Substantially reducing marginal tax rates was the principal objective of the two most prominent tax acts during the period: the Economic Recovery Tax Act of 1981 (ERTA) and the Tax Reform Act of 1986. Marginal tax rates were at the heart of the debates both before and after enactment of the controversial Omnibus Budget Reconciliation Act of 1990. They were also a key issue in the 1992 presidential campaign.

Despite this prominence, marginal tax rates are widely misunderstood, both in terms of their meaning and their significance. The focus of the policy discussions has been on the statutory marginal tax rates in the Federal individual income tax. But these rates are only one of the many determinants of the effective marginal tax rates actually faced by taxpayers. In the Tax Reform Act of 1986, for example, reduction and simplification of the statutory marginal tax rate structure was counted as a major achievement, but the broadened tax base left effective marginal tax rates -- on average -- virtually unchanged.

Determining the actual marginal tax rate faced by a taxpayer is, in many cases, not a simple task. The complex nature of the tax structure as well as the economic effects of taxes (shifting the tax burden from one taxpayer to another, for example) make discerning the marginal tax rate complex and difficult. Some tax structure complexity is inevitable in an advanced society with a complex financial system. Some of the complexity, however, could be avoided; it results from attempts to achieve multiple (and sometimes irreconcilable) goals in designing the tax system and/or efforts to disguise the true nature of the tax structure. Some of the tax benefit phaseouts, for example, are indirect and complex means of achieving the same thing that could be accomplished more directly by increasing the tax rate. Simplifying the tax structure would, in many cases, make the marginal tax rate easier to ascertain. This would improve taxpayer understanding of the revenue system used to finance the government and its impact on them. It would also facilitate the millions of decisions made by taxpayers every year -- involving work, spending patterns, saving, and investment -- in which the marginal tax rate is a relevant consideration.

An implication of the complexity in determining marginal tax rates is that some of the research on the behavioral effects of the tax system may be at least partially erroneous. Measuring the effects of the tax system on behavior requires estimating marginal tax rates. Most studies of tax effects take account of some of the complexities in calculating marginal tax rates, for example the phaseout provisions. But many of the studies do not take full account of all the factors determining marginal tax rates. In particular, virtually none of the studies take account of likely changes in deductions resulting from higher income. Furthermore, most of the studies do not allow for taxpayer confusion about marginal tax rates. The tax rate that governs behavior is the tax rate the taxpayer THINKS he is subject to. If the estimated marginal tax rates used in the study differ from the perceived rates that actually govern behavior, then the conclusions of the research will be in error.

While much of the attention to effective marginal tax rates in articles and in policy discussions focuses on rates which are higher than the statutory rates, in fact, taxpayers on average are subject to effective marginal rates which are lower than the statutory rates. The lower marginal tax rates are attributable primarily to the sources of marginal income -- which may be completely or partially exempt -- or to changes in deductions which accompany the higher income.

Nonetheless, some taxpayers do face marginal tax rates which exceed the statutory rates, in some cases by substantial amounts. Many of those with the highest effective marginal tax rates are in the lowest income brackets and may be subject to the phase out of the earned income tax credit as well as phaseouts or limitations under various welfare programs. It is not uncommon in these circumstances to find effective marginal tax rates exceeding 60 percent.

Despite the substantial attention given to marginal tax rates, their effect on real economic behavior is debatable. Changes in marginal tax rates frequently have significant effects on the timing of transactions such as capital gains realizations or investments, suggesting the importance of effective dates and transition rules in tax legislation. Tax rate changes also affect portfolio decisions, such as the demand for tax-exempt bonds or the corporate debt/equity ratio, suggesting the importance of striving to "level the playing field." But, with some exceptions such as the labor force participation of secondary earners, most evidence suggests that marginal tax rate changes have relatively little effect on most real economic phenomena, such as levels of work effort, saving, and investment. Hence, changes in marginal tax rates, even big changes, cannot be expected to have dramatic effects on the economy. ²⁶

This does not imply, however, that marginal tax rates are unimportant. All taxes distort economic decisions and thereby reduce economic welfare. The distortions occur because taxes alter relative prices from those that would occur in the private marketplace absent the effects of taxation. The welfare losses are smaller with lower tax rates and smaller differentials between various tax rates. Hence, low tax rates and a "level playing field" are objectives consistent with a goal of maximizing public welfare. Unlike observable economic phenomena such as the amount of saving or investment, however, welfare losses are for the most part discernable only through the application of economic theory and estimation, which makes it difficult to construct a winning argument in the political arena based on this concept.

² Pure public goods are goods which are not divisible and which people cannot be excluded from consuming. National defense is the classic example; all citizens receive its benefits. Externalities are costs or benefits that are not included in the price of a good or service. Pollution given off as the by-product of a manufacturing operation is an example. An externality can be "internalized" by taking action resulting in the cost of the externality being reflected in the price of the output; requiring pollution abatement is an example of such an action.

³ The worker may THINK his marginal social security tax rate is 7.65 percent based on the tax and wage reported on his pay statement. This points out a difficulty in using marginal tax rates: the PERCEIVED marginal tax rate may differ from the actual rate. When the effect of a tax is complex and indirect, this possibility must be taken into account in assessing its effects.

⁴ Martin Feldstein and Andrew Samwick, Social Security Rules and Marginal Tax Rates, National Tax Journal, March 1992, p. 1-22.

⁵ The taxpayers in this group had 1990 wages ranging from $4,272 to $25,740. The net marginal tax rate depends on the discount rate assumed and the age at which the taxpayer is assumed to retire. Feldstein and Samwick used a discount rate of 4 percent and an assumed retirement age of 65.

⁶ As with social security, the net effects of the IRA deposit depend on the discount rate, the age of the taxpayer, and the assumed retirement age; see the analysis in U.S. Library of Congress, Congressional Research Service, IRA Alternatives: A Comparison of Taxpayer Benefits, Report No. 90-191 E by Donald W. Kiefer, April 6, 1990, 20 p.

⁷ This marginal tax rate is defined precisely in theoretical economics in terms of calculus as the derivative of tax liability with respect to income. For most applications, the difference between the simple definition in the text and this more precise definition is insignificant.

⁸ A larger effect would result if the effective marginal tax rate on the IRA contribution were negative. This would be the case if the taxpayer expected to be in a lower tax bracket during retirement than during the earning years.

⁹ For further discussions and detail, see Ellin Rosenthal and Pat Jones, Phasing Out Tax Benefits: Social Policy Meets Deficit Reduction, Tax Notes, September 12, 1988, p. 1105-1110; and Ellin Rosenthal, Phaseouts: Too Much of a Good Thing?, Tax Notes, September 19, 1988, p. 1228-1233.

¹⁰ See Tax Expenditures Compendium of Background Material on Individual Provisions, Committee on the Budget, U.S. Senate, Committee Print, 102nd Congress, 2nd Session, November 1992, 556 p.

¹¹ See Linda L. Johnson and James A. Fellows, Revisiting the Effective Marginal Tax Rate on Long-Term Capital Gains, Tax Notes, April 20, 1992, p. 403-404.

¹² Allen H. Lerman and Jim Cilke, The Distribution of Effective Marginal Tax Rates, in 1991 Proceedings of the Eighty- Fourth Annual Conference on Taxation, National Tax Association, Columbus, Ohio, 1992, p. 245-256.

¹³ The estimates are based on estimated 1991 income levels and 1991 tax law except that provisions to be phased in between 1991 and 1994 were assumed to be fully phased in for purposes of the calculations.

¹⁴ These two imputations are made to derive a total pre-tax income measure for use in analyzing the distribution of the total Federal tax burden. The issue is similar to the one discussed above in section II A.

¹⁵ Since the slope applies to a line segment rather than a point, points on the effective marginal tax rate line are plotted at mid-points (horizontally) of the line segments of the average tax liability line.

¹⁶ The figures in the table are based on data in 1992 Green Book: Overview of Entitlement Programs, Committee on Ways and Means, U.S. House of Representatives, 102nd Congress, 2nd Session, May 15, 1992, p. 628.

¹⁷ Norbert Lloyd Enrick, A Pilot Study of Income Tax Consciousness, National Tax Journal, June 1963, p. 169-173; Norbert Lloyd Enrick, A Further Study of Income Tax Consciousness, National Tax Journal, September 1964, p. 319-321; and D.A.L. Auld, Public Sector Awareness and Preferences in Ontario, Canadian Tax Journal, March-April 1979, p. 172-183.

¹⁸ Alan Lewis, Perceptions of Tax Rates, British Tax Review, n. 6, 1978, p. 358-366; and Bruno Bises, Income Tax Perception and Labour Supply in a Sample of Industry Workers, Public Finance, No. 1, 1990, p. 3-18.

¹⁹ See, for example, Charles T. Clotfelter, Federal Tax Policy and Charitable Giving, University of Chicago Press, 1985.

²⁰ Charles T. Clotfelter, The Impact of Tax Reform on Charitable Giving: a 1989 Perspective, [in] Joel Slemrod, ed., Do Taxes Matter: The Impact of the Tax Reform Act of 1986, Cambridge, MIT Press, 1990, p. 203-235.

²¹ See, for example, Harvey S. Rosen, What Is Labor Supply and Do Taxes Affect It?, American Economic Review, May 1980, p. 171-176.

²² Barry Bosworth and Gary Burtless, Effects of Tax Reform on Labor Supply, Investment, and Saving, Journal of Economic Perspectives, Winter 1992, p. 3-25.

²³ See, for example: Steven Venti and David Wise, Have IRAs Increased U.S. Savings? Evidence from Consumer Expenditure Surveys, NBER Research Paper 2217, 1987; Congressional Budget Office, How Capital Gains Tax Rates Affect Revenues: The Historical Evidence, 1988; Donald W. Kiefer, Lock-In Effect Within a Simple Model of Corporate Stock Trading, National Tax Journal, March 1990, p. 75-94; and U.S. Library of Congress, Congressional Research Service, Capital Gains Taxes, IRA's, and Savings, by Jane G. Gravelle, Report No. 89- 543 RCO, September 1989.

²⁴ Joel Slemrod, The Economic Impact of the Tax Reform Act of 1986, [in] Do Taxes Matter: The Impact of the Tax Reform Act of 1986, p. 1-12, (cited in footnote 21).

²⁵ The substitution effect resulting from a tax is the difference between 1) the behavior that would occur under a lump-sum or head tax raising the same amount of revenue and 2) the behavior under the actual tax. The difference between 1) and behavior prior to imposition of the tax is the income effect. If observed behavior does not change in response to imposition of the tax, then the income effect and substitution effect exactly offset each other, but they still may be substantial.

²⁶ There are obvious limitations to this statement. As represented in the so-called Laffer Curve, tax rates approaching 100 percent would have obvious effects on at least REPORTED economic activity.