NBER Examines Impact of AMT on Weighted Average Marginal Tax Rates

©2003 by Daniel Feenberg and James Poterba. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given to the source.

The Alternative Minimum Tax and Effective Marginal Tax Rates

Daniel Feenberg

[3] The minimum tax expired in 1981. Before its expiration, however, in 1979, a new "Alternative Minimum Tax" was established with a base that included all the components of Adjusted Gross Income (AGI) and the capital gains deduction in addition to preferences. The new AMT allowed the full amount of regular income tax as a credit. Aside from changes in the treatment of net operating losses and slight changes in the exclusion amount, this is the AMT that is currently in force, and that is scheduled to remain in force indefinitely. When a preferential tax rate for long-term capital gains was established in 1991, care was taken to avoid treating capital gains as a preference.

[4] A number of research studies published since the late 1990s have identified the AMT as a growing factor in the income tax structure. Harvey and Tempalski (1997) were among the first to point out that because the exclusion level for the AMT was not indexed for inflation, while most other key parameters in the tax code were indexed, the AMT would apply to a growing number of taxpayers. Rebelein and Tempalski (2000), the U.S. Congress Joint Economic Committee (2001), Tempalski (2002), and Burman, Gale, and Rohaly (2002) provide further analysis of the growing importance of the AMT.

[5] The AMT became more important after the passage of the Economic Growth and Taxpayer Relief Reconciliation Act of 2001 (EGTRRA), which reduced ordinary income tax liabilities for many households. Kiefer, et al. (2002) and Burman, Gale, and Rohaly (2002) are two examples of studies that show that because EGTRA lowered income tax liabilities for many taxpayers, it will raise the number of AMT payers. The Jobs and Growth Taxpayer Relief Reconciliation Act of 2003 reinforces this effect. For many upper middle income households who will face the AMT as a result of recent tax reforms, the AMT "takes back" a substantial share of the potential tax relief associated with these reforms. Burman, Gale and Rohaly (2003a, 2003b) suggest that in 2010, 33.8 percent of the EGTRRA tax cuts will be "recaptured" in the form of higher AMT liabilities. This percentage exceeds 65 percent for taxpayers with AGI between $100,000 and $500,000.

[6] Sullivan (2002) offers a careful review of the issues associated with the potential repeal of the AMT. Proponents of repeal point to the potential compliance burden of the tax, which can effectively require taxpayers to prepare two tax returns and to compare the results, and to the unanticipated impact on taxpayer liabilities. The potential revenue cost of AMT repeal is a substantial impediment to systematic reform.

[7] In this paper, we use the NBER's TAXSIM program to evaluate the impact of the AMT on incentives to work and to save. We present some estimates of the growing number of taxpayers who will face the AMT, but they are not the central focus of the study. We emphasize projections for calendar year 2010, since under current law, this is the year when the AMT will affect the largest number of taxpayers. A number of income tax provisions are currently scheduled to expire in 2010, so the tax environment in 2011 will be substantially different from that in 2010. Although our projections rely on strong assumptions to project the future number of tax returns and the level and composition of income on these returns, they offer some insight on the likely future course of the AMT relative to the ordinary income tax.

[8] The paper is divided into six sections. The first describes the basic structure of the AMT and compares alternative minimum taxable income with taxable income as defined under the ordinary income tax. Section two describes our procedure for projecting future AMT liabilities, and it presents our projections of the number of AMT taxpayers in aggregate, and in income subcategories, for future years. Our results from the TAXSIM analysis are broadly consistent with those from other recent studies that project future AMT liabilities. Section three describes the probability that taxpayers in various income categories will face the AMT. It also studies how various reform proposals would affect these probabilities. The fourth section investigates the impact of the AMT on the weighted average marginal tax rates that apply to wages, interest income, dividend income, and several income tax deductions. It considers marginal tax rates associated with the federal income tax, as well as the total effective tax rate that combines the federal income tax with the potentially-deductible state income tax. Section five explores the effect of various reform proposals on the weighted average marginal tax rates on income and deduction flows. There is a brief conclusion.

1. The Structure of the Alternative Minimum Tax

[9] Although the AMT is sometimes described as a complex and mysterious tax, for many taxpayers who face the AMT, the AMT calculation is straightforward. For taxpayers who do not itemize, the "tentative AMT" is 26 percent of AGI in excess of an exclusion amount. For taxpayers with alternative minimum income above $175,000, the AMT tax rate is 28 percent. The AMT is paid as an actual tax liability only to the extent that it exceeds the taxpayer's regular tax liability. For 2010, the year when the AMT is projected to generate the largest revenue flow, the exclusion equals $45,000 for joint filers and $33,750 for single individuals. For taxpayers with incomes above $150,000 on joint returns and $112,000 on single returns, the exclusion is phased out at the rate of 25 cents for each dollar of AGI above the threshold. This implies that the 28% AMT rate applies to joint filers with AGI above $206,000, and for single filers with income above $189,400. These AGI cutoffs are calculated by finding the AGI level at which alternative minimum income will equal 175,000. For example for joint filers, this requires AGI - (45000 -.25*(AGI - 150000)) = 175000.

[10] Several features of the AMT are scheduled to change between 2003 and 2010 under current legislation. For example, after 2005, the zero bracket amount is no longer added to the exclusion. For 2003 and 2004, the exclusion is $58,000 for joint filers and $40,250 for single filers. It declines in later years.

[11] For itemizers, the AMT calculation is more complicated. All itemized deductions except those for state and local taxes, medical expenses in excess of 2.5% of AGI, and miscellaneous deductions are available as deductions against alternative minimum income. The taxpayer must identify eligible deductions and subtract them from alternative minimum income. Itemized deductions under the AMT are not subject to the phase-out that is present in the regular personal income tax. In rare cases this may invalidate the usual rule that taxpayers should itemize if itemized deductions exceed the standard deduction. For a very few, the decreased AMT from itemization will compensate for a larger taxable income.

[12] For most taxpayers with long term capital gains, the AMT is calculated on non-gain income plus 20% of long-term gains. For low income taxpayers with gains, the 10% capital gains rate is applied to gains that would not push taxable income above the 15% bracket boundary. All tax credits except the foreign tax credit can be credited against AMT liability, as can the taxpayer's regular tax liability. Since tax year 2000, tax credits do not affect the total liability calculation. Prior to that year, the tax forms distinguished between AMT liability, recorded on form 6251, and credits whose value, if taken, would be fully offset by the AMT. These were referred to as "lost credits" and the credit forms included lines to avoid taking such credits.

[13] While the AMT may require taxpayers to compute their tax liability under two regimes, none of the calculations require any special record keeping beyond what is required by the regular tax. Moreover, the calculations will be completely automated by tax preparation software. It is true that some taxpayers with net operating losses, or accelerated depreciation or in other similar esoteric situations are subject to additional record keeping, and may be subject to additional AMT by 2010, but there are few such taxpayers. For most taxpayers who will face AMT liability, the AMT calculation is not particularly burdensome.

[14] Figure 1 provides information on the pattern of AMT and ordinary income tax liability for a taxpayer filing a joint return with two dependents and claiming the standard deduction. Tax liabilities under both tax schedules are shown at different levels of adjusted gross income. The figure graphs AMT and ordinary income tax liability for both 2003 and 2010, with the taxpayer's real AGI held constant in 2003 dollars. The graphs show that AMT liability exceeds income tax liability for a substantial range of incomes, beginning at roughly $50,000, and ending at roughly $375,000, in 2010. The figure also shows that the range of incomes over which AMT exceeds the income tax is wider in 2010 than in 2003. The four lines demonstrate how similar the shapes of the four tax schedules are, so that a small shift leftward in the real AMT schedule, such as the one that takes place between 2003 and 2010, shifts much of the income distribution onto the AMT schedule.

2. Projecting Future AMT Liabilities: The TAXSIM Model

[15] The NBER TAXSIM model is a computer program that calculates federal income tax and payroll tax liabilities for a representative sample of U.S. families. It analyzes data from the Statistics of Income (SOI) Public Use File, a stratified random sample of U.S. taxpayers that oversamples high-income tax returns. Feenberg and Coutts (1993) describe the personal income tax sections of the TAXSIM model in some detail.

[16] The TAXSIM algorithm includes detailed computer code based on the personal income tax system that, based on current legislation, will be in force in future years. To construct households that will face the tax system in 2010, we "age" the most recent publicly available individual income tax return file, the 1999 SOI data file, to 2010. This "aging" process requires forecasts of the growth rate of the aggregate amount of various income components and deduction flows. We assume annual growth rates of 1.2 percent for population, 2.2 percent for real incomes and deductions, and 2.5 percent for the price level for all income and deduction items. These growth rates are drawn from CBO projections. Long term gains are normalized to 5.6 percent of AGI, which was their average value over the years 1981- 1999.

[17] When the growth rates described above are applied to the 1999 cross-section data on tax returns, the resulting income tax revenue forecasts for 2004-2013 are substantially higher than those made by the Congressional Budget Office. A one-time ten percent reduction in all nominal magnitudes, applied for example in 1999, brings our aggregate revenue estimates into much closer agreement with the CBO. We therefore reduce all nominal 1999 magnitudes by ten percent before we make our future projections.

[18] TAXSIM calculates federal as well as state marginal income tax rates. AMT payers cannot deduct state income taxes from their taxable income, so their effective state income tax rate is higher than the comparable rate facing those who pay ordinary federal income tax and claim an itemized deduction for state taxes. To recognize the impact of the effective state tax rate on the total tax burden on wages, interest, dividends, and other income components, we therefore calculate federal marginal tax rates under the AMT as well as the combined tax rate that equals the federal marginal tax rate plus the net-of-federal-deductibility state income tax rate.

[19] Table 1 presents information on our projections regarding the AMT under current law. The table shows the total number of taxpayers who we project will face the AMT, the total amount of AMT revenue, and the ratio of revenue from the AMT to revenue from the ordinary income tax. The table shows that projected AMT revenue peaks in 2010, when it accounts for nearly nine percent of total income tax revenue. We project that 37 million taxpayers -- nearly one in four -- will pay the AMT in 2010. Of this group, almost fifteen million of the AMT payers will have no itemized deductions, no capital gains, no phase-out of the AMT exclusion, and no other preferences. This group of taxpayers will face relatively simple AMT calculations.

[20] Table 1 shows that the number of AMT filers and the revenue collected from the AMT decline after 2010, when various provisions of EGTRRA expire. If this legislative change occurs in 2011, the number of AMT filers will fall by more than one third, and AMT revenues will decline by more than half.

[21] The projections in Table 1 are similar to those in several other studies of the AMT. For example, Burman, Gale, and Rohaly (2003b) project that there will be 33.1 million AMT filers in 2010. They note that in the absence of EGTRRA, they would have projected 14.3 million AMT filers. Their projection suggests AMT revenue of $124 billion in 2010, only $1 billion less than our projection. Kiefer, et al. (2002) project that there will be 35.1 million AMT filers in 2010 and that the AMT will yield $133 billion in revenue that year. Their projections include the effect of EGTRRA, but they do not incorporate any of the 2003 tax changes.

[22] The prospective growth of AMT liabilities and the number of AMT taxpayers has generated a number of proposals for legislative reforms that would slow the growth of the AMT. We use the TAXSIM model to evaluate the impact of six potential reforms on the number of AMT taxpayers and the revenue generated by the AMT. Table 2 presents these findings. The first row reports baseline calculations for 2010, corresponding to the information in Table 1. The next six rows consider reform proposals.

[23] The first proposal we consider would allow personal exemptions to be subtracted from other income in defining alternative minimum income. Such exemptions are permitted under the ordinary income tax, thereby reducing ordinary taxable income relative to alternative minimum taxable income. The second row of Table 2 shows that this reform would sharply reduce the number of AMT taxpayers in 2010, from 37 million to 12 million, and it would also reduce revenue from $126 to $48 billion.

[24] A second potential reform calls for the extension of income tax provisions that expire between 2005 and 2010 so that they would remain in force in 2010. The two most significant provisions are one that would raise the AMT exclusion in 2010 to $13,000 for joint filers and $6,500 for single filers, and one that would continue the preferential taxation of dividends. The former greatly reduces the AMT, and the latter reduces both the regular income tax and the AMT by similar amounts. The third row of Table 2 shows that extending all expiring provisions would reduce the number of AMT filers from 37 million to 14.1 million in 2010, with a decline in total AMT revenue from $126 to $58 billion.

[25] The third AMT reform that we consider involves changes in the relative treatment of single and married taxpayers. The 2010 joint taxpayer exclusion of $45,000 under the AMT is substantially less than twice the single taxpayer exclusion of $33,750. In principle, the single exclusion could simply be doubled to $67,500, but that would substantially increase the marriage bonus for couples with non-working spouses. We have therefore modeled a less generous option which offers the couple the greater of the current joint deduction, or the single deduction plus the secondary earner's wages up to the single exclusion. This means that the full $67,500 is offered only to two earner couples, with both earners making more than $33,750. Our analysis of this reform shows that it leads to a more modest reduction in AMT taxpayers than the two earlier reform proposals. The number of AMT filers in 2010 drops from 37 million to 25 million in this case and AMT revenue declines by roughly one third, from $126 to $87 billion.

[26] The last three proposals we consider involve changes to the AMT exclusion level. One involves indexing the exclusion, while the second and third involve raising the exclusion by $3,000 and $10,000 respectively. The nominal, un-indexed character of the AMT exclusion is a key contributor to the projected growth of AMT revenues and taxpayers. We project 14.0 million AMT taxpayers in 2010 if the exclusion is indexed, just over one third of the number without indexation. AMT revenues in this case are $48 billion. Indexing the exclusion therefore reduces AMT revenue in our projected peak revenue year by over sixty percent.

[27] The findings on the impact of indexing the AMT underscore the importance of the inflation rate in determining the relative revenue yield of the regular income tax and the AMT. Figure 1 showed that small changes in the real threshold at which taxpayers are affected by the AMT can have substantial effects on revenues and on the number of AMT filers. By comparison, real growth in the aggregate economy has a much more modest effect on AMT revenue, because real bracket creep raises the regular income tax as well as the AMT base.

[28] The last two reforms that we consider involve specific nominal changes in the AMT exclusion. Raising the exclusion by $3000 in 2010 reduces the projected number of AMT taxpayers by 6.4 million, from 37 million to 31 million. Raising the exclusion by $10,000 reduces the number of filers to 17 million. Revenues drop from $126 billion in the status quo, to $100 billion if the exclusion is raised by $3000, to $65 billion if the exclusion rises by $10,000.

[29] The burden of the AMT does not fall equally across the income distribution. Table 3 shows the AMT tax burden as a share of AGI for households stratified by AGI. The entries show total AMT liability divided by total AGI for the households in each category; they are not restricted to households with AMT liability. The calculations apply to 2010, and they assume no changes between now and then in the federal tax law. The AMT burden is highest for those with AGI between $200,000 and $500,000. This group faces AMT liability that averages 2.7% of AGI. The burden is half as great, 1.3% of AGI, for taxpayers with AGI of $50,000 to $75,000. It is lower for AMT taxpayers with higher -- greater than $500,000, and lower -- less than $50,000 -- AGI.

[30] Figure 2 plots information similar to that in Table 3. The figure shows average tax rates, with and without the AMT, in 2002 and 2010. The AMT has its largest impact on the average tax rate for taxpayers with AGI between $200,000 and $500,000, and very little impact on average tax rates for households whose 2003 AGI is below approximately $50,000.

3. The Probability of Facing the AMT: Current Tax Policy and Various Reforms

[31] The results in Table 3 and Figure 2 underscore the importance of disaggregating taxpayers by income class when analyzing the AMT. One useful way to illustrate how the AMT's impact varies across households with different characteristics and income levels is to project the probability that a taxpayer will be an AMT filer. Burman, Gale, and Rohaly (2003a, 2003b) and other earlier studies estimate the AMT-filing probabilities for taxpayers in various income categories. Our analysis explores how various reform proposals would affect the income-specific likelihood of facing the AMT.

[32] Table 4 presents results based on our projections of the AMT in 2010. The table reports probabilities under the assumption that current tax legislation remains in force through 2010 and under the various reforms described in Table 2. The results in the first column indicate the projected AMT payment probabilities under the status quo. The table presents results for three sets of taxpayers: the whole taxpayer population, the thirty percent of taxpayers who are projected to itemize in 2010, and the subset of all taxpayers with two or more dependents. The probabilities of facing the AMT differ across these groups.

[33] The results in the first panel of Table 4 show that the taxpayers who have the greatest chance of entering the AMT regime in 2010 are those with AGI between $75,000 and $500,000. These income thresholds are specified in constant $2003. At lower income levels, the AMT exclusion makes it unlikely that a taxpayer will face AMT liability. At AGI levels above $500,000, the progressivity of the personal income tax schedule makes it likely that for most taxpayers, personal income tax liability will exceed AMT liability. The results underscore the substantial differences across AGI categories in the importance of the AMT. The probability of facing the AMT is 56.3% for a $50-75,000 AGI taxpayer, 77.2% for a taxpayer with AGI between $75,000 and $100,000, and more than 90% for taxpayers with AGI between $100,000 and $500,000.

[34] The results in the second panel of Table 4, for itemizers, are broadly similar to those for the universe of all taxpayers. The distribution of itemizers across AGI categories is shifted toward higher incomes, however, so that while only 25.7% of all taxpayers are projected to face the AMT, 52.5% of all itemizers are projected to face the AMT. The last panel of Table 4, which presents results for taxpayers with two or more dependents, shows that the presence of dependents changes the AGI-specific probability of facing the AMT. In the $25-50,000 AGI category, for example, 15.8% of all taxpayers are projected to pay the AMT, while more than half of the taxpayers in this AGI category with two or more dependents are projected to face the AMT. There is also a sharp increase in the probability of facing the AMT at incomes between $50,000 and $75,000. The overall probability of facing the AMT is greater than 50% for the subset of taxpayers with two or more dependents.

[35] The next six columns of Table 4 present results similar to those in the first column, but corresponding to the various AMT reform options described above. The second column considers the impact of allowing personal exemptions to be subtracted from alternative minimum taxable income. In this case, the overall probability of facing the AMT drops from 25.7% to 8.4%. The effect is strongest for taxpayers with AGI below $100,000. For the $75-100,000 income category, for example, the probability of facing the AMT drops from 77.2% to 15.8%. The effect of this reform is also very powerful for taxpayers who have positive income and two or more dependents. This reform reduced their probability of facing the AMT in 2010 from 56.1% to 15.6%.

[36] The third column of Table 4 considers the impact of extending the expiring EGTRRA provisions through 2010. For all taxpayers, this reform reduces the probability of paying the AMT from 25.7% to 9.8%, with the most pronounced declines again found among the lower income AMT payers. For taxpayers with two or more dependents, the impact of this reform is smaller than allowing personal exemptions to be deducted from alternative taxable income. The probability of facing the AMT drops from 56.1% under the status quo to 27.8% with the extension of expiring provisions.

[37] The next column considers the "AMT marriage penalty reform" described in the last section. This reform reduces the percentage of taxpayers facing the AMT from 25.7% to 17.4%. The three panels of Table 4 show that the impact of this reform on the probability of paying AMT is less concentrated at lower income levels than at higher levels. In particular, the probability that taxpayers in the $50-100,000 AGI range face the AMT declines less for this reform than for either of the previous reforms that we considered.

[38] The last three columns consider changes in the AMT exclusion level. Comparison of the findings in these columns shows that indexing generates the largest reduction in the probability of facing the AMT for taxpayers with AGI between $50,000 and $200,000. Even with indexation, the probability of paying the AMT remains 83.7% for taxpayers with AGI of between $200,000 and $500,000. The probability is 92% for this group when the exclusion is raised by $3000 or by $10000. The third panel of Table 4 shows that virtually all of the taxpayers in this AGI category with two or more dependents face the AMT. Indexing the AMT exclusion has a smaller effect on the percentage of taxpayers with two or more dependents who pay the AMT than it does on the fraction of all taxpayers facing this tax.

[39] Figure 3 shows the impact of raising the AMT exclusion on number of AMT taxpayers and on the revenue yield of the AMT. The marginal effect of an incremental increase in the exclusion declines as the exclusion rises. A $10,000 increase in the exclusion in 2010, for example, is predicted to reduce AMT revenue by nearly fifty percent, and it has a similar proportionate effect on the number of AMT taxpayers. A $20,000 increase is predicted to reduce the number of AMT filers by 28 million, from 34.8 million to 6.7 million, and it reduces the total revenue generated by the AMT from $121.2 billion to $41.5 billion.

4. The AMT and Marginal Tax Rates

[40] Most of the popular discussion surrounding the AMT focuses on the number of taxpayers who may face the tax, or on the impact of the AMT on federal tax revenues. The effect of the AMT on incentives for working, saving, and engaging in various activities that generate tax deductions has received much less attention. To explore the incentive effects of the AMT, we compare the ordinary income tax rate and the AMT rate facing taxpayers who are projected to be subject to the AMT in 2010. Since a taxpayer may face different marginal tax rates on different income flows, we present tabulations for various components of taxable income. For each income component, we compute the fraction of that component that is received by AMT taxpayers. We then compute the percentage whose tax rate rises, and the percentage whose tax rate declines, as a result of the AMT.

[41] We construct marginal tax rates by calculating the incremental tax that a taxpayer would pay if one of his income elements or deductions was one percent greater than the reported value. The tax rate on a specific income component for a given household is defined as the change in tax liability divided by .01 times the initial tax return entry, i.e., the change in tax divided by the change in income or the change in the deduction amount. We calculate these marginal tax rates under the assumption that all currently legislated tax rules remain in force unless they are changed by legislation that has already been enacted. We therefore assume that phase-outs of some of the tax provisions that were enacted in 2001 and 2003 will take effect as planned.

4.1 Marginal Tax Rate Increases and Decreases with the AMT

[42] Table 5 presents our findings with regard to the disparity between a taxpayer's ordinary income tax rate and AMT rate. It focuses on the effect of the AMT on the taxpayer's federal marginal income tax rate. Table 6 reports parallel calculations that consider the impact of the AMT on the taxpayer's combined federal and state tax rate, net of federal tax deductibility. The first column in Table 5 considers wages and salaries. Nineteen percent of the wages received by AMT taxpayers faces a lower marginal tax rate as a result of the AMT, while the remaining 81 percent faces a higher tax rate. The most common outcome is a tax rate increase of between zero and five percentage points, for example reflecting a move from the 33 percent bracket on the ordinary income tax schedule to 28 percent on the AMT schedule. Almost one tenth of the wage and salary income received by AMT payers is taxed at a rate 5 to 10 percentage points lower as a result of the AMT, with a roughly equal amount of wage income taxed at a rate 5 to 10 percentage points higher.

[43] The next two income categories that we consider are dividends and interest income. For both of these income categories, there is a greater chance of a decline in the marginal tax rate than there was for wage and salary income. More than 32% of dividends, and 26% of interest income, faces a lower marginal tax rate as a result of the AMT. A large share of both dividends and interest -- 35.1 and 41.2 percent, respectively -- faces marginal tax rates that are higher, but no more than five percentage points higher, as a result of the AMT.

[44] The remaining columns of Table 5 present similar calculations for long term gains, state income tax deductions, and charitable contributions. Each of these three income components or deductions is treated in a different way under the AMT. For long-term gains, 12 percent of the long-term gains reported by AMT taxpayers face the same tax rate with the AMT as without it. Because state and local taxes are not deductible in computing alternative minimum income, but they are deductible for ordinary income tax calculations, virtually all taxpayers face much higher marginal tax rates on these deductions under the AMT than under the ordinary income tax. For such taxpayers, the change in the marginal tax rate is large -- the full value of the AMT rate. Finally, the last column shows the impact of the AMT on the marginal income tax rate at which charitable contributions can be deducted. In part because contributions tend to be made by taxpayers in upper income brackets, for whom the AMT tends to reduce marginal tax rates, nearly three quarters of these contributions are deductible at a lower marginal tax rate as a result of the AMT.

[45] Table 6 shows the impact of the AMT on the combined federal and state marginal tax rates that taxpayers face on income and deductions. The effective marginal state income tax rate rises from (1-t_federal)*t_state to t_state for an itemizer under the ordinary income tax who switches to become an AMT payer. Comparing the results in Tables 5 and 6 shows how substantial this effect can be. For example, the percentage of wage income that experiences a 0 to 5 percentage point tax increase as a result of the AMT drops from 55 to 44%, while the percentage with a 5-10 percentage point increase rises from 10 to 21%.

[46] The effect of recognizing state income tax rates on the distribution of marginal tax rate changes for interest income and for dividend income is similar to that for wage income. For longterm gains, the share of gains received by AMT payers with no change in their marginal tax rates declines from 42% in the upper panel of Table 5 to 11.7% in Table 6, and the share received by those with marginal tax rate increases between zero and five percent rises from 16% to 72%. For state tax deductions, there is very little change as a result of the inclusion of state taxes.

4.2 Weighted Average Marginal Tax Rates

[47] Tables 5 and 6 offer some insights on how the AMT affects taxpayer incentives. The tables do not consider how many taxpayers face the AMT, however, and they do not yield single summary measures of the impact of the AMT. To address these needs, we compute weighted average marginal tax rates on a range of different income components received by all taxpayers. We include both AMT payers and other taxpayers, and thereby obtain a measure of the AMT's overall impact. We calculate the weighted average marginal rate on each income or deduction category by averaging marginal tax rates across taxpayers with weights equal to the share of aggregate income or deductions in that category received by each household.

[48] Table 7 shows the weighted average marginal income tax rates on four income components and on four deduction categories. The table shows weighted average marginal tax rates every three years between 2001 and 2013 both with and without the AMT. For the "without AMT" case we assume that the AMT is repealed but that there are no other changes in the personal income tax schedule. In practice, AMT repeal would presumably coincide with changes in the income tax law that would restore at least part of the revenue that is currently collected by the AMT. Because there are many possible ways in which the income tax could be modified to recoup lost AMT revenue, however, we decided not to make an arbitrary choice and report the associated results.

[49] Comparing the columns for 2001 and 2004 in Table 7 shows that the AMT has a relatively minor effect on weighted average marginal income tax rates in these years. For 2004, for example, the estimated weighted average marginal tax rate on wages is 22.6% with the AMT, and 22.3% without it. The only income or deduction category for which the weighted average marginal tax rate diverges by more than one percentage point is state income taxes, where the disparity in 2004 is nearly six percentage points.

[50] The weighted average marginal tax rates with and without the AMT diverge by more in 2010 than in the earlier years of the decade. For wages, the weighted average marginal tax rate is projected to be 1.5 percentage points higher in 2010 as a result of the AMT. Because the maximum AMT rate of 28 percent is much lower than the maximum regular bracket rates, it may be surprising that the AMT raises average marginal tax rates. However, the lowest AMT bracket rate of 26 percent is higher than the income tax bracket of 15 percent that a non-itemizing family of four will leave only at $78,500 ($2003), and it is slightly higher than the 25 percent ordinary income tax rate that is in effect through $175,000 of AGI. This explains why our earlier tables showed a substantial fraction of wage income received by AMT taxpayers facing a higher tax rate with, rather than without, the AMT.

[51] The marginal tax rate patterns for interest income and for dividends are similar to those for wages. In 2004, the AMT has a negligible effect on the weighted average marginal tax rate for interest income, and it raises the dividend tax rate by 0.3 percentage points. By 2010, however, the effects are larger. There is a 0.9 percentage point increase in the marginal tax rate on interest income, and a 0.7 percentage point increase in the tax rate on dividends. The distribution of dividend income is more skewed toward high-income households than the distribution of interest income, so a higher fraction of dividend recipients have marginal income tax rates that are above their marginal tax rate under the AMT.

[52] The fourth row in Table 7 shows the impact of the AMT on the marginal tax burden on capital gains. The weighted average marginal tax rate with the AMT is higher than that without the AMT, with the difference reaching its maximum at 0.7 percentage points in 2010. The difference is 0.2 percentage points in 2004. The table also shows that the weighted average marginal tax rate under current law is lower in 2004 than in 2001, reflecting the reduction in marginal tax rates that was enacted in 2003. For realized long-term capital gains, JGTRRA reduced the weighted average marginal tax rate by about four percentage points. The same tax act lowered the marginal tax rate on dividend income, and the third row of Table 7 shows the impact of this change: a ten percentage point drop in the weighted average marginal tax rate.

[53] The last four rows present information on how the AMT affects the weighted average marginal tax rates that apply to various personal income tax deductions. For state and local income tax deductions, shown in row five, the effect is dramatic. These tax payments are not deductible from alternative minimum taxable income, but they are deductible from taxable income under the ordinary income tax. In 2004, the weighted average marginal tax subsidy on state and local taxes was 18.2 percent. If the AMT had not been in force, this subsidy rate would have been 24.1 percent. The disparity is the result of roughly one third of state and local tax deductions being claimed by taxpayers who are in the AMT regime, and for whom these taxes do not generate a deduction. The disparity between the average marginal tax rate with and without the AMT is projected to grow over the next decade. In 2010, the weighted average marginal tax rate on state and local tax deductions is projected to be 9.4 percent with the current AMT in effect, compared with 26.9 percent if the AMT did not exist. The projections also show that expiration of the EGTRRA provisions in 2011 brings the two sets of weighted average marginal tax rates into closer proximity, with the weighted marginal tax rate under the status quo, with the AMT, rising from 9.4% in 2010 to 18.2% in 2013.

[54] The AMT has a smaller impact on the weighted average marginal tax rates for the other deductions that we consider. In 2004, for example, it has no effect on the weighted average marginal tax rate on medical deductions, and it affects the mortgage interest deduction by 0.4 percentage points and charitable contributions by 0.7 percentage points. In 2010, all of these disparities are larger. The weighted average marginal tax rate for medical deductions is 0.5 percentage points lower with the AMT than without it, while both the mortgage interest and charitable contributions deductions are at weighted averages that are nearly two percentage points greater under the AMT than without it.

[55] Table 8 presents weighted average marginal tax rate calculations similar to those in Table 7, but it includes the effect of the AMT on net-of-deductibility state income tax rates as well as the effect on federal rates. The level of the marginal tax rates in this table is higher than the level in Table 7, and the differences that result from the AMT are also larger. For wages, for example, the projected impact of the AMT in 2010 is an increase in the weighted average marginal tax rate of 2.2 percentage points, compared with 1.5 percentage points when only federal tax rates are considered. The impact of the AMT on some deduction items also increases. For mortgage interest, the weighted average marginal tax rate on deductions rises to 2.6 percentage points in Table 8, compared with 2.2 percentage points in Table 7.

[56] The weighted average federal marginal income tax rates in Table 7 are projections. To provide a base for comparison with both the level and the variation in similar marginal tax rates in the past, Table 9 presents data for years since 1960 on the weighted average marginal tax rates on wages, several different components of capital income, and several deductions. The data in the appendix represent an updated version of the time series for marginal tax rates in Poterba (2002). The table shows that the changes associated with the AMT, even in 2010, are in many cases smaller than the changes associated with significant tax reforms in the past. The Tax Reform Act of 1986, for example, reduced the weighted average marginal income tax rate on wage income by 4.3 percentage points over the 1986 to 1988 period. The marginal income tax rate reductions associated with ERTA, the 1981 tax legislation, were comparable in magnitude.

5. Reform Options and Their Impact on Weighted Average Marginal Tax Rates

[57] The foregoing analysis considered how various reform proposals would affect the number of taxpayers facing the AMT and the distribution of AMT payers across AGI categories. We have not considered how reforms would affect weighted average marginal tax rates. Table 10 presents such information for two income components, wages and interest income, and for state income tax deductions. This deduction is the one with the largest change in the weighted average marginal tax rate as a result of the AMT.

[58] The results in Table 10 show that all of the proposed reforms have the effect of reducing the marginal tax rates on wages and on interest income, although the magnitude of these effects vary across proposals. The proposals to subtract personal exemptions from alternative minimum taxable income and to index the AMT exclusion for inflation are the two that have the largest impact on the marginal tax rate on wages. In both cases the marginal tax rate in 2010 falls by over a percentage point as a result of these reforms. For interest income, these two proposals along with extending the expiring provisions in EGTRRA and JGTRA have the greatest marginal tax rate impact. Indexing and allowing personal exemptions under the AMT have significant effects on the weighted average marginal income tax rate that applies to state income tax deductions. Indexing, for example, would raise this marginal tax rate by nearly five percentage points in 2010. This is the largest absolute effect of any of the proposed reforms on any of the marginal tax rates in Table 10. Given that the AMT has modest effects on most marginal tax rates, it is not surprising that the effect of most of the reform proposals is also modest.

6. Conclusions and Future Directions

[59] This paper presents new evidence on the growth of AMT liabilities over the next decade, and the incentive effects associated with this growth. Our baseline projections confirm the widely documented pattern that because the AMT exclusion level is not indexed, there will be rapid growth in the number of AMT taxpayers, and in the amount of revenue collected by the AMT, until 2010. After 2010, when a number of provisions in the 2001 tax reform are scheduled to phase out, the number of AMT taxpayers will decline, but it will rise again in subsequent years from the lower post-2010 base.

[60] Although the AMT creates substantial changes in many aspects of the income tax system, we find that the average marginal tax rates on many income components, such as wages and interest income, are affected only modestly by growth of the AMT. In 2010, for example, we project that the AMT will raise the weighted average marginal tax rate on wage income by 1.3 percentage points relative to what it would be if the AMT were repealed and no other tax changes were enacted. For interest income, the effect on the weighted average marginal tax rate is just below one percentage point. These changes conceal larger changes in marginal tax rates for individual households. Some are pushed from marginal tax rates of 15 or 25 percent under the ordinary income tax into the AMT brackets of 26 and 28 percent. Others drop from marginal tax rates above 30 percent under the income tax to the AMT rate of 28 percent.

[61] Our results are based on stylized assumptions about the rate at which aggregate income, population, and the price level will grow over the next decade. All of our analysis uses the same underlying assumptions to "age" income tax records from 1999 through 2013. It would be useful to gauge the sensitivity of our findings to these assumptions, and to explore how differences in growth rates or more importantly in the rate of inflation between 2003 and 2010 would affect the magnitude of AMT liabilities.

[62] One of the under-studied issues associated with the AMT concerns the impact of this tax on the tax liability of single households and married couples. The effect of a change in household status on tax liabilities is different when the taxpayers face the AMT than when they face the regular income tax schedule. Gravelle (2001) has noted this potentially important effect. The growing significance of the AMT may therefore call into question the traditional analysis of the "marriage penalty" and related features of the income tax system. Future research should consider this aspect of the distribution of AMT liabilities, along with potential reforms to address this issue.

REFERENCES

Burman, Leonard, William Gale, and Jeffrey Rohaly. "The AMT:Problems and Potential Solutions," National Tax Journal 55 (September 2002), 555-596.

Burman, Leonard, William Gale, and Jeffrey Rohaly. "The Expanding Reach of the Alternative Minimum Tax." Journal of Economic Perspectives 17 (Spring 2003a), 173-186.

Burman, Leonard, William Gale, and Jeffrey Rohaly. "The AMT: Projections and Problems," Tax Notes (July 7, 2003b), 105-117.

Feenberg, Daniel R., and Elisabeth Coutts. "An Introduction to the TAXSIM Model." Journal of Policy Analysis and Management 12 (Winter 1993), 189-194.

Gravelle, Jane. "The Individual Alternative Minimum Tax: Interaction with Marriage Penalty Relief and Other Tax Cuts." Congressional Research Service, June 12, 2001.

Harvey, Robert P. and Jerry Tempalski. "The Individual AMT: Why it Matters," National Tax Journal 50 (1997), 453-473.

Kiefer, Donald, Robert Carroll, Janet Holtzblatt, Allen Lerman, Janet McCubbin, David Richardson, and Jerry Tempalski. "The Economic Growth and Tax Relief Reconciliation Act of 2001: Overview and Assessment of Effects on Taxpayers." National Tax Journal 55 (March 2002), 89-118.

Poterba, James. "Taxation, Risk-taking, and Portfolio Behavior." In Alan Auerbach and Martin Feldstein, eds.: Handbook of Public Economics: Volume 3 (North Holland, Amsterdam, 2002).

Rebelein, Robert, and Jerry Tempalski. "Who Pays the Individual AMT?" Washington: U.S. Treasury Department, Office of Tax Analysis Paper 87, 2000.

Sullivan, Martin. "AMT Changes Everything." Tax Notes (September 23, 2002), 1678-1683.

Tempalski, Jerry. "The Impact of the 2001 Tax Bill on the Individual AMT." Proceedings of the 94th Annual National Tax Association Conference on Taxation. National Tax Association: 2002. 340-348.

U.S. Congress, Joint Economic Committee. The Alternative Minimum Tax for Individuals: A Growing Burden. Washington: Joint Economic Committee, 2001.

Table 1: Projected Number of Alternative Minimum Tax Returns

                            Percent of                   AMT Revenue/Total

               Filers                        Tax Liability

Table 3: AMT Liability as a Percentage of Adjusted Gross Income, By AGI Category, 2001-2013

AGI Class           2001      2004      2007      2010     2013

             Allow      Extend     Marriage  Index    Add $3,000  Add $10,000

Marginal Tax

Marginal Tax

          With AMT                      Without AMT

          With AMT                      Without AMT

Year    Wage   Interest  Dividends            Mortgage  Pension

               Wages

 

[Figure 1]

 

Regular and Tentative AMT Effective Rate in 2003 and 2010

 

Joint Filing, No Deductions and 2 Dependents

 

 

 

 

[Figure 2]

 

Evolution of Average Tax Rates 2002-2010

 

Fixed Distribution of Income 1995-1999

 

 

 

 

[Figure 3]

 

Effect of Raising Exclusion in 2010