Tax policy and income inequality in the united states, 1979-2007.

Bargain, Olivier ; Dolls, Mathias ; Immervoll, Herwig 等

I. INTRODUCTION

Over the past decades, incomes have become more unequally distributed in most Organisation for Economic Co-operation and Development (OECD) countries and especially in the United States (OECD 2011). In particular, the increase of the top 1%'s income share has received considerable attention (e.g., Piketty and Saez 2003) resulting in numerous calls for higher taxes on the rich (e.g.. Diamond and Saez 2011; Piketty, Saez, and Stantcheva 2014). Yet, very little is known about the actual impact of past tax policy changes on inequality. The reason is that the usual evaluation approach--comparing income inequality measures before and after taxes (see, e.g., Gottschalk and Smeeding 1997 or Heathcote, Perri, and Violante 2010)--is not able to isolate the pure policy effect because tax burdens are determined by both tax policy and pretax income distribution. For instance, a given progressive income tax schedule redistributes more when the distribution of taxable incomes becomes more dispersed, and not at all if everybody earns the same (Dardanoni and Lambert 2002; Musgrave and Thin 1948). Hence, it is unclear how much of an observed change in tax liabilities (and resulting inequality) is due to policy reforms and what part is due to other factors, notably the change in the underlying pretax income distribution.

This paper is the first to isolate and quantify the pure tax policy effect on inequality in the United States for an extended time period spanning almost three decades (1979-2007). It can be seen as a natural follow-up of the study by Piketty and Saez (2007) who analyze changes in the progressivity of the federal income tax over time but cannot disentangle policy changes from other factors. We also use tax return micro-data from the Statistics of Income (SOI) division of the Internal Revenue Service (IRS) (1) and the NBER's TAXSIM calculator for our analysis. In a first step, we perform a series of detailed counterfactual simulations by applying current and next year tax policy rules to current and next year incomes, respectively. The resulting decomposition shows how the distribution of posttax income would have looked like if either tax policy (federal and state level income and payroll taxes) or the distribution of pretax incomes had remained unchanged between two given years. This allows us to quantify the direct tax policy effect on inequality and to provide (descriptive) evidence on the extent to which the increase in inequality, particularly the surge in top income shares, is market driven or related to major U.S. tax reforms during the past three decades. (2) In addition, we extend the baseline decomposition by using estimates of the elasticity of taxable income (ETI, see Saez, Slemrod, and Giertz 2012 for a survey) in order to account for indirect policy effects due to behavioral responses (such as labor supply, income shifting, or migration).

The main findings are as follows. The baseline decomposition shows that the size of the policy effect corresponds to 11%-29% of the total change in income shares of different income groups. The simulated impact of tax policy is largest for taxpayers in the 95th to 99th percentile of the income distribution, but smallest for those in the top 1%. This suggests that tax policy had a non-negligible effect on changes in inequality, but explains only a small fraction of the sharp increase of the top 1%'s income share, where other forces played a much more important role. Extending the baseline decomposition and accounting for indirect policy effects (i.e., behavioral responses) do not affect our results qualitatively, but yields a larger overall policy effect on inequality (up to 41% of the total change). We also find that reforms in the 1980s and early 2000s exacerbated trends of growing inequality, those in the early 1990s benefited lower-income taxpayers. Taken together, the cumulative policy effect over the entire period contributed to the increasing income share of taxpayers in the top quintile (and especially the top decile) at the cost of middle-class taxpayers. Hence, without any tax policy changes, inequality levels in the mid-late 2000s would likely have been significantly lower than those that were actually observed.

II. RELATED LITERATURE

Rising income inequality in the United States has stimulated a large body of research examining the underlying driving factors. In this literature, several strands have emerged which focus on different types of inequality. While the focus of this paper is on redistribution and the impact of tax policy on trends in posttax income inequality, this cannot be comprehensively assessed without taking into account trends in pretax inequality.

The existing evidence on pretax inequality in the United States points to a widening gap in the last 30 years, in particular at the top of the distribution. In a seminal contribution, Piketty and Saez (2003) (updated 2012) build a series of pretax income shares based on tax return data from the IRS. They find that inequality grew relatively smoothly in the time period considered here. Further studies relying on IRS tax return data are, among others, Slemrod (1992), Feenberg and Poterba (1993), Bakija, Cole, and Heim (2012), and DeBacker et al. (2013). Similar trends are found in analyses using CPS data. (3) A general conclusion from these studies is that total income inequality, i.e., inequality in pretax, post-transfer income rose sharply in the 1980s, and that this growth continued at a reduced pace in the 1990s and early 2000s. Burkhauser et al. (2012a) seek to reconcile findings from IRS SOI and CPS data. They use internal CPS data which are--compared with public-use CPS--much less affected by topcoding (although a number of other measurement and conceptual differences remain) and apply similar income definitions as Piketty and Saez (2003) do, namely pre-transfer, tax-unit income. They conclude that the rise in inequality from 1993 onwards is mainly due to gains made by the top 1% of the income distribution. A related strand of the literature examines how pretax or taxable income is affected by behavioral responses to tax policy changes. For instance, it has been shown that the Earned Income Tax Credit (EITC) reforms had a substantial impact on participation rates of married couples and single mothers (cf. Eissa and Hoynes 2006 and Eissa, Kleven, and Rreiner 2008, among others). In addition to adjustments in participation or hours worked, tax reforms may affect other margins such as tax evasion or the timing of capital gains realizations with the latter two of particular relevance at the period around the Tax Reform Act (TRA) of 1986 (Auerbach 1988; Auerbach and Slemrod 1997; Feenberg and Poterba 1993; Slemrod 1996). More disputed is the question to what extent the increase in inequality, especially at the top of the distribution, is due to responses such as income shifting from corporate to personal income, that is, if the documented increase is real or caused by behavioral responses (see in particular Saez 2004 and Reynolds 2007). The recognition of the importance of these responses has led to the growing tax responsiveness literature focusing on the ETI which shall capture all these behavioral responses (see Saez, Slemrod, and Giertz 2012 for a survey).

We contribute to the literature which examines the impact of tax policy on posttax income inequality. By extracting the direct policy effect through counterfactual simulations, we complement the analyses conducted by Piketty and Saez (2007) or the Congressional Budget Office (2010). In these studies, shares of posttax income and average federal tax rates are calculated for various income groups and similar time periods. However, their estimates do not isolate the direct policy effect since they reflect both legislative changes as well as other factors that influence income and tax rates. Some studies have conducted so-called "what if' calculations (Poterba 2007), but to the best of our knowledge, none of these papers have sought to identify a policy effect on a year-by-year basis over a long time period. We are aware of two early contributions which explicitly consider--via counterfactual simulations--the impact of tax policy on the posttax income distribution. Lindsey (1987) applies this methodology to estimate taxable income elasticities in response to the Reagan tax reform in the early 1980s. Gramlich, Kasten, and Sammartino (1993) apply tax and transfer policies of 1980 and 1985 to the pretax income distribution of 1990. They report that 16% of the increase in the Gini coefficient from 1980 to 1990 is due to changes in taxes and transfers. More recently, Poterba (2007) conducts conceptually similar policy swaps by applying 2004 effective tax rates to the 2000 pretax income distribution and vice versa and examines the resulting effects on the share of posttax (but before payroll tax) income accruing to various income groups. A key finding from his analysis is that the impact of changes in the pretax income distribution is approximately four times as large as the policy effect of changes in effective tax rates. (4)

III. EMPIRICAL APPROACH

A. Decomposition Methodology

In order to decompose inequality changes into the effect of tax policy and all other factors, we follow and extend the approach suggested by Bargain and Callan (2010). It is important to note that, by construction, in the baseline decomposition the "tax policy effect" measures only the direct effect of tax policy on the (given) income distribution abstracting from any behavioral responses (such as changes in labor supply or income shifting) or general equilibrium effects as a consequence of tax policy changes (e.g., due to differential growth or changes in labor demand or immigration). (5) These are captured by the "other effect" which additionally includes all exogenous changes to the income distribution which occur independent of tax policy. In an extension of our baseline decomposition, we try to additionally quantify the endogenous indirect tax policy effects.

Consider a data matrix y containing information on individuals' pretax income from different sources as well as various individual and household characteristics which are relevant for the calculation of income and payroll taxes. The tax function d represents the rules and structure of the tax system (e.g., marginal tax and contribution rates), while vector p accounts for all the monetary parameters (e.g., tax brackets). The distribution of posttax income is represented by [d.sub.i] ([P.sup.i], [y.sup.l]) for tax rules of year i, tax parameters of year j, and nominal incomes (and characteristics) of year l. For counterfactual simulations, it is necessary to nominally adjust income levels by an uprating factor a accounting for nominal changes (e.g., inflation) between base and end year.

For the decomposition, two different approaches are possible: it can be conducted either on base year or on end year incomes while applying tax policy of the respective other year. The two approaches usually lead to identical results. If they differ, this is an indication of (1) general equilibrium effects of tax policy, (2) changes to the tax base (with some items available only pre- or posttax base change), or (3) income shifting from other income sources to personal income, as discussed below. In the first approach, the counterfactual situation [d.sub.t+1] ([p.sup.t+1], [[alpha].sup.t+1] + [y.sup.t]) represents post-tax incomes obtained by applying tax rules and parameters of year t + 1 on year t data nominally adjusted to year t + 1. Here the policy of end year t + 1 is applied while holding the pretax incomes of year t constant. In the second approach, the initial policy from year t is applied to the pretax income distribution in t + 1. For this, we need to construct a counterfactual [d.sub.t][[p.sup.t], ([y.sup.t+1]/ [[alpha].sup.t+1])], where pretax incomes are adjusted with the same factor [[alpha].sup.t+1] used to scale up the distribution of pretax income between period t and t + 1. (6) As further explained below, policy changes usually combine changes in policy structure d and changes in parameters p (the "uprating policy").

In the empirical part, we are interested in distributional measures M, computed as a function M[[d.sub.i] -([p.sup.i], [y.sub.l])] of the simulated distribution of posttax income. The advantage of the present approach is that we can use any measure and not only those with specific properties (i.e., decomposable inequality indices). More generally, it is possible to decompose any scalar M such as inequality indices, (top) income shares, average and marginal tax rates, or measures of redistribution. Characterize the total change [DELTA]M in measure M between initial and final period as

(1)

[DELTA] M = M [[d.sub.t+l] ([p.sup.t+1], [y.sup.t+1])] - M [[d.sub.t] ([p.sup.t], [y.sup.t])]

and notice that the last term can also be written M[[d.sub.t] ([[alpha].sup.t+1], [p.sup.t], [[alpha].sup.t+1] [y.sup.t] since function d is linearly homogenous in p and y. (7) Then, the total change between periods t and t + 1 can be decomposed into a change in tax policy and a change in the pretax income distribution. We refer to the last change as the "other effect." The "policy effect" can be assessed on base-year data followed by a change in the underlying data conditional on the new policy. Decomposition I can thus be

written as

(2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Here, the end-period tax system is evaluated on nominally adjusted base period data [[alpha].sup.t + 1] [y.sup.t]. Symmetrically, the decomposition can be written as a policy effect assessed on end-period data [y.sup.t+1], and in this case, the other effect is assessed on the base period tax system, yielding decomposition II:

(3) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].

Other effect II

In this case, base period tax parameters are applied to end-period data [y.sup.t+1] after nominal adjustment, that is, [y.sup.t + 1] / [[alpha].sup.t + 1].

Note again that, by construction and in line with Piketty and Saez (2007), both policy effects in the baseline decompositions capture only the direct effect of tax policy on the income distribution. Behavioral responses to tax policy changes are attributed to the other effect (together with exogenous changes to the income distribution). In Section IV.C, we further decompose the other effect into an indirect policy effect and a residual effect.

As the decompositions are path-dependent, we suggest to simply average both policy and other effect over the decompositions I and D. This corresponds to Shorrocks's (1999) reinterpretation of the Shapley value procedure. In the empirical part, we verify that results based on decompositions I and II usually do not differ (much). Exceptions indicate that significant (behavioral or conceptual) changes between base and end year occurred, which were not captured in one year's data but present in the other years' data (such as income shifting between the corporate and private sector in anticipation of TRA86 or realizations of capital gains, as discussed below).

Notice that policy and other effect are affected by the choice of the uprating parameter a. The way tax brackets are uprated by governments can have important implications for the income distribution in the long run. Usually there are three options: (1) no uprating, (2) uprating according to the level of price inflation, and (3) uprating according to the level of earnings growth. With non-indexation of tax brackets in progressive systems, or price indexation when incomes rise faster than prices, the total number of taxpayers (and the number of higher-rate taxpayers) increases. This phenomenon of bracket creep is likely to affect the final distribution of post-tax income. In our empirical application, we use changes in the consumer price index (8) to adjust pretax incomes in the counterfactuals which is equivalent to an indexation of tax brackets. This reference situation is extensively used in policy analyses of tax reforms (compare discussion in the study by Clark and Leicester 2004). In a robustness check, we rely on a more conservative approach based on nominal wage growth, that is, a distributionally neutral scenario (Bargain and Callan 2010).

B. Data

Several data sources have been used in studies on the impact of taxation on income inequality, in particular, tax return data (e.g., Piketty and Saez 2007) and household surveys such as the CPS (e.g., Aim, Lee, and Wallace 2005). It is well known that there are advantages and disadvantages for both types of data (Poterba 2007). In brief, tax return data allow to precisely calculate top income shares, but do not contain information about non-filing households (typically at the bottom of the distribution) and lack certain (not tax-relevant) components of household income. In this study, we use large public-use files of tax return micro-data from the SOI division of the IRS. (9) Annual cross-sectional micro-data are available from the SOI since 1960, but given that TAXSIM is able to simulate state level income

taxes only from 1979 onwards, we start our analysis in 1979. (10)

We follow Piketty and Saez (2007) in terms of sample selection and include both filing and non-filing tax units so that income groups such as quintiles or top percentiles are based on the total population. Non-filing tax units are imputed following Piketty and Saez (2007), assuming that they earn 20% of the average income of filing units. Because of this imputation, we usually do not report the decomposition results for the bottom quintile (P0-20) which consists of households with low market incomes including nonfiling tax units (their share varies between 4% and 8% of all taxpayers). (11)

Throughout this paper, we focus on pre- and posttax income inequality. Tax units are ranked based on pretax incomes excluding capital gains as they are not a regular stream of income. For all subsequent calculations, capital gains are added back to pre--and posttax incomes. Pretax income includes all sources of market income which are reported on tax returns, that is, wages and salaries; bonuses and exercised stock-options; employer and private pensions; self-employment income; business income; dividends, interest, and rents; and realized capital gains. Posttax income is defined as pretax income minus the simulated components of the income tax system including federal and state level income taxes, employee social insurance contributions (payroll taxes), and tax credits (e.g., EITC). As is common in the literature, we thus assume that the burden of the taxes is borne by those who remit them and is not shifted elsewhere through adjustments in pretax wages and prices. Our measures of income do not include imputed corporate or federal estate and gift taxes. It is important to note that the policy effect, which is the focus of this study, is not affected by (omitting) these taxes given that we simulate them neither in the baseline nor in the counterfactual scenarios. (12)

In order to calculate (federal and state) income and payroll taxes, we use NBER's simulation model TAXSIM. (13) We use simulated taxes for all computation of taxes and post-tax incomes--including the observed case of current year income and current year tax policy. A comparison of tax liabilities observed in the income tax return data with the simulated TAXSIM output leads to a perfect match in more than 99% of all cases for each year with no systematic differences across the income distribution. The simulation approach allows us to conduct controlled experiments by changing the parameters of interest while holding everything else constant. When assessing the isolated role of tax policy on income inequality (i.e., the policy effect), we are thus able to account for changes in federal and state level income taxes as well as payroll taxes and tax credits. Importantly, the estimated policy effect is solely affected by the changes in these taxes. However, it should be noted that this approach can identify causal effects of tax policy on inequality only under the assumption of an exogenous pretax income distribution. Owing to (potential) changes in the distribution of market income through indirect effects of tax policy, our results should be interpreted as association rather than causation. We address this issue in Section IV.C when we discuss indirect policy effects.

C. U.S. Tax History

In this section, we briefly outline the major changes in the U.S. income tax system from 1979 to 2007 which are also summarized in Table S2 in the Supporting Information. We concentrate on large legislative changes which drive the tax policy effect. Reforms of interest are the Economic Recovery Tax Act of 1981 (ERTA81), the TRA of 1986 (TRA86), the Omnibus Budget Reconciliation Act of 1990 and 1993 (OBRA90 and OBRA93), the Taxpayer Relief Act of 1997 (TRA97), the Economic Growth and Tax Relief Reconciliation Act of 2001 (EGTRRA01), and the Jobs and Growth Tax Relief Reconciliation Act of 2003 (JGTRRA03).

ERTA81 introduced the indexation of individual income tax parameters which became effective in 1985. Tax cuts were phased in over the years 1982-1984, with a reduction of top marginal tax rates from 70% to 50% in 1982 and of other tax rates by 23% in three annual steps. Further, the income threshold for the top rate substantially increased from $85,600 in 1982 to $109,400 (1983) and $162,400 (1984) for married couples filing jointly. Similarly, thresholds were increased for couples filing separately and for singles. The reduction in tax revenue amounted to 2.89% of GDP (4-year average, cf. Tempalski 2006 for estimates of revenue effects mentioned in this section). Key aspects of TRA86 were the broadening of the tax base and reductions in marginal tax rates. Overall, the reform was almost revenue neutral. (14) TRA86 further lowered the top marginal rate to 38.5% in 1987 and to 28% in 1988, reduced the number of tax brackets from 15 in 1986 to four in 1988, and also substantially expanded the EITC with financial benefits for low-income households.

OBRA90 contained increases in income taxes as well as expansions of the EITC and other low-income credits. Furthermore, payroll taxes were increased by lifting the taxable maximum for Medicare which was finally abolished in 1994. OBRA93 then led to the largest single expansion of the EITC (cf. Eissa and Hoynes 2011), and further increases in income tax rates were implemented, for example, the top rate rose from 31% to 39.6% in 1993. The EITC became much more generous in 1994 with higher maximum credits and an expansion to single workers with no children. The EITC was further expanded in the following years. The revenue effect of OBRA90 and OBRA93 was--again evaluated on a 4-year average--positive and amounted to 0.5% and 0.63% of GDP, respectively. TRA97 lowered capital gains tax rates and introduced additional tax credits (child and education tax credits).

EGTRRA01 and JGTRRA03 were characterized by reductions in marginal tax rates, both for low- and high-income families, expansions of the child tax credits, and reductions in taxes on dividends. In 2003, JGTRRA accelerated those provisions of EGTRRA which were not set to become effective until 2006. Both reforms had a revenue-decreasing effect (-0.71% and -0.57% of GDP, 4-year average).

IV. DECOMPOSITION RESULTS

A. Major Tax Reforms

We start our analysis by illustrating the decomposition procedure for each major tax reform in our sample period. In Tables 1-4, we compare average tax rates (including federal and state level income as well as payroll taxes) and posttax income shares for various income groups before the start of the reform and after it was fully phased-in (base and end year). We decompose the total change into two components as explained in Section III. The first is due to tax reforms (policy effect) while the second is due to changes in the pretax income distribution (other effect) which may include indirect policy effects. The left part of the table reports the different components of the decomposition, including base and end-period baselines (columns (1) and (4), respectively), the two relevant counterfactuals as well as the total change. Columns (2) and (3) show the counterfactuals with average tax rates and income shares given end-period pretax incomes and base period tax legislation (column (2)), and base period pretax incomes and end-period tax legislation (column (3)). The right part of Tables 1 -4 reports both the policy and the other effect for decompositions I, II, and the Shapley value (i.e., the mean of the two decompositions). As they yield almost identical results in most cases, we will focus on the Shapley value. An exception is TRA86 (see Table 2) where the difference between decompositions I and II does matter which is discussed below. Reassuringly, the policy effect is (close to) zero in years without (major) tax changes and the observed change equals the other effect.

Policy versus Other Effect. The policy effect reveals how average tax rates and income shares would have changed under constant pretax incomes but changing policy. It is based on a counterfactual scenario in which the composition of pretax incomes remains constant, and pretax incomes grow in accordance with the inflation rate which is used for parameter adjustments in the counterfactuals. (15) Adding the policy effect to the baseline values yields counterfactual values of average tax rates and income shares under "constant pretax incomes," but changing policy parameters. A positive (negative) value of the other effect implies that the average tax rate of a given income group would have increased (decreased) in the absence of direct policy changes. This can either be due to pretax income growth above (below) the inflation rate or due to changes in the composition of pretax incomes or tax units. (16) In the case of posttax income shares, the interpretation slightly differs as an increase in the income share of one group automatically implies that the share of at least one other group must have decreased. Here, the other effect shows how income shares would have changed in the absence of direct policy changes.

From 1981 to 1984, the period around ERTA81 (see Table 1), average tax rates decreased for all income groups. Starting with the other effect, we observe that changes in pretax incomes have pushed average tax rates up only for the top 0.1% of the population. For all other income groups, average tax rate would have decreased even in the absence of the tax reform due to the recessionary period in the early 1980s or due to indirect policy effects. Results for the policy effect show that with the exception of the second quintile, legislative changes led to reductions in average tax rates which were largest for the upper part of the income distribution. As an example, the cumulative policy effect from 1981 to 1984 reduced the average tax rate for those in the top 0.01% by 5.6 points, while the negative policy effect for the third and fourth quintiles (-0.1 and -0.7, respectively) was only marginal. With regard to the absolute size of policy and other effect, the reduction in average tax rates due to changes in pretax incomes and indirect policy effects was larger than the reduction caused by the direct policy effect for taxpayers up to the fourth quintile. For those in the ninth decile and above, the absolute size of the policy effect was larger than the other effect. Moving to the effect of ERTA81 on inequality, we find that income shares for those below (above) the 80th percentile would have decreased (increased) in the absence of any (direct) policy changes. The direct policy effect strengthened this effect. We conclude that ERTA81 exacerbated the increase in inequality such that posttax income shares were more unequal in 1984 compared with a counterfactual of no policy changes.

Contrary to ERTA81, TRA86 (see Table 2) contained both inequality-increasing (reduction in top marginal tax rates) and inequality-decreasing elements (expansion of EITC, tax base broadening). Table 2 reveals that it was mainly the top 1 % which experienced substantial reductions in average tax rates and increases in their income shares. Furthermore, our decomposition for TRA86 shows that it makes a difference if the policy and other effect are evaluated on base or end-period data. This can be explained by behavioral reactions, in particular, income shifting and timing responses. Capital gains realizations peaked in 1986 in anticipation of the increase in the marginal tax rate on realized long-term capital gains from 20% to 28% in 1987. Furthermore, taxpayers shifted income from the corporate to the individual sector as a response to the reduction of the top marginal rate which fell from 1986 to 1988 in two steps from 50% to 28% and thus below the basic corporation income tax rate (see, e.g., Auerbach 1988; Feenberg and Poterba 1993; and Slemrod 1996). (17) For taxpayers in the top 1%, the hypothetical average tax rate with 1988 policy parameters, but 1986 pretax incomes (column (3)) would have been much higher than the observed average tax rate in 1988 (column (4)) due to the fact that a substantially larger share of their income in 1986 consisted of long-term capital gains which were taxed at a higher rate in 1988. Conversely, the hypothetical average tax rate with 1988 pretax incomes, but 1986 policy parameters (column (2)) picks up the effect of a larger share of wage and entrepreneurial income reported by affluent taxpayers in 1988. Differences between decompositions I and II are thus driven by behavioral responses of taxpayers which caused a dramatic change in their income composition around TRA86. (18)

OBRA90 and OBRA93 (see Table 3) counteracted the growing inequality at that time--at least to some extent. The other effect on post-tax income shares was negative for those below the 80th percentile implying that their income share would have declined substantially. Owing to expansions of the EITC the policy effect led to a considerable reduction in average tax rates of those in the lower half of the distribution, in particular, in the second quintile, while increases in marginal rates caused average tax rates to rise in the upper half of the distribution. This effect was strongest at the top of the distribution where the policy effect increased average tax rates, for instance, for those in the top 0.01% by more than 11 percentage points. Unsurprisingly, the cumulative policy effect of OBRA90 and OBRA93 on the income share of the top 1% was negative, while it was positive for the rest of the population and again largest for those in the second quintile.

Decomposition results for EGTRRA01 and JGTRRA03 (see Table 4) show that, similar to previous periods, the other effect pushed average tax rates up only for the top 1%. The tax cuts enacted in 2001 and 2003, however, led to substantial reductions in average tax rates across the distribution, with strongest policy effects--in absolute terms--at the top of the distribution. The positive policy effect on the income share of those at the top underlines the inequality-increasing effect of the Bush tax cuts.

[FIGURE 1 OMITTED]

B. Cumulative Effects over Time

In this section, we focus on the cumulative effect of changes in policy and pretax incomes on average tax rates and income shares over the entire period. (19) An extension of our baseline decomposition which additionally includes indirect policy effects resulting from behavioral responses to tax changes is presented in Section IV.C.

Average Tax Rates. We first turn to the results for average tax rates. Figure 1 shows how the total average tax rate developed from 1979 to 2007 (black diamond). Additionally, we consider two counterfactual scenarios. These counterfactuals reveal how the average tax rate would have changed if either policy parameters or pretax incomes would have remained as observed in the base year 1979. Over the entire period, the policy effect (other effect) pushed the average tax rate down (up) as can be seen by the hollow (black) triangles. In particular, policy changes implemented in the 1980s and early 2000s had a dampening impact on the total average tax rate, while the reforms in the early 1990s to some extent reversed the Reagan tax cuts. If policy parameters had remained constant on their 1979 level, the total average tax rate would have almost constantly grown from 1982 until 2000. This is due to the fact that total income grew faster than the inflation rate which is used to adjust pretax incomes in the counterfactuals.

Clearly, any diverging trends across income groups are hidden behind this aggregate average tax rate. Therefore, in Figures 2 and 3, we plot changes in average tax rates for income quintiles and fractiles of the top 1% comparing the actual change (left panel) with the counterfactual scenario of constant pretax incomes (right panel), respectively. Hence, the right panel shows how average tax rates would have developed if the pretax income distribution had remained on its 1979 level and only policy parameters had changed over our sample period. Importantly, the difference between these two series is given by the other effect capturing the impact of changes in pretax incomes on the average tax rate conditional on constant policy parameters.

[FIGURE 2 OMITTED]

We start with the income shares reported in Figure 2. Several important findings stand out. First, in absolute terms, the dampening policy effect on average tax rates was smallest for the third (P40-60) and fourth (P60-80) quintiles and largest for the second (P20-40) and fifth (P80-100) quintiles. Hence, it is the middle and upper middle class which benefited least from changes in tax policy. Second, taxpayers in the top quintile benefited more from tax policy than is visible in the left panel due to the fact that the other effect pushed their average tax rate up. The opposite is true for all other taxpayers for whom the other effect had a dampening effect on the average tax rate. Third, the right panel gives an indication of how the political cycle might have affected average tax rates at different parts of the income distribution. In short, the tax burden on high-income taxpayers (fourth and fifth quintiles) was reduced under the Republican administrations in the 1980s and early 2000s, whereas lowincome taxpayers (second quintile) faced largest reductions under the Democratic administrations in the 1990s. (20) The picture for the third quintile is different as their tax burden first rose under Republican administrations in the 1980s, but was subsequently reduced to a similar extent under Democratic (1990s) and Republican (early 2000s) administrations.

Figure 3 shows that policy changes affecting the top 1% of taxpayers had a much stronger impact on average tax rates than for the rest of the population. Even within this group, taxpayers were affected rather differently. Policy changes reduced the average tax rate of taxpayers located within the 99-99.5 fractile by roughly three percentage points, but by more than 12 points for those in the top 0.01% between 1979 and 2007. Observed changes in average tax rates were mainly driven by the policy effect. In contrast to the results for taxpayers in the top quintile (Figure 2), the other effect did not push up average tax rates of the richest taxpayers despite the tremendous income growth this group experienced over the sample period. As discussed above, the negative other effect on average tax rates for those in the top 1 % was largely due to the changing composition of their pretax incomes, partly caused by behavioral reactions around TRA86.

[FIGURE 3 OMITTED]

Income Shares. Now we turn to the effect of tax policy on inequality. The left panel of Figure 4 shows how posttax income shares of taxpayers in the second to fifth quintiles have changed relative to the base year pretax income share, while the policy effect on posttax income shares is shown in the right panel. We find a stark contrast between the observed change in income shares and the policy effect. The income share of those in the top quintile increased by roughly 24% over the whole sample period, whereas all other groups saw their income shares declining, with cumulative losses ranging from 22% (fourth quintile) to 28% (second quintile). Tax policy contributed to the increase (decrease) in the income share of those in the top (third and fourth) quintiles with an overall direct policy effect of roughly 1% (minus 2%). Remarkably, the cumulative policy effect on the income share of those in the second quintile almost canceled out over time. The direct tax policy effect was equalizing in some periods and disequalizing in others which is in line with the results for the policy effect on average tax rates. Again, the different sub-periods broadly coincide with the political cycle.

Figure 5 shows results for taxpayers in the top 1%. As for average tax rates (Figure 3), observed changes in income shares as well as the policy effect were much larger at the top of the distribution than for any other income group. For instance, from 1979 to 2007 the income share of those in the top 0.01% has risen by 350% with the direct policy effect contributing 18% to the increase. Interestingly, the highly disequalizing direct policy effect in the 1980s was almost completely reversed after OBRA93, but the tax cuts in the early 2000s reinforced the overall increase in inequality.

[FIGURE 4 OMITTED]

Relative Importance. The difference in scale of the left- and right-hand-side panels in Figures 4 and 5 suggests that changes in pretax incomes (the other effect) were the main driver of the total change in inequality. While this finding clearly confirms general perceptions about the roots of increasing inequality, it does not account for the fact, however, that the direct policy effect was equalizing in some periods and disequalizing in others and that these differential effects to some extent canceled out over the period of analysis. Calculating the mean of the absolute values of the policy effect and the total change, respectively, and expressing the former as a fraction of the latter, we find a non-trivial impact of policy changes. This is shown in Table 5 for average tax rates and income shares. Columns (1) and (3) present baseline results for the direct policy effect without behavioral reactions (see Section IV.C for the total policy effect including indirect policy effects). Unsurprisingly, column (1) reveals that policy changes matter more for average tax rates than changes in pretax incomes with the policy effect as a fraction of the total change ranging from 51% to 99% depending on the income group. Interestingly, the importance of the policy effect relative to the other effect is highest for taxpayers in the 95th to 99th percentiles. The corresponding value for those in the top 1% is much smaller (77.5%). The reason is that the other effect was larger for the top 1% than for those in the 95th to 99th percentiles due to the tremendous income growth experienced by the richest taxpayers.

For income shares the mean (absolute) policy effect expressed relative to the mean (absolute) total change is lower, but still substantial and ranges between 11% and 29%. Here, among all taxpayers, the relative importance of the policy effect is smallest for those in the top 1% which again reflects the fact that pretax income growth was the main driver of their rising post-tax income shares. These findings are in line with Poterba (2007) who shows that the effect of changes in pretax incomes on the posttax income distribution was four times as large as tax policy changes in the early 2000s.

[FIGURE 5 OMITTED]

C. Indirect Policy Effects

The baseline decomposition presented above singled out the direct policy effect, thereby ruling out potential behavioral responses to tax policy changes. These were captured by the other effect together with non-tax-related changes to the income distribution. However, behavioral responses to tax policy are potentially large. In order to additionally identify indirect policy effects in our decomposition framework, it is necessary to make assumptions about potential behavioral changes of taxpayers after policy changes as we are lacking a tractable full general equilibrium model. We proceed as follows. We extend our (mechanical) baseline decomposition and retrieve hypothetical pretax incomes for each income group in both counterfactuals under the assumption that all behavioral responses to taxation are captured by the ETI. If this is the case, the ETI is a sufficient statistic for welfare anlaysis (Feldstein 1995, 1999) and hence it is appropriate to use estimates of the ETI to assess the potential behavioral responses to tax policy changes.

Following Giertz (2009), who studies how tax revenues could be affected by behavioral responses after an expiration of the Bush tax cuts, we use stylized values of the ETI of 0.2, 0.5, and 1.0 (which are in the middle-low, middlehigh, and very high range of existing estimates for the United States--see Saez, Slemrod, and Giertz 2012 who report a preferred value of 0.25) in order to consider a reasonable range of values for the indirect policy effect. Note that our baseline decomposition can be considered as a lower bound as it is implicitly based on the assumption of a zero ETI. For the sake of simplicity and due to missing estimates, in a first step we follow Giertz (2009) and assume the ETI to be constant across income groups and over time. In a second step, we allow for different ETIs across the income distribution. We are thus able to quantify what fraction of the total change in pretax income from period t to t + 1 is due to behavioral responses and other factors, respectively. The indirect policy effect is derived such that it precisely corresponds to that fraction of the other effect (see the Appendix for a formal derivation). It is important to note that the decomposition of the other effect into an indirect policy effect and a residual effect rests upon the assumption that the ETI captures all indirect policy effects such as behavioral responses and general equilibrium effects. (21) If this is not the case, the residual effect still contains further indirect policy effects. Under the strong and arguably implausible assumption that there are no exogenous changes to the income distribution, the total change in inequality would be due to tax policy changes.

Tables A1-A4 show decomposition results for all major tax reforms in our sample period including indirect policy effects. Columns (1)-(5) correspond to the baseline decomposition (Tables 1-4), while columns (6)-(11) report results for indirect and residual effects. Note that the direct policy effect from our baseline (column (4)) is not affected by the extension as it mechanically captures changes in policy parameters, but no behavioral responses. On average, for an ETI of 0.2, indirect policy effects are much smaller than direct policy effects. They become larger if we assume an ETI of 0.5 and are often as important as direct policy effects for an ETI of 1.

In Figure 6, we relate the direct policy effect (equivalent to the right-hand-side graph in Figure 4) to the upper bound estimate of the total policy effect which is given as the sum of direct and indirect policy effects. Results based on an underlying ETI of 0.2 and 0.5 are presented in Figures S1 and S2 in the Supporting Information. Importantly, our baseline results are quantitatively affected, but not qualitatively. (22) Over the whole time period, taxpayers in the top quintile benefited most from tax policy and this effect is larger the higher the underlying ETI. The opposite effect can be observed for taxpayers in the second to fourth quintiles.

In columns (2) and (4) of Table 5, the upper bound estimate for the total policy effect (ETI = 1) is expressed in relation to the total change in average tax rates and income shares. While the fraction becomes smaller for average tax rates when indirect policy effects are accounted for (column (2) vs. (1)), the overall impact of tax policy on inequality becomes larger which amplifies the direct effect. It now ranges between 18% and 41%.

As it is plausible to assume that the ETI is higher for higher income groups who have more of a scope for avoidance and evasion opportunities, in the next step, we allow for different ETIs across the income distribution. In particular, we assume an ETI of 0.2 for taxpayers in the second quintile, an ETI of 0.5 for those in the third and fourth quintiles, and an ETI of 1 for those in the fifth quintile. The assumption of differential ETIs across the income distribution slightly reduces the disequalizing (total) effect of tax policy compared with the baseline of a uniform ETI of 1 (Figure S3).

[FIGURE 6 OMITTED]

D. Sensitivity Checks

In this section, we check the sensitivity of our results with respect to several choices made.

Choice of the Uprating Factor. As a first sensitivity check, we replicate the analysis with mean nominal wage growth as uprating factor (a in formulae (2) and (3)) in order to answer the question to what extent our results depend on the choice of the uprating factor. Over the whole sample period, mean nominal wages grew faster than the inflation rate which implies that taxpayers might move into higher/lower tax brackets when adjusting pretax incomes in our counterfactuals. (23) Figure S4 shows that the results do not change much with nominal wage indexation. The overall effect of tax policy is slightly more disequalizing than in our baseline. Cumulative policy effects on income shares are more beneficial for taxpayers in the fourth and fifth quintiles relative to those in the second and third quintiles.

Ranking of Tax Units. In our baseline, we follow the approach of Piketty and Saez (2007) and rank tax units based on their pretax incomes excluding capital gains given that realized capital gains are not a regular stream of income. Capital gains are added back to pre- and posttax incomes for the calculation of average tax rates and income shares. This might affect our results in particular for those periods in which significant changes in the amount of realized capital gains occurred, as can be observed around TRA86. Table S1 shows decomposition results for TRA86 when tax units are ranked based on pretax incomes including capital gains. For the base year 1986, average tax rates appear to be substantially higher for taxpayers at the top 0.1% of the distribution than in our baseline (Table 1), whereas for 1988 average tax rates for the richest tax units are similar to those in our baseline. These differences can be explained by the fact that realizations of long-term capital gains peaked in 1986, in particular, among affluent taxpayers, in anticipation of the tax increase in 1987.

[FIGURE 7 OMITTED]

The ranking of tax units also affects our decomposition results. This is especially evident when hypothetical average tax rates with 1986 pretax incomes and 1988 policy rules are compared (column (3) in Tables 1 and SI). The counterfactual average tax rate for taxpayers at the top 0.1% is substantially higher when taxpayers are ranked based on pretax incomes including capital gains. As a consequence, the beneficial effect of TRA86 for the richest taxpayers (top 1 %) appears to be much stronger in our baseline, while results for the bottom 99% do not critically depend on the way tax units are ranked. Decomposition results for all other tax reforms in our sample period are not affected by the way tax units are ranked.

SOI IRS versus CPS Data. In a previous version of this paper, we have performed the decomposition analysis using data from the CPS. Results are not directly comparable owing to various data issues such as the need to impute itemized deductions and top-coding of high incomes in the CPS. As a consequence, we have relied on percentile ratios such as the P90/P10, P90/P50, or P50/P10 and the Gini rather than (top) income shares for the calculation of the policy effect. Nevertheless, overall conclusions are the same. The policy effect is non-marginal, but smaller than the other effect. Tax policy was equalizing in the early 1990s, but highly disequalizing in the 1980s and early 2000s. A comparison of policy effects on the Gini coefficient based on these two data sources is shown in Figure S5. For most years of our sample, the policy effects are of similar size.

Imputation of Non-Filers. In our baseline analysis, we imputed non-filers (and their income) to our analysis. If we do not impute non-filers to our data, our results refer to the group of filers rather than to the full population of all taxpayers. In a further sensitivity check, we exclude the imputed non-filers. As shown in Figure 7, this only marginally changes the cumulative policy effect for the different income groups which is not surprising given that the share of non-filers is small varying between 4% and 8% of tax units.

V. CONCLUSION

In this paper, we have analyzed how tax policy has affected posttax income inequality in the United States from 1979 to 2007 based on counterfactual simulations. The decomposition analysis isolates and quantifies the direct effect of tax policy on the posttax income distribution. A main finding is that, over the whole sample period, tax policy aggravated the trend of growing inequality in pretax incomes: tax policy had a positive (negative) effect on the income share of taxpayers above (below) the 80th percentile. Hence, without any tax policy changes, observed inequality today is predicted to be lower. A second key result is that the policy effect corresponds to 11%-29% of the total change in income shares of different income groups. The effect was largest for taxpayers in the 95th to 99th percentiles but smallest for those in the top 1%. Thus, even though the surge in top incomes in the last three decades was to a large extent market driven, tax policy explains a substantial part of this trend. In addition, accounting for indirect policy effects due to behavioral responses does not change our results qualitatively, but raises the relative importance of the policy effect on inequality: the upper bound estimate for the total policy effect is 18%-41% (depending on the inequality measure) of the total change. The analysis also suggests that tax reforms in the 1980s and early 2000s exacerbated trends of growing inequality while those in the early 1990s benefited low-income taxpayers.

These results should be interpreted in the light of the following qualifications. First, our analysis is purely positive. Throughout this paper, we have abstracted from normative welfare considerations regarding the optimal amount of redistribution. Second, the calculation of indirect policy effects is based on stylized assumptions about behavioral responses to tax changes. However, we have used a range of plausible parameter values for the ETI and found qualitatively similar results. Third, we have focused the analysis on the United States. In future research, it would be interesting to replicate our analysis for other countries in order to investigate if tax policy affects inequality differently across different institutional settings.

ABBREVIATIONS

CPS: Current Population Survey

EITC: Earned Income Tax Credit

ETI: Elasticity of Taxable Income

IRS: Internal Revenue Service

NBER: National Bureau of Economic Research

OECD: Organisation for Economic Co-operation and

Development

SOI: Statistics of Income

TRA: Tax Reform Act

doi: 10.1111/ecin.12172

APPENDIX: DECOMPOSITION INCLUDING INDIRECT POLICY EFFECT

We extend decompositions I and II as follows: Decomposition I:

(A1) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Decomposition II:

(A2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Residual effect II

with [[??].sup.t] and [[??].sup.t + 1] as vectors of hypothetical pretax incomes after behavioral responses. The ETI-formula reads:

[epsilon] = ([[DELTA].sub.y]/[DELTA] (1 - T)) x (1 - T/y).

The behavioral response is calculated for decompositions I and II:

[DELTA][y.sup.t] = [epsilon] x [DELTA] (1 - T) x ([y.sup.t] (1 - T)) (decomposition I)

[DELTA] [y.sup.II] = [epsilon] x [DELTA] (1 - T) x ([y.sup.t+1] /[(1 - T).sub.t+1]) (decomposition II)

with [epsilon] = 0.2, 0.5 or 1, [DELTA] (1 - T) given by the policy effect on average tax rates, T average tax rates and [y.sup.t]/[y.sup.t+1] observed pretax incomes in t and t + 1. Averaging over both decompositions yields the Shapley value:

[DELTA][y.sup.S] = [DELTA][y.sup.1] + [DELTA][y.sup.II]/2.

For each income group, the indirect policy effect is calculated as a fraction of the total change in reported income from period t to t + 1:

(A3) IPE = ([DELTA][y.sup.S]/([y.sup.t+1] - [y.sup.t])) x OE

TABLE A1
Decomposition Results for Major Tax Reforms
Including Indirect Policy Effects: ERTA81

Data Year        1981         1984        Total
Policy Year      1981         1984        Change
                 (1)          (2)          (3)

Average Tax Rates

P20-40             15.1         14.9         -0.3
P40-60             22.3         21.2         -1.2
P60-80             26.9         25.5         -1.4
P80-90             29.4         27.9         -1.5
P90-95             30.6         28.9         -1.7
P95-99             32.5         30.0         -2.5
P99-99.5           36.0         32.4         -3.7
P99.5-99.9         39.8         35.9         -3.9
P99.9-99.99        44.5         42.2         -2.3
P99.99-100         46.9         44.5         -2.4
Top 20%            32.2         30.4         -1.8
Top 10%            33.7         31.6         -2.1
Top 5%             35.5         33.0         -2.4
Top 1%             40.1         37.2         -2.9

Posttax Income Shares

P20-40              8.6          8.0         -0.6
P40-60             15.7         15.1         -0.7
P60-80             25.4         24.9         -0.4
P80-90             18.0         18.1          0.2
P90-95             11.3         11.6          0.3
P95-99             12.1         12.8          0.6
P99-99.5            2.3          2.5          0.2
P99.5-99.9          2.7          3.0          0.4
P99.9-99.99         1.2          1.6          0.3
P99.99-100          0.5          0.8          0.3
Top 20%            48.1         50.4          2.3
Top 10%            30.1         32.3          2.2
Top 5%             18.8         20.7          1.9
Top 1%              6.7          7.9          1.2

Data Year     ETI = 0                   ETI = 0.2
Policy Year       PE           OE          IPE           RE
                 (4)          (5)          (6)          (7)

Average Tax Rates

P20-40              1.1         -1.3         -0.2         -1.1
P40-60             -0.1         -1.1          0.1         -1.2
P60-80             -0.7         -0.8          0.0         -0.8
P80-90             -0.9         -0.6          0.0         -0.6
P90-95             -1.1         -0.6          0.1         -0.6
P95-99             -1.8         -0.7          0.1         -0.8
P99-99.5           -2.8         -0.9          0.1         -1.0
P99.5-99.9         -3.3         -0.5          0.0         -0.6
P99.9-99.99        -3.9          1.6          0.1          1.5
P99.99-100         -5.6          3.1          0.2          2.9
Top 20%            -1.6         -0.2          0.0         -0.3
Top 10%            -2.0         -0.1          0.1         -0.2
Top 5%             -2.4          0.0          0.0         -0.0
Top 1%             -3.5          0.6          0.0          0.6

Posttax Income Shares

P20-40             -0.2         -0.4         -0.1         -0.3
P40-60             -0.2         -0.5         -0.0         -0.4
P60-80             -0.1         -0.3         -0.0         -0.3
P80-90              0.0          0.1          0.0          0.1
P90-95              0.1          0.2          0.0          0.2
P95-99              0.2          0.5          0.0          0.4
P99-99.5            0.1          0.1          0.0          0.1
P99.5-99.9          0.1          0.3          0.0          0.3
P99.9-99.99         0.1          0.3          0.0          0.3
P99.99-100          0.0          0.3          0.0          0.3
Top 20%             0.5          1.8          0.1          1.7
Top 10%             0.5          1.7          0.1          1.6
Top 5%              0.5          1.4          0.1          1.3
Top 1%              0.3          0.9          0.0          0.9

Data Year     ETI = 0.5                 ETI = 1.0
Policy Year      IPE           RE          IPE           RE
                 (8)          (9)          (10)         (11)

Average Tax Rates

P20-40             -0.5         -0.8         -0.9         -0.4
P40-60              0.1         -1.2          0.3         -1.4
P60-80              0.1         -0.8          0.2         -0.9
P80-90              0.1         -0.7          0.1         -0.7
P90-95              0.2         -0.7          0.3         -0.9
P95-99              0.2         -0.9          0.4         -1.1
P99-99.5            0.2         -1.1          0.4         -1.3
P99.5-99.9          0.1         -0.7          0.2         -0.8
P99.9-99.99         0.3          1.4          0.5          1.1
P99.99-100          0.6          2.5          1.2          2.0
Top 20%             0.1         -0.3          0.2         -0.5
Top 10%             0.1         -0.3          0.3         -0.4
Top 5%              0.1         -0.1          0.2         -0.2
Top 1%              0.0          0.5          0.1          0.5

Posttax Income Shares

P20-40             -0.2         -0.2         -0.3         -0.1
P40-60             -0.1         -0.4         -0.1         -0.3
P60-80             -0.0         -0.3         -0.0         -0.3
P80-90              0.0          0.1          0.0          0.1
P90-95              0.0          0.2          0.1          0.2
P95-99              0.1          0.4          0.1          0.3
P99-99.5            0.0          0.1          0.0          0.1
P99.5-99.9          0.0          0.2          0.1          0.2
P99.9-99.99         0.0          0.2          0.1          0.2
P99.99-100          0.0          0.2          0.0          0.2
Top 20%             0.2          1.6          0.4          1.4
Top 10%             0.2          1.5          0.4          1.3
Top 5%              0.2          1.2          0.3          1.1
Top 1%              0.1          0.8          0.2          0.7

Notes'. Results for ETI = 0 correspond to the baseline
results. If ETI = 0, total change = PE + OE. If ETI > 0, PE
is equal to (4). Total change = PE + IPE + RE. Average tax
rates include federal and state level income and payroll
taxes. Uprating according to the level of price inflation.
PE, policy effect: IPE, indirect policy effect; OE, other
effect; RE, residual effect; ETI, elasticity of taxable
income.

Sources: Own calculations based on SOI IRS income tax return
data and NBER TAXSIM calculator.

TABLE A2
Decomposition Results for Major Tax Reforms Including
Indirect Policy Effects: TRA86

Data Year              1986         1988        Total
Policy Year            1986         1988       Change
                        (1)          (2)          (3)

Average Tax Rates

P20-40                 14.7         14.3         -0.4
P40-60                 21.3         21.2         -0.1
P60-80                 25.9         25.6         -0.2
P80-90                 28.2         27.9         -0.4
P90-95                 29.5         29.3         -0.2
P95-99                 31.0         30.1         -0.9
P99-99.5               33.7         30.5         -3.2
P99.5-99.9             38.1         30.5         -7.5
P99.9-99.99            43.4         30.1        -13.4
P99.99-100             45.8         29.2        -16.6
Top 20%                31.3         29.3         -2.1
Top 10%                32.9         29.9         -3.0
Top 5%                 34.5         30.1         -4.4
Top 1%                 39.0         30.2         -8.8

Posttax Income Shares

P20-40                  7.8          7.5         -0.2
P40-60                 14.8         13.9         -0.9
P60-80                 24.6         23.0         -1.6
P80-90                 18.0         17.0         -1.0
P90-95                 11.6         11.1         -0.5
P95-99                 13.1         13.1         -0.0
P99-99.5                2.6          3.1          0.4
P99.5-99.9              3.4          4.7          1.2
P99.9-99.99             1.7          3.3          1.6
P99.99-100              1.0          2.2          1.2
Top 20%                51.5         54.3          2.8
Top 10%                33.5         37.3          3.9
Top 5%                 21.8         26.2          4.4
Top 1%                  8.8         13.2          4.4

Data Year           ETI = 0                 ETI = 0.2
Policy Year           PE           OE          IPE           RE
                     (4)          (5)          (6)          (7)

Average Tax Rates

P20-40                  -0.5          0.2         0.0          0.2
P40-60                 -0.3          0.2          0.0          0.2
P60-80                 -0.5          0.2          0.0          0.2
P80-90                 -1.5          1.1          0.1          1.0
P90-95                 -2.2          2.1          0.3          1.7
P95-99                 -2.9          2.0          0.2          1.7
P99-99.5               -3.1         -0.1          0.1         -0.1
P99.5-99.9             -5.2         -2.3          0.4         -2.7
P99.9-99.99            -9.4         -4.0          0.2         -4.2
P99.99-100            -10.1         -6.5          0.7         -7.2
Top 20%                -3.2          1.1          0.1          1.0
Top 10%                -3.9          0.9          0.1          0.9
Top 5%                 -4.6          0.2          0.1          0.2
Top 1%                 -6.4         -2.4          0.3         -2.7

Posttax Income Shares

P20-40                 -0.1         -0.1         -0.0         -0.1
P40-60                 -0.3         -0.6         -0.0         -0.6
P60-80                 -0.4         -1.2         -0.1         -1.1
P80-90                 -0.1         -1.0         -0.0         -1.0
P90-95                  0.1         -0.6          0.0         -0.6
P95-99                  0.2         -0.2          0.0         -0.2
P99-99.5                0.1          0.4          0.0          0.3
P99.5-99.9              0.2          1.0          0.1          0.9
P99.9-99.99             0.3          1.3          0.1          1.3
P99.99-100              0.2          1.0          0.1          0.9
Top 20%                 1.0          1.9          0.1          1.8
Top 10%                 1.1          2.8          0.1          2.7
Top 5%                  1.0          3.4          0.2          3.2
Top 1%                  0.8          3.6          0.2          3.5

Data Year         ETI = 0.5                 ETI = 1.0
Policy Year          IPE           RE           PE           RE
                     (8)          (9)          (10)         (ID

Average Tax Rates

P20-40                  0.0          0.1          0.1          0.1
P40-60                  0.0          0.2          0.0          0.2
P60-80                  0.0          0.2          0.1          0.2
P80-90                  0.3          0.8          0.6          0.6
P90-95                  0.8          1.3          1.2          0.8
P95-99                  0.6          1.4          1.0          1.0
P99-99.5                0.2         -0.3          0.4         -0.4
P99.5-99.9              0.9         -3.2          1.7         -4.1
P99.9-99.99             0.6         -4.6          1.2         -5.2
P99.99-100              1.8         -8.3          3.6        -10.0
Top 20%                 0.2          0.9          0.4          0.7
Top 10%                 0.2          0.8          0.3          0.6
Top 5%                  0.1          0.1          0.3         -0.1
Top 1%                  0.7         -3.1          1.4         -3.8

Posttax Income Shares

P20-40                 -0.0         -0.1         -0.0         -0.1
P40-60                 -0.0         -0.5         -0.1         -0.5
P60-80                 -0.2         -1.0         -0.3         -0.9
P80-90                 -0.1         -0.9         -0.1         -0.9
P90-95                  0.1         -0.7          0.2         -0.8
P95-99                  0.1         -0.3          0.1         -0.3
P99-99.5                0.0          0.3          0.0          0.3
P99.5-99.9              0.1          0.9          0.2          0.8
P99.9-99.99             0.2          1.2          0.3          1.0
P99.99-100              0.1          0.8          0.2          0.8
Top 20%                 0.2          1.7          0.4          1.5
Top 10%                 0.4          2.5          0.7          2.1
Top 5%                  0.5          2.9          1.0          2.5
Top 1%                  0.5          3.2          0.7          2.9

Notes: Results for ETI = 0 correspond to the baseline
results. If ETI = 0, total change = PE + OE. If ETI > 0, PE
is equal to (4). Total change = PE + IPE + RE. Average tax
rates include federal and state level income and payroll
taxes. Uprating according to the level of price inflation.
PE, policy effect; IPE, indirect policy effect; OE, other
effect; RE, residual effect; ETI, elasticity of taxable
income.

Sources: Own calculations based on SOI IRS income tax return
data and NBER TAXSIM calculator.

TABLE A3
Decomposition Results for Major Tax Reforms Including
Indirect Policy Effects: OBRA90/OBRA93

Data Year          1989         1994        Total
Policy Year        1989         1994       Change
                    (1)          (2)          (3)

Average Tax Rates

P20-40             14.4          9.8         -4.5
P40-60             21.3         19.9         -1.4
P60-80             25.7         26.1          0.5
P80-90             27.9         28.6          0.7
P90-95             29.6         30.5          1.0
P95-99             30.1         31.9          1.8
P99-99.5           30.3         33.9          3.6
P99.5-99.9         30.3         37.3          7.0
P99.9-99.99        29.8         40.0         10.2
P99.99-100         29.3         39.8         10.5
Top 20%            29.3         31.9          2.6
Top 10%            29.9         33.4          3.5
Top 5%             30.1         34.6          4.5
Top 1%             30.0         37.4          7.4

Posttax Income Shares

P20-40              7.6          7.5         -0.1
P40-60             13.9         13.7         -0.1
P60-80             23.1         22.8         -0.2
P80-90             17.1         17.5          0.4
P90-95             11.2         11.6          0.3
P95-99             13.3         13.8          0.5
P99-99.5            3.1          3.1          0.0
P99.5-99.9          4.5          4.2         -0.3
P99.9-99.99         3.0          2.5         -0.5
P99.99-100          1.9          1.5         -0.4
Top 20%            54.1         54.1          0.0
Top 10%            37.0         36.7         -0.3
Top 5%             25.8         25.1         -0.6
Top 1%             12.5         11.4         -1.1

Data Year     ETI = 0                   ETI = 0.2
Policy Year       PE           OE          IPE           RE
                 (4)          (5)          (6)          (7)

Average Tax Rates

P20-40             -3.4         -1.1          0.6         -1.7
P40-60             -0.6         -0.8          0.2         -1.1
P60-80              0.3          0.2         -0.0          0.2
P80-90              0.3          0.4         -0.0          0.4
P90-95              0.5          0.4         -0.0          0.4
P95-99              1.0          0.8         -0.0          0.8
P99-99.5            2.6          1.0         -0.2          1.1
P99.5-99.9          6.2          0.8         -0.2          0.9
P99.9-99.99        10.0          0.2         -0.3          0.5
P99.99-100         11.2         -0.7         -0.4         -0.3
Top 20%             2.1          0.5         -0.0          0.5
Top 10%             2.9          0.5         -0.1          0.6
Top 5%              3.9          0.6         -0.1          0.7
Top 1%              7.0          0.4         -0.2          0.6

Posttax Income Shares

P20-40              0.3         -0.4          0.0         -0.5
P40-60              0.2         -0.4          0.0         -0.4
P60-80              0.1         -0.4          0.0         -0.4
P80-90              0.1          0.2          0.0          0.2
P90-95              0.1          0.3          0.0          0.3
P95-99              0.0          0.5          0.0          0.5
P99-99.5           -0.0          0.1         -0.0          0.1
P99.5-99.9         -0.3          0.0         -0.1          0.1
P99.9-99.99        -0.4         -0.1         -0.1         -0.0
P99.99-100         -0.3         -0.1         -0.1         -0.0
Top 20%            -0.8          0.9         -0.0          0.9
Top 10%            -0.9          0.6         -0.3          0.9
Top 5%             -1.0          0.4         -0.4          0.7
Top 1%             -1.0         -0.1         -0.3          0.2

Data Year     ETI = 0.5                 ETI = 1.0
Policy Year      IPE           RE          IPE           RE
                 (8)          (9)          (10)         (id

Average Tax Rates

P20-40              0.9         -2.0          1.2         -2.3
P40-60              0.4         -1.2          0.5         -1.3
P60-80             -0.0          0.2         -0.0          0.2
P80-90             -0.0          0.4         -0.0          0.4
P90-95             -0.0          0.4         -0.1          0.5
P95-99             -0.0          0.8         -0.0          0.8
P99-99.5           -0.3          1.3         -0.5          1.4
P99.5-99.9         -0.3          1.1         -0.4          1.2
P99.9-99.99        -0.5          0.7         -0.7          0.9
P99.99-100         -0.9          0.3         -1.3          0.6
Top 20%            -0.1          0.6         -0.1          0.6
Top 10%            -0.1          0.7         -0.2          0.7
Top 5%             -0.2          0.8         -0.3          0.9
Top 1%             -0.3          0.8         -0.5          1.0

Posttax Income Shares

P20-40              0.1         -0.5          0.1         -0.5
P40-60              0.0         -0.4          0.0         -0.4
P60-80              0.0         -0.4          0.0         -0.4
P80-90              0.0          0.2          0.0          0.2
P90-95              0.0          0.3          0.0          0.3
P95-99              0.0          0.5          0.0          0.5
P99-99.5           -0.0          0.1         -0.0          0.1
P99.5-99.9         -0.2          0.2         -0.2          0.3
P99.9-99.99        -0.2          0.1         -0.3          0.2
P99.99-100         -0.1          0.0         -0.2          0.1
Top 20%            -0.1          1.0         -0.1          1.0
Top 10%            -0.4          1.0         -0.5          1.1
Top 5%             -0.6          0.9         -0.7          1.0
Top 1%             -0.5          0.4         -0.8          0.7

Notes: Results for ETI = 0 correspond to the baseline
results. If ETI = 0. total change = PE + OE. If ETI > 0, PE
is equal to (4). Total change = PE + IPE + RE. Average tax
rates include federal and state level income and payroll
taxes. Uprating according to the level of price inflation.
PE. policy effect; IPE, indirect policy effect; OE, other
effect; RE, residual effect; ETI, elasticity of taxable
income.

Sources: Own calculations based on SOI IRS income tax return
data and NBER TAXSIM calculator.

TABLE A4
Decomposition Results for Major Tax Reforms Including
Indirect Policy Effects: EGTRRA01/JGTRRA03

Data Year          2000         2004        Total
Policy Year        2000         2004       Change
                    (1)          (2)          (3)

Average Tax Rates

P20-40             10.5          2.3         -8.2
P40-60             20.9         16.7         -4.2
P60-80             25.9         23.2         -2.7
P80-90             28.5         25.6         -2.9
P90-95             30.6         27.7         -2.9
P95-99             31.9         29.2         -2.7
P99-99.5           33.9         31.8         -2.1
P99.5-99.9         35.8         32.6         -3.2
P99.9-99.99        36.6         32.9         -3.7
P99.99-100         37.2         31.8         -5.5
Top 20%            32.1         28.9         -3.2
Top 10%            33.3         30.1         -3.2
Top 5%             34.2         30.9         -3.2
Top 1%             35.9         32.3         -3.5

Posttax Income Shares

P20-40              7.3          7.1         -0.2
P40-60             12.4         12.4         -0.0
P60-80             20.5         20.9          0.3
P80-90             15.8         16.4          0.6
P90-95             10.9         11.2          0.4
P95-99             14.2         14.2          0.0
P99-99.5            3.6          3.4         -0.2
P99.5-99.9          5.6          5.2         -0.3
P99.9-99.99         4.5          3.9         -0.5
P99.99-100          2.8          2.8         -0.0
Top 20%            57.4         57.2         -0.2
Top 10%            41.5         40.8         -0.7
Top 5%             30.7         29.6         -1.1
Top 1%             16.5         15.4         -1.1

Data Year     ETI = 0                   ETI = 0.2
Policy Year          PE           OE          IPE           RE
                    (4)          (5)          (6)          (7)

Average Tax Rates

P20-40             -2.0         -6.2          0.2         -6.4
P40-60             -2.5         -1.7          0.5         -2.3
P60-80             -2.1         -0.7          0.4         -1.0
P80-90             -2.3         -0.5          0.1         -0.6
P90-95             -2.5         -0.4          0.0         -0.5
P95-99             -2.3         -0.4          0.0         -0.5
P99-99.5           -2.3          0.2          0.0          0.1
P99.5-99.9         -3.4          0.2          0.0          0.1
P99.9-99.99        -4.3          0.6          0.1          0.5
P99.99-100         -4.8         -0.7          0.1         -0.7
Top 20%            -2.7         -0.5          0.0         -0.5
Top 10%            -2.8         -0.4          0.0         -0.4
Top 5%             -2.9         -0.3          0.0         -0.3
Top 1%             -3.6          0.1          0.0          0.0

Posttax Income Shares

P20-40             -0.1         -0.1         -0.0         -0.1
P40-60             -0.1          0.0         -0.1          0.1
P60-80             -0.1          0.5         -0.1          0.5
P80-90             -0.0          0.6         -0.0          0.6
P90-95              0.0          0.3          0.0          0.3
P95-99              0.0          0.0          0.0         -0.0
P99-99.5            0.0         -0.2          0.0         -0.2
P99.5-99.9          0.1         -0.4          0.0         -0.4
P99.9-99.99         0.1         -0.6          0.0         -0.7
P99.99-100          0.1         -0.1          0.0         -0.2
Top 20%             0.3         -0.4          0.0         -0.5
Top 10%             0.3         -1.0          0.1         -1.1
Top 5%              0.3         -1.4          0.1         -1.5
Top 1%              0.3         -1.4          0.1         -1.5

Data Year     ETI = 0.5                 ETI = 1.0
Policy Year         IPE           RE          IPE           RE
                    (8)          (9)         (10)         (11)

Average Tax Rates

P20-40              0.4         -6.6          0.7         -7.0
P40-60              0.8         -2.5          1.0         -2.7
P60-80              0.4         -1.1          0.5         -1.1
P80-90              0.1         -0.6          0.1         -0.7
P90-95              0.1         -0.5          0.1         -0.5
P95-99              0.1         -0.5          0.1         -0.6
P99-99.5            0.0          0.1          0.1          0.1
P99.5-99.9          0.1          0.0          0.2         -0.0
P99.9-99.99         0.1          0.5          0.2          0.4
P99.99-100          0.2         -0.9          0.4         -1.1
Top 20%             0.0         -0.5          0.1         -0.6
Top 10%             0.1         -0.4          0.1         -0.5
Top 5%              0.1         -0.4          0.1         -0.4
Top 1%              0.1         -0.1          0.2         -0.1

Posttax Income Shares

P20-40             -0.0         -0.0         -0.0         -0.0
P40-60             -0.0          0.1         -0.1          0.1
P60-80             -0.1          0.6         -0.2          0.6
P80-90             -0.0          0.6         -0.0          0.6
P90-95              0.0          0.3          0.0          0.3
P95-99              0.0         -0.0          0.0         -0.0
P99-99.5            0.0         -0.2          0.0         -0.2
P99.5-99.9          0.0         -0.5          0.0         -0.5
P99.9-99.99         0.1         -0.7          0.1         -0.7
P99.99-100          0.1         -0.2          0.1         -0.3
Top 20%             0.1         -0.5          0.1         -0.5
Top 10%             0.2         -1.2          0.2         -1.2
Top 5%              0.2         -1.6          0.3         -1.6
Top 1%              0.2         -1.6          0.3         -1.7

Notes: Results for ETI = 0 correspond to the baseline
results. If ETI = 0, total change = PE + OE. If ETI > 0, PE
is equal to (4). Total change = PE + IPE + RE. Average tax
rates include federal and state level income and payroll
taxes. Uprating according to the level of price inflation.
PE, policy effect; IPE, indirect policy effect; OE, other
effect; RE, residual effect. ETI, elasticity of taxable
income.

Sources: Own calculations based on SOI IRS income tax return
data and NBER TAXSIM calculator.

REFERENCES

Aim, J., F. Lee, and S. Wallace. "How Fair? Changes in Federal Income Taxation and the Distribution of Income, 1978 to 1998." Journal of Policy Analysis and Management, 24(1), 2005, 5-22.

Auerbach, A. "Capital Gains Taxation in the United States." Brookings Papers on Economics Activity, 19(2), 1988, 595-638.

Auerbach, A., and J. Slemrod. "The Economic Effects of the Tax Reform Act of 1986." Journal of Economic Literature, 35, 1997, 589-632.

Bakija, J., A. Cole, and B. Heim. "Jobs and Income Growth of Top Earners and the Causes of Changing Income Inequality: Evidence from U.S. Tax Return Data." Working Paper No. 2010-22, Williams College, 2012.

Bargain, O., and T. Callan. "Analysing the Effects of Tax-Benefit Reforms on Income Distribution: A Decomposition Approach." Journal of Economic Inequality, 8(1), 2010, 1-21.

Burkhauser, R., S. Feng, S. Jenkins, and J. Larrimore. "Trends in United States Income Inequality Using the March Current Population Survey: The Importance of Controlling for Censoring." Journal of Economic Inequality, 9(3), 2011, 393-415.

--. "Recent Trends in Top Income Shares in the USA: Reconciling Estimates from March CPS and IRS Tax Return Data." Review of Economics and Statistics, 94(2), 2012a, 371-88.

Burkhauser, R., J. Larrimore, and K. Simon. "A 'Second Opinion' on the Economic Health of the American Middle Class." National Tax Journal, 65(1), 2012b, 7-32.

Clark, T., and A. Leicester. "Inequality and Two Decades of British Tax and Benefit Reform." Fiscal Studies, 25(2), 2004, 129-58.

Congressional Budget Office. "Average Federal Tax Rates and Income, by Income Category, 1979-2007." 2010. Accessed July 1, 2013. http://www.cbo.gov/publications/collections/collections.cfm?collect=13. Dardanoni, V., and P. Lambert. "Progressivity Comparisons." Journal of Public Economics, 86, 2002, 99-122.

DeBacker, J., B. Heim, V. Panousi, S. Ramnath, and I. Vidangos. "Rising Inequality: Transitory or Permanent? New Evidene from a Panel of U.S. Tax Returns." Brookings Papers on Economics Activity, Spring, 2013, 67-142.

Diamond, P., and E. Saez. "The Case for a Progressive Tax: From Basic Research to Policy Recommendations." Journal of Economic Perspectives, 25(4), 2011, 165-90.

Eissa, N., and H. Hoynes. "Behavioral Responses to Taxes: Lessons from the EITC and Labor Supply," in Tax Policy and the Economy, Vol. 20, edited by J. M. Poterba. Cambridge, MA: MIT Press, 2006, 74-110.

--. "Redistribution and Tax Expenditures: The Earned Income Tax Credit." National Tax Journal, 64(2), 2011, 689-730.

Eissa, N., H. Eleven, and C. T. Kreiner. "Evaluation of Four tax Reforms in the Unites States: Labor Supply and Welfare Effects for Single Mothers." Jounal of Public Economics, 92, 2008, 795-816.

Feenberg, D. R., and E. Coutts. "An Introduction to the TAXSIM Model." Journal of Policy Analysis and Management, 12(1), 1993, 189-94.

Feenberg, D., and J. Poterba. "Income Inequality and the Incomes of Very High-Income Taxpayers: Evidence from Tax Returns, " in Tax Policy and the Economy, Vol. 7, edited by J. M. Poterba. Cambridge, MA: MIT Press, 1993, 145-77.

Feldstein, M. "The Effect of Marginal Tax Rates on Taxable Income: A Panel Study of the 1986 Tax Reform Act." Journal of Political Economy, 103(3), 1995, 551-72.

--. "Tax Avoidance and the Deadweight Loss of the Income Tax." Review of Economics and Statistics, 81(4), 1999, 674-80.

Giertz, S. "The Elasticity of Taxable Income: Influences on Economic Efficiency and Tax Revenues, and Implications for Tax Policy," in Tax Policy Lessons from the 2000s, edited by A. Viard. Washington, DC: AEI Press, 2009, 101-36.

Gottschalk, P., and S. Danziger. "Inequality of Wage Rates, Earnings and Family Income in the United States, 1975-2002." Review of Income and Wealth, 51(2), 2005, 231-54.

Gottschalk, P., and T. Smeeding. "Cross-National Comparisons of Earnings and Income Inequality." Journal of Economic Literature, 35, 1997, 633-87.

Gramlich, E., R. Kasten, and F. Sammartino. "Growing Inequality in the 1980s: The Role of Federal Taxes and Cash Transfers," in Uneven Tides: Rising Inequality in America, edited by S. Danziger and P. Gottschalk. New York: Russell Sage Foundation, 1993, 225-49.

Heathcote, J., F. Perri, and G. Violante. "Unequal We Stand: An Empirical Analysis of Economic Inequality in the United States, 1967-2006." Review of Economic Dynamics, 13(1), 2010, 15-51.

Kasten, R., F. Sammartino, and E. Toder. "Trends in Federal Tax Progressivity, 1980-93," in Tax Progressivity and Income Inequality, edited by J. Slemrod. Cambridge: Cambridge University Press, 1994, 9-50.

Lambert, R, and T. Thoresen. "Base Independence in the Analysis of Tax Policy Effects: With an Application to Norway 1992-2004." International Tax and Public Finance, 16, 2009, 219-52.

Leigh, A. "Do Redistributive State Taxes Reduce Inequality?" National Tax Journal, 61(1), 2008, 81-104.

Lindsey, L. "Individual Taxpayer Response to Tax Cuts: 1982-1984. With Implications for the Revenue Maximizing Tax Rate." Jounal of Public Economics, 33, 1987, 173-206.

Meyer, B. "The Effects of the Earned Income Tax Credit and Recent Reforms," in Tax Policy and the Economy, Vol. 24, edited by J. Brown. Cambridge, MA: MIT Press, 2010, 153-180.

Mitrusi, A., and J. Poterba. "The Distribution of Payroll and Income Tax Burdens, 1979-99." National Tax Journal, 53(3), 2000, 765-794.

Musgrave, R" and T. Thin. "Income Tax Progression, 1929-1948." Journal of Political Economy, 56. 1948, 498-514.

OECD. Divided We Stand: Why Inequality Keeps Rising. Paris: OECD Publishing, 2011.

Piketty, T., and E. Saez. "Income Inequality in the United States. 1913-1998." Quarterly Journal of Economics, 118(1), 2003, 1-39.

--. "How Progressive Is the U.S. Federal Tax System? A Historical and International Perspective." Journal of Economic Perspectives, 21(1), 2007, 3-24.

Piketty, T., E. Saez, and S. Stantcheva. "Optimal Taxation of Top Labor Incomes: A Tale of Three Elasticities." American Economic Journal: Economic Policy, 6(1), 2014, 230-71.

Poterba, J. "Income Inequality and Income Taxation." Journal of Policy Modeling, 29, 2007, 623-33.

Reynolds, A. "Has U.S. Income Inequality Really Increased? Policy Analysis." Cato Institute, Policy Analysis No. 586, 2007.

Saez, E. "The Effect of Marginal Tax Rates on Income: A Panel Study of 'Bracket Creep'." Journal of Public Economics, 87, 2003, 1231-58.

--. "Reported Incomes and Marginal Tax Rates, 1960-2000: Evidence and Policy Implications," in Tax Policy and the Economy, Vol. 18, edited by J. Poterba. Cambridge, MA: MIT Press, 2004, 117-73.

Saez, E., J. Slemrod, and S. Giertz. "The Elasticity of Taxable Income with Respect to Marginal Tax Rates: A Critical Review." Journal of Economic Literature, 50(1), 2012, 3-50.

Shorrocks, A. "Decomposition Procedures for Distributional Analysis: A Unified Framework Based on the Shapley Value." University of Essex and Institute for Fiscal Studies, Wivenhoe Park, 1999.

Slemrod, J. "Taxation and Inequality: A Time-Exposure Perspective," in Tax Policy and the Economy, Vol. 6, edited by P. James. Cambridge, MA: MIT Press, 1992, 105-27.

--. "High-Income Families and the Tax Changes of the 1980s: The Anatomy of Behavioral Response," in Empirical Foundations of Household Taxation, edited by M. Feldstein and J. Poterba. Chicago: University of Chicago Press, 1996, 169-92.

Tempalski, J. 2006. "Revenue Effects of Major Tax Bills." Office of Tax Analysis Working Paper No. 81.

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article:

Appendix S1. Additional results: Tables S1-S2, Figures S1-S5.

(1.) Note that as a robustness check, we perform the decomposition analysis on the Current Population Survey (CPS) and show that results are in line with those based on tax return data. This comparison of how the policy effect on inequality differs between IRS SOI tax return and CPS data complements the analysis by Burkhauser et al. (2012a) who reconcile estimates on top income shares between these two data sources.

(2.) Our approach formalizes analyses of policy effects as performed, for instance, by Clark and Leicester (2004) for the United Kingdom. See also Bargain and Callan (2010) for France and Ireland. A related concept for the comparison of tax regimes with respect to progressivity--the transplant-and-compare procedure (Dardanoni and Lambert 2002)--is applied by Lambert and Thoresen (2009) for Norway. They isolate the tax policy effect by comparing pretax income distributions that have been adjusted to a common base.

(3.) See for example, Gottschalk and Danziger (2005), Heathcote et al. (2010), and Burkhauser et al. (2011). Differences between these studies exist with regard to the definition of the income unit (family vs. household), sample selection (full population vs. working-age population), and whether topcoding in the public-use CPS is accounted for. Burkhauser et al. (2012b) show that differences in inequality trends based on income tax return data and the CPS can be explained by different income measures and sharing units. Most notably, median income growth is shown to be significantly higher with a broader income measure including cash and in-kind transfers and with economies of scale in household consumption taken into account.

(4.) Further studies examining the degree of redistribution of the U.S. income tax system by means of policy swaps are Kasten et al. (1994), Mitrusi and Poterba (2000), Aim et al. (2005), Leigh (2008), and Meyer (2010). However, these studies do not quantify how much of an observed change in posttax income inequality is due to policy changes. Instead, the focus of these contributions is on the changing importance of income and payroll taxes over time (Mitrusi and Poterba 2000), on the progressivity of the income tax (Aim et al. 2005; Kasten et al. 1994), the redistributiveness of state taxes (Leigh 2008), and the distributional effect of the EITC reform enacted through the American Recovery and Reinvestment Act of 2009 (Meyer 2010).

(5.) This approach is supported by Piketty and Saez (2007) who argue that given the controversy about behavioral responses to taxation [... ] considering the basic case with no behavioral response is a useful starting place (p. 9).

(6.) A measure [d.sub.t] ([p.sub.t], [y.sup.t+1]) would not be consistent since base-period parameters would be artificially applied to endperiod income levels. For instance, previous tax band thresholds would be applied to new and possibly higher income levels, thereby generating artificial "fiscal drag" or "bracket creep" (Saez 2003).

(7.) Converting tax parameters and income from dollars into Euros does not change the relative location of households in the distribution of posttax income.

(8.) Following Piketty and Saez (2007), we use the CPIU-RS (http://www.bls.gov/cpi/cpirsdc.htm) which is the commonly used CPI measure for comparisons over time.

(9.) The number of observations varies between 90,000 and 200,000.

(10.) In a previous version of this paper, we have performed our calculations with data from IPUMS-CPS (Integrated Public Use Microdata Series, CPS) which is a rich micro-data set of U.S. households and a primary data source for investigating income distribution trends. However, it does not contain information about itemized deductions and capital gains which are important in any analysis on top incomes. Further, for confidentiality reasons, the U.S. Census Bureau "top codes" all income sources, with differences in methods between some years. This can cause a downward bias of income inequality estimates (cf. Burkhauser et al. 2011). We compare the result from SOI to those obtained from CPS data in Section IV.D.

(11.) We use the information from the study by Piketty and Saez (2003) on the total number of tax units in each year and are thus able to impute those who do not file (number of non-filer = total number tax units - number of tax returns observed in the data). Income of non-filers is imputed as 20% of average income. Groups are defined relative to all tax units (filers and non-filers). The total number of tax units in the United States ranges from 97.5 million in 1979 to almost 150 million in 2007. Over the sample period, the share of tax units that file a tax return is roughly between 92% and 96% (see online appendix of Piketty and Saez 2003, updated to 2010).

(12.) Some of the caveats discussed by Piketty and Saez (2007) apply to our study as well. In particular, we ignore the redistributive effect of government transfers and untaxed income such as in-kind benefits (except tax credits such as the EITC). Furthermore, our data are repeated cross sections and we therefore abstract from any lifecycle perspective.

(13.) For more information on TAXSIM see Feenberg and Coutts (1993) or visit http://www.nber.org/taxsim/.

(14.) As part of the tax burden was effectively shifted from the individual to the corporate sector which is not part of our analysis, TRA86 constitutes a tax cut in the context of this paper.

(15.) Results are robust when the adjustment of pretax incomes in the counterfactuals is based on mean nominal wage growth instead of the inflation rate (see Section IV.D).

(16.) A shift in the income composition to sources that are taxed by a lower rate ceteris paribus leads to a negative other effect.

(17.) For those in the top 1%, entrepreneurial income made up 11.1% of their total income (excluding capital gains) in 1986, but 21.2% in 1988. Conversely, capital gains made up 38.8% of their total income (including capital gains) in 1986, but only 14.6% in 1988 (see updated tables to Piketty and Saez 2003, accessible at http://elsa.berkeley.edu/~saez/).

(18.) Note that the way we rank tax units, that is, based on their pretax incomes excluding capital gains which are added back for the calculation of average tax rates and income shares, might critically affect our decomposition results when significant changes in the amount of realized capital gains occur from one period to the other. In Section IV.D, we show how results change for TRA86 when tax units are ranked based on pretax incomes including capital gains, which provides additional evidence for the impact of behavioral changes around TRA86 on our decomposition results.

(19.) This can be done in two ways. First, one can look at several year-by-year changes (e.g., from 1979 to 1980 and then from 1980 to 1981) and then add them up. Second, one can hold a base-year constant (e.g., 1979) and look at the changes to various end years (e.g., 1980, 1981, ...). In addition, various combinations are possible (e.g., adding up 3-year- or 5-year changes). This can lead to an almost infinite number of potential results. In our empirical analysis, it turned out that the different approaches led to very similar results (both in terms of magnitude of the effects and trends). Hence, we decided to focus on the first approach.

(20.) It should be noted, though, that having a Republican or Democratic administration in power does not mean that the initiative for a tax reform comes from this administration. For instance, TRA86 was started as a bi-partisan initiative.

(21.) By assuming a value for the ETI, it is clear that we cannot separate real responses from timing or income shifting responses. In addition, the ETI is not supposed to capture general equilibrium effects. Yet, an ETI of unity seems very high. Hence, one may argue that this serves as an upper bound potentially capturing also (some) general equilibrium effects. Furthermore, some of those effects might actually work in the other direction dampening the effect of tax policy on inequality.

(22.) An exception is the upper bound estimate on the total policy effect for taxpayers in the second quintile which turns out to be more negative than for taxpayers in the third quintile (Figure 6). For the total policy effect, it is important to note that a substantial part of the behavioral responses, in particular around TRA86, consisted of avoidance and timing responses which do not imply any additional income.

(23.) We choose the National Average Wage Index according to which the taxable maximum for Social Security is automatically adjusted. See http://www.ssa.gov/OACT/ COLA/AWI.html for further information. If the consumer price index (CPI-U-RS) and the National Average Wage Index are normalized to 1 for the base year 1979, their 2007 values are 2.66 and 3.52, respectively.

OLIVIER BARGAIN, MATHIAS DOLLS, HERWIG IMMERVOLL, DIRK NEUMANN, ANDREAS PEICHL, NICO PESTEL and SEBASTIAN SIEGLOCH *

(a) This paper uses TAXSIM v9. TAXSIM is continually being improved and updated and the results presented here represent the best available at the time of writing. Our version of TAXSIM is based on IRS SOI tax return data. We would like to thank Daniel Feenberg for granting us access to NBER's TAXSIM and helping us with our simulations. We are grateful to James Ziliak, the editor, an anonymous referee as well as Alberto Alesina, James Aim, Felix Bierbrauer, Denvil Duncan, Clemens Fuest, Carl Klarner, Wojciech Kopczuk, Jeff Larrimore, Erzo F.R Luttmer, Jim Poterba, Ronald L. Oaxaca, Andrew Oswald, Torsten Persson, Emmanuel Saez, James Sullivan, Tim Smeeding, as well as seminar and conference participants in Bari (ECINEQ), Bonn (IZA), Buch (ESSLE), Cambridge (NBER SI), Canazei (IT8), Chicago (SOLE), Dublin (IMA, ESRI), Gottingen (VfS), Oslo, Providence (NTA), and Zurich (EPCS) for helpful comments and suggestions. Peichl is grateful for financial support by Deutsche Forschungsgemeinschaft (PE1675). Any errors as well as the views presented in this paper are the responsibility of the authors alone. In particular, the views do not represent the official positions of organizations to which the authors are affiliated.

Bargain: Aix-Marseille School of Economics, CNRS EHESS, France; IZA, Germany. Phone +33 4 91 14 07 70, Fax +33 4 91 90 02 27, E-mail Olivier.BARGAIN@ univmed.fr

Dolls: ZEW and IZA, Germany. Phone +49 621 1235-395, Fax +49 621 1235-4220, E-mail dolls@zew.de

Immervoll: OECD, France; IZA, Germany. Phone +33 4524 9214, E-mail immervoll@oecd.org

Neumann: CORE (Universit catholique de Louvain), Belgium; ZEW and IZA, Germany. E-mail dirk.neumann@ uclouvain.be

Peichl: ZEW, University of Mannheim, IZA and CESifo, Germany. Phone +49 621 1235-389, Fax +49 621 12354389, E-mail peichl@zew.de

Pestel: IZA and ZEW, Germany. Phone +49 228 38 94 160, Fax +49 228 38 94 180. E-mail pestel@iza.org

Siegloch: University of Mannheim, IZA and ZEW, Germany. Phone +49-621-181-1818, E-mail siegloch@uni-mann heim.de

TABLE 1
Decomposition Results for Major Tax Reforms: ERTA81

Data Year        1981       1984       1981       1984
Adjusted to                 1981       1984                Total
Policy Year      1981       1981       1984       1984    Change

                  (1)        (2)        (3)        (4)    (4)-(1)

Average Tax Rates

P20-40           15.1       13.8       16.2       14.9       -0.3
P40-60           22.3       21.2       22.2       21.2      -1.2
P60-80           26.9       26.1       26.2       25.5      -1.4
P80-90           29.4       28.7       28.5       27.9      -1.5
P90-95           30.6       30.0       29.4       28.9      -1.7
P95-99           32.5       31.7       30.7       30.0      -2.5
P99-99.5         36.0       35.0       33.1       32.4      -3.7
P99.5-99.9       39.8       39.0       36.2       35.9      -3.9
P99.9-99.99      44.5       46.0       40.5       42.2      -2.3
P99.99-100       46.9       50.3       41.6       44.5      -2.4
Top 20%          32.2       31.9       30.5       30.4      -1.8
Top 10%          33.7       33.6       31.7       31.6      -2.1
Top 5%           35.5       35.4       33.0       33.0      -2.4
Top 1%           40.1       40.6       36.6       37.2      -2.9

Posttax Income Shares

P20-40            8.6        8.2        8.3        8.0       -0.6
P40-60           15.7       15.2       15.5       15.1      -0.7
P60-80           25.4       25.0       25.2       24.9      -0.4
P80-90           18.0       18.1       18.0       18.1       0.2
P90-95           11.3       11.6       11.4       11.6       0.3
P95-99           12.1       12.6       12.3       12.8       0.6
P99-99.5          2.3        2.4        2.4        2.5       0.2
P99.5-99.9        2.7        2.9        2.8        3.0       0.4
P99.9-99.99       1.2        1.5        1.3        1.6       0.3
P99.99-100        0.5        0.8        0.5        0.8       0.3
Top 20%          48.1       49.9       48.7       50.4       2.3
Top 10%          30.1       31.8       30.7       32.3       2.2
Top 5%           18.8       20.2       19.3       20.7       1.9
Top 1%            6.7        7.6        7.0        7.9       1.2

Data Year     Decomposition I       Decomposition II
Adjusted to   PE         OE         PE         OE
Policy Year

              (3)-(1)    (4)-(3)    (4)-(2)    (2)-(1)

Average Tax Rates

P20-40             1.1       -1.3       1.1       -1.3
P40-60           -0.1       -1.1       -0.1       -1.1
P60-80           -0.7       -0.7       -0.6       -0.8
P80-90           -0.9       -0.6       -0.8       -0.7
P90-95           -1.2       -0.5       -1.1       -0.6
P95-99           -1.8       -0.7       -1.7       -0.7
P99-99.5         -2.9       -0.8       -2.6       -1.0
P99.5-99.9       -3.5       -0.3       -3.1       -0.8
P99.9-99.99      -4.0        1.7       -3.9        1.6
P99.99-100       -5.3        2.9       -5.8        3.4
Top 20%          -1.6       -0.2       -1.6       -0.3
Top 10%          -2.0       -0.1       -1.9       -0.2
Top 5%           -2.5        0.1       -2.4       -0.0
Top 1%           -3.5        0.6       -3.4        0.5

Posttax Income Shares

P20-40            -0.2       -0.4      -0.2       -0.4
P40-60           -0.2       -0.5       -0.2       -0.5
P60-80           -0.1       -0.3       -0.1       -0.3
P80-90            0.0        0.1        0.0        0.1
P90-95            0.1        0.2        0.0        0.2
P95-99            0.2        0.5        0.2        0.5
P99-99.5          0.1        0.1        0.1        0.1
P99.5-99.9        0.1        0.3        0.1        0.3
P99.9-99.99       0.1        0.3        0.1        0.3
P99.99-100        0.0        0.3        0.1        0.3
Top 20%           0.5        1.8        0.6        1.8
Top 10%           0.5        1.7        0.5        1.7
Top 5%            0.5        1.4        0.5        1.4
Top 1%            0.3        1.0        0.3        0.9

Data Year         Shapley-December
Adjusted to       PE           OE
Policy Year      Mean         Mean
                (4)-(2)      (2)-(1)
                (3)-(1)      (4)-(3)

Average Tax Rates

P20-40              1.1         -1.3
P40-60             -0.1         -i.i
P60-80             -0.7         -0.8
P80-90             -0.9         -0.6
P90-95             -1.1         -0.6
P95-99             -1.8         -0.7
P99-99.5           -2.8         -0.9
P99.5-99.9         -3.3         -0.5
P99.9-99.99        -3.9          1.6
P99.99-100         -5.6          3.1
Top 20%            -1.6         -0.2
Top 10%            -2.0         -0.1
Top 5%             -2.4          0.0
Top 1%             -3.5          0.6

Posttax Income Shares

P20-40             -0.2         -0.4
P40-60             -0.2         -0.5
P60-80             -0.1         -0.3
P80-90              0.0          0.1
P90-95              0.1          0.2
P95-99              0.2          0.5
P99-99.5            0.1          0.1
P99.5-99.9          0.1          0.3
P99.9-99.99         0.1          0.3
P99.99-100          0.0          0.3
Top 20%             0.5          1.8
Top 10%             0.5          1.7
Top 5%              0.5          1.4
Top 1%              0.3          0.9

Notes: Average tax rates (%) include federal and state level
income and payroll taxes. Uprating according to the level of
price inflation. PE, policy effect; OE, other effect.

Sources: Own calculations based on SOI IRS income tax return
data and NBER TAXSIM calculator.

TABLE 2
Decomposition Results for Major Tax Reforms: TRA86

Data Year        1986       1988       1986       1988
Adjusted to                 1986       1988                 Total
Policy Year      1986       1986       1988       1988     Change
                  (1)        (2)        (3)        (4)    (4)-(1)

Average Tax Rates

P20-40           14.7       15.0       14.3       14.3       -0.4
P40-60           21.3       22.0       21.5       21.2       -0.1
P60-80           25.9       26.7       26.0       25.6       -0.2
P80-90           28.2       30.6       27.9       27.9       -0.4
P90-95           29.5       33.8       29.5       29.3       -0.2
P95-99           31.0       35.8       31.0       30.1       -0.9
P99-99.5         33.7       36.5       33.4       30.5       -3.2
P99.5-99.9       38.1       39.3       36.4       30.5       -7.5
P99.9-99.99      43.4       42.7       37.3       30.1      -13.4
P99.99-100       45.8       43.4       39.8       29.2      -16.6
Top 20%          31.3       34.9       30.6       29.3       -2.1
Top 10%          32.9       36.8       32.0       29.9       -3.0
Top 5%           34.5       38.0       33.2       30.1       -4.4
Top 1%           39.0       40.0       36.1       30.2       -8.8

Posttax Income Shares

P20-40            7.8        7.8        7.8        7.5       -0.2
P40-60           14.8       14.4       14.7       13.9       -0.9
P60-80           24.6       23.7       24.5       23.0       -1.6
P80-90           18.0       17.1       18.1       17.0       -1.0
P90-95           11.6       10.9       11.6       11.1       -0.5
P95-99           13.1       12.6       13.0       13.1       -0.0
P99-99.5          2.6        2.9        2.6        3.1        0.4
P99.5-99.9        3.4        4.3        3.5        4.7        1.2
P99.9-99.99       1.7        2.9        1.9        3.3        1.6
P99.99-100        1.0        1.9        1.1        2.2        1.2
Top 20%          51.5       52.7       51.8       54.3        2.8
Top 10%          33.5       35.5       33.8       37.3        3.9
Top 5%           21.8       24.6       22.2       26.2        4.4
Top 1%            8.8       12.0        9.1       13.2        4.4

Data Year     Decomposition I       Decomposition II
Adjusted to     PE          OE        PE          OE
Policy Year
              (3)-(1)    (4)-(3)    (4)-(2)    (2)-(l)

Average Tax Rates

P20-40           -0.4        0.0       -0.7        0.3
P40-60            0.1       -0.3       -0.8        0.7
P60-80            0.2       -0.4       -1.1        0.9
P80-90           -0.3       -0.1       -2.7        2.4
P90-95            0.1       -0.2       -4.5        4.3
P95-99            0.0       -0.9       -5.8        4.9
P99-99.5         -0.2       -3.0       -6.0        2.8
P99.5-99.9       -1.7       -5.8       -8.7        1.2
P99.9-99.99      -6.1       -7.2      -12.6       -0.7
P99.99-100       -6.0      -10.6      -14.3       -2.4
Top 20%          -0.7       -1.4       -5.6        3.6
Top 10%          -0.9       -2.1       -6.9        3.9
Top 5%           -1.3       -3.1       -7.9        3.5
Top 1%           -2.9       -5.9       -9.9        1.0

Posttax Income Shares

P20-40            0.0       -0.3       -0.3        0.0
P40-60           -0.1       -0.8       -0.5       -0.3
P60-80           -0.1       -1.5       -0.7       -0.9
P80-90            0.0       -1.1       -0.2       -0.9
P90-95           -0.0       -0.5        0.2       -0.7
P95-99           -0.1        0.0        0.4       -0.5
P99-99.5          0.0        0.4        0.1        0.3
P99.5-99.9        0.1        1.1        0.4        0.9
P99.9-99.99       0.2        1.4        0.4        1.2
P99.99-100        0.1        1.1        0.3        0.9
Top 20%           0.3        2.5        1.6        1.2
Top 10%           0.3        3.6        1.8        2.1
Top 5%            0.3        4.1        1.6        2.8
Top 1%            0.4        4.0        1.2        3.3

Data Year     Shapley-December
Adjusted to      PE            OE
Policy Year     Mean          Mean
               (4)-(2)      (2)-(1)
               (3)-(1)      (4)-(3)
Average Tax Rates

P20-40             -0.5          0.2
P40-60             -0.3          0.2
P60-80             -0.5          0.2
P80-90             -1.5          1.1
P90-95             -2.2          2.1
P95-99             -2.9          2.0
P99-99.5           -3.1         -0.1
P99.5-99.9         -5.2         -2.3
P99.9-99.99        -9.4         -4.0
P99.99-100        -10.1         -6.5
Top 20%            -3.2          l.l
Top 10%            -3.9          0.9
Top 5%             -4.6          0.2
Top 1%             -6.4         -2.4

Posttax Income Shares

P20-40             -0.1         -0.1
P40-60             -0.3         -0.6
P60-80             -0.4         -1.2
P80-90             -0.1         -1.0
P90-95              0.1         -0.6
P95-99              0.2         -0.2
P99-99.5            0.1          0.4
P99.5-99.9          0.2          1.0
P99.9-99.99         0.3          1.3
P99.99-100          0.2          1.0
Top 20%             1.0          1.9
Top 10%             1.1          2.8
Top 5%              1.0          3.4
Top 1%              0.8          3.6

Notes: Average tax rates include federal and state level
income and payroll taxes. Uprating according to the level of

price inflation. PE, policy effect; OE, other effect.

Sources: Own calculations based on SOI IRS income tax return
data and NBER TAXSIM calculator.

TABLE 3
Decomposition Results for Major Tax Reforms: OBRA90/OBRA93

Data Year       1989      1994      1989      1994
Adjusted to               1989      1994                 Total
Policy Year     1989      1989      1994      1994      Change

                 (1)       (2)       (3)       (4)     (4)-(1)

Average Tax Rates

P20-40          14.4      13.3      10.9       9.8        -4.5
P40-60          21.3      20.5      20.7      19.9        -1.4
P60-80          25.7      25.9      25.9      26.1         0.5
P80-90          27.9      28.3      28.2      28.6         0.7
P90-95          29.6      30.0      30.1      30.5         1.0
P95-99          30.1      30.9      31.2      31.9         1.8
P99-99.5        30.3      31.3      33.0      33.9         3.6
P99.5-99.9      30.3      31.3      36.7      37.3         7.0
P99.9-99.99     29.8      30.2      40.0      40.0        10.2
P99.99-100      29.3      28.9      40.8      39.8        10.5
Top 20%         29.3      29.9      31.5      31.9         2.6
Top 10%         29.9      30.6      33.0      33.4         3.5
Top 5%          30.1      30.8      34.1      34.6         4.5
Top 1%          30.0      30.7      37.2      37.4         7.4

Posttax Income Shares

P20-40           7.6       7.2       8.0       7.5        -0.1
P40-60          13.9      13.5      14.1      13.7        -0.1
P60-80          23.1      22.7      23.2      22.8        -0.2
P80-90          17.1      17.4      17.3      17.5         0.4
P90-95          11.2      11.5      11.3      11.6         0.3
P95-99          13.3      13.8      13.3      13.8         0.5
P99-99.5         3.1       3.1       3.0       3.1         0.0
P99.5-99.9       4.5       4.5       4.1       4.2        -0.3
P99.9-99.99      3.0       2.9       2.6       2.5        -0.5
P99.99-100       1.9       1.8       1.6       1.5        -0.4
Top 20%         54.1      54.9      53.3      54.1         0.0
Top 10%         37.0      37.5      36.0      36.7        -0.3
Top 5%          25.8      26.1      24.7      25.1        -0.6
Top 1%          12.5      12.3      11.4      11.4        -1.1

Data Year     Decomposition I     Decomposition II
Adjusted to       PE        OE        PE        OE
Policy Year

              (3)-(1)   (4)-(3)   (4)-(2)   (2)-(l)

Average Tax Rates

P20-40          -3.4      -1.1      -3.4      -1.1
P40-60          -0.5      -0.9      -0.6      -0.8
P60-80           0.3       0.2       0.3       0.2
P80-90           0.3       0.4       0.3       0.4
P90-95           0.5       0.4       0.6       0.4
P95-99           1.0       0.8       1.0       0.8
P99-99.5         2.6       1.0       2.6       1.0
P99.5-99.9       6.4       0.6       6.1       0.9
P99.9-99.99     10.2      -0.0       9.8       0.5
P99.99-100      11.5      -1.0      10.8      -0.3
Top 20%          2.2       0.4       2.0       0.6
Top 10%          3.0       0.4       2.8       0.6
Top 5%           4.1       0.4       3.8       0.7
Top 1%           7.2       0.2       6.7       0.7

Posttax Income Shares

P20-40           0.3      -0.4       0.3      -0.4
P40-60           0.2      -0.4       0.2      -0.4
P60-80           0.1      -0.4       0.1      -0.3
P80-90           0.1       0.2       0.1       0.3
P90-95           0.1       0.2       0.0       0.3
P95-99           0.0       0.5       0.0       0.5
P99-99.5        -0.1       0.1      -0.0       0.1
P99.5-99.9      -0.3       0.0      -0.3       0.0
P99.9-99.99     -0.4      -0.1      -0.4      -0.1
P99.99-100      -0.3      -0.1      -0.2      -0.1
Top 20%         -0.9       0.9      -0.8       0.8
Top 10%         -1.0       0.6      -0.9       0.5
Top 5%          -1.0       0.4      -1.0       0.3
Top 1%          -1.1      -0.1      -1.0      -0.2

Data Year     Shapley-December
Adjusted to          PE           OE
Policy Year        Mean         Mean

                (4)-(2)      (2)-(1)
                (3)-(1)      (4)-(3)

Average Tax Rates

P20-40             -3.4         -1.1
P40-60             -0.6         -0.8
P60-80              0.3          0.2
P80-90              0.3          0.4
P90-95              0.5          0.4
P95-99              1.0          0.8
P99-99.5            2.6          1.0
P99.5-99.9          6.2          0.8
P99.9-99.99        10.0          0.2
P99.99-100         11.2         -0.7
Top 20%             2.1          0.5
Top 10%             2.9          0.5
Top 5%              3.9          0.6
Top 1%              7.0          0.4

Posttax Income Shares

P20-40              0.3         -0.4
P40-60              0.2         -0.4
P60-80              0.1         -0.4
P80-90              0.1          0.2
P90-95              0.1          0.3
P95-99              0.0          0.5
P99-99.5           -0.0          0.1
P99.5-99.9         -0.3          0.0
P99.9-99.99        -0.4         -0.1
P99.99-100         -0.3         -0.1
Top 20%            -0.8          0.9
Top 10%            -0.9          0.6
Top 5%             -1.0          0.4
Top 1%             -1.0         -0.1

Notes: Average tax rates include federal and state level
income and payroll taxes. Uprating according to the level of
price inflation. PE: policy effect; OE, other effect.

Sources: Own calculations based on SOI IRS income tax return
data and NBER TAXSIM calculator.

TABLE 4
Decomposition Results for Major Tax Reforms: EGTRRA01/JGTRRA03

Data Year        2000       2004       2000       2004
Adjusted to                 2000       2004                 Total
Policy Year      2000       2000       2004       2004     Change

                  (1)        (2)        (3)        (4)    (4)-(1)

Average Tax Rates

P20-40           10.5        4.4        8.7        2.3       -8.2
P40-60           20.9       19.3       18.6       16.7       -4.2
P60-80           25.9       25.3       23.9       23.2       -2.7
P80-90           28.5       28.0       26.2       25.6       -2.9
P90-95           30.6       30.3       28.2       27.7       -2.9
P95-99           31.9       31.6       29.8       29.2       -2.7
P99-99.5         33.9       34.1       31.7       31.8       -2.1
P99.5-99.9       35.8       36.1       32.6       32.6       -3.2
P99.9-99.99      36.6       37.4       32.5       32.9       -3.7
P99.99-100       37.2       37.0       32.8       31.8       -5.5
Top 20%          32.1       31.7       29.5       28.9       -3.2
Top 10%          33.3       33.1       30.6       30.1       -3.2
Top 5%           34.2       34.0       31.4       30.9       -3.2
Top 1%           35.9       36.2       32.5       32.3       -3.5

Posttax Income Shares

P20-40            7.3        7.2        7.2        7.1       -0.2
P40-60           12.4       12.5       12.4       12.4       -0.0
P60-80           20.5       21.0       20.4       20.9        0.3
P80-90           15.8       16.5       15.8       16.4        0.6
P90-95           10.9       11.2       10.9       11.2        0.4
P95-99           14.2       14.2       14.2       14.2        0.0
P99-99.5          3.6        3.4        3.6        3.4       -0.2
P99.5-99.9        5.6        5.1        5.7        5.2       -0.3
P99.9-99.99       4.5        3.8        4.6        3.9       -0.5
P99.99-100        2.8        2.7        2.9        2.8       -0.0
Top 20%          57.4       56.9       57.7       57.2       -0.2
Top 10%          41.5       40.5       41.8       40.8       -0.7
Top 5%           30.7       29.3       30.9       29.6       -1.1
Top 1%           16.5       15.1       16.8       15.4       -1.1

Data Year     Decomposition I       Decomposition II
Adjusted to      PE         OE         PE         OE
Policy Year

              (3)-(1)    (4)-(3)    (4)-(2)    (2)-(1)

Average Tax Rates

P20-40           -1.8       -6.4       -2.1       -6.1
P40-60           -2.3       -1.8       -2.6       -1.6
P60-80           -2.0       -0.7       -2.1       -0.6
P80-90           -2.3       -0.6       -2.4       -0.5
P90-95           -2.4       -0.5       -2.6       -0.3
P95-99           -2.2       -0.5       -2.3       -0.4
P99-99.5         -2.2        0.1       -2.4        0.3
P99.5-99.9       -3.2       -0.0       -3.5        0.3
P99.9-99.99      -4.0        0.3       -4.6        0.9
P99.99-100       -4.4       -1.0       -5.2       -0.3
Top 20%          -2.5       -0.6       -2.8       -0.4
Top 10%          -2.7       -0.5       -3.0       -0.2
Top 5%           -2.8       -0.4       -3.1       -0.1
Top 1%           -3.4       -0.2       -3.8        0.3

Posttax Income Shares

P20-40           -0.1       -0.1       -0.1       -0.1
P40-60           -0.1        0.0       -0.1        0.0
P60-80           -0.1        0.4       -0.1        0.5
P80-90           -0.0        0.6       -0.1        0.6
P90-95            0.0        0.3        0.0        0.3
P95-99            0.0       -0.0       -0.0        0.0
P99-99.5          0.0       -0.2        0.0       -0.2
P99.5-99.9        0.1       -0.4        0.1       -0.4
P99.9-99.99       0.1       -0.6        0.1       -0.7
P99.99-100        0.1       -0.1        0.1       -0.2
Top 20%           0.3       -0.4        0.3       -0.4
Top 10%           0.3       -1.0        0.3       -1.1
Top 5%            0.3       -1.4        0.3       -1.4
Top 1%            0.3       -1.4        0.3       -1.4

Data Year     Shapley-December
Adjusted to       PE           OE
Policy Year      Mean         Mean
               (4)-(2)      (2)-(1)
               (3)-(I)      (4)-(3)

Average Tax Rates

P20-40             -2.0         -6.2
P40-60             -2.5         -1.7
P60-80             -2.1         -0.7
P80-90             -2.3         -0.5
P90-95             -2.5         -0.4
P95-99             -2.3         -0.4
P99-99.5           -2.3          0.2
P99.5-99.9         -3.4          0.2
P99.9-99.99        -4.3          0.6
P99.99-100         -4.8         -0.7
Top 20%            -2.7         -0.5
Top 10%            -2.8         -0.4
Top 5%             -2.9         -0.3
Top 1%             -3.6          0.1

Posttax Income Shares

P20-40             -0.1         -0.1
P40-60             -0.1          0.0
P60-80             -0.1          0.5
P80-90             -0.0          0.6
P90-95              0.0          0.3
P95-99              0.0          0.0
P99-99.5            0.0         -0.2
P99.5-99.9          0.1         -0.4
P99.9-99.99         0.1         -0.6
P99.99-100          0.1         -0.1
Top 20%             0.3         -0.4
Top 10%             0.3         -1.0
Top 5%              0.3         -1.4
Top 1%              0.3         -1.4

Notes--. Average tax rates include federal and state level
income and payroll taxes. Uprating according to the level of
price inflation. PE, policy effect; OE, other effect.

Sources: Own calculations based on SOI IRS income tax return
data and NBER TAXSIM calculator.

TABLE 5
Relative Importance of Policy Effect

              ETI = 0    ETI = 1    ETI = 0    ETI = 1

               Average Tax Rates       Income Shares
                  (1)        (2)        (3)        (4)

P20-40           50.8       41.9       21.4       32.2
P40-60           77.9       62.5       17.4       29.4
P60-80           87.0       78.2       15.8       26.1
P80-90           96.2       87.6       12.4       20.9
P90-95           98.3       83.6       20.4       33.5
P95-99           99.0       85.1       29.3       40.5
P99-99.5         80.4       71.6       11.9       18.1
P99.5-99.9       73.4       64.5       13.9       23.1
P99.9-99.99      75.5       68.2       12.1       21.6
P99.99-100       69.3       57.3       10.9       20.8
Top 20%          94.6       87.9       19.2       28.3
Top 10%          92.1       85.8       15.9       26.0
Top 5%           89.0       83.0       14.2       24.8
Top 1%           77.5       69.4       12.5       22.4

Notes'. The mean of the absolute values of the policy effect
is expressed in % of the mean of the absolute values of the
total effect. Average tax rates (%) include federal and state
level income and payroll taxes. Uprating according to the
level of price inflation. Ranking based on pretax income excl.
capital gains. ETI, elasticity of taxable income.

Sources: Own calculations based on SOI IRS income tax
return data and NBER TAXSIM calculator.