Taxation and Inequality: A Time-Exposure Perspective Author(s): Joel Slemrod Source: Tax Policy and the Economy, Vol. 6 (1992), pp. 105-127 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/20061811 Accessed: 28/09/2009 09:27 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ucpress. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that promotes the discovery and use of these resources. For more information about JSTOR, please contact [email protected].

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Tax Policy and the Economy.

http://www.jstor.org

TAXATION

AND

INEQUALITY: A TIME-EXPOSURE PERSPECTIVE Joel Slemrod The University

of Michigan

and NBER

I. INTRODUCTION consensus There is a wide economists that the distribution of among income in the United States recently has become more unequal. Most trace that this trend was well under way by 1980; many observers believe its origin to as early as 1970. There is less consensus about the causes of are a shift in in inequality. Among the growth the competing hypotheses demand labor and away from unskilled toward high-skilled labor, an a increase in the relative of low-skilled and shift in workers, supply are more sectors toward where individuals' vari output productivities able and more easily identified. There growth policy

is virtually no consensus about the role of the tax system in the in inequality, and swirling controversy tax about the appropriate to this trend. For example, and Gr?mlich, Kasten, response

on Tax Bureau of Economic for the National Research Conference Policy and the to be held November 19 in Washington, I am grateful to Marianne D.C. Page for exemplary research and to Shlomo Yitzhaki for providing assistance, Rodney on an earlier draft were in the analysis. comments used programs key computer Helpful Ned Gr?mlich, Rick Kasten, Donald Kiefer, Jim provided by Don Fullerton, Roger Gordon, Frank Sammartino, and Shlomo Yitzhaki. Bob Williams, Poterba, Prepared

Economy, and Alec

106

Slemrod

Sammartino came slightly

that, in the 1980s, the tax system be (1991) have claimed in reducing in pre-tax incomes. less effective the inequality has that the (1990) precipitous Lindsey argued drop at the start of the tax rate imposed on higher income individuals was 1980s in the marginal a cause of the increased in of pre-tax apparent principal inequality it has to because these work more, individuals comes, encouraged report more income, realize more capital gains, and convert nontaxed compen sation to taxable compensation. of the tax policy changes Finally, many in the child credit financed advocated, currently being particular by on are tax rates higher taxpayers, higher-income justified in part as an attempt to offset the increase in income inequality. all of the conclusions about the trend in inequality and the role Nearly of the tax system have been based on cross-sectional of an snapshots that conclusions nual income. Yet it is well known based on snapshots can give a misleading of the inequality of a more permanent picture across annual notion of income, due to the mobility of individuals in come classes. will income the of appearance transitory give Specifically, a over a in than If one-year greater inequality snapshot longer period. income has changed over time, then the relative importance of transitory the trends in the inequality of annual income may misstate the trends in a more permanent most of the conclu notion of income. Furthermore, or the sions about inequality have been based on either labor earnings Current definition of which income, (CPS) money Population Survey other capital income, excludes and is capital gains, poorly measures to maximum values. coded subject In this paper Imake use of tax return data that has extensive informa two longitudinal tax return tion on property income, and in particular data bases that follow an unchanging from 1979 to sample of taxpayers set of taxpayers from 1967 to 1973, in order to 1986, and a distinct reassess some of the conclusions about taxation and inequality. The a as is II brief Section review what is of follows. paper arranged provides over the past two decades. of inequality based on snapshots, known, some of these same tax return data to investigate III utilizes Section on decomposing an emphasis the trends by source of in issues, with come.

data file of In Section IVI bring to bear the 1979-1986 longitudinal some of the earlier results; Section V compares tax returns, and reassess the findings these results to the 1967-1973 panel. Section VI summarizes to these findings. and restates the caveats to be applied

Taxation

and Inequality

107

IL INEQUALITY AND TAXES: A BRIEF REVIEW OF THE LITERATURE A.

Trends in Inequality in There is a broad consensus that income inequality has been increasing There is evidence of increased dispersion since the the last two decades. in both the lower and upper tails of the distribution mid-1970s for fami and in the distribution lies and for individuals, of labor income for work ers. At least since 1979, inequality has grown both between less and more educated workers and also among apparently similar workers. As a great vari to this conclusion is robust has documented, (1991) Karoly children and race ety of disaggregations, including by families with in inequality cannot simply be The increase among workers ethnicity. or in the education shifts experience composition explained by gender, to of the work the sample is restricted force, and is evident even when full-time

workers.

to the increased have been offered alternative explanations to attention has been labor market Most factors, as labor paid inequality. market for about 70 percent of family income. Among income accounts co the supply explanations offered are shifts in the size of worker-age of these cohorts. Among horts and the educational distribution the de mand explanations offered are shifts in the composition of final output, in the occupational in skill requirements, mix within and in industries, Several

the density of unionization. The strength these not mutually exclusive explanations Murnane (1991).

of the evidence supporting is assessed in Levy and

as an

to Growing Inequality Offset measure it is clear that of income is before taxation, the CPS Although of well-being. With it after-tax income is a superior measure that inmind, is interesting to know to what extent the tax system has offset, or exacer in pre-tax income. Using CPS data supple bated, the increased inequality tax mented data and Sammartino from returns, Gr?mlich, Kasten, by B.

Taxes

that during the 1980s tax and transfer policy changes (1991) conclude in reducing pre-tax income inequality. As a result, became less effective increased by even more after-tax income inequality than pre-tax income a in 16.5 increase the Of Gini coefficient of post-tax, percent inequality. income between 1980 and 1990, they attribute 40 percent to post-transfer in the redistributive the decline of the federal tax and trans effectiveness fer system, with the 40 percent about equally divided between taxes and in the distribution transfers. Some of the change of transfers and taxes

108 Slemrod and demographic the result of economic rather was, however, changes than explicit changes in policy. When hold the distribution of pre they income at its 1990 level but adjust transfers and taxes to tax, pre-transfer reflect 1980 law, only 16 percent of the increase in the inequality of post income was due to policy changes. Most of these policy tax, post-transfer a decrease in the progressivity of federal taxes, due to involved changes the increased of the regressive importance nificant was a fall in the share of transfers

security taxes; also sig to low-income people.

social going

as an Inducement

to Growing Inequality in the that the tax system The previous calculation ignores possibility in pre-tax has that in the income. Lindsey duces changes (1990) argued tax rates at the top (which fell in early 1980s the sharply reduced marginal in 1987 to 39 percent, and in 1988 to 28 to 50 percent, 1981 from 70 percent to work more hours, induced report a taxpayers percent) high-income more to income Internal Revenue realize their the share of Service, higher more in taxable and take form, compensation generally capital gains, income or leisure. substitute taxable income for either nontaxed increase in the dis of the apparent This argument implies that much were in The incomes is of income the 1980s there, high illusory. persion were not in the either in that to would but forms 1981, appear prior they

G

Taxes

standard

data

sets

on

income,

not

be

realized

income

sources,

or were

in the form of leisure. This important and income consumed potential in the analysis that follows, claim is not addressed controversial directly next can help to assess its validity. the data sources described although

D. Public Finance Implications

of Inequality

of income and its sources is criti An accurate picture of the distribution is the in public finance: How two at cal for least important questions and how tax should burden of any given system shared by individuals, the tax burden be shared across income classes? how the burden of taxation is There are two steps to understanding as the question to economists of tax In the first step, known borne. of supply and demand is of the price sensitivity incidence, knowledge a tax will be borne by the statutory the extent to which used to predict the extent to which the "check writer"?and bearer of the burden?the in prices. In the second burden will be shifted to others through changes is translated into about how prices will be affected step the conclusions information This step requires is changed. how families' well being labor income, and tastes. For about the distribution of wealth, potential a once about how a cigarette tax will change ismade judgment example, to know and the value of tobacco land, one needs the price of cigarettes

Taxation

and Inequality

109

unskilled the distribution for cigarettes, of demand labor, and owner ship of tobacco land. a critical of income-earning The distribution ability is also input into tax which the prob the modern of poses theory optimal progressivity, the social benefit of a more equal distribution lem as a tradeoff between cost of the tax and transfer system of well-being and the disincentive to effect a more equal distribution. In the standard models needed of this a more in of Slemrod described distribution literature, (1983), unequal inherent abilities?or the return to these abilities?increases the optimal it increases the social value of the redis because degree of progressivity tribution accomplished for any given degree of tax progressivity. An other strand of the literature, exemplified the (1980), stresses by Varian an social insurance function of progressive taxation. In this framework, increase in exogenous of income would also increase the variability opti it increases mal degree of progressivity, because the insurance value of the progressivity.

III. SNAPSHOT DISTRIBUTIONS A. Why Snapshots May Mislead about the distribution of these factors are based Many of the conclusions on a at the distribution taken of point of time. It is critical to snapshots can give a that snapshots of distributions realize, however, highly mis a more the of notion of distribution of income permanent leading picture or well-being. an economy As an illustration, consider of individuals each individual who are identical except for their age, where follows the same life-cycle pattern of increasing labor income over the working years, followed by a period of retirement with no labor income. A snap shot of the distribution of labor income would conclude that to the extent a tax lowers after-tax wage itwill have a highly skewed burden income, some individuals distribution, (the working severely affecting young) and leaving others unscathed (the retired elderly). But, of course, from a lifetime perspective is equally effects) everyone (except for transitional as everyone will at some time be at the high labor income age affected, and also the retirement to life-cycle effects, a measure of age. In addition a on will based intercohort to due effects capture inequality snapshot real income growth over time. reason why a This paper focuses on a separate snapshot may give a of income sources?transitory in picture of the distribution misleading come. An example will best convey an the problem. economy Imagine where individuals have but two sources of income. One source, which I

110 Slemrod across individuals will call labor income, is equally distributed and stable over time. The other source, which Iwill call capital gains, also is equally over time; every so often an exoge distributed but is also highly unstable nous process A snap conveys capital gains income to some individuals. of income sources will reveal two "facts": shot of any year's distribution of total income ismore unequal than that of labor (1) that the distribution source of income alone and (2) that capital gains are a more important even These facts will appear income for high-income individuals. though, over the long run, capital gains income is distributed no differently than have equal income. labor income and all individuals it was There is one more twist to this story. In the above illustration was assumed that the time pattern of capital gains income exogenously In reality the timing of the realization determined. of income for tax a to deal In of control by the individual. purposes may be subject great nature tax the the income because of of system, particular, graduated over the to use their discretion there is an incentive for individuals taxable income over time. Not only timing of some income to smooth some to extent the taxable income from but capital gains realizations, can to kind this be of Other transitory business, "self-averaging." subject or income as to those due such of income, components unemployment are essentially to the taxpayer. windfalls, exogenous out the effect of is to separate The goal of this research program as on so to income measured get a better measure inequality, transitory notion of inequality. of the nature of, and trends in, a more permanent

Tax Return Data B. Measuring Inequality Using Most analysis of trends in U.S. inequality have relied on the data in the Bureau in Current Population (CPS), conducted by the Census Surveys that follows is based on the data from March of most years. The analysis the Internal Revenue Service (1RS) Statistics of Income Division's public use files of tax return information. This shift in data sources has its I detail below. which and disadvantages, advantages to be a the is that it is designed An important CPS of advantage exam not is and for therefore entire of the restricted, population sample a wide variety of demographic to tax informa It collects filers. also ple, tion, including details about family structure and earnings disaggregated of the CPS is its poor coverage of One disadvantage by family member. Realized income. capital capital gains are not included at all, and other are apparently of realized severely under components capital income as total family as well all of In income, addition, components reported. no on so are information the upper tail that there is income, top-coded,

Taxation

and Inequality

111

of the distribution of income or its components. Clearly both of these are in of the high-income the important analysis especially problems population. The strengths and weaknesses of the tax model data mirror those of the that file tax returns, so it systematically CPS. It covers only households low income makes them exempt from filing excludes households whose credit. It also and who do not file a return in order to receive a refundable a return. There is very little demo to fail file excludes those who illegally data has very rich information. On the plus side, the tax model graphic on taxable sources of capital income, and is not top-coded. information

G Trends in Snapshot Inequality some summary in the information about the changes Table 1 presents over the last two distribution of after-tax income, and its components, The income concept used here, called expanded decades. income, be to with income and then adds it excluded gins adjusted gross long-term an after-tax and all adjustments. When capital gains, excluded dividends tax liability net of credits is subtracted from expanded is used, concept are designed to make income. These changes the income concept more across

comparable

years.

an ideal measure income is by no means of annual in Expanded are the failure to correct capital income for come. Among the problems of the rental value of owner-occupied the exclusion homes inflation, and other consumer the failure to subtract real interest pay durables, rather than and the inclusion of capital gains on a realization, ments, an accrual, basis. Nevertheless, because many studies of tax return data use

to under income concepts with these characteristics, it is important that are based on this type of stand the nature of inequality measures information.

as a measure In Table 1 Imake use of the Gini coefficient of inequal row a in increase in the The numbers the document first ity.1 striking 1972 income between and of after-tax the Gini 1988; inequality expanded coefficient rises from 0.445 to 0.544 over this period. For pre-tax income the increase is from .468 in 1972 to .496 in 1980 and .567 in 1988. This is consistent with the Gr?mlich, and Sammartino Kasten, (1991) broadly rose result that the Gini of pre-tax, income in 0.473 from pre-transfer

1 I am grateful coefficient

Gini

random stratified Statistics

to Shlomo Yitzhaki and decomposition

for providing discussed

the computer for calculating the programs on The figures of Table 1 are based of approximately Tax Model 15,000 per year of the Individual Files, a subsamples tax returns made random of about available 100,000 sample by the publicly of Income Division of the Internal Revenue Service. here.

112 Slemrod TABLE and Decomposition Gini Coefficients its Components and Pre-Tax

1.

of After-Tax Expanded Income Income, 1972-1988. Expanded

1972 Gini income

Expanded after tax

Share

and

1976 Rho

Gini

Share

Rho

0.4450.458

and salaries

0.489

0.927

0.856

0.515

0.948

0.856

1.052

0.055

0.760

1.066

0.042

0.733

0.884

0.041

0.438

0.883

0.051

0.457

0.981

0.026

0.712

0.982

0.024

0.734

Schedule

C 1.188

0.053

0.562

1.286

0.047

0.542

Schedule

E 1.713

0.029

0.509

2.144

0.024

0.422

Schedule

F 2.465

0.006

0.320

6.664

0.003

0.326

0.030

0.303

Wages

Capital gains Interest Dividends

Pension 0.976 Other

income

income Expanded before tax

0.016

0.279

0.962

+2.716

-0.007

+0.006

-1.937

-0.014

+0.654 Tax

-0.146

+0.957

+0.682

-0.155

0.029 +0.960

0.4680.485

0.523 in 1990. Table 1 also pres 1980 to 0.513 in 1985 and an estimated ents the Gini coefficient and shares of each of several components of after-tax income.2 It also lists, under the columns headed the Gini "rho," and after-tax income.3 As discussed the component correlation between in Lerman of income can be and Yitzhaki (1985), the Gini coefficient so over to all is that it the of the sum, components, equal decomposed own in its and the share the Gini of Gini, income, product component's correlation.

about 90 percent of after-tax and salaries Because wages represent it is reasonable to look income (and about 80 percent of pre-tax income), in inequality, and this strategy is there for the origin of the change 1972 and 1988 rewarded. The second row of Table 1 reveals that between and salaries increased from 0.489 to 0.578. the Gini coefficient of wages a in is associated with the overall of trend large part inequality Clearly the increased

inequality

of wages

and salaries.

2 The

for income the prevalence astute reader will notice of Gini coefficients components is due to the presence of negative values that are in excess of one. This for these compo curve lies the x-axis. that the Lorenz nents, which means partly below 3 The Gini correlation of component tion's rank in the income distribution -1 and tion of k. It ranges between and the ranking by k are identical.

k is the ratio of the covariance of k with the observa of k with the rank in the distribu and the covariance +1, taking on a value of +1 if the ranking by income

Taxation

and Inequality

113

1980 1984 1988 Gini

Share

Rho

Gini

Share

Rho

Share

Rho

0.544

0.502

0.467

Gini

0.526

0.937

0.854

0.554

0.909

0.853

0.578

0.882

0.867

1.088

0.053

0.781

1.061

0.070

0.780

1.086

0.062

0.781 0.487

0.886

0.074

0.489

0.869

0.091

0.457

0.888

0.069

0.975

0.030

0.773

0.963

0.026

0.686

0.967

0.029

0.666

1.408

0.040

0.490

1.446

0.039

0.433

1.271

0.046

0.494

3.582

0.020

0.497

4.581 -8.925 0.955 289.38 +0.682

0.496

0.053 -0.002 0.034 0.000 -0.179

0.270 0.315 0.255 0.385 +0.957

653.00 -1.844 0.945

0.000 -0.007 0.041

-3.805

-0.012

+0.687

-0.159

0.524

0.210 0.167 0.348 0.226 +0.956

-37.74 0.932

0.000

0.182

0.067

0.490

-3.578

-0.011

+0.731

-0.164

0.520 +0.962

0.567

Source: Random subsamples 15,000 per year) of Individual Income Tax Model files. (of approximately Note: Gini coefficients of sources of income with a share close to zero are unreliable indicators of the skewness of the distribution.

tax liability reduces For all the years, the dispersion of subtracting In the is small. income as measured the Gini, although change fairly by 1972 the reduction is 0.023. It peaks in 1980 at 0.029, when the ratio of tax income also reaches its peak of 0.179. It liability to after-tax expanded in 1984 back to 0.022 and then turns slightly up in 1988 to 0.023. declines to offset Thus the federal income tax liability has done little, if anything, in the 1980s it has slightly the increasing dispersion of pre-tax incomes; exacerbated the trend toward increased inequality of after-tax incomes, incomes holding pre-tax unchanged. the Gini coefficient of pre-tax and post-tax incomes an Comparing swers the question a total elimination of what of federal income taxes is how a marginal do to inequality. Another meaningful question taxes in would affect overall income proportional change inequality. Lerman and Yitzhaki (1985) show that a 1 percent change in component k changes the overall Gini by sk(pkGk sk, pk and Gk are G)/100, where own the share, Gini correlation, and Gini of component k, respectively, and G is the overall Gini. Performing this calculation reveals that a 10 percent increase in taxes would have reduced the overall proportional Gini in 1972 by 0.00264, or 0.59 percent of its actual value. By 1988, the

would

114

Slemrod

same

have reduced would experiment 0.48 percent of its actual value.4

the overall

Gini

by 0.00262,

or

IV. TIME-EXPOSURE INCOME D.

The Importance Income of Transitory The presence of transitory of taxpayers across income, and the mobility that inequality income classes, means of income in any given snapshot can exceed that of a more permanent notion of income, and that conclu of any source of income to inequality based on sions on the contribution can be misleading. some evidence In this section I present on snapshots the importance of these issues for the period 1979 to 1986. tax returns. is a panel of individual The source of this information in 1RS the Statistics Income Division of has been collect 1979, Beginning a tax returns of information from the selected group of ing randomly as was This known the Continuous Work File, taxpayers. History panel,

this longitudinal for internal use, but the 1RS has made data developed a set available to academic researchers with through special arrangement at the University in con the Office of Tax Policy Research of Michigan, with the Ernst & Tax Research Database. The junction Young panel now soon. to and 1979 with 1987 1988 The 1986, spans expected panel is a nonstratified random sample chosen on the basis of the last four digits of the primary social security number (SSN). Of those numbers taxpayer's a return is in included the chosen, anyone sample. The first three filing in excess of 45,000 returns, years of the panel each contain though the in last three years of the panel show a substantial the number of drop in 1983 observations and 1986,19,000 9,000 in 1982,1984, (approximately at the 1RS.5 Pooling and 1985) due to budgetary limitations all observa a tions in the panel gives sample size of 177,177. Due largely to the small in 1982, 1984, and 1986, the number of individu number of observations 4Note

that this procedure that a 100 percent in taxes (i.e., complete reduction implies in 1988 by 0.0262, whereas the Gini the actual difference increase would elimination) is due to the fact that the pre-tax and after-tax Gini is only 0.023. The discrepancy between at the margin. in in the text is precise nonincremental the calculation Any only change a taxes would of taxpayers the increase income, which would require reranking by after-tax to the complete calculation elimination Gini. For this reason the marginal of taxes applied overstates 5 The

the actual

difference

between

pre-

and

after-tax

Gini

coefficients.

on the basis of the last four digits of the social security is drawn number of sample In 1979 through 1981 the sample the primary (first listed on the tax return) taxpayer. In 1982 only all returns filed in a calendar includes year with any of five four-digit endings. were one of the five endings two of the in 1983 returns with those returns with drawn; was were continued chosen. The endings alternating one-ending, two-ending cycle 1986. through

Taxation

and Inequality

115

in all eight years of the panel is limited to 5,780 taxpayers. als present in each observation The information is a subset of the informa contained tion on the standard forms filed by the taxpayer, and varies slightly from year

to year.

to from the panel may occur for a number of reasons unrelated a in in the marital change sample size, including death, change that would status, income below the minimum trigger filing, or simply the two married, choice of which spouse (between joint filers) is listed first on the tax form (and thus becomes the "primary" taxpayer whose SSN is the A taxpayer who files sufficiently basis of selection). late (in the calendar year after the return is due) will also escape inclusion. to suspect It is not unreasonable that a panel of this sort may exhibit some drift relative to the population as a whole. each year's Although as a in the be of the taxpayers panel may representative population cause a or the sampling method "attri bias whole, may "survivorship" tion" among in more those observations than one year of the present Attrition

deliberate

and Frischmann the first six years of the (1989) analyzed panel. Christian attrition and that for bias concluded the panel sample of taxpayers pres ent in all those years shows statistically from popu significant differences lation averages. Average income is about 20 percent higher and married to Also, compared joint filers) are more numerous. couples (specifically, a random returns the for which fraction of the sample, primary taxpayer claims an aged exemption (for being sixty-five or older) is lower in the initial year of the panel, rises more rapidly, and is higher in the final year of the panel. to the tax model files used in Section 3, the critical advan Compared the is its data of allows the re nature, which tage panel longitudinal to identify on searcher the opportunity effects income. Its transitory ran other than attrition the is the bias, principal disadvantage, purely dom nature of the sample, to the stratified random character compared of the tax model that heavily over-samples upper-income taxpayers. that there is a significant amount of mobility Table 2 documents in and out of the upper-income classes from year to year, at least for this particu lar definition In a typical year in the first half of the 1980s, of income. more than 20 percent of taxpayers in the top decile of expanded income had not been in the top decile the previous year. For the top one percen the figure rises to 33 percent for 1982 through 1985. Mobility into in 1986, presumably due (and out of) the top classes was extraordinary to the large spurt in capital gains realizations that occurred in anticipa tion of tax increases due in 1987. The presence of transitory income means those with that, in general, are probably in a snapshot not really as badly off as one low income

tile,

116

Slemrod

TABLE

2.

Income Percentage Expanded of Those in Top Percentiles ofPre-Tax in That Group in the Previous 1980-1986. Who Were Not Year, 1980

5% 1%

22 26 28

Source: 1979-1986

20%

1981

1982

1983

1984

24 28 30

21 26 33

22 26 33

20 24 33

27

Panel Returns

only.

Panel of Individual

Tax Returns,

Balanced

1985

1986

23 27 33

31 40

and those with high income are probably not as year's income suggests In order to investigate the aggregate well off as the snapshot suggests. I for the calculate each these of effects, average real taxpayer magnitude 1979 to income over the seven-year from 1985. available period Although in the data set, 1986 is excluded from the calculations of the because in to that I realizations refer amount of year. extraordinary capital gains as "time-exposure" to contrast it to this concept income, "snapshot" income.

or "lifetime" I purposely this concept from "permanent" distinguish is a separate and literature income. There (e.g., Fitzgerald Maloney, to calculate and Rogers, the in 1990; Fullerton 1991), which attempts or in from of lifetime income annual permanent by purging equality come not only the effect of transitory income but also the life-cycle and entails first esti intercohort effect. The calculation procedure generally a as an annual income of vari that function equation predicts mating data is available, ables such as age, education, race, and, if longitudinal an the discounted fixed effect, and then calculating individual-specific life. Using annual incomes over the expected working value of projected data from 1969 to 1981 from the Panel Study on Income Dynamics, for in and Maloney that, not correcting (1990) calculated Fitzgerald the Gini coefficient for lifetime income was only 1.4 tercohort effects, intercohort effects were lower than that of 1979 income; when percent the Gini of lifetime income was 19.1 percent lower than that eliminated, of 1979 income. transfers reduce the fiscal system

to which taxes and They also found that the degree a snapshot, understated that is suggesting by inequality income over the life than smooth household does more

cycle.

of the magnitude Table 3 illustrates income by arraying and time-exposure clear evidence toward the of reversion classes below $20,000, time-exposure

the difference

between snapshot the latter by the former. There is mean. In 1983 snapshot income income is greater than 1983 in

Taxation

TABLE Pre-Tax

Time Exposure

Expanded Expanded

1983 Snapshot income

expanded class

0-5000 5-10,000 10-15,000 15-20,000 20-25,000 25-30,000 30-50,000 50-75,000 75-100,000 >100,000 Source: 1979-1986

Panel of Individual

Average

1983 income

-22923