Average Marginal Tax Rates from Social Security and the Individual Income Tax Author(s): Robert J.

Barro and Chaipat Sahasakul Source: The Journal of Business, Vol. 59, No. 4, Part 1 (Oct., 1986), pp. 555-566 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/2353008 Accessed: 06/11/2009 09:57
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 service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Business.

http://www.jstor.org

Robert J. Barro
University of Rochester

Chaipat Sahasakul
Rutgers University

Average Marginal Tax Rates from Social Security and the Individual Income Tax*
In our previous paper (Barro and Sahasakul 1983)we provided estimates of average marginal tax rates from the federal individualincome tax for 1916-80. Now we extend these figures to 1983 and supplementthem to include the social security tax on labor earnings. With this addition, in 1983the includedtaxes constitute 73%of federal and 45%of total governmentreceipts. If some nontax items are excluded, the values are 78%and 53%, respectively.1 In the main, the social security levy is a flatrate tax, paid partly by workers, partly by employers, and partly by self-employed persons. The computationof average marginaltax rates is simpler than in the case of the federal income tax, which has a graduated-rate structureand allows for numerous deductions from taxable income. The main complicationsthat arise for the social security tax are the following. (a) For workers and self-employed persons with earnings above a ceiling value the marginaltax rate is nil; (b) the tax applies only to labor earnings(and to earningsfrom self-employment)ratherthan to total income; (c) the employer and employee
* This research was supportedby the National Science Foundation.We appreciatethe data that were providedon income taxes by Dan Holik and on social security by Anthony Pellechio and Wayne Long.
1. The data are from U.S. Survey of Current Business,

We extend previous estimates of the average marginaltax rate from the federal individual income tax to include social security. Our computationsconsider the tax rates on employers, employees, and the self-employed; the income that accrues to persons with earningsbelow the ceiling; and the effective deductibilityof employers' social security contributionsfrom workers' taxable income. The net effect of social security on the average marginaltax rate is below .02 until 1966but then rises to .03 in 1968, .04 in 1973, .05 in 1974, .06 in 1979, and .07 in 1982.

March 1986.
(Journal of Business, 1986, vol. 59, no. 4, pt. 1) ? 1986 by The University of Chicago. All rights reserved. 0021-9398/86/5904-0002$01 .50 555

556

Journal of Business

parts of the tax differ because the employer's payments are not counted as part of the employee's taxable income; and (d) an individual's future social security benefits depend positively on that person's historyof contributions.The last element reduces the effective tax rate thatan individualfaces. In fact, Gordon(1982)arguesthat this considerationis importantfor people who are close to retirementage. Generally, the inclusion of this effect would require forecasts of benefit schedules as well as survivalprobabilities.It would also be necessary to include various complexities of the social security law, such as the decliningmarginaleffect of past covered earningson benefits, the exclusion of some years of earningsfrom the formula,and the treatment of spouses and dependents. In any event, our subsequentcalculations do not take account of the effects of social security contributionson futurebenefits. Thus, by includingonly the tax aspects of these "contributions," we somewhat overstate the effective marginaltax rates from the social security program. I. TheoreticalConsiderations Let Sf be the social security tax rate (marginaland average)paid by a firmon workers' earnings. If profits are taxed at the rate T, then the firm's after-taxprofits are
I =

(1

TU)

[F(L) - wL(1 +

Sf)],

(1)

whereL is the quantityof laborinput, w is the real wage rate, and F(L) is the productionfunction. Maximizationof profit implies F' = w(1 +
Sf),

(2)

where F' is labor's marginalproduct. The representativeworker's total real income, Y, equals wL + I, where I is nonlaborincome. As in our previous paper, this income is spent on consumption, C, or on income taxes, T.2In addition,there is now the worker's social security tax, Se - wL, where Se is the employee's (marginaland average) contributionrate. Thus we have + Y = wL + I = C + T Se(WL).
(3)

As before, income taxes, T, depend on taxable income, Y - D, where D is a broad concept of deductions. If utility depends positively on consumptionand negativelyon work, then the first-order conditionfor
2. For present purposes it is unnecessaryfor us to considertwo categoriesof consumption-depending on the treatmentby the tax law-as we did in the earlierpaper. We also do not allow here for efforts aimed at avoidingincome taxes.

Average Marginal Tax Rates

557

maximizingutility can be written as

-

a U/aL adUtac = w(l - T' -

Se),

(4)

where T' is the marginalincome-tax rate. Substitutingfor w from equation (2)4nto equation (4) implies -aU/aL F'(l - T' - Se) au/ac 1 + Sf Thus equation (5) shows how the tax system creates a positive wedge between labor's marginalproduct, F', and the utility rate of substitution between consumptionand leisure, -(adU/L)I(adUIC). Let v be the overall effective marginaltax rate on labor's marginal product, F'. Then equation (5) implies
-

T'
1+

-Se Sf

or 1
(Sf + Se

+ T').

(6)

Thus the tax system effectively deflates labor's marginalproduct, F', by the factor 1 + sf (see eq. [2]) and then applies the marginal rate, tax Sf + Se + T'.3 If the social security tax is not purely a flat-ratelevy (because of the ceiling on taxable earningsin the U.S. system), then we can interpretsf and Se in equation(6) as the marginalsocial securitytax rates. For self-employedpersons the formulais simpler;namely, if ss is the marginalcontributionrate to social security, then the effective marginal tax rate, Ts, iS4
Ts = ss + T'.

(7)

Previously, we calculated weighted averages, T', of the marginal income tax rates, T'. We weighted either by adjustedgross income or by numbers of returns, and we computed arithmeticand geometric averages. Here we consider only the series that we focused on earlier, which is the arithmeticaverage weighted by adjustedgross income.
3. Note that T does not depend solely on the sum Sf + Se* Thatis because, unlikethe worker'spayments, the employer'spaymentsare not part of the worker'stax base. 4. If the marginaltax rates, T', are equal, then the equationof Ts from (7) to T in (6) requires sS to be less than sf + Se, as was true in the United States until 1984. For example,if T' = .3 and sf = Se = .067 (the value for 1982),then the equalizingvaluefor Ssis .017.The actualvalue of sS for 1982was .0935.The social securitylaw passedin 1983 andeffective in 1984sets the self-employedrateequalto the sumsf + Se but providesfor some offsettingincome tax credits.

558

Journal of Business

Equations (6)-(7) tell us the necessary extensions to go from the previous measures, T', to weighted averages, T, that includethe social security tax; namely,5
T

T +

)+

Q2

Ss -

(8)

where Sf, Se, and sS are now the social security contributionrates for persons with earnings below the taxable ceiling;6 Q1 is the ratio to aggregate adjusted gross income of the wage and salary income of workerswith earningsbelow the ceiling; Q2 is the corresponding ratio for self-employed persons; and "'is the (weighted)average marginal tax rate for workers with earningsbelow the ceiling. II. Computations Tax Rates of Table 1 shows the salaries and wages (col. 1) and self-employment income (col. 3) that accrue in each year to persons with earningsbelow the ceiling. (In col. 4 the table shows the dollarvalue of the ceiling for each year.) These data, combined with values of aggregateadjusted gross income, allow us to calculate the weights Q1 and Q2, which appearin equation (8). These weights are in columns 5-6 of table 1. For subsequent purposes the importantvariable is Q1, the ratio to adjustedgross income of the salaries and wages of persons below the ceiling. This ratio can be divided into two parts-first, the ratio of salaries and wages of persons below the ceiling to the aggregate of salariesand wages (col. 2 of table 1) and, second, the ratio of aggregate salariesand wages to aggregateadjustedgross income. The latterratio is highly stable about its mean value of .84. Hence Q1fluctuatesmainly because of changes in the fraction of overall salaries and wages that accrue to persons below the ceiling. This fraction depends in turn on the ceiling earningsfor social security in relationto the distributionof nominal earnings in the economy. For example, the decrease in fQ1 from .46 in 1937to .24 in 1965correspondsto a decline in the ratio of salaries and wages for persons below the ceiling to total salaries and wages from .57 to .29. This behaviorreflects the relatively slow rise in the dollar ceiling on earnings, which increased from $3,000 in 1937to only $4,800 in 1965. However, the ceiling has advanced rapidlysince 1965, reaching $35,700 in 1983. Correspondingly,the ratio of salaries and wages for persons below the ceiling to total salaries and wages went from .29 in 1965to .68 in 1983. This change led to an increase in Q1from .24 in 1965 to .57 in 1983.
5. To get the last termwe approximate T'/(1 + Sf) T'(1 - Sf) in (6). This approximation is satisfactoryfor our data sample. 6. Note that the social security levy is a flat-ratetax in this range.

Average Marginal Tax Rates

559

The values for Sf = Se and s, for each year also appear in table 1. (These values are nonzero only since the start of the social security programin 1937.) Using these numbers we can calculate the second term, QI(sf + Se)I/I + Sf, and the thirdterm, Q2 ss, on the rightside of equation (8). The results appear in columns 2-3 of table 2. It is more complicated to calculate the final term of equation (8), which depends on the average marginaltax rate T"for workers with earningsbelow the ceiling. From the IRS's Statistics of Income, Individual Tax Returns for each year, we approximated "'by using the marginaltax rates and associated values of adjustedgross income for the followingfilingunits. First, we take all returnsfrom income classes for which the average of salaries and wages per return is below the ceiling value. (For example, for 1980, when the ceiling on earningsis $25,900, we go up to an adjustedgross income per returnof $30,000.) We then include enough additionaljoint returnsfrom income classes where the average of salariesand wages per returnis above the ceiling so as to exhaust the known total of salaries and wages that accrues to persons with earnings below the ceiling. However, we carry out this calculation by using the lowest possible income classes; that is, we assume that low numbers for individuals' salaries and wages correspond to low numbersfor adjustedgross income per return. There is some approximation here since some of the low values for salariesand wages may come from either multiearner families or families with high nonlaborincome, which would have high marginaltax rates. But some experimentationindicates that the potential error is quantitativelyunimportant.Column4 of table 2 shows the resultingcalculationfor the finalterm, - QlsfT"I,in equation(8). Note that this term, which reflects the exclusion of firms' social security paymentsfrom workers'taxable income, is always below .01 in magnitude. Our previous estimates of the average marginal tax rate when weighted by adjustedgross income, T', appearin column 1 of table 2. With the availabilityof more recent data we can now extend the series from 1980to 1983.For 1981,where the Reagantax cut appliedonly to a small extent, the effects of bracket creep actually raised the average marginaltax rate, T', from 30.4% in 1980 to 31.3%in 1981. But then there was a substantialdrop to 29.3%in 1982and 27.2%in 1983. The decline in the average marginaltax rate by 4.1 percentagepoints from 1981to 1983was much largerthan that (2.6 percentagepoints) for the tax Kennedy-Johnson cut in 1964.When later data are available,it will be interestingto see the extent to which the averagemarginaltax rates declined furtherin 1983 and 1984. The overall modificationsto incorporatethe social securitytax-the sum of columns 2-4 in table 2-appear in column5 of the table (labeled SS). Then the sum of columns 1 and 5 is the averagemarginaltax rate, T, from the federal individualincome tax and the social security tax.

560

Journal of Business

c)

c)

c)

c)cc

ooo

ooooo

ooooo

oNeee

ii8

-~~~~~~~~~~

00

00

00

oII

o ocici'

oomm ooo

O-

I00004 . .

. e0. nI . nr-enO I

000W)N . .> N . .> .> 0.> N e 1cN

) 00000 .b N n .b .b ) ONe .b ee

em

ff1N

El

I C) . I C)C

0 I

I I .

00 'I000

0 0.

0.00 . 0000 0

0)00

0O 00

00000-r-

0 C) )

eN C

ooo

ooooo

ooooo

ooooo

.=cE8 .
C

88
^^

80
^^^^^

x???
^^

ooooo~~~~~~~~~~~~~~~~~

????

rA Q

Ctt

'IC 'IC

C-0 0 \0 00OC

Average Marginal Tax Rates

561

14 vi W W; 116 116 116 r-:06 r: r-:r-:r- 06 06 06 C C 116 116 r

ON 0

0 0

e,; el; el; el; el; el;

00

00 N

4 vi vi vi vi vi vi V ,6 ,6 ,6 ,6 ,6 ,6

00 oo

oo oo oo 0

r. 00 0 u ml

04

. . . . .

. . . . .

. . . . .

. . . . .

mmmen

. . . .

ci

4.4

NC)OOr-W) enenNNN
4! Zs

CQ
0

cj

ot ot ot ot ot 4 4 4 4 4

ot c 4

"C 00 00

00 00 o 00 N

W) r-

C)

C)C)C)r6 r- 1*
0 C Cd C.)

C:0 0 C.)
ON

4.;-4

C-W)-r-

Noo-

ON N-W)MOO

-CCW)-

(ONNNN

00

0 C.)

C4 0

C.)

.C.)
r- W) en N 0 . . N "C (ON (ON N . . . . . "C (ON 0 . . . . . en . . . en en . . W) r- 00 00 . . . . Ei It:$

0 0 0 = 0 sa,
4.4 0 0

o C:
4-;4 0

0 0

-5 0 'a, Ei
'S
C:

o
>
4.4 0

C-) 0 C: 0

--

o oo

C C3

W) (ON ONON "C 0 ON- 00 (ON V C C; C; ,6 06 4 4 6 r- 06 -4 06 C-1C-1 N C-1N en r-

00 m N 00 06 C cn 00 (ON C)

0 " Ei I
00 04 0

Cd u , ;R 0 C.) 0

=+1;n 0

0 4.0 gu

0
-,
04 C.)

r.
0 C.)

00 (ON C)
ON ON ON ON ON
,.-I .-I .-I .-I .-I

CA en
ON
-.1 -.1 -.1 -.1 -.1

"C r- 00 (ON C) -.1 -.1 -.1 .-I .-I

ON ON ON ON ON
.-I .-I .-I -.1 -.1

ON ON ON ON ON
-.1

(ON(ON(ON (ON(ON(ON(ON
.-I .-I .-I

00 00 N 00 00 en

u ::) 0 0

J-Q
(/) ,
40.

ci 0
0

C)

562 TABLE 2 Average Marginal Tax Rates

i

Journal of Business

Sf + Se

I+
Years
1916

Sf

2*S

-f

(1)
.012

(2)
...

(3)
...

* SfST (4)
...

(5)
...

(6)
.012

1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964

.037
.054 .052 .046 .042 .046 .033 .035 .030 .028 .032 .041 .035 .023 .017 .029 .031 .034 .038 .052 .046 .034 .038 .056 .113 .192 .209 .252 .257 .226 .226 .180 .175 .196 .231 .251 .249 .222 .228 .232 .232 .229 .236 .234 .240 .244 .247 .221

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... .009 .009 .009 .010 .010 .009 .007 .007 .006 .007 .006 .006 .006 .008 .010 .009 .008 .010 .012 .012 .013 .013 .016 .018 .017 .017 .019 .018

... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... 0 0 0 0 0 0 0 0 0 0 0 0 0 0 .000 .000 .000 .001 .001 .001 .001 .001 .001 .001 .001 .001 .001 .001

... ... ... ... ... ...

...

... ... ... ... ... ... ... ... ... ... ... ... ...
...

.037
.054 .052 .046 .042 .046 .033 .035 .030 .028 .032 .041 .035 .023 .017 .029 .031 .034 .038 .052 .055 .043 .047 .065 .123 .200 .216 .258 .262 .233 .232 .185 .180 .202 .240 .259 .257 .231 .240 .243 .245 .242 .252 .253 .257 .260 .265 .238

... ... ... ...

...

... ... ... ... ... ... ... ...

-.000

... ...
...

... ... ... .009 .009 .009 .009 .009 .008 .007 .006 .006 .007 .006 .006 .005 .007 .009 .008 .008 .010 .012 .012 .013 .013 .016 .018 .017 .017 .018 .017

-.000 -.000 -.000 - .000 - .001 -.001 -.001 -.001 -.000 -.000 - .000 - .000 .000 .001 .001 .001 .001 .001 .001 .001 .001 .001 .002 .002 .002 .002 .001

Average Marginal Tax Rates TABLE 2 (Continued)

S, + Se

563

'I1

Years 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983

(1) .212 .217 .223 .252 .261 .243 .239 .242 .250 .257 .263 .273 .281 .310 .289 .304 .313 .293 .272

(2) .017 .028 .028 .032 .032 .031 .031 .034 .044 .050 .050 .050 .050 .052 .061 .062 .070 .071 .072

fl , S, (3) .001 .001 .001 .001 .001 .001 .001 .001 .002 .002 .002 .002 .002 .002 .003 .002 .003 .003 .003

-f11s

1 T"

(4) .001 .002 .002 .003 .003 .003 .003 .003 .004 .004 .005 .005 .005 .006 .007 .008 .010 .008 .008

SS (5) .016 .028 .027 .031 .031 .029 .029 .032 .041 .048 .047 .046 .047 .047 .057 .057 .063 .066 .067

(6) .229 .245 .250 .283 .292 .272 .268 .274 .291 .305 .310 .319 .328 .357 .346 .362 .376 .359 .339

grossincome,fromBarro incometax rate,weightedby adjusted NOTE.-T' is the averagemarginal and andSahasakul (1983,table2, col. 1). Valuesfor 1981-82are estimatesbasedon Thompson Hicks (1983)and Holik (1985);the value for 1983is estimatedfrom U.S. InternalRevenueService (1985, = col. 2 + col. 3 + col. 4; col. 6: T table3.4). Cols. 2-4: calculated with datafromtable 1; col. 5: SS = col. 1 + col. 5.

These values are in column 6 of the table. Figure 1 shows the average marginaltax rate from the individualincome tax, T' (col. 1 of table 2), the overall effect from social security, SS (col. 5), and the combined average marginaltax rate, v (col. 6). Considerthe overall effects from the inclusion of social security, as shown in column 5 of table 2 and in figure 1. The social security term, SS, is in the neighborhoodof 1%from 1937until 1958, reaches 2% in 1960,3%in 1966,4%in 1973,5%in 1974,6%in 1979,and almost7%in 1982. Thus the inclusion of this term produces a combined average marginaltax rate, T, that rises more steeply than does the income tax rate, T', especially since 1965. Instead of rising from 21% in 1965 to 31%in 1981and 27%in 1983, we find that the T goes from 23%in 1965 to 38%in 1981and 34%in 1983. The overall effect from social security on the average marginaltax rate is always much less than the rate of employees below the ceiling, + Sf). Primarily,this difference arises because Q1-the (Sf + se)I(l ratioof salariesand wages below the ceilingto aggregateadjustedgross income-is much less than unity. As mentionedbefore, the variations in Ql derive mainly from changes in the ratio of salaries and wages below the ceiling to total salaries and wages, which appearsin column

564
Marginal 0.40Tax Rate

Journal of Business

0.

35

300

1\
~ ~ ~ ~ ~ ~
'

.2 5

9
0.20]t

0.15

0.10-

0.05

1916

1920

1925

1930

1935

1940

1945

1950

1955

1960

1965

1970

1975

1980

1983

FIG. 1.-Average marginaltax rates

2 of table 1. For example, in 1965only 29%of total salariesand wages accruedto persons below the ceiling. If there had been no ceiling (and, unrealistically,if the rate of tax, Sf = Se, were unchanged),then the overall effect of SS would have increased by a factor of 3.5, from .016 to .056. On the other hand, the rapid increase of the ceiling in recent years has made this effect less important.In 1983, where 68%of total salariesand wages accruedto those below the ceiling, a removalof the ceiling (with contributionrates held fixed) would have raisedthe effect from SS by a factor of 1.5, from .067 to .100. Table 3 compares the social security tax with the federal individual income tax for selected years. Notice that the ratio of revenues raised by social security to that from the income tax (shown in col. 5) rises from .07 in 1945to .66 in 1983. Column 6 of the table shows a crude measure of the relative "efficiencies" of the two types of taxes. This measureis the revenue

Average Marginal Tax Rates

565

>1

-0

vNF<

r.ou C
a, cq -W F -W

x~~~~~
D

z4 x NN)N to

0{ dt c

8,

4a

a0 0
o P4-^

0
116

u
X

1~~~~~~~~0 19
t

Ct

o o o oo

\r}

Cb xC t-

o o

oo >L o \ o

566

Journal of Business

raised from social security divided by the contributionof this levy to the overall average marginaltax rate,7 expressed as a ratio to the correspondingfigurefor the income tax. On this basis the social security tax looks strikinglymore efficient. Specifically, in 1983the social security levy generates 2.5 times as much revenue per unit of average marginaltax rate as does the income tax, whereas in 1965 the corresponding number was 4.3. The main reason for the decline in this numbersince 1965is the sharprise in the ceiling on earnings,which has a positive effect on the average marginaltax rate from social security, relative to the revenue generated. The social security levy turns out to be relatively "efficient" because it combines two features of a tax-rate schedule that have been stressed in the literatureon optimaltaxation. First, it is a flat-ratelevy (on labor earnings and income from self-employment)in the range where the tax rate is positive. The shift to a flat-rateincome tax has been proposedby, amongothers, Friedman(1962, ch. 10)and Hall and Rabushka(1983). (Surprisingly,these authorsdo not seem to mention that, in the social security tax, we alreadyhave a close approximation to the flat-rateincome tax.) In comparisonwith a graduated-rate system, the flat-rate levy generates the same amount of revenues at a lower average marginaltax rate. Second, as advocated on theoretical groundsby Mirrlees(1971), the social security tax has a zero marginal rate at the top. However, as noted before, the rapid increase of the ceiling in recent years has made this featureless important than it used to be.
References
Barro,R. J., and Sahasakul,C. 1983.Measuring averagemarginal rate from the the tax individualincome tax. Journal of Business 56 (October):419-52. Friedman,M. 1962. Capitalism and Freedom. Chicago:Universityof ChicagoPress. Gordon,R. H. 1982.Social securityand laborsupplyincentives. NBER Working Paper no. 986. Cambridge, Mass.: NBER, September. Hall, R. E., and Rabushka, A. 1983. Low Tax, Simple Tax, Flat Tax. New York: McGraw-Hill. Holik, D. 1985. Individualincome tax rates, 1982. Statistics of Income Bulletin 4 (Spring):1-11. Mirrlees,J. A. 1971.An explorationin the theoryof optimum incometaxation.Review of Economic Studies 38 (April):175-208. Thompson, R., and Hicks, C. 1983. Average and marginaltax rates, 1981 individual income tax returns.Statistics of Income Bulletin 3 (Fall):41-49. U.S. Internal RevenueService. 1985.Individual Income Tax Returns, 1983. Washington, D.C.: U.S. GovernmentPrintingOffice. 7. It is unclearhow to allocatethe cross-term, - f11*Sf * T (col. 4 of table2), between the two levies, althoughthis term is quantitativelyunimportant. The figuresshown in table 3 allocate half of this term to each type of tax.