Oxford University Press The Quarterly Journal of Economics

Optimal Redistributive Taxation When Individual Welfare Depends Upon Relative Income
Author(s): Michael J. Boskin and Eytan Sheshinski

Source: The Quarterly Journal of Economics, Vol. 92, No. 4 (Nov., 1978), pp. 589-601
Published by: Oxford University Press
Stable URL: https://www.jstor.org/stable/1883177
Accessed: 20-07-2019 06:32 UTC
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.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access
to The Quarterly Journal of Economics
This content downloaded from 111.68.111.140 on Sat, 20 Jul 2019 06:32:06 UTC
All use subject to https://about.jstor.org/terms
OPTIMAL REDISTRIBUTIVE TAXATION WHEN
INDIVIDUAL WELFARE DEPENDS UPON RELATIVE
INCOME*
MICHAEL J. BOSKIN AND EYTAN SHESHINSKI
I. Introduction, 589.-II. An optimal negative income tax model, 591.-III. The

maximin criterion, 594.-IV. A utilitarian social objective, 597.-V. Conclusion,
598.
Our theory ... depends upon the validity of a single hypothesis, viz.: that the utility
index is a function of relative rather than absolute consumption expenditure.
-J. Duesenberry
Income, Saving and the Theory of Consumer Behavior
I. INTRODUCTION
The problem of the "appropriate" role of government in redis-

tributing income has a long and interesting history in economics.
Dating back in its modern form to Edgeworth, the original argument,
based on addition of the welfare of all citizens, diminishing marginal
utility of income, and no tax disincentive effects was for complete
equality, that is, a tax-transfer system that equalized incomes.
More recently, the rigorous examination of this problem by
Mirrlees (1971), Atkinson (1973), Fair (1971), Feldstein (1973),
Sheshinski (1971), and Stern (1976), focusing on the tax-induced
disincentive to supply labor, has concluded that optimal tax rates are
quite modest and that rather little public income redistribution is
socially desirable. While the results are fairly sensitive to assumptions
about labor supply elasticities1 and somewhat sensitive to the form
of the social welfare function, they stand in striking contrast to both
the Edgeworth analysis and the policies and policy proposals of many
advanced economies.
One aspect of the debate over income distribution that has not
been analyzed in the optimal income tax framework is the popular
* We would like to thank R. Dorfman, C. Chamley, R. Musgrave, L. Lau, and a

referee for valuable advice, and Louis Garrison for research assistance. This research
was supported by a contract with NBER from the Office of Tax Analysis, U. S. Treasury
Department (No. TO-76-13; OS-612).
? 1978 by the President and Fellows of Harvard College. Published by John Wiley & Sons, I
The Quarterly Journal of Economics, November 1978 0033-5533/78/0092-0589$01.00
590 QUARTERLY JOURNAL OF ECONOMICS
notion that welfare depends at least in part upon relative income or

consumption. We are familiar with the notion that poverty may be
a relative, as well as an absolute, income phenomenon. For example,
Galbraith (1958) argues that "people are poverty stricken when their
income ... falls markedly behind that of the community." Sen (1966)
has examined related issues. The relative income hypothesis has rarely
been analyzed or empirically tested in economics. Probably the most
important exception is Duesenberry (1949). A recent example is
Easterlin (1975). The purpose of the present paper is to explore the
structure of optimal income taxation-redistribution in an economy
where the welfare of individuals depends in part on relative after-tax
consumption; that is, we specify individual welfare as a function of
absolute and relative after-tax consumption, with diminishing mar-
ginal utility to each. With such a specification, of course, an additional
incentive for income redistribution from wealthy to poor citizens is
created, and the logical impossibility of increasing tax rates to the
point where disincentive effects actually reduce tax revenues is po-
tentially removed.2 The analysis highlights the importance of the
marginal valuation placed on upward social mobility in various ranges
of the income distribution and its interaction with the elasticity of
the marginal utility of consumption; of course, "labor supply" elas-
ticities, the form of the social welfare function, and the skill distri-
bution continue to play an important role.
In Section II we introduce the simplest model of optimal income
redistribution: linear redistribution plans (credits plus flat rate taxes,
i.e., archetypical negative income tax schemes) combined with in-
vestment in human capital (a proxy for all forms of increasing or de-
creasing income).3 We derive analytically the structure of optimal
linear negative income taxes for the case of individual welfare de-
pending upon both the absolute and relative levels of consumption.
We demonstrate that increased concern for relative consumption
levels leads to higher income guarantees and marginal tax rates.
In Section III we examine the extreme "maximin" case (in which
the government attempts to maximize the welfare of the worst-off
individual). Assuming for analytical convenience a Pareto distribution
of skills and logarithmically separable utility, we derive explicit for-
mulas for the optimal marginal tax rate and income guarantee. Of
course, these depend upon the usual parameters: the skill distribution,
the elasticity of the marginal cost of human investment, and the
elasticities of the private and social marginal valuation of absolute
and relative consumption. A numerical example reveals that some
OPTIMAL REDISTRIBUTIVE TAXATION 591
concern for relative status adds additional progre

policy, even substantial progression when concern f
is paramount.
In Section IV we report the results of some numerical calculations
of optimal marginal tax rates and income guarantees for a utilitarian
social objective. We examine how these policy parameters respond
to variations in concern for relative position as well as the other pa-
rameters of the problem. The optimal tax rate and relative income
guarantee in the utilitarian case turn out to be enormously sensitive
to concern for relative position. The optimal redistribution schemes
range from quite modest to substantial as concern for relative status
moves from nonexistent to virtually exclusive.
Thus, the often-quoted policy prescription, derived from optimal
income tax models, that optimal tax rates and income guarantees
should be quite modest, is very sensitive to individuals' concern for
relative income. The optimal rates reported below when individuals
are mostly concerned with relative, as opposed to absolute, income
are much higher than those derived in previous studies in the optimal
income tax tradition (see, for example, Fair, 1971; Mirrlees, 1971; and
Stern, 1976).
II. AN OPTIMAL NEGATIVE INCOME TAX MODEL
There is a variety of ways to parameterize the potential disin-

centive effects of income taxation (see, for example, Sandmo's, 1976
survey). For illustrative purposes, we consider the educational in-
vestment model suggested by Sheshinski (1971).
There is one consumption good, denoted by c, and an educationa
input (say, school years), denoted by x. Individuals are distinguished
by their ability to produce income from a given educational input. Let
this ability be indexed by n, (n > 0). Income, denoted y, is assumed
to depend on n and x: y = y(n,x). For simplicity, this relation is as-
sumed to have a multiplicative form: y = nx. The cost of educationa
investment, in terms of the consumption good, is denoted g(x). We
assume that g'(x) > 0, g" (x) > 0. In the absence of income taxation,
the individual's consumption is thus given by c = y - g(x).
Suppose that the government decides to redistribute income via
a linear negative income tax t(y),
(1) t(y) = -a + (1 - O)y)
where a is the income guarantee, or credit, and (1

tax rate.
Note that this tax rule, in accord with accepted practices, does
not allow a deduction for educational expenses. Accordingly, after-tax
income y - t(y) is equal to a + fy, and consumption is given by4
(2) c=a+ 0y-g(x).

Of course, thinking of c and y as lifetime variables, or their an-
nualized equivalents, begs several important problems. For example,
a minimal consumption is necessary for survival, and individuals may
not be indifferent to the same present value of income, or consump-
tion, if there is year-to-year fluctuation as opposed to constancy or
smooth, predictable growth. Indeed, many practical tax provisions
are designed to deal with these problems (e.g., income averaging, loss
carry forwards, etc.). A well-functioning capital market, of course,
mitigates some of these problems, but it is difficult to believe that
individuals can borrow as easily with potential human, as opposed
to actual nonhuman, capital as collateral.
Since our purpose here is to compare and contrast our results on
concern for relative position to an already existing literature that ig-
nores the point discussed above, we shall also do so. The reader should
keep in mind that our results should be compared to the analytical
results of others, and are not designed to capture fully the intricacies
of lifetime consumption and human investment decisions.
Each individual is assumed to maximize c with respect to x.5 The
first-order condition is
(3) g'(x)= fin, or x = x(On),

where x = g'1. We assume that the solution to (3) exists for all n.
Accordingly, the optimum before-tax income is also a function of n
and the tax rate, 9(n,) = 9(nx (fn)); optimum consumption is given
byc= a+ 9-g(x).
The individual's utility u is assumed to depend on two variables:
own consumption c, and the average consumption in the population,
denoted c:
(4) u = u(cC; 0),
where 0 is a parameter, 0 > 0,

individual's "relative position." In particular, we assume that when
0 = ., bu//c = u1 > 0, du/&c- = U2 =0, and (j2u/?/c(Ci = = = 0. For
any 0 > 0, we assume that u1 > 0, u2 < 0 and

and twice differentiable. As a special case we may have u(c,J; 0) =
v1(c) + Ov2(c/I), where vi, i = 1,2 are increasing, concave functions.
The variable c/c represents, of course, the individual's relative position
in terms of consumption. Clearly, 0 = 0 is the standard case when the
individual's utility is independent of other individuals' consump-
tion.
Let f (n) be a density function denoting the relative number of
individuals with ability n: f(n) > 0 and ff(n)dn = 1. In choosing the
optimum tax parameters (aAf, the government is assumed to follow
a utilitarian principle. Thus, its objective is to maximize the social
welfare function W,
(5) W = fu(cc; 0)f(n)dn
subject to a balanced budget constr
(6) a -(1- - y=0,
where 5 = 9f(n)dn is average income. Accordingly, average con-

sumption is given by
(7) c = a + b7 - fg(x)f(n)dn.
The first-order conditions for an interior ma

to (6) and (7) are
(8) f [u1(VT; 0) + u2(0, ; 0) - ji]f(n)dn = 0
(9) f[U1 (OX7; 0)9 + U2(',Z; 0)Y - A j)y]f(n)dn = 0,
where 1i > 0 is the shadow price (or Lagrangean coefficient) o

creased income redistribution, ui, (i = 1,2) denote partial deriva
and X = X(x) = g"(x)x/g'(x) is the elasticity of the marginal cost of
education. We assume that there exists a unique solution to these
equations, denoted by (a*,4*).
In order to find the effect of "concern for relative status" on the
optimum tax scheme, we differentiate (8), (9), and (6) totally with
respect to 0, evaluating these changes at 0 = 0. Assuming for simplicity
that X is constant (i.e., g(x) = x1+X/(1 + X), X > 0 a scalar), we
have
(10) fSuiif(n)dn f u119f(n)dn -1 ra
f u11J(n)dn f ul u 2f(n)dn-- - 1-
ly 0~2X k y /
I 0- at. by 0
=- f u2f(n
- f ufff(n
Denote the matrix on the lef

second-order conditions for max
(10) for 63*/60, we obtain
(1) aA fA*X
A *X O- J ut
Thus, as expected, concern for relative position, as well as ab-
solute position, always leads to an increase in the optimum marginal
tax rate compared with the case of sole concern for absolute
position.
It can be shown that ?30*/WO < 0 implies, at the optimum, that
6a*/19 > 0 and vice versa. Hence, increased concern for relative po-
sition will lead both to higher tax rates and to higher credits, or income
guarantees.
III. THE MAXIMIN CRITERION
As an alternative to the utilitarian welfare function, consider the

Rawlsian objective criterion (termed maximin), according to which
the government is concerned solely with the welfare of the worst-
off individual. Denote this individual's gross income (correspond-
ing to the lowest ability, say, n') by y', and his consumption by c':
c + _x (Any))
The government's objective is to maximize
(12) u' = u(c'7; 0)
with respect to a and /3, subject
guarantee and the marginal tax rate to maximize the welfare of the
worst-off individual, given the disincentive effects of the tax. The
first-order conditions now become
(13) ul+U2- =?
and
(14) uly'+ u2Y - A 1 - y- = 0,
where ul and u2 are evaluated at c = c'. One ma

solving for f-*t, as
(15) j 1+3*X1-y-
where X = u1/(u1 + u2), in > 1. Clear

0 = 0, and 7 increases with 0, which parameterizes the "concern for
relative position." Consequently, for a given ratio y'/y, (0 ? y'/y- <
1), as 0 increases from zero, d* will decrease; that is, the optimum tax
rate increases. Alternatively, for a given 0, as y'/y increases, f* de-
creases. These results should be expected: when the ratio of lowest
to average before-tax income increases, the optimum tax rate de-
creases. This reasoning, however, is not completely rigorous, since the
ratio y'/51 as well as aq, depends on the chosen 0*. Hence, equation (15)
cannot be directly used to calculate the latter. In special cases, how-
ever, when f(n) is further specified, an exact solution for d* is possible.
We shall now present such an example.
Let f (n) be the Pareto distribution:
(16) [(n) = bn'bn-6-1, n > n',
where 6 > 1 and n' > 0 are parameters, the latter denoting the lowest
ability level.
Suppose that the utility function is logarithmic:
(17) u(c'7; 0) = log c' + 0 log (c'Ac).
We continue to assume that g(x) = x1+X/(1 + X), where X > 0 is

the elasticity of the marginal cost of education.
Using the individual's first-order conditions, we find that
y 1+X
(18) 1-
v b
For Y to be finite, we assume that 1 + X < X6.

gives
(19) il (1 + O)J
U1 + U2 (1 + OF -OcO
Using (3), (6), (7), and (16), we find
(20) C' = il-/3\ ______ ____

r( IS -1- ) 1 + ]
and
(21) = [1 + A ] (X6 --A) (3n')(1+X)/X.
Substituting (20) and (21) into (19), we may write (15) as
(22)
13*=iP(f3*,O)_ 3(1+X-I+*)+0I3*(1+X)
Equation (22) is an implicit function for the optimum solution

13*
Two polar cases can be immediately identified. First, when 0 =

0, then 41(3, 0) is independent of 3 (d4A/d/3 = 0). Hence
(23) /3* = 42(0,0) = 1

1 + (1 + X)/b
Notice that the optimum marginal tax rate, as expected, is pos-

itively correlated with the elasticity of marginal cost X, and with the
dispersion of ability, represented by (the inverse of) the parameter
Second, as 0 a co, 41(, 0) - A/(1 + X). Hence, the unique solution

for d* = 42(1*, ao) is A* = 0. When individuals become exclusively
concerned with relative consumption, the optimum marginal tax rate
is 100 percent!
Intermediate cases, with 0 > 0, can be easily analyzed. Equation
(22) is a quadratic equation in A, which has a unique solution in the
range 0 < d < 1.8
The effect of an increase in 0 on the optimum 1* may be found
by differentiating VI with respect to 0, evaluating this effect at 2 =
13:
TABLE I
OPTIMAL MARGINAL TAX RATE AND RELATIVE INCOME GUARANTEE FOR

ALTERNATIVE PARAMETER VALUES (MAXIMIN OBJECTIVE)
0
6 A 0.0 1.0 5.0
2.5: 0.75 0.41 0.53 0.71

1.0 0.44 0.57 0.75
1.25 0.47 0.60 0.79
3.0: 0.75 0.37 0.48 0.68
1.0 0.40 0.52 0.72
1.25 0.43 0.56 0.76
(24) Xi (1 + A)o < 0.
As expected, the optimum marginal tax rate increases monotonically

with concern for relative position.9
For a range of reasonable values of X and 6, Table I presents some
sample calculations of the effect of changes in 0 on the optimum tax
rates.
Note also that
(25)
the "relative income guarantee"'equals the marginal tax rate. From

Table I we note that this extreme egalitarianism must be combined
with enormous concern by the worst-off individual over relative status
(much more than over own absolute consumption) to lead to a sub-
stantial income redistribution along the lines suggested by the
Edgeworth analysis.
We turn now to a consideration of whether concern for relative
position also becomes important when the government follows a
somewhat less egalitarian social objective.
IV. A UTILITARIAN SOCIAL OBJECTIVE
Returning to the utilitarian social objective discussed in Section

II, we may solve the system of equations (8) and (9) explicitly for th
case of logarithmically separable utility and a Pareto skill distribution.
This yields the following equations:
TABLE II
OPTIMAL MARGINAL TAX RATES AND RELATIVE INCOME GUARANTEES,

ALTERNATIVE PARAMETER VALUES (UTILITARIAN CASE)
0
0.0 1.0 5.0
6 = 2.5
X = 0.75 0.37 0.48 0.67
= 1.0 0.35 0.46 0.64
= 1.25 0.34 0.46 0.64
a = 3.0
X = 0.6 0.29 0.39 0.57
= 0.75 0.27 0.37 0.55
= 1.0 0.27 0.37 0.55
= 1.25 0.27 0.37 0.55
(26) (1 + 0) fIf(n)dn - -= ; c c
(27) (1 + 0) ,fWf(n)dn - 0 = t - 1- f)y.
Substituting from the individuals' first-order conditions for x, we

derive the expressions for y, c, ye, c, and a. This leaves us with two
nonlinear equations in the two unknowns, At and F3. We have solved
these equations numerically for certain values of the parameters 6,
X, and 0. The results are reported in Table II. Several characteristics
of the results are worth noting.
First, more than in the maximin case, the optimal marginal tax
rates and relative income guarantees are extremely sensitive to vari-
ations in 0, our parameter reflecting concern for relative versus ab-
solute consumption. The optimal marginal tax rates and income
guarantees range from a quite modest 27 percent to a very substantial
67 percent. For a given X and 6, the increase in 0 from zero (no concern
for relative position) to five (predominant concern for relative posi-
tion) approximately doubles the optimal marginal tax rate and income
guarantee. It is thus obvious that a strong relative income effect may
overcome somewhat the tendency of disincentive effects to hold down
marginal tax rates and income guarantees.10
Second, as in the maximin case, the results are only slightly af-
fected by modest variations in X, the (constant) elasticity of marginal
cost of human investment, and only modestly affected by variations

in 6, the skill dispersion parameter.1
The lesson to be learned from this exercise is quite important:
most attacks on the results of the optimal income tax literature have
suggested that a utilitarian social objective is "not egalitarian enough."
Our results indicate that such philosophical concerns are not necessary
to the defense of substantial income redistribution. Extreme concern
for relative status can yield substantial optimal redistributive taxation
with either a utilitarian or a maximin social objective. We merely add
the proviso that the perhaps intuitively appealing idea that people
care about their relative income has little scientific support at this
time. More research on the matter is urgently needed.
V. CONCLUSION
In considering the conjecture that individual welfare depends

in part upon relative consumption, we have explored some of the
implications for public policies that redistribute income via pro-
gressive income taxation. Briefly, these implications include the fol-
lowing:
1. The class of linear negative income taxes exhibits increasing
marginal tax rates and relative income guarantees as concern for
relative position increases.
2. The maximin criterion yields relatively modest redistribution
schemes that are made only slightly more progressive with modest
concern for relative position but extremely progressive with enormous
concern for relative position (recall the specific utility function and
skill distribution).
3. Strong concern for relative consumption may also substantially
increase the optimal marginal tax rate and relative income guarantee
with a utilitarian objective.
We conclude with a word of warning, lest we be misconstrued.

Our purpose here is to explore the potential implications of individual
concern for relative consumption on the design of optimal redistri-
butive taxation schemes. While it is clear that such concern can be
extremely important, it seems to be so in cases where concern for
relative consumption is extremely strong. Too often, such a concern
is considered "obvious." Evidence that this is indeed the case is vir-
tually nonexistent, let alone convincing. We hope that by demon-
strating the potential policy relevance of empirical information on
the "relative consumption effect," we shall encourage much additional
empirical research on the subject by economists and other social

scientists.
STANFORD UNIVERSITY AND NBER

STANFORD UNIVERSITY AND HEBREW UNIVERSITY
NOTES
1. Note the high tax rates obtained by Stern (1976) in the case of an extremely
inelastic labor supply.
2. Thus, Pigou (1962, p. 91) notes that the satisfaction which a man derives
from the possession of a given income depends not only on the absolute amount of in-
come, but also on the relation subsisting between it and the incomes of other people."
He goes on to conjecture that this may justify increased progression.
3. See Boskin (1976) for a discussion of the tax treatment of human invest-
ment.
4. We think of c and y as lifetime consumption and income, respectively.
5. If we also include the tax disincentives to hours of work and saving (see Boskin,
1978), the results reported below would be strengthened.
6. Following the usual study of this type, we focus on purely redistributive taxa-
tion. See Mirrlees (1971) and Atkinson (1973) for discussions of budgetary scale ef-
fects.
7. The maximin case has been analyzed by Phelps (1973); also see Atkinson (1973),
and Sheshinski (1972).
8. Write (22) as af32 + bf + c = 0. One can verify that, for any 0 > 0, -b/2a > 1,
using the assumption that 1 + X < XM. There is therefore only one solution in the interval
0 _ f 1.
9. In our models consumers maximize utility by maximizing lifetime consump-
tion even if they care exclusively about relative consumption. The original point of
Duesenberry's (1949) relative income hypothesis was to derive implications for con-
sumer behavior. Our model highlights the potential importance of concern for relative
income in the simpler case of no effect on consumer behavior. If, for example, there
is a substantial negative (positive) relative income effect on labor supply, it would
weaken (strengthen) the conclusions reported here.
If individuals are concerned with their position relative to parameters of the overall
distribution given parametrically and identically, there would be no observable market
implications of the relative income hypothesis across consumers. Market data could
not disentangle absolute from relative income effects.
10. Recall the provisos in note 5.
11. See the interesting comparison of alternative social welfare functions and the
comparison of the resulting optimal tax schedules to actual taxation in Cooter and
Helpman (1974).
REFERENCES
Atkinson, A. B., "How Progressive Should Income Tax Be?" in M

in Modern Economics (London: Longmans, 1973).
Boskin, M. J., "Notes on the Tax Treatment of Human Capital," Conference on Tax
Research, U.S. Department of the Treasury, 1976.
--" "Taxation, Saving, and the Rate of Interest," Journal of Political Economy,
LXXXVI, No. 2, Pt. 2 (1978), S3-S27.
Cooter, R., and E. Helpman, "Optimal Income Taxation for Transfer Payments," this
Journal, LXXXVIII (Nov. 1974), 656-70.
Duesenberry, J., Income, Saving and the Theory of Consumer Behavior (Cambridge:
Harvard University Press, 1949).
Easterlin, R., "Does Money Buy Happiness?" Public Interest, XXX (Winter 1973),
3-10.
Fair, R., "The Optimal Distribution of Income," this Journal, LXXXV (Nov. 1971),
551-79.
Feldstein, M., "On the Optimal Progressivity of the Income Tax," Journal of Public
Economics, II (4) (Nov. 1973), 357-76.
Galbraith, J. K., The Affluent Society (Boston: Houghton-Mifflin), 1958.
Lazear, E., "Family Background and Optimal Schooling Decisions," NBER Working
Paper No. 141, July, 1976.
Mirrlees, J., "An Exploration in the Theory of Optimal Income Taxation," Review of
Economic Studies, XXXVIII (1971), 175-208.
Phelps, E., "The Taxation of Wage Income for Economic Justice," this Journal,
LXXXVII (Aug. 1973), 331-54.
Pigou, A. C., A Study in Public Finance, 3rd ed. (London: Macmillan, 1962).
Sandmo, A., "Optimal Taxation-An Introduction to the Literature," Journal of Public
Economics, VI (July-Aug. 1976), 37-54.
Sen, A., "Labour Allocation in a Cooperative Enterprise," Review of Economic Studies,
XXXIII (4), No. 96 (Oct. 1966), 361-71.
Sheshinski, E., "On the Theory of Optimal Income Taxation," H.I.E.R. Discussion
Paper No. 172, February, 1971.
-, "The Optimal Linear Income Tax," Review of Economic Studies, XXXIX (3),
No. 119 (July 1972), 297-302.
Stern, N., "On the Specification of Models of Optimum Income Taxation," Journal
of Public Economics, VI (1976), 123-62.

Oxford University Press The Quarterly Journal of Economics

Uploaded by

Copyright:

Available Formats

You might also like

Oxford University Press The Quarterly Journal of Economics

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Oxford University Press The Quarterly Journal of Economics

Uploaded by

Copyright:

Available Formats

Optimal Redistributive Taxation When Individual Welfare Depends Upon Relative Income

Author(s): Michael J. Boskin and Eytan Sheshinski

MICHAEL J. BOSKIN AND EYTAN SHESHINSKI

I. Introduction, 589.-II. An optimal negative income tax model, 591.-III. The

The problem of the "appropriate" role of government in redis-

* We would like to thank R. Dorfman, C. Chamley, R. Musgrave, L. Lau, and a

notion that welfare depends at least in part upon relative income or

concern for relative status adds additional progre

II. AN OPTIMAL NEGATIVE INCOME TAX MODEL

There is a variety of ways to parameterize the potential disin-

(1) t(y) = -a + (1 - O)y)

where a is the income guarantee, or credit, and (1

(2) c=a+ 0y-g(x).

(3) g'(x)= fin, or x = x(On),

(4) u = u(cC; 0),

where 0 is a parameter, 0 > 0,

any 0 > 0, we assume that u1 > 0, u2 < 0 and

(5) W = fu(cc; 0)f(n)dn

subject to a balanced budget constr

(6) a -(1- - y=0,

where 5 = 9f(n)dn is average income. Accordingly, average con-

The first-order conditions for an interior ma

(8) f [u1(VT; 0) + u2(0, ; 0) - ji]f(n)dn = 0

(9) f[U1 (OX7; 0)9 + U2(',Z; 0)Y - A j)y]f(n)dn = 0,

where 1i > 0 is the shadow price (or Lagrangean coefficient) o

(10) fSuiif(n)dn f u119f(n)dn -1 ra

Denote the matrix on the lef

III. THE MAXIMIN CRITERION

As an alternative to the utilitarian welfare function, consider the

(12) u' = u(c'7; 0)

with respect to a and /3, subject

(14) uly'+ u2Y - A 1 - y- = 0,

where ul and u2 are evaluated at c = c'. One ma

where X = u1/(u1 + u2), in > 1. Clear

(16) [(n) = bn'bn-6-1, n > n',

(17) u(c'7; 0) = log c' + 0 log (c'Ac).

We continue to assume that g(x) = x1+X/(1 + X), where X > 0 is

For Y to be finite, we assume that 1 + X < X6.

Using (3), (6), (7), and (16), we find

(20) C' = il-/3\ ______ ____

(21) = [1 + A ] (X6 --A) (3n')(1+X)/X.

Substituting (20) and (21) into (19), we may write (15) as

Equation (22) is an implicit function for the optimum solution

Two polar cases can be immediately identified. First, when 0 =

(23) /3* = 42(0,0) = 1

Notice that the optimum marginal tax rate, as expected, is pos-

Second, as 0 a co, 41(, 0) - A/(1 + X). Hence, the unique solution

OPTIMAL MARGINAL TAX RATE AND RELATIVE INCOME GUARANTEE FOR

2.5: 0.75 0.41 0.53 0.71

(24) Xi (1 + A)o < 0.

As expected, the optimum marginal tax rate increases monotonically

the "relative income guarantee"'equals the marginal tax rate. From

IV. A UTILITARIAN SOCIAL OBJECTIVE

Returning to the utilitarian social objective discussed in Section

OPTIMAL MARGINAL TAX RATES AND RELATIVE INCOME GUARANTEES,

(27) (1 + 0) ,fWf(n)dn - 0 = t - 1- f)y.

Substituting from the individuals' first-order conditions for x, we

(20) C' = il-/3\ __