The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FOREWORD
On October 24 and 25, 1980, the Center for the Study of
American Business at Washington University and the Federal
Reserve Bank of St. Louis cosponsored their fifth annual
conference, "The Supply-Side Effects of Economic Policy." This
volume contains the papers and comments delivered at that
conference.
Proponents of "supply-side economics" have challenged the
policy recommendations that emerge from "Keynesian"
macroeconometric models. These models focus on the effects of
economic policy on the demand for output. Supply-side economics,
in contrast, emphasizes the response of output to changes in the
supply of inputs. Decisions affecting the capital stock and
employment-in particular, saving and investment decisions and
labor force participation and hours decisions-are the focus of the
supply-siders' attention.
The 1980 conference examined most of the major themes
associated with supply-side economics. The papers in Part I of this
volume develop the theory underlying various supply-side
propositions and present empirical evidence in support of some of
these propositions. In Part II, the effect of taxes on capital
formation and the effect of increased capital formation on output
growth and inflation are examined. The effect of tax and transfer
programs on labor supply, employment and unemployment are
examined in Part III. The final section contains the special
luncheon and dinner presentations.
Leading proponents of supply-side economics develop the
underlying theory and evidence in support of their propositions in
Part I. In "Tax Rates, Factor Employment, and Market
Production," Victor Canto, Douglas Joines, and Arthur Laffer
(CJL) develop a simple, static, one-good, two-factor general
equilibrium model in which taxes on factor incomes drive a wedge
between gross factor payments and net factor incomes. The authors
then derive the response of factor supplies, output, and tax revenue
to changes in tax rates. They demonstrate that the framework is
consistent with the existence of the so-called "Laffer Curve,"
according to which increases in tax rates initially increase
government revenue up to some revenue maximizing tax rate but
decrease tax revenue beyond this point.
CJL note that the theoretical analysis only suggests the possibility
that tax rate reductions may raise tax revenue. Empirical evidence is
required to demonstrate whether or not tax rates are in the
vii
VIII
/ FOREWORD
viii/
FOREWORD
prohibitive range of the Laffer curve. In the second half of their
paper, CJL therefore employ a time series analysis of tax revenues
1962 and 1964
to estimate the effects of the
the Kennedy tax cuts in 1962
1964 on
tax revenues. They conclude that the cumulative
cumulative revenue change
induced by the tax
tax cuts is approximately zero, with an equal chance
that the tax cuts increased revenue as that
that they reduced it.
“An Econometric
In "An
Econometric Model Incorporating the Supply-Side
Policy,” Michael Evans discusses the
Effects of Economic Policy,"
implications of the supply-side macroeconometric
macroeconometric model he recently
developed. According to Evans, stimulating investment is a key to
supply-side policy because it will
will both increase real growth and
moderate inflation. Evans finds that investment would
would be
significantly stimulated by reductions in tax rates, regardless of
whether
whether the tax cuts apply to corporate income, personal income,
or capital gains. He believes that a change in the corporate tax rate
has the most powerful effect on investment, and an increase in the
investment tax credit has the least impact. Evans also examines in
considerable detail the influence of personal tax cuts, cuts
cuts in capital
gains taxation, and a variety of other plans to stimulate saving.
These tax reductions raise the after-tax real rate of return and
increase saving; the increased saving
saving in turn increases demand for
assets, lowering interest rates and stimulating investment.
In the labor market equations, Evans finds important effects of
of
tax rates on both labor supply (participation rates and hours
worked) and on wage gains. The effect of taxes on wage gains is
particularly
particularly important because it permits tax declines to moderate
moderate
inflation.
Curve,” Alan Blinder
"Thoughts on the Laffer Curve,"
Blinder notes that the
In “Thoughts
proposition that the function relating tax rates to tax revenues rises
to a peak and then falls is both an old idea and a noncontroversial
noncontroversial
one. The important issue raised by the CJL paper, according to
Blinder, is whether or not current U.S. tax rates are in the
prohibitive range of the
the Laffer curve, implying that
that a decrease in
tax rates would increase tax revenues. Blinder presents a simple
model and employs alternative values of the critical labor supply
to provide some
and demand elasticities to
some hints as to whether
whether or not
it is plausible that we could be in this prohibitive range. He
“the revenue maximizing tax rate is very likely to be
concludes that "the
so high as to be considered ridiculous for any broad based tax.”
tax."
Steven Braun, who discusses the Evans paper, raises a number of
serious questions about the specifications of the key equations
in the Evans model: the Phillips curve
curve and the labor force
FOREWORD/
FOREWORD / ix
participation, hours, investment, and consumption (saving)
that each of Evans'
equations. Braun concludes that
Evans’ key policy
conclusions
conclusions is derived from an equation which is marred by serious
misspecification.
Albert Ando also discusses the Evans paper and reinforces
Braun's concern about misspecifications in the Evans model. He
Braun’s
on the two equations
focuses on Evans'
Evans’ productivity equation and on
in which the output of the productivity equation plays a role: the
manhours
manhours equation and the equation explaining maximum
production. Ando concludes that the Evans model is dominated by
a pattern of major defects, making it of questionable value as a
the effects of policy changes.
tool for examining the
Parts II and III provide evidence on the effects of economic
policy on investment and labor supply, respectively. In "Tax
“Tax Policy
and Corporate Investment,"
Investment,” Lawrence Summers evaluates various
arguments in support of policy measures to stimulate investment
response of investment
and then presents empirical evidence on the response
to an assortment of tax changes. Summers concludes that policies
to encourage investment will result in only a small increase in the
growth over the next decade, that
that tax policies to
rate of economic growth
to moderate
stimulate investment are unlikely to
moderate inflation, and that
that
fears of insufficient capital accumulation as a source of
of
unemployment are groundless. However, despite his pessimism
about increased economic growth or reduced inflation via tax
policies designed to stimulate investment,
investment, Summers concludes that
that
tax rate reductions may substantially reduce the deadweight
deadweight loss
associated with capital income taxation and substantially improve
economic welfare.
In “Estimates
"Estimates of Investment Functions and Some Implications
Growth,” Patric Hendershott evaluates the
for Productivity Growth,"
Evans’ macroeconometric model and discusses
investment sector of Evans'
the implications of the composition of investment for productivity
Evans’ treatment of
growth. Hendershott concludes that Evans'
nonresidential investment and residential investment does not
represent an advance relative to conventional treatments.
treatments.
Hendershott also considers ways in which economic policy can
affect economic growth by channelling investment into
into more
in implicitly mandated
productive uses. He notes that the surge in
investment in the last decade and the subsidy extended to ownerownerinvestment
occupied housing have tended to divert investment from its most
productive uses and, therefore, to lower the productivity associated
with a given capital stock.
with
X
x /7 FOREWORD
EOREWORD
Summers’ paper, Norman B. Ture takes issue
In his discussion of Summers'
with Summers’
Summers' conclusion regarding the effects of tax cuts
cuts on
investment. Ture questions the adequacy of the framework that
Summers used to investigate these issues. While he accepts
Summers'
associated
Summers’ view that there are substantial welfare gains associated
with decreasing taxes on
on capital income, Ture concludes that
that
Summers "grossly
underestimated" the gains in output and
“grossly underestimated”
employment which would result from reducing the existing tax bias
against capital formation and saving.
“Income and Payroll Tax Policy
In the first paper in Part III, "Income
and Labor Supply,”
Supply," Jerry Hausman presents evidence on the
supply. Hausman
effects of income and payroll taxes on labor supply.
emphasizes that while supply-side economics has focused attention
attention
on the labor supply and revenue effects of changes on tax rates, the
the
correct measure of the economic cost of taxation is the deadweight
loss associated with taxation. Hausman compares the implications
the Kemp-Roth
of JO"lo
10% and 30% tax cuts, along lines suggested by the
tax proposal, with
with a move to a linear progressive tax system (i.e.,
one with progressive average tax rates but constant marginal tax
rates). He finds that Kemp-Roth tax cuts
cuts decrease deadweight loss,
but they do so at the expense of a large decline in tax revenue. A
linear income tax which yields the
the same revenue as the current tax
system, on the other hand, can significantly reduce deadwcight
deadweight loss
as well as increase labor supply.
“Transfers, Taxes and the NAIRU,”
NAIRU," Daniel Hamermesh
In "Transfers,
presents a detailed examination of the effects of individual tax and
transfer programs on the unemployment rate (specifically, on the
the
NonAccelerating Inflation Rate of Unemployment, NAIRU), labor
supply and employment.
employment. He argues that this microeconomic
approach, building up from a study of individual programs, is
likely to be more reliable than an aggregate or
or macroeconomic
approach that ignores the programs’
complexities.
programs'
While Hamermesh concludes that the net effect of tax and
transfer programs on the NAIRU
NAJRU is approximately zero, he finds
they have a significant effect on labor supply, noting that all the
the
programs he examines are likely to decrease labor supply on net.
Hamermesh concludes that we cannot ease
ease program eligibility and
raise benefits without inducing change in labor supply and
employment, which further raise the costs associated with the
various transfer programs.
age for
programs. He suggests raising the eligibility age
Old Age and Survivors Insurance benefits back to its
previous
level
its
FOREWORD
/
FOREWORD/
xi
and preventing the evolution of Disability Insurance into a
retirement program.
Hausman’s paper, Jeffrey M. Perloff concludes
Commenting on Hausman's
that the paper provides the most reliable labor supply estimates to
date. Perloff does, however, raise a number of questions about
Hausman’s
Hausman 's methodology and examines some
some of the implications of
moving from Hausman's
Hausman’s partial equilibrium analysis to more of a
general equilibrium framework.
Commenting on Hamermesh's
Hamermesh’s paper, Fredric Raines questions
Hamermesh's conclusions about the overall effects of the various
Hamermesh’s
transfer programs on
on unemployment and labor force participation.
Raines agrees that the macro evidence is unreliable, but he
questions Hamermesh’s
Hamermesh's selectivity in accepting or rejecting evidence
from
from various studies of the effects of individual tax and transfer
programs. He also notes that it may be inappropriate to treat the
effects of the various programs as additive, as Hamermesh does in
his paper.
In his luncheon speech, "The
“The Power of Negative Thinking:
Performance," Murray L.
Government Regulation
Regulation and Economic Performance,”
Weidenbaum warns that, at a time when
when the importance of tax
incentives on
on economic activity is being debated, economists
economists should
increasing array of government
not overlook the continually increasing
regulation that
that impairs economic activity. In the current maze of
of a change in the after-tax rates
government regulation, the impact of
of return may, according to Dr. Weidenbaum, have little effect on
production. On the other hand,
hand, the response of the economy to
cuts can be greatly enhanced by simultaneously
supply-side tax cuts
reducing the burden of regulation on the economy.
“The Politics of Supply-Side Economics,"
Economics,”
In his dinner talk, "The
concludes that the establishment of a
Senator Orrin G. Hatch concludes
“budget process"
process” in Congress in the mid ‘7Os
"budget
'70s has not helped arrest
arrest
spending or the reliance on
on deficits.
the growth in government spending
war between supply-siders who seek substantial tax cuts
However, a war
and the various constituencies for
for federal government spending
spending is
that supply-side
unnecessary, according to the Senator. He believes that
tax cuts will sufficiently stimulate economic activity to pay for the
current rate of government expenditures.
Washington University
Laurence H. Meyer
Chairman,
Chairman, Dept. of
of Economics
ACKNOWLEDGEMENT
A
CKNO WLEDGEMENT
Many people associated with the Center for the
the Study of
American Business and the Federal Reserve Bank of St. Louis
have contributed to the planning
planning of this conference and to the
particularly like
production of this proceedings volume. II would particularly
to thank Marcia B. Wallace, of the
the Center for the Study of
American Business, who supervised the arrangements for the
conference and helped edit the manuscripts and prepare them for
publication, to Chris Varvares, who was my assistant, and
provided enormous help at all stages of the conference and with
production of the volume, and to Dan Brennan, of the Federal
Reserve Bank of St. Louis, who helped prepare this volume.
L.H.M.
Tax Rates, Factor Employment,
and Market Production
VICTOR A. CANTO
DOUGLAS H. JOINES
ARTHUR B. LAFFER
INTRODUCTION
An increasing amount of attention has recently been devoted to
the effects of alternative tax structures on the pattern of economic
activity, on the level of taxable economic activity, and on the
aggregate amount of revenue generated by the tax system. In this
paper, a static, one-sector, two-factor model is developed in order
to analyze the effects of taxes imposed purely for the purpose of
generating revenues. 1 For simplicity, these taxes are assumed to be
proportional taxes on the incomes of factors of production. We
derive some properties of the tax structure needed to maximize
output while raising a given level of government revenue. We then
examine empirically a specific instance of tax cuts, the Kennedy
cuts of the early 1960s, to determine their effect on revenues.
The model we present is a highly simpHfied one. While we call
our two factors of production capital and labor, we do not
distinguish one as fixed and the other as variable. Since the model
is static, we do not attempt to analyze the process of capital
formation.' Instead, we assume that at any point there exist fixed
stocks of capital and labor and that these stocks must be allocated
either to household production or to market sector production.'
Victor Canto and Douglas Joines are Assistant Professors of Finance and Busines1
Economics. and Arthur Laffer is Professor of Business Economics at the Graduate
School of Business, University of Southern California.
'More accurately, our model only has one market outpuc It is in facl a two-sector
model in lhe sense that it has a household production sector which also employs
capital and labor in proportions which depend upon their relative cost.
'For dynamic models which treat capital formation as the outcorne of au
intertemporal utility maximization process see Canto (1977) and Joines (1979).
'For a discussion of househQld production see, for examp!e, Becker and Ghez
(1975).
3
PRODUCTION
44 // TAX RATES
RATES AND
AND PRODUCTJON
THE MODEL
MODEL
Two factors are combined in the market sector according to a
Cobb-Douglas production function to produce
produce the market good Q:
Q:
(I)
Q = K°L0~),
(I)
where a and (l
elasticities of capital (Kl
(1 - a) are the partial output elasticities
(K)
and labor (L), respectively, and 0O <
< a<
a < I. The market good,
capital, and labor are inputs into the household production process.
Capital and labor thus have identical analytical properties except
that they are not
not perfect substitutes in either household or market
production.
We assume that
in the market sector are
that factors employed in
are paid
their marginal products and that the rental rate received by capital
(R *) and the wage rate received by labor (W*)
(W*) differ from the rates
(R*)
paid because of the taxation of factor income:
—~
(2)
W*
= W(l tL)
=
ti)
(3)
R*
= R(l t~)
=
tK)
—
—
where W and R are the gross-of-tax wage and rental rates on labor
and capital services, and t~
l1, and t~<
tK are the tax rates on income of
labor and capital, respectively. These tax rates are expressed as
percentages of the rental and wage
wage rates paid. The gross-of-tax
factor payments are denominated in terms of the market good Q.
A change in the ratio of W to R will cause a change in the ratio
of capital to labor demanded by
by firms for production of any level
of market goods. One of the characteristics
characteristics of the Cobb-Douglas
Cobb-Douglas
production function is the constancy of
of the shares of the factors of
production. Accordingly, the demands for labor and capital and the
optimal factor proportions are:
(4)
Kd = aQ
R
(5)
Ld
(6)
(l - IK) W*
1
Kd
aa
w
a
Wa( tK)V.T*
=
Ld
L’1
(I - a) R
(I
a)
tL)
(] - le)
(]
(l
a) (I
R*
=
(]
(1 -— a)Q
a)Q
W
w
—
—
—
CANTO,
AND LAFFER
LAFFER/
CANTO, JOINES,
JOINES, AND
/
the ratio of W* to R
A change in the
R*t will cause a change in the
ratio of capital to labor demanded by households for production of
any level of the household
household commodity. In addition, an increase in
*, given the same ratio of W* to
the absolute levels of W*
W’~and R
R*,
R*,
R
•, will cause households to substitute market goods for capital
and labor in the production of a given level of
of the
the nonmarket
commodity. In other words, an equiproportional increase in W*
and R* causes households to supply more of both capital and labor
to the market sector. Specifically, we assume that the supply
functions for capital and labor take the following
form:'4
following form:
R*
5
(7)
K
K'
(8)
LL'5
(R*\OK(R*\E
= (~:)
OK(R*Y
C +
=
\W*/
\ /
c::)°L(w•y
= (\V*\OL(w*Y
=
\R*/ ~ /
o~>0
C + 01
>0
It is assumed that the
the government derives its revenue entirely
from proportional taxes on factor income, that its budget is always
balanced, and that revenue collections
collections are returned to the economy
generated.’5
in a neutral fashion so that no income effects are generated.
4
that these
factor supply elasticities.
Notice that
these assumptions yield
yield positive
positive own-price factor
5,'
LR
LR
W* 8L
w*
ØL
-~L- ~
OW*
tKR
—
(o,
.. +
0
(°L
± t)
t)>> 0
~
c)> 0
+
The cross-price elasticities, however, could he
be either
either positive
positive or negative.
negative.
£
tk,.,,..
<~‘
—
tLR
C LR
=
w*
8K
K
aw•
R'
R
—
~
W* 8K
— --K aw*
L
L
BL
FL
ag~
BR'
>
.2:_o
-o
—0
K
-
<
> 0
.::,_
OL
L<
<
5
‘For
For simplicity
simplicity it is assumed that:
athe form of transfer payments to individuals,
a. government expenditure takes the
receipt
receipt of which is unrelated to factor supply,
b. there isis no waste or inefficiency on
government, and
on the part of the government,
costless to collect and distribute,
c. taxes and transfers are costless
distribute, respectively.
Under these conditions government spending will have
have no net income effect,
effect, only a
due to the relative price changes resulting from the taxes. Joines
substitution effect
effect due
(l
979) and Canto
Canto (1977) develop a similar analysis of government fiscal
policy in
(1979)
tiscal policy
Miles (1980)
which the possibility of deficit financing is presented. Canto and Miles
consider
consider the possibility of income effects resulting from different types of
government expenditure, collection
col!ection costs,
costs, and the government efficiency level.
5
S
66 //
TAX RATES
RATES AND
TAX
AND PRODUCTION
PRODUCTION
Combining equations 77 and 8, the
the ratio of factors supplied to the
Combining
market sector is:
(9)
(9)
K
L5
(~~\G
o,
\W*J
>0
-
where o,,
o~,tthe
he elasticity of substitution in factor supply, is assumed
to be positive and defined as 01<
+ oL
o~ +
+ e.
oK +
£. Equation 9 says that
that the
the
ratio of capital to labor supplied to the market sector depends only
On the other hand, equation 6
upon the after-tax wage-rental ratio. On
says that the proportion
proportion of capital to labor demanded by the
the
market sector depends only upon the gross-of-tax wage-rental ratio.
the equilibrium
Combining the two equations, one can solve for the
level of the gross- and net-of-tax wage-rental ratio as
as a function of
the tax rates:
(10)
W*
W*
R*
frI
=
—
a\
(I_—_tL~ l+o,
,/ \l
a
—
taJj
05
(11)
W(l_a\[(l_a~(ltL~~0s
R
\ a
a
J\l_txJj
I
Equations 10
IO and 11 show that both the net-of-tax wage-rental ratio
ratio
upon tax rates, factor
and the gross-of-tax wage-rental ratio depend upon
supply elasticities, and output elasticities of the two factors.
It can be shown that if producers maximize profits, the cost
function of the market good will also be of the Cobb-Douglas form:
(12)
I
=
(~)~
(w)(l~)
where the market good has been defined as the numeraire.
numeraire.
12 and substituting for the gross-of-tax
Rearranging equation 12
wage-rental ratio (equation 11), one can solve for the gross-of-tax
wage rate:
a0
5
(13)
w
(l - a)
W=6-a)
[(l_a\
a
J
(1-tL~]
\l
t~/
-
-
1+05
CANTO, JOINES, AND LAFFER/
Similarly, the gross-of-tax rental rate can be expressed as:
(14)
R
= a-
Substituting equations 13, 14, 2, and 3 into the factor supply
equation, one can determine the equilibrium quantities of each
factor and the proportions of capital to labor employed in the
market sector:
(16)
(17)
K ""'
-l+a5
L
The equilibrium level of market output as a function of the
tax rates is obtained by substituting equations 15 and 16 into
equation 1:
aL - a 5 (l + f)ct
(18)
l + a5
Q
EFFECTS OFTAXATION ON MARKET ACTIVITY
:*
Upon inspection of equations 13, 14, and 11, it is apparent that
an increase in the labor wedge {Le., a reduction in (TL
=
)J
will unambiguously increase the equilibrium levels of the gross-of-
1
8 // TAX
RATES AND PRODUCTJON
PRODUCTION
8
TAX RATES
rate (W) and wage-rental ratio (W/R)
tax wage rate
(W /R) and decrease the
rates.'
equilibrium levels of the gross-of-tax rental rates.’
The increase in the
the gross-of-tax wage-rental ratio will generate a
substitution effect away from labor into capital. The equilibrium
level of labor employed in the market sector will unambiguously
of the tax on the equilibrium level of capital
decline.'
decline.’ The effect of
will
employed will be ambiguous.'
ambiguous.’ However, the capital-labor ratio will
‘Defining
=
T =
= (I
"Defining E as the d log operator, TTL
= (1(I —- 5,)
t 1) and TK
(l —
- t~)
tK)
1
1< obtains
Differentiating logarithmically Equations
11 one
Equations 13,
13, 14
14 and ll
obtains
13) EW =-
14) ER =-
-—~——
I
(I
~
15)E(W/R)
*
E(T
/T,)
1<
— a)o.
1-4-n Eli N/Ti)
=
E(T /T~)
1<
~
-
Notice that ET
1<
=
dt
4- and ET
4TN
— —-
dt
T1
t.
=
‘Differentiating
"Differentiating logarithmically equation IS
16
EL
=
£
ET,
OL -·-
=
cw,
~L:1~±
!I -~+ o,
as
£
EO
1<
1~
rK+
(o~= aor
/T,)
+ o~+
I + n
(l_a)0 t]
5
(l+n~)
ET,
-
The cocfficicnr
coefficient for the
the ETK
ET K tern,
term is clearly ambiguous. This ambiguity is due to two
opposing effects. One is the substitution
subsritution effect generated
generated by
by an increase in the tax
rate on capital which leads to a higher
hig}1er proportion of labor services being used in
in the
production of
of market goods,
(reduction in
goods, and the other is a scale effect (reduction
in output)
which leads to a lower amount
amount of labor services being demanded. Whether
Whether
employment of labor
on the relative strength of the two
labor increases or not
nor depends on
effects. On the
the other hand, since
+ aL >
> 0,
0, °~
0, and ft >
effects,
since<-a +
a_,>> 0,
> 00 by assumption, the
the ET
positive. In this case, the
the scale and
coefficient for the
ETL term is unambiguously positive,
1
substitution effect reinforce each other.
8
‘Differentiating
IS
Differentiating logarithmically equation 15
ElK = a ET,~—EK=tETK
= l(J
=
t(l
f(I -— a)o -= a,.,
—~
S'
"K E(TK/TL)
E(T /T )
11 +
K
L
+ o,
a)o.
—o EI~ *
- cr)o,
~x -OK
lI+o,
+ o,
-
a + cr0 + a
-=-·-~-r:rw_., + OK
I+o,
l + o
--—-—-
~-
'
As in the previous footnote, the coefficient
coefficient
ET
1<
the second term is unambiguously
for the
that of the first term is clearly
tirst
positive, while that
dearly ambiguous. The ambiguity of the first
term
term is due to two opposing effects. One is the
the substitution effect which
which leads to a
higher proportion of capital per worker
work-er and the other is the scale effect (reduction
(reduction in
amount of capital
output) which leads to a lower amount
capita! being demanded. Whether
employment of capital increases or not depends on the relative strength of the two.
two.
CANTO, JOINE5,
AND LAFEER
/
CANTO,
JOINES, AND
LAFFER/
unambiguously increase, resulting in a net reduction of the level of
goods.’ The effects of an increase in the
production of the
the market goods.'
tax on income from capital can be analyzed in a similar manner.
Using the simplified model developed in the
the previous section, we
derive certain
certain propositions concerning
concerning the effects on output and
government revenue of changes in the two tax rates. The specific
forms taken by the proofs of these propositions depend upon the
structure we have assumed for our model. This structure allows us
to obtain a closed form solution for the
the variables of interest.
Despite its simplifications, we feel the present model is useful as a
pedagogic device for demonstrating the propositions. Most of these
propositions can be proved using less restrictive models which derive
the factor supply decisions as explicit
explicit results
results of utility maximization,
treat capital accumulation in a dynamic framework of intertemporal
choice, and allow for the possibility of government debt.
Proposition 1.
I. There exists a trade-off between taxes on labor
necessary to maintain
level.
and capital necessary
maintain output at a given level.
The percentage change in output is:
EQ =
=
(19)
E
ET
C ETL
L
- (°L
(0 L
—
\
o,(l +
o~(l+
(I + a,)
(l+o,)
-—
E)a)
E(T IT )
e)a\ E(TK/TL)
/
K
L
At a given level of output (i.e., on an isoquant), EQ
EQ =0.
= 0. Thus, the
the
previous equation implies that:
‘For a Cobb-Douglas production function, E(K/L)
"For
E(K/L)
was shown that
that
E(w/R)
E(TL/TK)
=
=
=
E(W/R).
E(W
/R). In footnote 6,
6, itit
<0
—
I
+
Differentiating equation 18
18 logarithmically
a,oL —- n/I
os(l + a)a
£)a
EQ == aET
EQ
EETL—
1
EQ
=
II++
r~.(I -— a)
(I + E)a,(l
.
Il*a,
+ a,
a’,
o,
—
a
NET
E(TN/Tj)
+
aØ + r)n — n
—
‘ETN
(l+o,)
T and T
TKK appear to be ambiguous.
The signs of the coefficients for TL
ambiguous. However, it isis
1
apparent that as long as the own
own price elasticities effects dominate the cross-price
supply, the
elasticities of factor supply,
the coefficients will be unambiguously positive. In
In the
is assumed that own effects dominate cross
remainder of this paper, it is
cross effects. This
assumption is consistent
consistent with available empirical evidence on factor supply. An
implication of this assumption is that an increase in any of the factor tax rates will
unambiguously reduce the level of market output.
9
10 // TAX
R0
OD
U CCT
ON
T A X RATE
R A T E S AND
AN 0 P
PR
0U
T II 0
N
FIGURE I
Tx
(20)
-!~-ii~ I
=
ETL
+
a
UI1
E(I + a,)
o,(l
oQ +
+ E)a
e)a
—
<O
'
from which one can derive the marginal rate of factor tax
substitution.” This is merely the rate at which the economy can
substitution.'"
substitute the tax on a given factor of production for a tax
tax on
another factor, while keeping output constant. The marginal rate of
factor tax substitution is the slope of an isoquant in the tL - tK
tK
1.
space. Such an isoquant is shown in Figure l.
The above assumptions ensure that only one isoquant will pass
through any point in the tax space. Also, the
the higher the level of tax
rates, the lower will be the level of output. Thus, the closer an
isoquant is to the origin, the higher is the
the level of output to which
range, isoquants are concave
it corresponds. Within the relevant range)
say, the isoquants exhibit a diminishing
from above; that is to say,
homothetic
marginal rate of factor tax substitution. They are also homothetic
—
“The
unambiguous given the assumption that own effects
''"The negative sign isls unambiguous
dominate cross effects. See n.
n. 9.
CANTO,
JOJNES, AND
AND LAFFER/
CANTO, JOINES,
LAFFFR / 11
11
in the tax space. Finally, since it is possible to produce some output
without one of the factors being taxed, the isoquants will intersect
each axis with a finite slope.
Proposition 2. There exists a tax structure that maximizes
Proposition
government revenue.
Here we seek to demonstrate that increases in tax rates are not
always accompanied by increases in tax revenues, and the
reverse
the reverse
fact be the case. Total government receipts can be
may in fact
expressed as:
as:
(21)
o
G
=
=
+ atd
atEI
Q[(l -— a)tL +
=
=
+ a(l -— TK)].
Q[(l —- a)Q
a)(] -— T~)
Tc) +
Differentiating
Differentiating logarithmically, we have:
(22)
EG
=
=
[(l[
+
[o+rxI
1
—
j FT1
a) 0~ — ~K
l+o,
1
+ [(l
+ r)ao, - oL]
+
[Q+e~os_—_~L
a,
lI + o_,
[
j
ETK —-
(I
a)(TL) —FT
l—Rl—a)TL+aTK]
F
—
-
—
aTK
aT~
ETK.
1—[Q—
Or)TL
l - [(l - a)
TL + aTRI
a T Kl
Equation
Equation 22 shows that the percentage change in tax revenue
induced by changes in tax rates depends on the output elasticity
with respect to tax rates (the first and third terms) and the levels of
the tax rates
rates on capital and labor. The equation implies that the
government tax revenue will increase initially with increases in the
tax rates, but at a decreasing rate. Thus, the
the marginal tax revenue
with increases in tax rates, finally reaching some
raised decreases with
point where the marginal tax revenue raised is zero. Beyond this
point, any tax rate increases will reduce revenue collection. Tax
revenue is maximized at the point at which the
the marginal tax revenue
zero, Figures 22 and 33 illustrate
is zero.
illustrate government tax revenues as
functions of the tax rates on labor and capital, respectively,
assuming that the tax rate on the other factor remains constant.
In Figures 2 and 3, two distinct stages can be identified. In Stage
I, the normal range,
!,
&tL
0 and
BtK
0.
FIGURE 2
G
Government
Revenues
n
Tax Rate
FIGURE 3
G
Government
Revenues
/
Tax Rate
CANTO.
JOINES, AND LAFFER/
13
CANTO. JOINES,
LAFFER / 13
In other words, lowering tax rates lowers government receipts and
vice versa. In
in Stage ll,
II, the prohibitive range,
JG <Oand
<O,
<0 and ilG <0,
a1L
otK
OIL
8t~(
and increases in tax rates on
on labor and capital decrease government
revenues, and vice versa.
In
in all the stages, the change in government revenues arising from
on the elasticities of the factor
changes in the tax rates depends on
supply curves, the output elasticities of the factors, and the
the level of
the taxes. The foregoing
foregoing analysis shows that there exists a tax
structure
structure at which government tax receipts are maximized.
The first-order conditions imply that G
maximized when
0 is maximized
(23)
—A + (I
(24)
—9 + (1— O)BTL + (B + 9aT~ = 0
—
cr)(A + DTL + OATK
=
(3
where
(25)
A
(26)
B
B
=
(] + ,)(! - a)o, - oK
(I+r)Q—a)o,—oK
I + 05
1+0
= J!-±
(l + ,) aa_, a)ao,
=
aL
0
L
—
Il ++ Os
a,
From equations 23 and 24, one can solve
solve for the factor wedge:
(27)
TL=
TL—
(28))
TK
A
A
(l—afl,A+B+I)
(l
- a )(A + B + l)
-
= ___B_ __
a(A + B —icr(A
+ I)l)
—
—
—
t)(l —- a)a,
(l
(I +
+ LXI
a)~~-— oK
(l+a)(l—a~J+o,)
(l + ,)(1 - a)(l + a,)
(l +
(I
+ E)ao,
e)ao, - o'-.c
(1(l + c)a(l +
+ a,)
o,)
—
illustrate the marginal wedges which maximize
quations 27 and 28 illustrate
vernment tax revenues. Using these results, one can then solve
,vernment
licitly for the tax rates, the maximum amount of revenue that
,licitly
output.
government can produce, and the corresponding level of output.
apparent also that
that these results depend on the supply and
‘ut elasticities of the
•ut
the factors of production.
-
CANTO, JO
JOINES,
CANTO,
INES, AND LAFFER
LAFFER //
IS
15
FIGURE 44
B
A
D
C
TK
If both factor income tax
tax rates are in the prohibitive range, an
increase in either tax rate, the other rate constant, leads to a
reduction in
in total revenue collected. Since both tax rates are in the
prohibitive range, the
the other factor tax rate must be reduced if
revenue is to remain unchanged. Hence the iso-revenue curve is also
downward
downward sloping in this region, which corresponds
corresponds to segment BC
in Figure 4.
In Case 3, one of the factor tax rates is in the prohibitive range
while the other is in the normal range. An increase in the
the
prohibitive tax rate leads to a reduction in revenue. If revenue is to
remain unchanged, the tax rate in the normal range must increase,
and the iso-revenue curve is therefore upward sloping. Case 33
corresponds to segments AB and CD in figure 4.
Higher valued iso-revenue curves lie inside lower valued curves.
In the limit, the iso-revenue curve shrinks to a point, the maximum
revenue point (Proposition 2).
Proposition 3: There exists a tax structure that
that maximizes output
output
at a given level of government expenditures.
CANTO,
/ 1$
CANTO, JOINES,
JOJNES, AND
AND LAFFER
LAFFER/
15
FIGURE 44
TL
B
A
C
If both factor income tax rates are in the prohibitive range, an
increase in either tax rate, the other rate constant, leads to a
rates are in the
reduction in total revenue collected. Since both tax rates
prohibitive range, the other factor tax
tax rate must be reduced if
revenue is to remain unchanged. Hence the iso-revenue curve is also
Segment BC
downward sloping in this region, which corresponds to segment
in Figure 4.
In Case 3, one of the factor tax rates is in the prohibitive range
while the other is in the normal range. An increase in the
prohibitive tax rate leads to a reduction in revenue. If revenue is to
remain unchanged, the tax rate in the normal range must increase,
and the iso-revenue curve is therefore upward sloping. Case 33
corresponds to segments AB and CD in figure 4.
4,
Higher valued iso-revenue curves
curves lie inside lower valued curves.
In the limit, the iso-revenue curve shrinks to a point, the maximum
revenue point (Proposition 2).
Proposition 3: There exists a tax structure that maximizes output
at a given level of government expenditures.
16 // TAX RATES
RATE5 AND PRODUCTION
PRODUCTION
FIGURE 55
tL
Isorevenue
lsoquant
*
t.
tK
tK
K
The graphical solution to this problem is quite simple."
simple.’’ The level
Once this is
of revenue collection determines
determines the iso-revenue curve. Once
known, the objective becomes to find the lowest
lowest possible isoquant
that satisfies the revenue constraint. At this point the two curves
curves
two loci has the
are tangent. The question becomes which of the two
largest curvature at the tangency point. It is obvious that the isorevenue curve can never be below the isoquant. If it were, a lower
isoquant (higher output level) could be found that
that yields the same
amount
amount of revenue. The graphical solution is presented in Figure 5.
The design of an optimal tax system has long been a matter of
concern to economists.
economists.’122 In order to design an optimal
optimal tax system
(since value judgments must be made as to the objective function to
be maximized), some sort of social welfare function has to be
specified. Our discussion of Proposition 3
3 implicitly assumes that
‘‘For a formal derivation of this proposition, see Canto, Laffer, and Odogwu
"For
(1978).
“For an illustration see Harberger (1974), Mirlees
"For
Mirlees (1971), Stiglitz
Stig!itz (1972),
(1972), Cooter
(1978).
CANTO,
JOINES, AND LAFFER
/ 17
17
CANTO, JOINES,
LAFFER/
policymakers have somehow arrived at a social welfare function
into which both transfer payments and market output enter with
the transfers, some cost
cost in terms
positive signs. In order to finance the
of market output is incurred. Thus, a trade-off exists and the
optimum will be at a point where the marginal social gain from the
government expenditure equals the marginal social loss from the
fall in output.
EMPIRtCAL EVIDENCE FROM THE KENNEDY TAX CUTS
CUTS
EMPIRICAL
In the previous section, we demonstrated that there is a tax
structure which maximizes government revenue (Proposition 2) and
that it is possible for tax rates to be so high as to generate less
revenue than would be raised from lower tax rates. Whether any
real-world governments have ever operated in the prohibitive range,
however, is an empirical issue. There are several ways of analyzing
this question, the most common of which is what might be called
“elasticities” approach. This approach consists of examining
the "elasticities"
existing estimates of, for example, factor supply elasticities and tax
rates. These estimates are applied to some theoretical model in
order to simulate the revenue effects of tax rate changes. In
general, the higher the elasticities and the tax
tax rates, the more likely
it is that the tax rates are in the prohibitive range. One recent study
conducted along these lines is that of Fullerton (1980).
While this approach can undoubtedly provide valuable
information on the revenue effects of tax cuts, it has several
shortcomings. The first of these is that the effective tax base may
be smaller than total economic activity. Some economic activity
may escape taxation because it is legally exempt from taxation or
because of outright tax evasion. The factor supply elasticities
relevant for an analysis of revenue effects are the elasticities
elasticities of
supply of factors to taxable activities. If there is a reasonable
degree of substitutability between taxable and nontaxable activities,
then these elasticities
elasticities may well be higher than the conventionally
measured overall factor supply elasticities. This problem can be
quite severe as concerns saving, since there are many uses to which
saving can be put which involve
involve a partial or complete tax exemption
of
the
resulting
income.
Notable
among these are residential capital
of
resulting income.
and municipal bonds. Recent discussions of the “underground
"underground
economy”
economy" suggest that under-reporting of income may well make
18 // TA
TAX
PRODUCTION
18
X RATES
RATE S AND
A N D PRO
D U CT I ON
the distinction between taxable and nontaxable activity important
for labor supply as well.”
well."
Another difficulty with employing this elasticities approach in a
highly aggregated model is that there are in fact many tax rates
which apply to different types of economic activity and also many
categories of productive factors, each of which potentially has a
different elasticity of supply to taxable economic activity. Given
this multiplicity of tax rates and of types of factors, it seems quite
likely that some tax rates somewhere in the system are in the
prohibitive range. This, in fact, is the very essence of certain tariffs
on international transactions which are imposed for protectionist
protectionist
purposes rather than for revenue generation. Certain features of the
domestic U.S. tax system may also result in a high tax rate being
imposed
imposed on an elastically supplied factor. For example, the federal
personal income tax imposes a “marriage
Hmarriage penalty”
penalty" which taxes the
income of a secondary worker at the marginal rate of the primary
worker in the family. This fact, combined with evidence that
that
married women have substantially higher labor supply elasticities
elasticities
than do prime-age males, makes it at least reasonable to conjecture
that some features of the current tax system result in prohibitive
taxation. Also, recent evidence indicates that
that proprietors of small
businesses, who have more control over hours worked than do most
employees, may have a considerably higher supply elasticity
elasticity than do
4
14
general.’ Finally, effective marginal tax rates can be quite
males in general.
quite
for
high for those in upper income brackets and can be even higher for
“The
JJThe factor supply functions (equations 77 and 8) attempt to take these
these effects into
account. As tax rates alter the relative price of factors of production, they also alter
alter
the relative price of the nonmarket (i.e., nontaxed) activities. The change in
the
in the
to the market sector thus depend.s
factor supply to
depends on two effects, a substitution
substitution effect
household production and a scale effect. The substitution effect ls
is captured by
in household
by the
factor supply equations.
E term in both factor
These
to own
own and cross factor supply elasticities, as shown in n. 4.
4.
These effects give rise lO
The
The own effects are always positive, and the cross effects are ambiguous.
ambiguous.
It can be shown
sho,vn that if the
rhe product of the own-price elasticities is larger than that
that
,vtkp > ttRrK\v).
of the cross-price elasticities frI
(El,wrh
ri,R<=kw), the effects of taxes on
on output
outpur are
the cross effects. However,
qualitatively similar to those that neglect the
Hmvever, the magnitude
of the change will be different. Whether the total effect
effect is larger or smaller depends
or not the cross-price
upon whether
wherher or
cross-price elasticities offset or reinforce the own-price
own-price
etfects. In
In the
the latter case, it is easily shown that the market-output price elasticity
effects.
will he
the case in which the
the cross-price elasticities are
wil!
be larger than the
arc zero. Thus, the
the factor markets)
markets) could
neglect of these cross elasticities (the interaction
interaction between the
lead one to underestimate the economy’s
economy's responsiveness to tax
tax rate changes. See
Canto (I
977) and Joines
( 1979).
(1977)
Joines (1979).
“See
"Sec Wales (1973).
C A N T O , J O l N E S . A N D L A F F E R / 19
the poorest workers and those receiving Social Security, who stand
to lose benefit payments as their earnings increase.
The relevant question to ask is thus not whether the United States
or some other real-world economy is operating in the prohibitive
range. It is quite Hkely that somewhere in the system there exists a
tax rate on some type of activity which results in less revenue than
would a lower tax rate. The relevant issue concerns the revenue
effects of a specific set of tax rate changes.' 5 Of particular interest
are recent proposals for broad-based cuts in federal personal and
corporate income tax rates. While the elasticities approach might be
employed to simulate the effects of such a tax cut, another method
suggests itself. 10 This method consists of examining past instances
of similar tax cuts to determine their effects on revenue,
The Kennedy tax cuts of 1962 and 1964 offer a natural
experiment. Following their enactment, the economy experienced a
greater than normal expansion of real economic activity. A
comparison between measures of economic act1vity prevailing
before (l 96 l) and after ( 1966) the tax cuts were enacted indicates
that unemployment declined from 6.7 percent to 3.8 percent and
capacity utilization as measured by the Federal Reserve Board
increased from 77 .3 percent to 91.9 percent. During this period,
real GNP grew at an average annual rate of 5.9 percent. The
average annual growth rate in nominal GNP was 7 .5 percent, while
federal government expenditures grew at a rate of 6.2 percent.
Consequently, the ratio of government expenditures to GNP fell. It
thus seems unlikely that the increase in economic activity can be
attributed entirely to the stimulus of increased government
spending.
Another issue concerns whether the apparent expansion of
economic activity was sufficiently large to offset the negative effect
on tax revenues of the tax rate reductions themselves. Alternatively
stated, the issue concerns whether the economy was in the normal
or the prohibitive range of the Laffer curve. Michael K. Evans'
"Fullerton recognizes the multiplicity of tax rates and factor supply elasticities to
which we refer. He i~ also careful to simulate the effects of a specific tax cut-a
broad-base<l cut in tax rates on labor income.
"In using the elasticities approach to simulate the effects of proposals such as the
Kemp-Roth bill, one must be careful not to treat them as cuts only in labor income
tax rates. They also entail reductions in personal tax rates on income from capitaL
The elasticity of supply of saving and factor demand ela&ticities, as well as labor
supply elasticities, are important in su<:h a model. In addition, there may be
important cross elasticities of factor supply, as discussed in n. 13 above.
20 I/ TAX
TAX RATES
AND PRODUCTlON
PRODUCT(ON
RATES AND
(1978) examination of revenue data for this time period indicates
that revenues from individuals with taxable incomes
incomes in excess of
$100,001) increased from $2.3 billion in 1962 to $2.5 billion in 1963,
$100,000
to $3 billion in 1964, and to $3.8 billion in 1965. Total personal
income tax revenues, however, declined between 1963 and 1964.
Although high-income individuals would appear to have been in the
the Laffer curve, the evidence concerning
prohibitive range of the
overall personal tax revenue suggests that the weighted average of
the individual personal income tax rates was in the normal range.
the
range.
That is, a reduction in the overall personal tax rate led to a
loss in tax
reduction in revenues. This can be attributed to a loss
revenues from individuals at low
low income levels in excess
excess of the gain
in tax revenues
revenues from
from individuals at high income levels.
Other
Other casual evidence on the revenue effects of the Kennedy tax
tax
cuts
cuts exists, but there is some dispute as to the interpretation of this
evidence. Representative Kemp and Senator Roth have asserted that
federal tax revenues during the fiscal
fiscal years 1963 through 1968
federal
the 1962
showed a cumulative increase of $54 billion over the
1962 level of
annual
annual receipts, whereas the Treasury Department had estimated a
cumulative revenue loss of $89 billion over the
the same period as a
result of the tax cuts."
cuts,’ Heller (1978) and others have pointed out
that these two numbers are not comparable,
comparable, however. The $54
billion refers to the increase in actual revenues between the earlier
and later years. The $89 billion figure is the Treasury Department's
Department’s
estimate of the difference between actual revenues during the later
period and what they would
the same period if
wonld have been during the
the tax reduction had not occurred. That there is no
no necessary
inconsistency
inconsistency between these two numbers can be seen by examining
set of estimates reported by Pechman (1965). Pechman
a similar
similar set
forecast that actual individual income tax liability on returns filed
for 1965 would be $46.4 billion, or $10.7 billion lower than his
estimate of 1965
I 965 liability
liability with no tax cut, but $$1.6
l.6 billion higher
than actual liability on 1962
1962 returns. Furthermore, if the $89 billion
figure cited by Kemp and Roth were adjusted to include similar
estimates of
of the effects of the Tax Adjustment Act of
Treasury estimates
estimate would be
1966, the Treasury’s
Treasury's cumulative revenue loss estimate
be only
$83 billion.
It is quite possible that the Pechman and Treasury estimates
overstate the size of the actual revenue loss resulting from the tax
of the early 1960s.
1960s. These estimates are derived by comparing
cuts of
“See Kemp
Kemp (1977).
"Sec
CANTO, JOINE5,
AND LAFFER/
LAFEER /
CANTO,
JOINES, AND
21
the revenues which would result from applying alternative tax
structures to a given level of economic activity. Such “static”
"static"
estimates thus ignore any feedback effects of tax rates on economic
activity and revenues. If these feedback effects are quantitatively
estimates may considerably
important, then the static estimates
considerably overstate the
true revenue loss.
It would be desirable to obtain an alternative set of revenue loss
for any
any actual feedback of tax rates on
estimates which allow for
economic activity. Such estimates would not be based on any
prescribed level of economic activity. In the next section, we report
such a set of estimates derived from univariate time series analysis
of various revenue series and reported in Canto, Joines, and Webb
(1980).
TIME SERIES ESTIMATES
ESTIMATES
There are several ways of obtaining revenue estimates without
first prescribing a level of aggregate economic activity.
activity. The
the true
desirability of these estimates rests on the belief that the
structure of the economy is such that tax rate changes affect
economic activity.
activity. An obvious way of incorporating
incorporating any existing
existing
feedback effects would be to estimate aa structural model which
includes such
This model could be used to obtain forecasts
such effects. This
would have been in the absence of tax rate cuts,
of what revenues would
and these forecasts could in turn be compared with actual revenues.
could be used to simulate the
the effects of
Alternatively, the model could
of
various tax changes.
difficulties with this approach, however. Aside
There are several difficulties
from the sheer effort required to design and estimate aa complete
structural model, the resulting forecasts would be subject to certain
sources of error in addition
addition to the parameter estimation errors
which affect all
all attempts at statistical inference. The most
important of these sources is misspecification of the structural
model, either
either through an incorrect choice of variables to be
included in the
the model or through the imposition of
of incorrect
incorrect
identifying restrictions. In addition, Lucas (1976) points out that
policy simulations
simulations based on such structural models
models are inherently
suspect because the parameters of the model will in general be
functions of policy variables and will change in response to shifts in
those policy variables.
Palm (1974) provide an exhaustive taxonomy of the
the
Zellner and Palm
22 /
TAX RATES
RATEs AND
AND PRODUCTION
PRODUCTION
TAX
types of equations associated with dynamic simultaneous
various types
equation systems and discuss the uses
uses and limitations
limitations of each. It is
of particular interest to note that the univariate time series
properties of the system’s
system's endogenous variables are implied by the
structure of the model and the time series properties of the
exogenous variables. It is thus meaningful to fit time series models
to each of the endogenous series over periods when both the
structure of the complete model and the time series properties of
exogenous variables are stable. One of the primary uses of such
the exogenous
a simple univariate model is in forecasting
forecasting the series to which it is
fit. In addition, these models make much more modest demands in
terms of data requirements and a
a priori knowledge of the system's
system’s
structure than would full-blown structural estimation. Furthermore,
as Nelson (1973) points
points out, univariate time series models are not
subject to errors
errors in specifying the structure of the complete model,
and hence in theory need not yield less accurate forecasts than
would structural estimation. The results reported in Nelson (1972)
indicate that this conclusion holds in practice as well as in theory.
1950 to the
the early 1960s there existed the most stable federal
federal
From 1950
tax policy of any period of comparable length
length since the end of
World War I. There were no important changes in personal or
corporate income tax rates from 1951 to 1964. Compared to the
fluctuations in tax rates during the Great Depression, World War II,
II,
and the
the Korean War, the stability during the later period is quite
striking. It thus seems reasonable to regard this period as one
during which the underlying structure of the economy was fairly
stable. Furthermore, the period of stability is long enough to
provide a minimal number of observations for estimation of
univariate time series models.
models. Canto, Joines, and Webb used this
period to fit univariate models to various revenue series of interest
and employed these models to forecast revenues into the mid-l960s
under the assumption that there would
would be no changes in tax rates or
the underlying structure of the economy. The forecast errors from
these models
models can
regarded as
revenue
these
can be
be regarded
as point
point estimates
estimates of
of the
the revenue
changes resulting from the tax rate
rate cuts of the early 1960s.
The two federal revenue series to which univariate models were
fit are denoted FPR and FCR. They represent,
represent, respectively,
quarterly federal personal income tax receipts and quarterly federal
corporate income tax receipts, each deflated by the Consumer Price
Index. The base period for the price deflation is the fourth quarter
of 1963. None of these series has been seasonally adjusted.
CANTO, JOINES, AND LAFFER/
23
The models which fit these two series are:' s
vv .FPRl
""' 0.0026 +
lt
(0, 11)
6£ "" 0.60
t "" 1956:l - 1963:4
and
0.2403 + 0.156. +
{0.12)
(0.12)
[I + O.2OB 4 ]
(0.15)
0.326, + 0.4162
(0.l3)
(0.12)
au
"" 0,47
di
""'
l, quarter i, i ""1, ... , 4
0, otherwise
- 1962:4
of the residuals t 1and 1\ yielded
t
"" 1952:4
Examination
no indication of
model inadequacy.
The forecast errors which result from applying these models to
the immediate post-estimation observations may be regarded as
"Standard errors appear in parentheses below parameter estimates. The model for
FPR for the longer period 1952:2 to 1963:4 is slightly complicated due to an
"intervention" which occurred in the first quarter of 1955. The Internal Revenue
Code of 1954 moved the filing deadline for the federal personal income tax from
March 15 to April l5 of ead1 year. This change noticeably altered the seasonal
pattern of personal income tax receipts, shifting revenues from the first quarter to
the second quarter of each calendar year from !955 onward. Such an intervention
could be represented by the model in the differenced serie,
'v'v,FPR,
~
µ, + {wo - w,B -
w,B'J
!,
+ ,,
where
l _
' -
1, t = 1955: t
0, otherwise.
One would expect a priori to find w,, w,
yielded the equation
1/V .FPR, "" -0.049 +
(0.091)
o,
=
< 0 and
w,
> 0. Estimation of this model
I - 2.00 + 5.998 - 2.278'!
(0.61)
(0.61)
I, +
£1
(0.61)
0.60
Examination of the residuals t, gave no indication of model inadequacy. Since !he
intervention term does not affect forecast, for the posH963 period, Canto, Joines,
and Webb chose to base their analysis on the simpler model reported in the text. See
Box and Tiao (l 975) for a description of intervention analysis.
24 /
TAX RATES AND PRODUCTION
TABLE l
Estimates of Cumulative Change in Federal
Personal Income Tax Receipts
(Billions of Dollars)
Cumulative
Change Through
1964
Time Seriesa,b
-2.93
Treasuryb,c,d
Pechmanc,~
-2.4
-9,9
-20.6
( 1.32)
1965
- 9.31
(6.76)
-11.l
1966
-14.43
-23A
( 18 .00)
aconstant (]%3:4) dollars. Standard errors appear in parentheses below estimates.
0Fisca! year.
"Current dollars.
<lSource: H.J. Fowler, "Statement Bdore the Committee on Banking and
Currency." /',feelings Wirh Department and Agency Officia{s: Hearings Before the
Commitlee on Banking and Currency, House of Representatives Washington: U.S.
Government Printing Office, 1967, p. 12.
•Cumulative change in rnx liability on returns filed for relevant tax year. Source:
J. Pechman, "The lndividual Income Tax Provisions of the Revenue Act of 1964."
Journal of Finance 20 (May 1965), p. 259.
point estimates of the revenue changes resulting from the 1962 and
1964 tax reductions. These estimates may then be compared with
other published estimates of the revenue changes.
Table 1 contains alternative estimates of the cumulative change in
federal personal income tax receipts. The time series and Treasury
estimates are for the cumulative change from the time the rate
reductions became effective until the end of selected federal
government fiscal years. Pechman's estimates are for the cumulative
change in tax liability on returns filed for selected tax years, and
hence do not cover time periods strictly comparable to those of the
other estimates. 19
Comparison of the time series estimates with the various static
estimates shows very little discrepancy for 1964, Furthermore, while
"The lime series estimates which correspond most closely to the periods covered
by Pechrnan are --9.07 (wilh standard error of 4.SJ} for 1964 and -14.77 (with
standard error of 14.50) for 1965.
CANTO, JOlNES, AND LAFFER/
25
TABLE 2
Estimates of Cumulative Change in Federal
Corporate Income Tax Receipts
{Billions of Dollars)
Cumulative Change
Through Fiscal Year
aTime Series
bTreasury
1963
-0.06
(1.06)
-2.4
1964
L70
(4.34)
-4.1
1965
4.77
(8.47)
-6,9
1966
10,74
(13.43)
-9.5
aconstant (1%3:4) dollars. Standard errors appear in parentheses below estimates.
"Current dollars. Source: H. J. Fowler, "Statement Before the Committee on
Banking and Currency." Meetings With Depar!ment and Agency Officials: Hearing5
Before the CommiUee 011 Banking and Currency, House of Representatives.
Washington: U.S. Government Printing Office, !967, p. 12.
the point estimates are indistinguishable from the various static
estimates for that year, they are more than two standard errors
below zero. This would seem to indicate that the initial feedback
effects on the tax base were negligible.
Examination of Table I shows that for years after 1964, the time
series estimates show smaller revenue losses than do the static
estimates, and by l 966 the difference between the time series and
Treasury estimates is considerable. It should be noted that the
standard error associated with the time series estimate for 1966 is
quite large. Nevertheless, these results, if taken at face value,
indkate that there is only about a twenty percent probability that
the cumulative change through 1966 was positive. They also
indicate, however, that there is only about a thirty percent chance
that the cumulative loss was as large as the Treasury estimated.
Table 2 contains alternative estimates of the cumulative change in
federal corporate income tax receipts resulting from the various
corporate tax changes legislated in 1962 and 1964. Whereas the
26
TAX RATES
RATES AND
AND PRODUCTION
PRODUCTION
26 /I TAX
Treasury estimates show a steadily growing
growing revenue loss between
1963 and 1966, the time series estimates show a negligible revenue
loss in I1963
963 followed by a steadily increasing revenue gain between
1964 and 1966. As was the case with federal personal
personal income tax
receipts, the standard error associated with the cumulative revenue
receipts,
change through 1966 is somewhat large. Nevertheless, these results
indicate that there isis only a twenty-five percent chance that there
ten percent
was aa cumulative revenue loss, and less than a ten
probability that there was a loss as great as the Treasury estimated.
Thus far we have examined only federal government receipts
taxes which were actually reduced in the early
early 1960s.
As
from the taxes
1960s. As
Bronfenbrenner (1942, p. 701) points out, however, the notion that
reduction in tax rates may increase revenues takes two forms.
A direct form limits attention to the specific levy under consideration. As
As
argument applied to the tax on beer states simply
applied in
it1 direct form, the
the argument
that an increased rate would
would decrease revenues from the tax on beer, and vice
vice
versa.
tax system.
applied to
versa. An
An ittdirect
indirect form
form applies
applies to
to the
the general
general .... , tax
system. As
As applied
to the
the
beer tax, it states that even
though an increased rate may increase receipts
receipts
even though
from beer, it will decrease receipts
receipts from other taxes by more than
rhan enough to
offset the gross increase.
If the federal personal and corporate income tax cuts did in fact
expand economic activity, if the base for other taxes is positively
related to economic activity, and if the rates of
of these other taxes
remained constant, then one should observe higher than expected
revenues from these other taxes during the years immediately
following the federal
if such
federal income tax reductions. Furthermore, if
indirect effects do exist,
exist, they should be taken explicitly into account
in estimating the revenue effects of proposed tax
tax changes.
In order to determine whether any
any indirect revenue
revenue increases
resulted from the federal income tax cuts, Canto, Joines, and Webb
fit a univariate time
time series model
model to quarterly
quarterly state and local
local income
income
tax receipts deflated by the Consumer Price Index, neither of which
had been seasonally adjusted. The model appropriate to this
variable,
SL!, is
variable, denoted SLI,
2}e,
V
SLJ, =
= 0.11 + [I
[1 + 0.258
v,su,
0.25B + 0.548
0.54B'Je,
4
(0,11)
(0.020)
(0.11)
(0.11)
(0,020)
6,, = 0.089
0.089
o,
=
t
=
1948:1 - 1963:4
1948:l
—
ê~gave no indication
Examination of the residuals e,
indication of model
inadequacy.
C A N T O , J O l N E S , A N D L A F F E R / 27
TABLE 3
Estimates of Cumulative Change in State
And Local Income Tax Receipts
(Billions of Dollars)
Cumulative Change
Through Fiscal Year
1964
1965
1966
2
3
Time Series
Estimate
Standard
Error
0.49
1.48
0.14
0.45
3.28
0.86
Corn,tant (1963:4) dollars.
Table 3 contains estimates of the cumulative change in state and
local income tax receipts for selected fiscal years. For each year the
point estimate is positive and large relative to its standard error. It
is possible that part of this increase could have arisen because state
and local tax rates increased faster between I 964 and I 966 than
they did during the period used to construct our forecasts. To check
this possibility, we computed a weighted average of state personal
income tax rates for years before and after the federal rate cuts.
This average actually increased more slowly during the three years
after the federal rate cuts than during the preceding three years.
This evidence therefore strongly suggests that the federal tax cuts
did entail the predicted indirect revenue increases.
In summary, analysis of these three types of revenues yields a
point estimate for the cumulative loss in the three types of revenues
combined of $0.41 billion through 1966. Given the uncertainty
attaching to this estimate, it is virtually indistinguishable from zero.
Furthermore, it contrasts sharply with the Treasury's estimate of
the federal revenue loss of $33 billion. 1t thus seems quite likely
that the static revenue estimates used by the Treasury greatly
overstate the revenue effects of federal tax rate changes. In
addition, it seems almost as likely that the federal tax cuts increased
revenues as that they reduced them.
If the Kennedy tax cuts did result in revenue losses smaller than
those implied by simple static calculations, this suggests that tax
rate reductions may in fact be effective in stimulating economic
activity. One qualification to this line of reasoning is in order,
however. lt was noted above that if tax shelters are expensive, a
28 // TAX
TAX RATES AND PRODUCTION
PRODUCTION
TABLE 4
Estimates of Cumulative
Estimates
Cumulative Changes in
Real Gross National Product
Cumulative Change
Thro ugh Fiscal Year
Through
aTime Series
Estimate
Standard
Standard
Error
1964
1965
1966
5.25
29.05
84.34
4.81
4.81
18.03
33.68
33.68
aConstant
(1963:4) dollars.
dollars.
aconstant (1963:4)
reduction in tax
tax rates might result
result in a decrease in tax revenues
without necessarily
necessarily being accompanied by an increase in economic
activity. The expansion of
of the tax base might instead occur as
people transfer economic activity from nontaxable to
to taxable forms.
Product
Examination of some variable such as real Gross National Product
would allow
allow a separate
separate check on the influence
influence of the Kennedy tax
tax
cuts on economic activity.
The following multiplicative seasonal time series model was
identified and estimated for quarterly
quarterly data on real Gross National
Product:
206 + 0.0956,, + 8.3656,,
6
VGNP —= -9.366"
5.206,,
VGNP
—9.366,, ++ (0.627)
S.
2t + (0.624)
0.09563~ + (0.626)
S~ Sd4~
T — (0.652)
T
(0.652)
(0.627)
(0.624)
(0.626)
+
11 —- 0.350B'Ja,
0.350B1Ja,
+ [I
(0.140)
6a
&,
= 2.15
=
—
it
—
=
l,quarteri,i
0, otherwise
otherwise
0,
1, quarter i, i
=
I
4
l, ... , 4
1951:2 - 1963:4
—
The price index was the Consumer
Consumer Price Index, and the series was
not seasonally adjusted. Diagnostic checks of the residuals did not
indicate any significant departures from a white noise process.
This time series
series model was used to develop forecasts of real
output which
which were then compared with post-sample realized values.
The results are summarized in Table 4. The point estimates reported
there provide evidence that an unforecast expansion in economic
activity followed the tax rate cuts, with most of the effect occurring
C A N T O , J O I N E S , A N D L A FF E R /
29
in fiscal years 1965 and 1966. This is consistent with the evidence
from the analysis of tax revenues. The point estimate of the
cumulative gain through 1966 is $84 bjl]ion and is about two and a
half times its standard error.
CONCLUSION
Our analysis shows that increases in taxes reduce the returns to
the factors as well as factor employment and market output. A
firm's decision to employ a factor is based partly on the total cost
to the firm of the factor's services. The more it costs to hire
factors, the lower the quantity of factor services the firm will
demand. The lower the costs to the firm to hire factors, the more
factor services the firm will demand. Increases in tax rates increase
the cost of hiring factors. Therefore, increases in tax rates will
result in fewer factor services demanded.
For the owners of factors, the decision to offer factor services to
the market is based in part on the earnings the factor receives net
of taxes. The more the factor receives net, the larger will be the
quantity of services offered to the market, and vice versa. Increases
in tax rates reduce the net-of-tax returns to factors. Increases in tax
rates reduce the quantity of factor services supplied. Thus, both the
firms' desire to employ factors and the factors' willingness to work
are diminished by increases in tax rates. The foregoing analysis
applies equally to either capital or labor employment and their
respective returns. The net effect is that the level of factor
employment and output fall as tax rates increase.
Our analysis also indicates that increases in tax rates could as
well reduce as increase government tax revenues. In fact, there
exists a tax rate structure which maximizes government tax receipts.
This tax structure depends on the supply and output elasticities of
the factors of production. The set of tax rates which creates
conditions such that increases in the rates are accompanied by
increases in government tax revenues are referred to as the normal
range. The tax rates where increases in the rates are accompanied
by decreases in tax revenues are said to be in the prohibitive range.
Except at a corner solution, whenever tax rates are reduced, total
revenue is never reduced in the same proportion as the tax rate
reduction. The more elastic factor supplies are, the more likely it is
that any given tax rates will fall into the prohibitive range. Also,
the higher the level of tax rates, the more likely tax rates are to be
in the prohibhive range.
30 / TAX
TAX RATES
RATES AND
AND PRODUCTION
PRODUCTION
Our simple static model shows the government tax policy affects
output which can be obtained from a given stock
the market-sector output
of resources. In particular, increases in tax rates reduce market
employment and output. Such a tax rate increase, however, would
would
also have long-term effects on the size of the resource stock. Both
human and nonhuman capital are
are reproducible resources which can
any
be augmented only at some cost. The stocks of such capital at any
point
point in time depend upon past investment decisions, and the future
stocks
upon current
current investment
A change
in afterstocks depend
depend upon
investment decisions.
decisions. A
change in
aftertax factor rewards will affect not
not only the intensity of utilization of
existing factors, but also the decision to invest in new
currently existing
resources, and thus the size of the future stock of factors of
production. A dynamic model is required to analyze such questions.
We merely
We
merely note in closing that increases in tax rates are likely to
cause reductions in future output potential, which reinforce the
reductions in current output predicted by our static model.
The proposition that increases in tax rates beyond a certain level
may actually reduce tax revenues and hence market-sector output is
an empirical issue. Data on tax revenues and real per capita output
before and after the Kennedy tax cuts of 1962
1962 and 1964
l 964 were
examined in order to ascertain whether this proposition has
empirical support. The evidence suggests that a significant
expansion of
of economic activity and no significant loss of revenue
occurred as a result of the Kennedy tax cuts. The point estimate of
occurred
the cumulative unexpected expansion in output through 1966 is $84
billion, which is large relative to its standard error. Our evidence on
revenues is
The point
cumulative
is less
less conclusive.
conclusive. The
point estimate
estimate of
of the
the cumulative
revenues
revenue change is virtually identical to zero, and it is thus almost
equally likely that the Kennedy tax cuts increased revenues as it is
that they decreased them.
CANTO, JO INES, AND l.. AF FER /
31
REFERENCES
Atkinson, A. B., and J. E. Stiglitz. "The Structure of Indirect
Taxation and Economic Efficiency." Journal of Public
Economics I (April 1972): 97-119.
Becker, G. S. and G. Ghez. The Allocation of Time and Goods
over the Lijecycle. Columbia University Press for the National
Bureau of Economic Research, New York, 1975.
Box, G. E. P., and G. C. Tiao. "Intervention Analysis with
Applications to Economic and Environmental Problems."
Journal of the American Statistical Association, 70 (March
1975): 70- 79.
Bronfenbrenner, Martin. "Diminishing Returns in Federal
Taxation?" J.P.E., 52 (October 1942): 699-717.
Canto, V. A. "Taxation, Welfare and Economic Activity." Ph.D.
Dissertation, University of Chicago, 1977.
Canto, V. A., D. H. Joines, and R. L Webb. "The Revenue
Effects of the Kennedy Tax Cuts." Unpublished Working Paper.
Graduate School of Business Administration, University of
Southern California, August 1980.
Canto, V. and M. Miles. "The Missing Equation: An Alternative
Interpretation," Journal of Macroeconomics, Vol. 3, No. 2,
Spring 1981.
Canto, V., A. Laffer, and 0. Odogwu. "The Output and
Employment Effects of Fiscal Policy in a Classical Model."
University of Southern California, Working Paper, I 978.
Cooter, R. "Optimal Tax Schedules and Rates: Mirrlees and
Ramsey." American Economic Review, 68 (December 1978):
756-68.
Evans, M. "Taxes, Inflation, and the Rich." The Wall Street
Journal, August 7, 1978, p. 10.
Fowler, Henry J. "Statement before the Committee on Banking
and Currency." Meetings with Department and Agency Officials:
Hearings before the Committee on Banking and Currency,
House of Representatives. Washington: U.S. Government
Printing Office, 1967.
32 / TAX RATEs
RATES AND
AND PRODUCTION
PRODUCTION
Fullerton, Don.
Don. “On
"On the Possibility of
of an Inverse Relationship
Revenues.” NBER
NBER Working
Between Tax Rates and Government Revenues."
Paper No. 467, April 1980.
1980.
Harberger, A.C. Taxation and Welfare. Little Brown and Company,
Boston 1974.
“The Kemp-Roth-Laffer Free
Heller, Walter. "The
Free Lunch.”
Lunch." The Wall
Street Journal, July 12,
12, 1978, p. 20.
“Government Fiscal Policy
Joines, D.
D. H. "Government
Policy and Private Capital
Formation."
Dissertation, University of Chicago, 1979.
Formation.” Ph.D. Dissertation,
1979.
Kemp, Jack. "The
“The Roth-Kemp Tax Reduction Act of
1977 Parallels
of 1977
the Kennedy Tax Reductions of the Early Sixties."
Sixties.” Congressional
Record, July 14, 1977: H7156-58.
87156-58.
Lucas, R. “Econometric
"Econometric Policy Evaluation: A Critique.”
Critique." In
K. Brunner and
and A. Meltzer, eds. Phillips Curve and Labor
Markets. North Holland, 1976.
1976.
Mirrlees, J. "An
“An Exploration
Exploration in the Theory of Optimum Income
Taxation.” Review of Economic Studies 38 (April 1971):
197111:
Taxation."
175-208.
Nelson,
Nelson, C. R. Applied Time Series Analysis for Managerial
Forecasting. San Francisco: Holden Day, 1973.
1973.
“The Predictive Performance
______ . "The
Performance of the FRB-MITFRB-MITEconomy.” American
PENN Model of the U.S. Economy."
American Economic
Review 62 (October 1972): 902-17.
____________
Pechman, J.
J. "The
“The Individual
individual Income
Tax Provisions
Pechman,
Income Tax
Provisions of
of the
the
Revenue
Act
of
1964.”
Journal
of
Finance
20
(May
1965):
Reve@e Act of 1964." Journal of Finance 20 (May 1965):
247-72.
247-72.
Terence J.
of aa Labor
Curve for
for SelfSelf"Estimation of
Labor Supply
Supply Curve
Wales, Terence
J. “Estimation
Proprietors.” International Economic Review
Employed Business
Business Proprietors."
14 (February 1973): 69-80.
Zellner,
“Time Series Analysis and Simultaneous
Zellner, A. and F.
F. Palm. "Time
Models." Journal of Econometrics 22 (1974): 17-59.
Equation Models.”
An Econometric Model Incorporating
The Supply-Side Effects of
Economic Policy
MICHAEL K. EVANS
This paper summarizes the principal findings of the new
macroeconomic supply-side model which I have recently completed
at Evans Economics. Rather than describe each individual equation
or even blocks of equations, I have selected an alternative
approach. Since the main thrust of the supply-side model is to
examine the ways in which total productive capacity can be
increased, I first examine the determinants of productivity, and
then show how these determinants are estimated within the confines
of the model. The bulk of this paper is devoted to the discussion of
the productivity function, the investment functions, and the labor
market functions. The concluding section then examines some
alternative solutions generated by changes in monetary and fiscal
policies. Rather than examine the usual full-scale multiplier tables,
I have chosen to concentrate on a specific set of policy alternatives
which should be able to increase productive capacity and
employment while at the same time reducing inflation.
DETERMINANTS OF PRODUCTIVITY
As part of the supply-side model, we have estimated an econometric
equation to explain changes in productivity on an endogenous basis.
Previous attempts to explain productivity reached the conclusion
that while some of the decline could be tied to the reduction in the
investment ratio and other endogenous factors, part of it could not
be explained by economic variables. However, we have found that
not to be the case.
The function we have estimated relates the annual percentage
change in productivity to two sets of variables: short-term cyclical
variables and long-term secular factors. The short-term variables
are a) percentage change in real GNP, and b) a nonlinear term of
capacity utilization which takes the form (95 - CP)'ic. Essentially
this term represents the fact that productivity growth slows down as
Mirhael Evans is Presiden1 of Evans Economics, Inc., Washing!on, D.C.
33
34 // SUPPLY-SIDE
5UPPLY~5tDE ECONOMETRIC
ECONOMETRIC MODEL
34
the economy approaches
approaches full employment and full capacity
capacity because
of shortages and bottlenecks, more overtime and hence more
worker errors, and hiring of less skilled and trained workers.
The long-term
long-term secular factors which we consider, together with
with
the weights which we have assigned to each of them, are as follows:
I. Decline in the investment ratio
1.
2. Costs of government regulation
3. Increase in secondary workers in the
the
labor force
4. Increase in relative price of
of energy
energy
5. Reduction in ratio of R&D
expenditures to GNP
1%
1%
½o/o
½%
#I;
(included in #1;
not measured
separately)
in the function
While the last factor was not explicitly included in
because of the very long lag times involved, it enters the function
function
indirectly through its eventual effect on investment. This point is
discussed in
in more
detail in
next section.
discussed
more detail
in the
the next
section.
The actual equation used in our supply-side model isis as follows:
Independent
Variable
Variable
-CSECWKO1
SECWK0I
INVXCOI
INVXC0I
REG
ENERGYC
GNP72
CAPUTIL
Estimated
Coefficient
Coefficient
Standard
Error
T-Statistic
T ~Statistic
Contribution
Contribution
To H’
R2
—7.51592
-7.51592
—0.839850
-0.839850
0.625840
—0.208791
-0.208791
—4.11652
-4.11652
0.524536
1.11549
1.11549
4.57449
0.35581
0.355811I
0.419031
0.419031
0.170205
2.49016
2.49016
0.108446
0.108446
0.440452
0.440452
-1.64301
—1.64301
-2.36038
—2.36038
1.49354
- 1.22671
—1.22671
-—1.65311
1.6531 I
4.83686
2.53261
0.6983770-01
0.698377D-01
0.379613D-Ol
0.379613D-0I
0.188628D-0l
0.188628D-0I
0.3425540-01
0.342554D-0l
0.293259
0,804009D-01
0.804009D-0I
= 0.7368
R-Squared =
0. 7368
R-Squared (Corrected) = 0.6616
= 0.2122
Multicollinearity Effect =
Durbin-Watson Statistic
Statistic =
= 1.3901
1,3901
Durbin-Watson
Number of Observations =
= 28
Number
28
Sum of Squared Residuals == 17.7071
Standard Error
of the
the Regression
Regression = 0.918256
Standard
Error of
0.918256
=
The dependent variable is:
PRDT
PRDT
=
i\PRD
=
41±RP
PRD,
PRO,
where PRD
= Private
Private nonfarm
business productivity.
productivity.
where
PRD =
non farm business
EVANS/35
EVAN5 / 35
The independent variables are:
22
= IJ_ ~:l: SECWORK_;
SECWORKS
SECWK0I
SECWKOI
=
22.0
i=0
where SECWORK
INVXC0I
INVXCO1
I
=
2 .
2.
2
:l:
X
=
Secondary workers
Total employment
INVXC;
INVXC~
I
where
INVXC
where INVXC
ENERGYC
cars and trucks
Gross National Product
= A11 ( PWIFP )
=
PGNP
PGNP
where PWIFP
PGNP
PGNP
GNP72
GNP72
=
Business Fixed
Fixed Investment
=
Business
Investment less
less investment
investment in
in
=
=
Producer Price Index, fuel and power
Implicit Deflator, Gross National Product
=
LiGNP
= AGNP
GNP_,
GNR
where GNP
= Gross National Product, billions of 1972
=
l 972 dollars
= (95 - CP) 112
where CP =
= Index of Capacity Utilization, manufacturing
CAPUTJL
CAPUTIL
=
—
to GNP has
At first glance, the ratio of fixed business investment to
remained roughly constant over the postwar period and in fact
posted an above-average value for
for 1979. However, this ratio is
misleading and must be adjusted for several factors.
calculated in constant rather than
First, the ratio should be calculated
current dollars. Just
Just because the price of capital goods has
increased faster than other prices does not mean that we are
devoting more of our resources
resources to capital formation.
Second, the investment figure should exclude capital spending
undertaken to meet
federally-mandated standards. The only figures
meet federally-mandated
available in this category are those for pollution abatement and
control, so our estimate obviously understates total capital spending
in this area. However, removal of these figures makes a noticeable
difference to the investment ratio.
36 / SUPPLY-SIDE ECONOMETRIC MODEL
FIGURE I
RATIO OF FIXED INVESTMENT TO GNP
11.0
10.5
10.0
I
-
\
I
I
9.5
6.5
-
,
\.
I
--
-
7,0
I'
I II
1
R
A 9.0
T
I 8.5
0
8.0
7.5
I
\ '
~
- ~~ -
.,._., ~ -
CL'.FiJi.C'lt POl i.AR~
{ ON'HANT DOI LAR:,
I
Jr,.:\.!..STMf-:-..;· !l-:',~ 1-'();J LTJO!\o- "-f:IATl.'-H:",I t-;XPfNDIH:R}!,., ("ON'i-TA~l l){)\ 1 AFCS
_ ._______,_,. _ _ l~\"f:';.7Mf:S:1 f 1.:S5 P(ll U. not-, AB.>\ n-.'\ff~T LXPI" Nlnn:RE'.1 A"ifJ ,'WIO ;\:,JD
Ll(i.}-1) F~U(K f\:.E'"°;,iD/Tl:lU:$, CON~TM-.1 Dl:)[LAR';
6.0
1956
!960
1964
1968
1972
1976
1980
YEAR
Third, "investment" in cars and light trucks should be excluded
from the total investment figures. Most of these purchases are made
for personal or quasi-business reasons, and do not represent
investment in the traditional sense.
We have adjusted the investment ratio for all of these factors,
and the very considerable difference which it makes is shown in
Figure 1. Thus although the nominal ratio may not have declined,
the real ratio of capital spending to GNP properly adjusted exhibits
a striking demise for the past five years.
Our productivity equation suggests that a 1OJo increase in the
investment ratio, or a switch of about $25 billion (in I 980 dollars)
from consumption to investment would raise productivity by about
0.611/o per year and thus lower inflation by about twice that amount.
We defer discussion of the ways in which this could be
accomplished until the next section, turning now to the other
principal determinants of productivity.
The second factor causing reduced growth in productivity,
namely increased investment to meet federally-mandated standards,
is summarized in Table I. This table should also include investment
TABLE 1
Fixed Investment and Capital Stock
Adjustment for Inflation and Pollution Control Equipment
Year
Fixed
Business
Investment
{Current$)
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980E
100.5
104. l
ll6.8
136.0
150.6
150.2
164.6
190.4
221.1
254.9
264.2
•Pollution
Control
2.2
2.9
4.1
5.3
5.8
6.5
6.8
7.5
6.9
7. t
7.7
"Health
and
Safety
Productive
Fixed
Business
Investment
(Current $2
l.7*
1.8*
2.5
2.6
3.1
2.7
2.4
2.9
4.3
2.9
3.7
96.6
99.4
110.2
128.1
141.7
141
155.4
180
209.9
244.9
252.8
Productive
Fixed
Business
Investment
~Constant $2
105.8
103.2
110.2
123.4
122.7
106.6
112.2
122.9
143.3
155.3
147.6
•June, 1980 Survey of Current Business.
bAnnual Survey of Investment in Employee Safety and Health, McGraw-Hill Publications Company, 1980.
cAugust, 1979 Survey of Current Business.
*Extrapolated by Evans Economics, Inc.
All figures are in billions of dollars.
cNet
Capital
Stock
{Constant $}
833.7
859.5
889.8
929.5
965. l
981.2
1000.8
1029.0
1060.2
1089.3*
1110.7*
Net
Productive
Capital
Stock
(Constant $2
830.0
851.4
875.8
908.5
936.7
944.9
956.1
973.7
993.7
1024.8
1044.3
MODEL
38 // SUPPLY.SIDE
SUPPLY-SIDE ECONOMETRIC
ECONOMETRIC: MODEL
undertaken by the automobile industry to meet pollution, fuel
economy, and safety standards, but
but we were unable to find
find even
approximate
estimates
for
these
figures.
Even
without
them,
approximate
however, we note that adjusted capital stock has grown at an
annual rate of only 2.4% since 1970,
I 970, compared to 3.0% as
calculated from the investment figures
figures before adjustment.
Because pollution control costs represent
share of nonrepresent the lion’s
lion's share
nonproductive investment, we have presented them in greater detail in
Table 2. As shown there, investment in private sector pollution
control for stationary source emissions (i.e., excluding motor
vehicles) will average about 4% of investment over the 1973-1984
period. Public sector spending for pollution control will average
between 15%
20% of total public sector investment, while
150/o and
and 200/o
pollution control devices will represent about 10%
100/o of the cost of
purchasing aa new car.
costs associated with pollution control
We also repeat
repeat the annual costs
investment; they are defined to include interest, depreciation, and
operation and maintenance costs. According to Council on
Environmental Quality (CEQ) estimates, the total annual costs for
Environmental
the 1975-1984 period will be $486
$486 billion in 1975 dollars, or
approximately $750 billion in current dollars. These costs will
amount to between 2%
ONP during the
20/o and 3%
30/o of total GNP
forthcoming decade, representing a very significant
significant economic
and solid waste.
burden for the costs of clean air, water, and
additional comments should be appended to these
these figures.
Two additional
First, the cost of regulation appearing in the government budgets is
private sector of the
only a tiny fraction of the cost imposed on the private
economy; Murray Wiedenbaum and others have estimated that it is
50/o. Second, while pollution abatement probably does
only about 5%.
represent
represent the lion’s
lion's share of these costs, the burden of occupational
safety and health standards, consumer product safety, toxic
control act, and
and other programs are substantial and
substances control
should not be assumed to be zero just because no definite figures
are available for these categories,
categories.
We do not think itit is reasonable to expect
expect society to turn
turn back
the clock on the
the massive changes in social policy which produced
the federally-mandated standards of the 1970s. Yet it certainly
certainly
should at least be possible to rationalize these regulations so that
firms are charged with attaining the ends rather than the means. If,
for example,
example, one national goal is to reduce air pollution, utilities
ought to be able to decide on their
their own whether this is to be
accomplished through
through choice of fuel, use of scrubbers, less
TABLE 2
Total Actual and Expected Investment for Pollution Control, 1970-1984
(I)
(2)
(3)
Capital Investment
Stationary Source
Mobile
Source•
Private
Public
0
Year
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
2.2
2.9
4.1
5.3
5.8
6.5
6.8
7.5
8.9
11.0
11.7
12.2
13.6
15.0
16.5
0.1
0.1
0.2
0.5
3.7
6.6
8.0
6.0
6.7
7.0
7.4
7.8
8.2
8.6
9.1
0.3
0.4
0.4
I.I
1.2
2.3
2.9
3.5
5.6
6.3
6.6
7.2
7.8
8.4
8.9
(4)
(5)
Annual Costs••
"Stationary Source
Private
Public
I.I
1.7
2.4
3.5
5.4
8.2
11.4
15.3
20.6
25.3
31.0
37.7
45.2
53.3
62.9
0.0
0.0
0.1
0.3
1.4
3.3
7.4
10.3
14.2
17.2
20.5
25.0
29.3
37.4
42.9
(6)
(7)
(10)
Mobile
Source•
(8)
Pollution
Control
Investment
(Percent)
(9)
Total
Fixed
Business
Investment
Total
GNP
Pollution
Control
Resources
(Percent)
1.3
2.0
2.8
4.1
5.3
5.7
6.0
6.4
8.2
11.3
100.5
2.2
982
l04.1
116.8
136. ()
150.6
150.2
164.6
190.4
221.1
242. l
2.8
3.5
3.9
3.9
4.3
4.1
3.9
4.0
4.5
1063
1171
1307
1413
1529
1700
1887
2104
2281
0.2
0.3
0.5
0.6
0.9
1.1
1.5
1.7
2.0
2.4
12.1
12.2
12. l
11. 7
11.3
262.7
299.7
337.5
376.6
417.8
4.4
4.1
4.0
4.0
4.0
2479
2730
2980
3256
3551
2.6
2.7
2.9
3 .1
3.3
Source: Figures are interpolated from ten-year to,als given in the CEQ Annual Report. All figures are converted from constant to current
dollars. Numbers are based on total rather than incremental polJution control expenditures.
**lnlcrcsl, Depreciation, Operation. and Maintenance CosL, of Po!lu1.ion control.
"Air, water, and solid waste, excludes motor vehicle,
*Includes additional fuel costs, motor vehicle\
(8) = (J) / (7)
S U p p L Y - 5 I D P FECONOMETRIC
C 0 N0 M FT R C M
0 0F
40 / SUPPLY-SIDE
MODEL
production during "air
alerts," building plants in new locations,
“air alerts,”
and so forth,
best guess is
forth, rather than by administrative fiat. Our best
that the use of common sense in these areas could reduce the loss in
½
¾per
productivity growth due to regulation
regulation from 1¾to
l "lo to ½
"lo per year,
tVo per
thus reducing the
per year. If
the overall rate of inflation by about II "lo
$50 billion
in addition this reduction from $100 billion to $50
billion per year
would
would
would free resources for increased
increased capital spending,
spending, the
the gains would
be even larger.
The third factor which has accounted for the
the slowdown in
productivity growth, although it will be reversed during
during the 1980s,
is the sharp growth of secondary workers in the labor force. In
460/o of
of the
the total labor
1964, males aged 25 to 54 accounted for 46¾
force; in 1980
1980 the figure will be 38¾.The
380/o. The major increases have
54 and in teenagers of both sexes.
occurred in women aged 25 to 54
The problem has been compounded not only by rapid increases in
force participation rates but in the population aged under 25.
labor force
Many of these secondary workers have less education, vocational
training, or on-the-job experience than their primary counterparts
when
first hired.
As aa result,
result, they
initially less
less productive.
productive.
when first
hired. As
they were
were initially
This does not necessarily imply that such individuals will continue
to have a lower level of productivity over the lifetime of their jobs,
but rather that their productivity was lower when they initially
entered the labor market.
During the 1980s, however, the size of the population aged 16
16 to
24 will shrink by a full 66 million persons. Thus even if labor force
rates continue to rise for teenage workers, the number of potential
potential
rates
25
employees will decline significantly. Second, many women aged 25
to 54 in the labor force will have had the full complement of
education, vocational training, and on-the-job experience as their
male counterparts, so they
they will be just as productive.
productive. As a result,
we look for
for this factor to improve, hence raising the growth rate of
productivity for the 1980s by about ½¾
½ "lo per
per year.
The fourth factor retarding productivity, the skyrocketing cost of
of
energy, is only too well known to anyone associated with
with the
the utility
industry, but the increase as shown in Figure 2 isis striking
nonetheless. Furthermore, we find little if any reason to expect this
ratio
course over
10 years.
In the
U.S.,
ratio to
to reverse
reverse course
over the
the next
next IO
years. In
the U.S.,
consumption of petroleum products remains at a high level,
although not as much as previously, and production is stagnant.
Under these two sets
sets of circumstances it is clear that the long-run
trend for oil imports continues
continu_es in the upward direction, which gives
need to continue to raise
OPEC all the economic justification they need
prices in
real terms.
terms. In
this respect
respect itit is
is noteworthy
noteworthy that
OPEC was
was
prices
in real
In this
that OPEC
EVANS
EV ANS // 41
41
FIGURE 22
RATIO OF
OF PPI:
PP!, FUEL TO TOTAL
TOTAL PP!
PP1
I
220 t=:;:=;:::::::;::::::::;:::::;:::::;:=;:::::::;::::::::;:::::;:::::;:=;:::::::;::::::::;:::::;::::;::::l
I
I
I
I
I
I
210
—
200
—
190
—
180
ISO
R 170 —
~I70
A
TI6O—
Tl60
I
0 150
0150—
140
—
130
130
—
120
—
110
I 10
—
—
100100
90
II......- ~......- ~......
liii
1111111
- ~......- ~......
- ~......- ~......- ~......
1952
1952
1956
1960
1964
1%4
1968
1968
1972
1972
1976
1976
1980
1980
YEAR
price increase in June in spite of
of
able to push through yet another price
the fact that the U.S.
U.S. is definitely in the midst of a fairly serious
the
the world economy is also slowing
recession and the rest of the
significantly.
Jong-run effects
effects of energy prices on productivity are
The long-run
undoubtedly understated.
understated. Indeed, it has
has become increasingly
apparent that the long-term effects of changes in energy prices on
and
productivity are greater
greater than had been generally appreciated, and
larger than would be determined by empirical techniques which are
are
by nature restricted to
to the period since 1973. The productivity
productivity
equation which we have
model
have estimated in our supply-side model
indicates that
that the increase in energy
energy costs has lowered productivity
¾ per year. While that is probably
probably the appropriate
growth by ½ "lo
long-run figure is considerably greater.
figure for the short run, the long-run
greater.
of how higher energy costs reduce
The standard explanation of
manufacturing sector. With a
productivity is usually confined to the manufacturing
shift in relative prices, firms use
energy and more labor,
labor, raw
use less energy
materials, and capital. This shift is borne out by the increase in
SUPPLY-SIDE ECONOMETRIC
ECONOMETRiC MODEL
42 / SUPPLY-SIDE
MODEl,
employment throughout 1979 during a period of virtually stagnant
output, and while some of the excess workers are being disgorged
now that we
we are in a recession, the demand for labor still
still has
has
shifted to a higher plane,
plane.
This shift is an important change and one which cannot be
treated lightly.
lightly, Yet in
in the longer run it will
will probably
probably turn out to be
less important than the changes in
in productivity
productivity which affect the
the
transportation and
and distribution network. Some of these changes
changes are
are
already obvious, such
such as the 197475
decline
in
productivity
in
the
1974-75
transportation industry
industry when higher fuel prices led to lower speeds
speeds
by airlines (voluntary) and trucking (mandatory). However, these
short-run changes are already included in
in our measurements of the
½¾
½
% yearly decline. Here we consider
consider the longer term changes
brought about by higher energy
energy prices as they affect the entire
production and
and distribution
distribution system
system of the economy.
Let us first consider a world in which transportation and
distribution
distribution costs are negligible. If that
that were the case, the location
of manufacturing plants would be largely independent of markets
except
except for those products that gain weight or bulk during
during
manufacturing or those processes
processes which utilize large quantities of
raw materials.
materials. Most important, all plants would
would be large enough to
take full advantage of
of economies of
of scale. Hence there would be
relatively few plants in those industries where economies of scale
are significant, particularly metals, machinery, transportation
equipment,
equipment, and power
power generation. Competition would
would thrive
could not
because one firm could
not obtain an
an advantage merely by accident
of
be the only part
of location. The manufacturing sector would not be
part
of the economy
economy to benefit from this arrangement. Consumers
Consumers
would also benefit; they could comparison shop at several locations
locations
since the cost of a reasonable amount of travel to obtain better
prices would
would be small.
While transportation costs have always been
substantial portion
been a substantial
of the
the total price for some goods, such as cement,
cement, it is not
not too
farfetched to
to say
say that
that many elements of
of the
the economy
economy described
farfetched
described
1973. Indeed, itit should be clear in
above applied to
to the
the U.S.
U.S. before 1973.
general
productivity
general that cheap transportation and distribution aids productivity
and retards inflation.
inflation. ItIt encourages
encourages greater
greater efficiency
efficiency through
through
economies of scale in manufacturing, and it encourages greater
greater
competition through a wider range of choice in retail markets.
markets.
After all,
if consumers
consumers had
had no transportation and
were virtually
virtually
After
all, if
and were
forced to shop only at the closest store, the storekeeper would
would have
to cut costs through higher productivity.
far less incentive 10
Thus the higher cost of energy,
energy, through reducing the
tbe amount of
EVANS/
43
transportation utilized, raises prices by much more than the cost of
the more expensive fuel alone. Furthermore, this is not reflected in
higher profits; it is the deadweight loss of productivity which does
not benefit anyone. Manufacturing plants gradually become less
efficient, and retail outlets become less competitive and less
productive.
Obviously these events change only very slowly over time, which
is precisely why we cannot yet measure them very well. Existing
plants do not shrink when energy costs rise, alchough they may run
at lower rates of capacity utilization. Consumers do not change
their driving or living habits overnight, and so on. But over time
these gradual changes, almost imperceptible within the time frame
of a quarter or even a year, cumulate and eventually represent a
potent force affecting productivity.
Offsetting this to a certain degree is the fact that if capital
spending is stimulated during the 1980s, much of the new
investment may be used for energy-saving plant and equipment,
thus diminishing our dependence on imported oil. This would
eventually cause OPEC to reduce their price in real terms, hence
removing one of the major hurdles to higher productivity growth.
In other words, higher investment may have benefits far greater
than the traditional methods of raising productivity through
expanded capital stock; the new mix of capital stock may be more
energy-efficient as well, representing savings which would not come
about were new investment to proceed at a slower pace. However,
the entire relationship between energy prices and investment is a
very complicated one, wel! beyond the scope of this modest report.
The fifth factor which we believe influences the long-term growth
rate of productivity is the proportion of resources devoted to R&D
compared to GNP. As is shown in Figure 3, from a peak of 3%
reached in the mid-1960s at the height of the space program, this
ratio has declined to slightly over 2% in 1976, although it has
recently improved as private industry has stepped up its R&D
spending. The long lags between R&D spending and productivity
growth, which average up to five years, mean that this relationship
is not quite as precise as the other factors determining productivity.
However, as discussed in the next section, it is thought to have an
effect on investment, albeit with this very long lag.
To summarize this section, output/manhour in the private sector
increased at an annual average rate of 3% for the period from 1948
to 1965, but has declined to almost 0% currently. Table 3 contains
the tabulation of the postwar record for increases in output/
manhour in the private non farm sector. We have taken three-year
44 /I SUPPL
5 U P P L Y - 5SIDE
I D F ECONOMETRIC
F C 0 N 0 M F T R I C MODEL
M 0 D E t~
44
0
FIGURE 3
OF GNP
R AND D SPENDING AS A PROPORTION OF
I
I
I
I
I
I
I
I
3.0
p
R 2.5
0
TOTAL
p
0
2.0
R
R2.0
TT
l
0
1,5
N l.5
1.0
1,0
—
FEDERALLY FUNDED
-.
-
-
-·-·-.
- ~.
—
-
—
-
INDUSTRY FUNDED
0.5
I
1960
1960
I
1964
1964
I
I
1968
1972
I
I
1976
1976
I
1980
YEAR
YEAR
averages rather than yearly figures in order
order to smooth
smooth out the
the
output.
fluctuations in productivity caused by sharp changes in output.
While
\\Fhile some traces of recessions still remain in these numbers, the
overall swings in productivity emerge much more
more clearly than is
is the
case in
in the series for annual changes.
As shown in Table 3, productivity
productivity rose
in the
the years
rose very
very rapidly
rapidly in
years
immediately following World War II (no figures are available
before 1948)
GNP devoted to
1948) because of the large proportion of GNP
to
equipment. Productivity
investment to replace obsolete plant and equipment.
increases then declined to
to the
the 2.0%
range for
for the period 1956-1961,
increases
2.0"7o range
1956-1961,
considerably below the long-term average. This was due in
in large
part to the severity of the 1958 recession. Productivity then rose
1962 to l1968,
rapidly from
from tlte
tl1e period 1962
968, due to the
the increase in capital
capital
spending spurred by
by the investment tax credit, liberalized
liberalized
depreciation allowances,
allowances, and
and the
the reduction
reduction in
the corporate
depreciation
in the
corporate income
income
by the substantial
tax rate; productivity gains were also
also increased by
increases in federal spending for research and development.
of these driving forces toward
Beginning in 1969,
l 969, both of
toward higher
growth were removed. The investment tax
cancelled, and
tax credit was cancelled,
and
TABLE 3
Long-Term Trends in Productivity Growth
Three-Year Period
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981-1990
Average Annual Growth Rate
in Productivity
(Private Nonfarm Sector)
4.2
4.0
3.5
2.2
2.0
2.4
1.6
1.6
1.7
2.8
2.3
2.4
2.7
3.5
3.5
3.0
3.1
2.4
2.4
1.3
1.5
2.1
2.3
2.9
0.8
0.2
0.8
2.3
1.9
0.3
-0.6
1.0
S U P P t. V - S I D E FECONOMETRJC
C 0 N 0 M F T R I C !\.10DEL
M0 DE E
46 // SUPPLY-SIDE
recurring
recurring financial crises reduced the amount of money
money available
available
for new investment
investment spending. The reinstatement of
of the investment
1971 did raise investment above the levels which would
would
tax credit in
in 1971
otherwise have been reached, but this was offset by the substantial
expenditures required
required for environmental and
and safety
safety standards. As a
result, productivity actually declined for the first time in the
the
postwar period in 1974 and for the three-year period 1973-1975
postwar
1973-1975
showed virtually no improvement. While the 1977-78
1977-78 figures
indicate a rebound, that was
was due mainly to cyclical factors, as
as
shown by
1979 and 1980.
1980.
by the subsequent slowdown
slowdown in
in 1979
The
980s clearly depends on
The growth rate of productivity in the I1980s
what happens to the factors we
we enumerated at the beginning of this
this
section.
DETERMINANTS
DETERMINANTS OF INVESTMENT
INVESTMENT
It is generally agreed that an increase in the production of
spending will raise productivity, hence
resources devoted to capital spending
increasing real growth and lowering
lowering inflation. However, less
agreement exists concerning
concerning the determinants of investment.
generally divided into two groups:
groups: those who
Economists are generally
believe in the “trickle-down”
"trickle-down" theory, and those who claim that the
primary variable
variable is expected
expected rate of
of return.
The trickle-down theory
theory states that a rise in consumption is
sufficient to increase investment to the desired
desired level. Once the
demand for goods increases, businessmen, ever alert and eager for
increased opportunities,
opportunities, will expand capacity sufficiently to create
the productive capacity for these new goods. In somewhat
somewhat
oversimplified
oversimplified terms,
terms, demand creates its own supply.
return theorists would argue that no such
such automatic
The rate of return
mechanism exists to equilibrate demand and
and supply. Capital
spending will not increase
increase unless the expected rate of return is
sufficient
to cover
cost of
investment. To
To be
be sure,
an increase
increase in
in
sufficient to
cover the
the cost
of investment.
sure, an
demand does raise the rate
rate of return, other things being equal-but
equal—but
it does not
not in and of itself guarantee an adequate rate of return.
return.
Thus
the
tax mechanism must be used to insure that demand and
and
Thus
has
supply are kept in
in balance. Obviously the choice of theory has
tremendous implications
in determining
determining the
the appropriate
appropriate tax
tax policies
tremendous
implications in
policies
to stimulate growth and productivity.
productivity.
The investment functions which we have estimated in the
the Evans
Evans
Economics macro model
on the cost of capital-rate of
model rely heavily on
return variable originally introduced
introduced by
by Jorgenson. However, the
than
approach which we have used permits much greater flexibility than
EV ANS /
47
his original construction. By using a two-step procedure in which
we estimate equations for orders and investment separately, we are
able to measure the separate contributions for a change in the
corporate income tax rate, investment tax credit, and depreciation
allowances. Furthermore, since the index of stock prices is included
as one of the variables in the rental cost of capital term, we can
also examine how changes in the capital gains tax rate wiU affect
investment.
We can summarize the results here by listing the impact effects of
changes in these tax laws. By impact effects we mean simply the
marginal coefficients times the change in the tax law in quesHon,
These coefficients do not take into account the interactive and
dynamic effects, for which we need to solve the entire model, but
they do give some idea of both the absolute and relative importance
of each type of tax change.
Our results in the supply-side model have shown that, for the
same revenue-producing change, the corporate income tax rate cut
has greater efficacy than a change in depreciation allowances, which
in turn has a greater effect than a change in the investment tax
credit. Furthermore, a change in the stock prices has a substantially
greater effect than a proportional change in interest rates. Since
these findings are not universally accepted, a further word of
explanation is in order.
We have found that the corporate income tax cut has the highest
efficacy because it is a "pure" tax cut; it does not contain any of
the restrictions that the other types of tax changes contain. For
example, an investment tax credit can be used only for equipment,
but not for plant; a certain amount of the credit must be carried
over into future years and in certain circumstances companies
cannot use all the credit, which means they must find other
investors who use the credjt as a tax shelter. Jn addition, at least
until recently many investors believed that the investment tax credit
was a "gimmick" to be suspended or terminated at will by
Congress, and hence they were less .viUing to use it as a basis for
long-term investment planning,
While we do think that these three changes in the tax law will
have somewhat differing effects on investment, it should be stressed
that all of them will have a significantly positive response. Indeed,
the post-war history of capital spending in the U.S. economy is
largely tied to changes in the effective tax rate on corporate income.
The relationship between changes in capital spending (in constant
prices) and the effective corporate income tax rate lagged one year
is given in Figure 4.
48 / SUPPLY-SIDE
ECONOMETRIC MODEL
SUPPLY-SIDE ECONOMETRIC
FIGURE 44
REAL PRODUCTIVE INVESTMENT V EFFECTIVE TAX RATE
20
II
I
I
II
I
II
I
I
15
‘5
p
P
E
£
R
C
E
£
N
T
T
JO
10
—
55
00
C
H
A —5
-5
N
0
G
E -JO
- 15
—IS
'',,
REAL tNVESTMENT -—
REAL!NVESTMENT
-,
EFFECTIVE TAX RATE • - - .
I
—20
-20
1956
1960
1964
1968
I
I
1972
1972
I
1976
1980
1980
YEAR
YEAR
To summarize the information given in Figure 4, the U.S.
economy has undergone three investment booms in the postwar
period: 1955-1956, 1964-1966, and 1972-1973. Each of
of these booms
has a common characteristic:
characteristic: it was preceded in the previous year
by a major change in the tax code which was favorable to
investment. Hence 1954 marked the end of the
the excess profits
profits tax
from the Korean War and the first liberalization of depreciation
7"1o rate
allowances. The investment
investment tax credit was
was introduced at a 7%
20% reduction in
in late I1962
962 and was accompanied by a 20"/o
accounting tax lives; when this was followed by a reduction in the
48% in 1964, capital
corporate income tax rate from 52%
520/o to 480/o
200/o in constant prices in 1965, the
the only time in
spending climbed 20%
the postwar period that has occurred. Finally, in 1972 the
7% and accounting tax lives
investment tax credit was reinstated at 70/o
200/o.
were reduced by an additional 20%.
We also note that
that the sharp increase in tax rates in 1969, caused
10% income tax surtax and the suspension
by the imposition of the 100/o
of the investment tax credit, was sufficient to cause a decline in
in
EV
ANS / 49
FIGURE 5
PRODUCTIVE INVESTMENT AND COST RATIOS
:,
R
E
"r
I
"VE
s
r
v1
E
'I
r
I
d
'I
p
9.8
9.6
9.4
9.2
9.0
8.8
8.6
8.4
8.2
8.0
7.8
7.6
7.4
7.2
7.0
6.8
6.6
6.4
1.4
l.3
R
A
T
1.2
0
1.1
T
l
s
l.O
0.9
0
C
K
p
R
I
C
E
PRODUCTIVE INVESTMENT /REAL GNP - ,
0.8
STOCK PRICE/CONSTRUCTION COST-----
0.7
C
0
0.6
T
I
0.5
1960
1964
1972
1968
1976
1980
YEAR
investment in I 970 even though the economy was still operating at
high utilization rates.
However, the correlation between changes in investment and
changes in the effective corporate income tax rate is not perfect. In
particular, the sharp declines in investment in 1958 and 1975 appear
to be unrelated to changes in the tax code, and were indeed caused
by the severe recessions which occurred in those years.
This anomaly disappears when we correlate the investment ratio
and the ratio of stock prices to construction costs, lagged one year.
As shown in Figure 5, this ratio captures both the cyclical and
secular movements in the investment ratio. This fact has received
bipartisan support, as it was prominently discussed in both the 1977
and 1978 issues of the Economic Report of the President.
The theory behind this ratio is fairly straightforward. When stock
prices are high relative to construction costs and equity capital is
relatively inexpensive, businesses will expand by building new plants
and filling them with new equipment. However, when stock prices
are relatively depressed, businesses will expand by buying smaller
s
SO
50 I/
SUPPLY-SIDE ECONOMETRIC
SUPPLY-SIDE
ECONOMETRIC MODEL
FIGURE 66
1960
=
100
100
OUTPUT PER MANHOUR FOR MAJOR COUNTRIES
450
Japatl
414 --
378
378
342
342
Belgium
/Netherlands
306
306
-
270
270
;
- _/
,/"
~t~ice
Germany
234
Canada
198
198
162
162
126
----- - .,, .,,~~:~-~-------~-:
- ..----: •
, .....- _ .....-
__..-
Great Britain
- • -~nited
------States
90 '--''---'---'--....l.--.....j._ _'---'---'---'------'--'--'
1968
1976
1976
1960
1964
1980
1972
YEAR
YEAR
capital
existing businesses, rather than
than by investing more in new capital
assets. The course of the stock
stock market
market is thus of extreme
importance in determining the growth in investment, and explains
why this term is relatively more important than the interest rate.
We can never be absolutely positive that the slowdown in
I 966 was due to the reduced rate of growth in
productivity after 1966
investment. However,
However, additional supporting evidence can be
and growth patterns of
of the
the
gathered by
by examining the investment and
U.S. economy with those of other
other leading
leading industrialized countries
of the world. These comparisons are provided in the next two
graphs. In Figure 6 we find almost a perfect correlation between the
proportion of GNP spent on fixed investment and the growth in
productivity.
the extent
to which
in
productivity. Figure
Figure 77 documents
documents the
extent to
which increases
increases in
output/manhour in the U.S. have fallen behind growth in the rest
of the
world. Even
when one
one adjusts
for lower
wage gains
in
of
the world.
Even when
adjusts these
these for
lower wage
gains in
the evidence explaining the weakness of the dollar
dollar
this country, the
seems compelling.
compelling.
seems
It often comes as a shock to realize that in the past 15
15 years the
GNP going to
proportion of GNP
to fixed business investment and the rate
51
EV ANS /
FIGURE 7
p
E
R
C lO
E
N
T 9
PRODUCTIVITY GROWTH V. INVESTMENT, BY COUNTRY
I
-
A
N
G
7
-
E
6
I
5
f-
4
f-
3
-
D
u
C
T
I
V
I
T
y
* Netherlands
-
-
-
ltaly
"
* France
* Germany
-
* Canada
l-t
~
-
*
-
Great Britain ·
United States
-
0
l
Belgium~
¼t
2
I
-
8
p
R
0
I
Japan
C
H
N
I
l
(1965-1979)
I
I
!
I
10
15
20
25
I
30
INVESTMENT AS A PERCENTAGE OF REAL GNP
I
35
of increase· in productivity for the United States are below even
those of the United Kingdom. It is this below-par performance
which has been at the root of the weakness of the dollar since 1970.
The oil embargo and subsequent quintupling of OPEC oil prices
may result in some relative shift in these relationships during the
next decade. As shown in Figure 7, productivity declined in Japan
and all major European countries except Germany during 1974, the
first time this has occurred in the entire postwar period.
Furthermore, wage gains in Europe and Japan have been well
above increases in the U.S.; if this continues and is not offset by
continuing relative increases in productivity, these areas could lose
much of their allure for investors.
So far we have been discussing plans to stimulate investment
directly through lower taxes. However, investment can also be
stimulated indirectly, namely by increasing personal saving. A
decline in the tax rate on income generated from saving-such as
interest and dividend income-would result in more personal
saving, and eventually more investment.
52 / SUPPLY-SIDE
MODEL
SUPPLY-SIDE ECONOMETRIC
ECONOMETRIC MODEL
previous empirical work on
on the
the consumption
The vast majority of previous
implies that the interest rate has no significant effect on
function implies
the proportion of
of disposable
disposable income which
which is consumed or saved. It
is true that a simple correlation between
between the saving rate and the
the
interest rate reveals no relationship. However, we have
have found aa very
strong link between the real after-tax rate of
of return and personal
personal
saving. After substantial testing, we have determined
determined that this rate
can best be represented by
by the long-term bond yield
yield multiplied by
by
average tax rate on personal income) minus the
(1
the average rate
this rate of
of inflation over the past four years. Thus defined, this
return isis found to have an important effect on consumption
return
consumption and
saving. Specifically,
Specifically, a 1IWo
% increase in the rate of return—e.g.,
return-e.g., from
3%
4%-would raise saving
saving by $12 billion.
billion. Furthermore, we
3% to 4%—would
find that the importance
importance of the after-tax rate of
of return on savings
has been increasing
increasing in recent years as interest rates and inflation
to higher levels.
move to
An across-the-board $10 billion personal income tax
tax cut from,
30% to 29% would have relatively little
say 30%
little effect on saving over
over
although as
and above the increase stemming from higher income, although
we
note later
would have
on labor
market
we note
later itit would
have aa much
much larger
larger effect
effect on
labor market
behavior. However, the increase
increase in saving from this tax cut
cut due
due to
the increased rate of return would be only about $1 billion.
billion. On the
other hand,
cut of
of the
the same
targeted only
other
hand, aa tax
tax cut
same size
size which
which was
was targeted
only to
to
increase saving
saving through a higher
higher rate of return would
would result in aa rise
in saving of some
some $13 billion. Thus the form of the tax cut is allimportant in determining the effect on consumption and saving.
We now consider some of the ways in which saving and
investment are stimulated in the
the high-productivity simulation
simulation
this
calculated for th
is report.
As mentioned
mentioned above, the
the simplest and most direct approach isis a
reduction in the corporate income tax rate. A decrease from the
Sb
present level of 46% to 40% would cost
cost the Treasury about $11
billion per
per year
year before
before reflows;
reflows; these
these figures obviously increase
increase over
over
time
as the economy expands and
time as
and profits rise in nominal terms. The
impact effect on
on investment
investment would
would be
be to raise it $9 billion
billion after the
lagged effects were fully considered. The multiplier effects are
discussed in more detail in the final section.
Changes in depreciation lives could take several different forms,
Changes
and in general the analysis is somewhat more complicated than for
for
the
major plans
the simple cut in corporate income
income taxes. The two
two major
plans
which have been suggested
suggested for changing depreciation
depreciation allowances are
(a) replacement cost accounting, and (b) shortening tax lives, which
—
EVANS/53
has recently been popularized as 10-5-3, although clearly other
variants of shorter lives are possible.
The theoretical justification to adjust depreciation allowances, in
addition to the fact that this would stimulate investment, is that
these allowances fall far short of replacement needs in a period of
inflation.
Since that is the avowed objective of all such plans, it seems most
reasonable to us to meet the ravages of inflation by an adjustment
which compensates for inflation. This would be accomplished very
simply as follows. Depreciation allowances would be set equal to
the appropriate proportion of investment for each year times the
ratio of capital goods prices in the present year to capital goods
prices in the year during which the investment was originally
undertaken. Symbolically this can be expressed by:
where
PkT
= price of capital goods (implicit deflater, national income
accounts, business fixed investment) in year T;
DT
SL-r
=
=
depreciation allowances in year T;
proportion of investment depreciated by the straight line
method in year T;
t-r = average length of depreciable life of assets depreciated by
the straight line method in year T;
ACCr = proportion of investment depreciated by accelerated
methods in year T; and
IT
=
investment in year T.
The drawback to replacement cost accounting, according to the
proponents of 10-5-3, is that it is too complicated. However, we
feel that such a plan could be implemented very simply by having
all depreciation allowances increase by the average rate of inflation
of capital goods, as published by the Bureau of Economic Analysis
(BEA). Some distinction could be made for equipment and
structures, but as a first order of approximation 900'/o of the
inequities caused by inflation would be wiped out by linking to one
or two overall indexes.
The reduction in the maximum capital gains tax rate from 49 .1 0/o
54 I SUPPLY,SIDE ECONOMETRIC MODEL
to 28% in late 1978 brought forth anguished cries from some critics
who claimed that it benefited the speculator rather than the longterm investor. While we believe that all capital gains taxes should
eventually be abolished, the remaining tax burden could be
restructured to benefit more directly those members of society who
contribute most to spending on R&D, venture capital, and
investment in new companies.
One plan to restructure the capital gains tax laws states that
anyone investing venture capital into a new or fledgling company
and then holding on to the stock for five years or more would not
have to pay any capital gains taxes at all. Furthermore, capital
gains would be calculated on an indexed basis, so that investors
would not have to pay tax on the phony profits which are due only
to inflation. For purposes of calculation, the implicit GNP deflator
or some other broad-based price index would be used.
In order to relate the relationship between stock market prices
and investment to tax policy, we must determine how much a
change in capital gains taxes will affect the stock market. Here we
have found a significant relationship, namely that a I% change in
the maximum tax rate on capital gains (i.e., from 48% to 47%)
would raise stock prices by approximately 1 ½ % . Hence one of the
most important ways to stimulate investment and productivity is to
reduce the capital gains tax rate further.
Although no specific figures are available, it is likely that the
reduction in capital gains taxes will also contribute to a renaissance
of the venture capital industry, which was approximately a $3
billion a year industry in 1968 before higher capital gains taxes and
the decline of the stock market combined virtually to wipe out this
industry. R&D spending has also been hampered by the lack of
venture capital, and while this does not show up immediately in
declining productivity, it is thought to have a very substantial effect
over a five to IO year period.
A number of plans have emerged to reduce the burden on the
individual saver, and although these are not0'as far advanced
through the Congressional labyrinth, they still merit some
discussion and inclusion in our model simulations.
The formation of Individual Retirement Accounts (IRAs) four
years ago permitted individuals not covered by pension plans to
invest $1,500 each year tax-free, providing the money was not
withdrawn before retirement age. Our planned Individual Saving
Account (ISAs) would have some elements in common with this
general idea, in that they would encourage savings, but the scope
EV ANS /
55
would be much more broad-based. Each taxpaying unit could treat
up to $1,500 per year in interest income, dividend income, or
capital gains rollover as tax-exempt income. Thus, for example, if
an individual had a savings account of $10,000 on which he earned
an average interest rate of 9% and dividend income of $1,000,
$1,500 of that $1,900 income would not be included in his gross
taxable income. The plan would have certain strictures; taxpayers
would have to keep thefr principal fully invested, although they
could switch assets just as is the case for IRAs now. Any capital
gains would have to be reinvested (rolled over) into other similar
investments in order for that part of the exemption to qualify.
However, the basic idea of an ISA would be that income generated
from stocks, bonds, savings accounts, money market funds, or
similar assets would be tax exempt as long as the principal
remained invested in this class of assets. We estimate that this
would cost about $6 billion per year in ex ante revenue loss.
Clearly the establishment of ISAs would have many advantages.
It would reduce the tax burden for savers, particularly smaller
savers, and thus would be politically as well as economically
popular with the vast majority of voters. It would stimulate savings
and investment, and would pull the U.S. closer to being able to
compete with other major industrialized nations in terms of gains in
investment and productivity.
The disadvantages which are likely to be raised are threefold.
First, such a move would definitely increase the size of the federal
budget deficit; no backward-bending supply curves would operate
here. Second, it could be argued that mosr of the tax break would
simply go to taxpayers who would save and invest in any case; i.e.,
it would attract very little new savings. Third, someone is sure to
complain that most of the tax breaks will go to the "rich", which
to a certain extent cannot be refuted because most of the poor
don't save.
These objections suggest an alternative plan which would affect
marginal savings more directly. Under this alternative, taxpayers
would not receive an exemption or credit unless their savings in any
given year were greater than the average savings rate for that
income bracket. For example, if the average savings rate was 5%
for a $30,000 per year income, taxpayers at that level would not
receive any exemption unless they saved over $1,500 in that year. It
is difficult to estimate the ex ante revenue loss, but it would
certainty be under $5 billion per year.
A third alternative plan would be to "start the tax table over"
56 / SUPPLY-SIDE ECONOMETRJC MODEL
for nonwage income. For example, if a taxpayer had $50,000 in
wages and salaries and $10,000 in interest and dividend income, the
nonwage income would be taxed at marginal rates applying to
$ I0,000 of income, rather than $60,000. Thus if a wealthy
individual had, say, $250,000 of interest and dividend income he
would still pay high marginal tax rates-although in this case the
top marginal bracket would be limited to 50%, just as it is for wage
income, rather than the current top level of 70%. Indeed, we
estimate that lowering the top tax bracket from 70% to 50% would
actually net the Treasury about $3 billion per year as taxpayers
would shift out of tax-sheltered or tax-exempt sources of income.
Other alternative plans are available as well. The original concept
of the IRA could be expanded to allow much more of a deduction
than $ l ,500. The principal and interest on money put aside to buy a
home could be declared tax-exempt. In any case, all these schemes
would have the net effect of reducing the net tax rate on saving. In
the model we have assumed that some combination of these
reductions would result in lowering the marginal tax rate on savings
from its current level of 40% to 30%, which would result in a net
loss of revenue to the Treasury of $8 billion per year before reflows.
As a result of these findings, we have also introduced some
personal income tax cuts in the high-growth simulations, and some
personal tax increases-mainly through bracket creep rather than
actual rate hikes-in the low-growth simulation. While the changes
in laws affecting investment behavior are the most important
movers of the differential rate of growth, we should not ignore the
effect of changes in personal income tax rates on labor force
participation, the amount of labor offered by those already in the
labor force, the level of productivity, and the increase in wage
rates. We now examine these relationships in greater detail.
RELAT!ONSH!P BETWEEN LABOR AND TAX RATES
The theoretical literature of microeconomics has always posited
significant relationships between the demand and supply of labor
and the cost of that labor, including tax rates. A tax on labor
(such as a social security tax) would raise the cost of this labor,
thereby reducing its use. Similarly, an increase in taxes would
reduce the supply of labor offered, although this effect is
sometimes thought to be offset by the so-called backward bending
supply curve. However, these linkages have been almost entirely
absent from previous macroeconomic models, even though
microeconomic studies, including several funded by the federal
EV A NS / 57
government, have shown significant elasticities for various
classifications of employees, particularly secondary workers in the
labor force.
In addition to the beneficial aspects of tax cuts on saving and
investment in our new macroeconomic model, we have also found
significant relationships between changes in personal income taxes
and labor market conditions. These can be subdivided into three
areas: labor force participation, amount and quality of work
offered, and increase in wage rates.
Typical macroeconomic labor supply functions have been
estimated in the form of labor force participation rates by
demographic composition, with the principal independent variable
being the lagged value of the unemployment rate. Both theory and
microeconomic results suggest that the real wage should be included
as an additional determinant of labor force participation. However,
on an empirical basis the problem of separating out the income and
substitution effects has proved baffling. In general we would expect
that an increase in the wage rate would have offsetting effects. The
higher wage would induce an increase in labor supply, following the
usual upward-sloping supply curve for factors of production.
However, an increase in income would result in substitution of
leisure for work, following the so-called backward-bending supply
curve. Furthermore, an increase in prices generally reduces the real
income of the wage earner, so that a higher rate of inflation would
draw more people into the labor force in an attempt to make ends
meet.
The major problem in estimating labor force participation rate
equations with the wage rate has always been the difficulty in
sorting out the difference between the substitution and income
effects, since they should have different signs. Furthermore, most
of the theoretical work has been done under assumptions which
assume constant prices, whereas in reality fluctuations in the real
wage due to inflation are among the major determinants of labor
force participation.
Let us first turn to the problem of the income and substitution
effects. Musgrave has suggested that this problem can be handled
by considering the average and the marginal tax rates separately.
He argues that work effort will decline if the marginal rate is raised
(substitution effect) but will increase if the average rate is raised
(income effect). From a theoretical point of view, therefore, the
problem is solved by entering both of these tax rates.
From an empirical point of view, however, it is perfectly obvious
58 I
SUPPLY-SIDE ECONOMETRIC MODEL
that these rates mo~e together over time, and that it is not possible
to measure the empirical effects separately on a time-series basis.
One way around this problem is to introduce an income term
together with the marginal tax rate in the labor force participation
rate equations. Thus we have included the wage bill deflated by the
CPI, thus incorporating elements of both the wage rate and
income. While not a perfect solution, this combined variable does
enable us to estimate more robust estimates of the effect of tax
rates on labor force participation, separate the average and
marginal tax rate effects, and include the theoretical desirable
income term.
Thus the key variable used in the labor force participation rate
equation is:
where: W
=
wage and salaries;
= consumer price index;
and
trn = marginal tax rate as calculated by Evans Economics,
Inc. (EEi).
CPI
We now turn to the distinction between primary and secondary
workers in the labor force. In general economists have found a
modest if not insignificant relationship between labor force
participation rates for males aged 25 to 54 and either the real wage
or the rate of inflation. On the other hand, we would expect both
of these variables to be quite significant for secondary workers in
the labor force.
We also need to consider the effect of changes in the marginal
tax rate on labor supply. Again one can raise the question of
whether the substitution or income effect dominates; as tax rates
rise, it could be argued, labor supply increases in order to hold real
income constant. However, the overwhelming evidence of the
microeconomic studies suggest that the substitution effect
predominates, and that an increase in tax rates reduces the supply
of labor offered. Thus we have combined the tax term with the real
wage term in all of these equations.
We thus expect the standardized labor force participation rate
equation to contain the following terms: the unemployment rate,
the wage bill divided by the price level, the marginal tax rate on
personal income, and the rate of inflation.
It is often claimed that the minimum wage has contributed to an
EV ANS /
59
increase in the unemployment rate among teenagers, since they are
the potential employees whose marginal product is most likely to
be lower than the minimum wage. While this is undoubtedly the
case, the relationship has another dimension, which is that the
existence of the minimum wage barrier also deters many teenagers
from entering the labor force in the first place. Thus we find a
significant negative correlation between labor force participation
rates for those aged 16 to 24 and the minimum wage in real terms.
A l OJo increase in the minimum wage wiU reduce labor force
participation by approximately 0.2%.
At the other end of the age spectrum, we find a very strong
negative correlation between social security benefits in real terms
and labor force participation for those 55 and older. Since the
benefits are tied to the cost-of-living and in fact are one of the very
few types of personal income to outstr1p inflation over the past
deeade, it is clear that an increase in the rate of inflation raises real
income for recipients, especially when it is considered that social
security benefits are tax-free whereas earned income is subject to
personal and social security taxes. Hence the situation for
retirement-age individuals is unlike the situation for the rest of the
work force, for whom an increase in inflation lowers real income
and thus leads to greater labor force participation. One might argue
that real income remains constant for those on social security, but
actually very few people over 55 are buying or financing new
homes, and hence the CPI increase clearly overstates the increase in
their cost of living. Also, those over 65 receive medical care free of
charge; hence those rapidly rising prices are also not indicative of
the costs faced by older citizens.
The empirical results for labor force participation are best
divided into primary and secondary members of the work force.
The effects on primary workers, defined here as males aged 25 to
54, are significant but small. A one percentage point (p.p.)
reduction in the marginal personal income tax rate would result in
only a 0.05% increase in the primary labor force. However, it
would result in a 0.37% increase in the secondary labor force.
However, total increase in the labor force caused by a l p.p.
reduction in the tax rate would be 0.26%, or approximately 270,000
workers at the present size of the labor force.
The labor force participation equations also indicate that a I%
increase in the real minimum wage {adjusted for inflation) would
decrease labor force participation for those aged 16 to 24 by 0.2%.
At the other end of the age scale, a l % increase in real per capital
60 /
SUPPLY-SIDE ECONOMETRIC MODEL
social security benefits would diminish labor force participation of
those 55 and over by 0.4%.
The equations relating the amount of utilized labor to output,
capital stock, and productivity are usually known as inverted
production functions or labor demand functions. However, they are
actually a reduced form of labor demand and supply equations,
since the amount of labor used depends both on the demand for
labor by business and the degree of willingness to offer their labor.
These combined effects are very significant. We find that a 1%
increase in the average personal income tax rate including social
security taxes will reduce the amount of labor utilized by 0.5%.
This decline is caused by several factors. First, an increase in the
cost of labor through higher social security taxes will reduce the
demand. Second, an increase in tax rates will reduce hours worked
per week; we find that this effect accounts for slightly over half of
the total reduction in labor offered. Third, higher taxes lead to a
rise in vacation time, absenteeism, and unwillingness even to work
at alI by some members of the labor force.
The results we have found on the effect of changes in taxes on
work effort are quite striking. Yet they are corroborated by some
cross-section studies which we performed for the years 1962 and
1966. These years were chosen because they bracketed the major
1964 tax cut. We used the IRS tapes and stratified the income tax
returns by income classification in order to determine what
happened to work effort when taxes were reduced.
Basically the approach we have taken is the following. We know
that tax rates were reduced significantly between 1962 and 1966.
For any given level of adjusted gross income (AGI), we examined
what happened to the proportion of income accounted for by the
sum of wages and salaries and business and professional incomein other words, income earned from current work efforL If this
proportion remained unchanged we could conclude that the
reduction in tax rates had no significant influence on work effort.
If it increased, however, we could conclude that the tax reduction
heightened work effort. Note that by holding AGI constant in the
regressions we have automatically excluded any increase in work
effort which might have accrued from the overall growth in the
economy or rise in productivity. Our analysis is strictly a marginal
one for any given level of income.
We found the following results for a 1% reduction in tax rates.
For lower-income workers, such a reduction would raise work
effort by about 0. l %. For middle- and upper-middle workers, the
EV A NS /
61
increase was about 0.25%. For upper-income workers-those with
taxable income of $120,000 or more-we found that elasticities
were in excess of 2.0. The upper-income elasticities are probably
overstated for the following reason. When the top marginal tax rate
dropped from 91 \\1o to 70%, many individuals simply shifted some
of their compensation from capital gains and stock options back
into earned income. As a result, tax revenues in the top bracket
more than doubled from 1964 to 1966 after accounting for growth
in the economy even though the top bracket rates dropped
drastically.
We now consider the wage rate functions in the supply-side
model, for they play a critical role in determining the rate of
inflation. From the point of view of supply-side economics, the
view that we cannot simultaneously have full employment and
stable prices is anathema, for it is just this combination which our
model shows how to achieve. The problem is that a decline in
unemployment is usually triggered by policies which increase
aggregate demand but do not raise aggregate supply. When this
happens, it is small wonder that inflation eventually rises. However,
balanced growth policies, which raise both demand and supply at
the same rate, wiH lead to lower unemployment without increasing
inflation.
Yet if we accept the empirical proposition that a strong negative
relationship exists between wages and unemployment, how can we
then claim that a decline in unemployment will not result in higher
wages, unit labor costs, and prices?
Several possibilities can be considered. The main ones are as
follows:
1. The decline in unemployment is accompanied by an increase in
productivity, thus offsetting higher wage rates. This would occur,
for example, if the reduction in unemployment were due to greater
capital spending.
2. The decline in unemployment is accompanied by a reduction in
personal income tax rates, thereby causing wage earners to accept
smaller pre-tax pay increases.
3, The increase in output is accomplished by increasing labor
force participation and lengthening the work week, thereby leaving
the unemployment rate almost unchanged. This solution is
preferable mainly when the economy is near full employment; but
as indicated in our previous discussion, that is when the trade-off
between wages and unemployment becomes most severe. When
slack still exists in the labor markets, the increase in wage rates
62 /
S U P P L Y~S I D E E C O N O M E T R IC MO D E L
stemming from a decline in unemployment is much smaller.
4. An increase in output could be accompanied by declining
prices for other factors, such as an improvement in the value of the
dollar and hence lower import prices.
To be sure, these changes will not happen automatically. In fact,
it is probably the rule rather than the exception that wages, unit
labor costs, and prices will rise as unemployment falls. However, to
state that this is a general empirical rule because of past experience
does not necessarily imply that policies which will offset or mitigate
this trade-off cannot be fashioned. In fact, we have just proposed
four solutions which would accomplish just that.
It should be stressed that the lags on all of these variables are
substantial. The unemployment rate is included with an average lag
over the past two years. The lag on the CPI is at least one year in
all cases and ranges as far back as three years in the construction
equation. Similarly, the personal tax rate is averaged over the past
two to three years. Thus the effects which we are describing clearly
do not happen instantaneously. They do, however, point out that
delayed wage demands may be viewed as somewhat of a ''ticking
time bomb" in the aftermath of sharp increases in inflation or tax
rates. Just because wage demands do not spiral up immediately
after inflation and taxes increase does not necessarily mean that
they will never catch up, for the lag process can take up to three
years to become fully effective.
The generalized wage rate function which we estimate is of the
form:
w
w -
w
p
where: w = average wage rate;
Un = unemployment rate;
p = consumer price index;
tr = average tax rate on personal income;
4
x "" ¼
i= I
X_,; and
g = a generalized nonlinear function, e.g., ~1 ~
Un
Both the unemployment and inflation terms are in common use
in macroeconomic wage rate equations. However, the last major
term which we use in these equations, namely the average tax rate
on personal income, definitely is not. Yet its inclusion should not
EVANS/63
be considered particularly surprising. An increase in tax rates will
cause workers to bargain for wage increases in excess of the rise in
inflation in order to keep their real income constant. Similarly, a
tax reduction will permit them to accept gains which are less than
the rate of inflation because their take-home pay will still be at the
same or higher levels.
The elasticities for the various sectors of the economy, and for
total private nonfarm business, are given in Table 4. We see that a
I% increase in the CPI eventually results in a 0.62%, or 5/8%, rise
in wage rates. While this figure is high, it is not unity. Even after a
lag of up to three years, wage earners do not recoup the full
increase in the reported CPI. This fact has been fairly evident ever
since I973, when the real wage was some lOIIJo higher than current
levels in spite of two tax cuts in the intervening years.
TABLE 4
Elasticities for Wage Rate Equations
Manufacturing
Construction
Nonmanu facturing
Total private
nonfarm
1% Change
in CPI
• 1% Change
in Un
blOJo Change
in Un
qeyo Change
in tr
0.58
0.87
0.62
0.25
0.67
0.00
0.82
2.23
1.17
0.50
0.46
0.37
0.62
0.11
1.13
0.41
afrom 8% !O 7%
bfrom 5% to 4%
CJ p.p. change, Le,, from 30% to 3 I eyo
The elasticity with respect to unemployment is nonlinear, as we
think it should be. Above 8% unemployment we do not find any
effect at all on wage rates from a change in unemployment. The
change in each percentage point below 8% then becomes
progressively larger. We have selected two points on this
unemployment/wage trade-off curve: a change from 8% to 7%,
and a change from 5% to 4%. As can be seen, a change in the first
case results in a change in wage rates well below I%, whereas a
change in the second case results in a change in wage rates
somewhat above 1% .
We finally turn to the change in wage rates resulting from a
change in the average tax rate. It is encouraging to find that the
coefficients in all of the three equations are bunched closely
64 / SUPPLY-SIDE ECONOMETRIC MODEL
together. While we might expect differences in the unemployment/
wage rate trade-off in different industries because of varying
institutional and union structure, we would expect that workers
would respond similarly to changes in tax rates regardless of the
particular industry in which they were employed. We find that for
the overall economy, a I p.p. change in tax rates (i.e., from 30% to
31 %) would result in a 0.4% change in wage rates. However, this is
only an impact multiplier although it does take place over as much
as three years; we also need to consider the total effect after
including the interaction between wages and prices.
In order to understand the dynamics of the wage-price-tax
interaction, let us aggregate the equations in the wage sector. We
then find that a I p.p. reduction in personal income tax rates will
reduce prices by about 0.45% and wage rates by about 0.70%_
Since wage rates rise a full I OJo because of lower taxes, the after-tax
increase in the real wage rate stemming from the tax reduction
is 0.9%.
To summarize the results of this section, we find that:
I. A 1 p.p. change in the tax rate will change labor force
participation in the opposite direction for primary workers by a
minuscule 0.05% but will change the participation rate for
secondary workers by 0.37%.
2. A 1 p.p. change in the tax rate will change employment-hours
in the opposite direction by 0.5%. Much of this change stems from
the change in hours worked.
3. A 1 P-P· change in the tax rate will change the average wage
rate in the same direction by 0.4% on impact, and 0.7% when the
interaction between prices and wages is considered.
Thus a reduction in the personal income tax rate would increase
the supply of labor, increase the number of hours worked, and
reduce the gain in average wage rate. An increase in the demand
and supply of labor would expand the maximum productive
capacity of the economy. Thus inflation would be reduced both
through a lower wage rate and a higher level of maximum capacity,
thus widening the gap between actual and maximum capacity_
MAJOR LINKAGES IN THE SUPPLY-SIDE MODEL
One of the reasons that demand-oriented policies have been used
almost exclusively in the past 15 years is that all of the current large
scale econometric models have indicated that these policies will
benefit the economy more than supply-side changes. Embedded in
EVANS/65
these models is the implicit assumption that an increase in demand
will automatically trickle down to increase aggregate supply, thus
insuring balanced, noninflationary growth.
However, there is nothing magical about the balance between
aggregate demand and supply. If incentives are lacking for
investment, capital formation will stagnate. If incentives are lacking
for labor, labor force participation will decline, the amount of
labor offered by those already in the labor force will be reduced,
and productivity will diminish. As a result, total productive
capacity of the economy will grow more slowly than total demand,
and bottlenecks, shortages and higher inflation will eventually
result.
According to Keynesian demand economics, this higher inflation
must then be fought by causing a recession and reducing aggregate
demand. It is true that the gap between aggregate demand and
supply must be widened in order to diminish inflationary pressures.
However, surely there are two ways to accomplish this aim. One is
indeed to diminish demand, thereby causing higher unemployment.
The other is to increase aggregate supply, thereby raising the
production possibility curve of the economy and increasing jobs
and output at the same time that inflation is being lowered. This is
the fundamental hypothesis underlying our supply-side modeling.
As already noted, most fiscal policy analysis of the past I 5 years
has been based on the belief that an increase in government
spending will lead to a larger rise in demand and output than an
equivalent reduction in taxes. The reasoning which leads to this
conclusion is straightforward if inaccurate. If the government
increases its spending, the entire dollar is used to raise aggregate
demand. If taxes are cut, however, some of each dollar is used for
saving. Since existing Keynesian models do not incorporate the links
between saving and investment, demand does not rise as much.
Furthermore, these models also state that a personal income tax
cut has a larger effect than a corporate income tax cut, and for
much the same reason. Individuals spend a larger proportion of the
extra money they receive from reduced taxes than do corporations,
and that left-over saving does not contribute to economic growth or
prosperity.
The supply-side model which we have built gives exactly the
opposite result: an income tax cut has a larger effect on the
economy than an increase in government spending. The supply-side
mechanisms which support this conclusion can be qualitatively
summarized as follows. In particular, a reduction in personal and
66 /
SUPPLY-SIDE ECONOMETRIC MODEL
corporate income taxes will set in motion the following chain of
events.
1. An increase in the after-tax rate of return on personal saving
occasioned by a reduction in personal income tax rates raises the
incentives of individuals to save. This increase in saving leads to
lower interest rates and higher investment.
2. A reduction in the effective corporate income tax rate, either
through lower tax rates, a higher investment tax credit, or more
liberal depreciation allowances, improves capital spending directly
by increasing the average rate of return.
3. An increase in both personal and corporate saving leads to
greater liquidity and less loan demand, thereby lowering interest
rates. These effects help both capital spending and residential
investment.
4. A rise in the ratio of investment to GNP leads to higher
productivity, which means that more goods and services can be
produced per unit of input. As a result, unit costs do not rise as
fast and inflation grows more slowly.
5. A reduction in personal income tax rates leads to a rise in
labor force participation and work effort, thereby increasing the
supply of labor necessary to produce more goods and services.
6. Thus labor supply, capital stock, and productivity are all
increased by lower tax rates, thereby expanding the maximum
productive capacity of the U.S. economy.
7. As a result of higher maximum capacity the inflationary
pressures of shortages and bottlenecks diminish, thereby reducing
the rate of inflation.
8. An increase in maximum capacity also permits the production
of more goods and services for export markets. This improves our
net foreign balance and strengthens the dollar, thus leading to lower
inflation because imported goods decline rather than advance in
price.
9. Lower personal income tax rates lead to smaller wage gains,
since wage bargaining is based at least in part on the level of
after-tax income. This in turn reduces inflation further.
10. Thus lower tax rates cause a reduction in inflation through
several channels. Inflationary pressures decline as the gap between
actual and maximum potential GNP rises; productivity increases,
thereby lowering unit labor costs; the donar strengthens, causing
less imported inflation; and wage rates rise more slowly.
I I. Lower inflation leads to higher real disposable income, since
EV ANS/
67
bracket inflation is mitigated. The rise in income leads to an
increase in consumption, output, and employment.
12. Lower inflation leads to lower interest rates, stimulating
investment in both plant and equipment and in housing.
13. The increased demand for goods and services stemming from
lower inflation is matched by the rise in the maximum potential
capacity of the economy to produce these goods and services,
thereby resulting in balanced, noninflationary growth.
One of the most important sets of linkages in the supply-side
model is the relationship between saving and investment. For if
saving rises and these funds are just used to increase idle cash
balances, investment may not expand. However, these links are well
documented in our model.
A $10 billion increase in personal saving raises time deposits by
$3.0 bH!ion and thrift institution deposits by $L6 billion. In
addition, it reduces loan demand by $3.6 bilHon.
As a result of these changes in the balance sheet of commercial
banks, demand for U.S. government securities by the banks
increases by $11.5 billion. This results in approximately a l %
decline in interest rates and a 3.2% increase in stock market prices.
These changes have two related effects on investment. First,
lower interest rates and higher stock prices stimulate fixed business
investment. Second, easier credit increases housing starts and
mobile homes and, to a lesser extent, producers durable equipment.
As would be expected, nonresidential construction is more
sensitive to changes in interest rates and stock prices than is
equipment. Thus we find a $2.5 billion increase in structures, as
compared to a $1.3 billion rise in producers durable equipment
from a $IO billion increase in personal saving. Residential
construction rises $ l .5 billion because of credit easing and $1.2
biHion because of lower interest rates. These are, of course, only
first-round effects which do not take into account the increase in
investment stemming from higher income and outpuL However,
these results do document the strong linkages between saving and
investment which exist in the supply-side model. For if these
linkages are not strong, the second-round effects will not be
observable either.
Another important breakthrough i.n our supply-side model is the
endogenous explanation of productivity, which we have already
d1scussed 1n the first section.
A l % increase in productivity will not only expand maximum
68 / SliPPLY-SlDE ECONOMETRlC MODEL
potential GNP by that amount; it will initially lower prices by
2/ 3%, since labor costs consist of 2/3 of total factor costs. This is
only the first-round effect, since lower prices will lead to lower
wages and further declines in unit labor costs and prices. The total
effect of a l % increase in productivity is to reduce prices by
about 2%.
We are also able to introduce other innovations into the supplyside model because of the endogenous treatment of maximum
capacity. In particular, the model introduces the concept of the
cumulative gap, already discussed in the first section, which we
define as the cumulative difference between 99% of maximum GNP
and the actual level of GNP when this gap is negative. When it is
positive-Le., actual GNP is below maximum potential outputinflationary pressures do not build because of bottlenecks and
shortages. However, when it is negative, prices start to rise faster
than would be indicated by the cost of factor inputs alone.
So far this term does not sound greatly different than an index of
capacity utilization, although it is much more inclusive in that it
covers all sectors of the economy. However, we have cumulated this
gap for all periods when the gap is negative. This term therefore
indicates that inflationary pressures build up over many years and
do not disappear every time a mild recession occurs. The
inefficiencies and distortions which occur when the economy is
operating near full capacity are not reversed overnight, and remain
as a legacy until the cumulative gap once again returns to zero. This
term may also represent the gradual buildup of inflationary
expectations.
The final area of the model in which supply-side economics has
been incorporated is the integration of the international sector with
the U.S. economy. Again, this is an area where theoretical
economists have long posited strong links, but they have never been
empirically documented within the context of a macroeconomic
model.
Supply-side effects are important in two specific areas, First, an
increase in the gap between actual and maximum potential GNP
raises exports, since the greater capacity of the U.S. economy
permits the production of more goods and services for export
markets as well. A I% increase in this gap raises net exports by
about $0. 7 billion per year; since the gap is cumulative, this figure
continues to increase linearly and is, for example, $2.1 billion after
three years.
The second major effect is the link between the trade-weighted
EV ANS /
69
average of the dollar, which is itself closely tied to the size of the
net foreign balance, and the overall rate of inflation. We find that
a 10% decline in the value of the dollar relative to a trade-weighted
average of the Deutschemark, French franc, Belgian franc, Dutch
guilder, and Japanese yen raises the producer price index 1.3% and
the consumer price index about half that much after a period of
two years.
Thus we can document several supply-side relationships that have
a significant effect on inflation as well as the rate of growth. All
these figures refer to the change in the CPI and are impact
estimates only. First, a 1 p.p. decline in the personal income tax
rate will lower wage rates and thus prices by about 0.5%. Second,
a I% increase in productivity will lower prices by 2/3%. Third,
a 10% improvement in the trade-weighted average of the dollar will
reduce inflation by about 0.6%. Fourth, after a three-year period,
a 1% increase in the gap between actual and maximum GNP will
lower prices by 0.4%. It is worth repeating that all of these figures
are impact estimates only and do not take into account the
interaction between wages, prices, productivity, and other factors of
production. Indeed, the final changes in prices are between two and
three times the initial impacts, depending on cyclical conditions at
the time.
Thus we find that the nemesis of demand-side economics, namely
that output must be reduced and unemployment increased in order
to dampen the rate of inflation, is only one of several alternatives.
Inflation can also be reduced by increasing productivity, reducing
personal and corporate tax rates, and strengthening the value of the
dollar. We would not quarrel with the statement that the size of the
gap between actual and maximum potential GNP is one of the
factors determining the rate of inflation, but do believe that other
factors must be considered as well.
The actual reduction in the implicit GNP deflator for the highgrowth, high-deficit case is only 1.3% by 1990, although even this
represents a marked change from the usual finding that inflation
would be higher. The two principal reasons for this discrepancy are
a) the lag structure and b) the large deficit. The changes in
productivity do not immediately translate into lower prices, since
both changes in wages and prices react to change in economic
stimuli with a substantial lag. In addition, the benefits to higher
productivity from higher investment are not felt immediately.
The second and more important reason is that the huge budget
deficit pushes up interest rates, thereby contributing to higher costs
70 / SUPPLY-SIDE ECONOMETRIC MODEL
of doing business and also raising the CPI through higher mortgage
interest rates.
Because of the fact that the implicit GNP deflator declines in this
high growth scenario, we find that the reflows are rather modest.
Hence the ex post deficit in 1990 is approximately $500 billion in
spite of the higher growth generated. While such a deficit is
economically feasible because the dissaving by the government is
funnelled into saving by the private sector, we do not think it
would be politically feasible, nor do we consider it the optimal
solution.
For this reason we have calculated another high-growth scenario,
one with a balanced budget, which is generated by reducing transfer
payments. This alternative high-growth scenario, which we then
adopt as our preferred run, also provides additional information
about the timing and magnitude of government spending
multipliers.
GENERATING A HIGH-GROWTH SCENARIO: THE BALANCED
BUDGET CASE
To generate this simulation, we made only one change from the
previous high-growth run: we reduced transfer payments enough to
generate a balanced budget. This resulted in transfer payments
increasing only 2.2% per year (current dollars) instead of the 11 .4%
per year increase which is included in both the baseline and high
growth-large deficit scenario. The total reduction in transfer
payments by l 990 is approximately $500 billion per year.
Before examining the economic ramifications of such a reduction,
it certainly is worth asking whether it would be possible to cut
transfer payments by this amount while still retaining the present
social fabric of the United States. Figures on the projected growth
of transfer payments over the next decade under alternative
assumptions are given in Table 5.
For purpose of this analysis, we can divide transfer payments
into three categories: retirement benefits, medical care payments,
and other transfers, which are largely veterans benefits and welfare
payments. Under the baseline case, retirement benefits are expected
to grow at a rate equal to the annual average increase in the CPI
plus the average increase in the population over 65. A similar
formula would apply for medical care benefits, although there we
use the increase in the CPI for medical care. Other transfer
payments are expected to grow at a rate of increase equal to the
TABLE 5
Projected Growth of Transfer Payments
1980
(billions)
Annual Increase Due To:
Change in
Inflation Pop. Coverage
Total
Annual
Change
1990
(billions)
A. Baseline
Retirement Benefits
Medical Care
Other
TOTAL
Retirement Benefits
Medical Care
Other
TOTAL
$157
38
98
293
9.9%
I0.1
2.0%
2.0
8.3
LO
0.0%
1.0
0.0
12.1%
13.4
9.4
11.4
B. Adjustment for Lower Inflation Only
0.0%
8.2%
6.1%
2.0%
$157
7.8
2.0
ll.O
38
1.0
8.5
7.5
o.o
98
l.O
293
8.7
C. Lower Inflation and Cutbacks in Program
Retirement Benefits $157
6.30/o
2.0% -9.0%
-0.7%
Medical Care
38
7.8
2.0
-5.0
5.0
Other
98
7.5
1.0
-3.7
4.6
TOTAL
293
2.2
aJmplicit Constant Def1ator instead of CPI
$490
134
241
865
$344
108
222
674
$147
62
154
363
72 / SUPPLY-SIDE ECONOMETRIC MODEL
average rise in the implicit GNP deflator plus the average gain in
total population. These figures are all given in Table 5A.
The figures in Table 5B are adjusted for lower inflation, and also
incorporate the assumption that retirement benefits would be
indexed to the implicit deflator for consumption rather than the
CPI, since the tendency of the latter to overstate price increases
because of its overdependence on the cost of buying and financing
a home is now weil known. Thus switching to the higher-growth
lower-inflation scenario, plus this one sensible adjustment in the
indexation scheme for social security benefits, reduces transfer
payments by almost $200 billion per year by 1990.
While this $200 billion is indeed an impressive saving, it is far
less than the $500 billion which is needed to balance the budget.
Table 5C provides the arithmetic to indicate how these remaining
savings are achieved. From an economic point of view, the
following changes are instituted:
1. The retirement age is raised from 65 to 70. There is nothing
sacrosanct about the number 65 for a retirement age; indeed, if we
use the most recent actuarial tables, we find that a retirement age
of 65 in the mid-1930s (when social security was originally
implemented) now corresponds to an age of almost 70, and that
figure will probably rise to 72 by the end of this decade.
As might be expected, the savings in postponing the retirement
age are substantial. Each additional year of postponcment-e.g.,
from 65 to 66-saves the government $18 billion at current levels of
benefits and population. If we adjust this figure upward for the
increase in the implicit consumption deflator and the growth in
population over 65, by 1990 this figure amounts to $40 billion for
each year the retirement age is postponed. Thus raising the
retirement age to 70 would save a whopping $200 billion, in which
case retirement benefits would actually be somewhat below present
levels.
The other cuts are less drastic. The reduction in medical care
benefits could be accomplished, we believe, by simply adding a
deductible and coinsurance whereby the patient would pay the first
$100 per year of medical expenses and 90% of the remainder up to
some fixed limit which might be equal to, say, 10% of his annual
income. For example, if an individual had an income of $20,000, be
would be required to pay no more than $2,000 in medical premiums
that year regardless of the extent of his actual bills. This would
provide 100% coverage for catastrophic illness while alerting
patients to the substantial cost of medical services which is borne
EVANS/
73
by society at large. We estimate that the deductible and coinsurance
as outlined above would cut the growth of medical care payments
in half.
The remaining cuts would occur in the phasing back of existing
programs, such as food stamps for college students, a cap on black
lung payments, reduction in the Aid to Families with Dependent
Children as these parents returned to work, and other similar
welfare programs. Of the three major areas, these cuts are
proportionately the smallest and the most politically feasible.
It should be made quite clear that workers who no longer receive
retirement benefits at ages 65 through 69 will remain in the labor
force, but the higher growth rates will certainly provide the
additional jobs necessary to support these older workers. As we
have already mentioned above, the U.S. economy will shift from a
labor surplus to a labor shortage economy by 1990, and the jobs
which these older workers retain will mitigate the labor shortage
problem. Hence the gradual raising of the retirement ageincreasing it, for example, six months every year over the next
decade-would fit hand in glove with the need for more workers
and the redirection of resources from the public to the private
sector.
COMPARISON OF THE TWO HIGH-GROWTH SCENARIOS
Based on traditional multiplier analysis, one might expect that the
$500 billion decrease in transfer payments would result in a far
slower rate of growth because of the resulting decline in
consumption. However, this is not at all what happens. The
reduction in the federal government budget deficit lowers interest
rates, thereby stimulating capital formation. Furthermore, the lower
rate of inflation which stems from higher productivity growth also
reduces interest rates. Finally, since income is redistributed to those
who are working away from those who are not, labor force
participation rises, which provides the additional labor inputs
needed to complement increased capital spending.
The comparison for several key variables is given in Table 6. In
particular we note that while real growth is about ½ OJo per year
higher for the largest deficit case in the early 1980s, the pattern is
completely reversed in the second half of the decade, and by I 990
real GNP is increasing almost 1/2 OJo per year faster for the balanced
budget case. As can be seen, the rate of inflation is approximately
1% per year lower for the balanced budget case after 1985.
74 / S U P P L Y · S I D E E C O N OM E T R l C M O D E L
TABLE 6
Comparison of Two High-Growth Scenarios
198119821983 1984 1985 1986 1987 1988 1989 1990
-- -- -- -- -- -- -- -- -- --
Real GNP,% Growth
Large deficit
No deficit
2.6
2.5
6.2
5.9
4.4
3.8
1.0
0.2
2.1
1.6
3.4
3.2
3.9
4.1
4.4
4.7
4.8
5.2
5.0
5.4
7.6
6.9
6.6
5.7
6.1
5.0
5.6
4.5
5.3
4.2
4.9
3.8
Implicit GNP Deflator, OJo Growth
Large deficit
No deficit
9.2
9.2
8.7
8.7
8.8
8.6
8.6
8.2
Federal Budget Surplus or Deficit, billions of $
Large deficit
No deficit
-78
-65
-70
-19
-92 -148 -199 -239 -284 -348 -416 -508
-2 -15 -16
-2
13
15
16
-4
Government Spending/GNP, ratio
Large deficit
No deficit
37.1 35.5 34.5 34.8 35.1 35.2 35.2 35.2 35.2 35.2
36.6 34.0 32.2 31.9 31.6 31.1 30.4 29.9 29.4 29.0
AA Utility Bond Rate, %
Large deficit
No deficit
11.5 11.3 11.7 13.0 13.6 14.l 14.6 15.5 16.6 18.0
t t.5 11.0 IO. 9 11.8 11.5 11.3 11.0 11.2 I 1.5 12.2
LOW -OROWTH SCENARIO
We have generated a high-growth scenario with a balanced
budget by cutting corporate and personal income tax rates
dramatically and then balancing the budget through lower transfer
payments. The low-growth alternative, however, cannor realistically
be generated by raising tax rates the same amount they were cut in
the high-growth alternative, for no one expects the statutory tax
rates to be raised during the l 980s, although rates may drift up
EV ANS /
75
because of bracket creep. Thus we must lower growth directly by
reducing growth in the labor force and by lowerfog the rate of
growth in productJvity. This can be done by a combination of
a) higher tax rates through bracket creep, b) higher costs of
government regulation, and c) higher relative energy prices.
Thus we have approached the 1ow-growth scenario in a much
different manner, and have changed those variables which impact
directly on labor force growth and productivity other than income
tax rates. The changes which we have introduced to generate this
scenario are the following;
1. Energy prices, both imported and domestic, grow at a faster
rate.
2. The cost of government regulation doubles over the decade.
3. Labor force participation rates grow more slowly.
4. Transfer payments grow 15.6% per year instead of 11.4%.
The average tax rate increases from 24.9% to 38.3% by 1990-but
that is entirely due to bracket creep and does not reflect any rise in
the statutory rate.
In addition to these four changes, we have also cancelled any
personal or corporate income tax cuts over the decade which are
included in the baseline, held depreciable lives at I 980 levels, and
terminated the investment tax credit. However, it should be stressed
that these do not account for the bulk of the decline in growth
which occurs in this scenario- that is due to the four factors listed
above.
COMPARISONS OF THE ALTERNATIVE SCENARIOS
We now compare the performance of the economy, on a decadelong average and for year-by-year changes, for the baseline, high
growth with balanced budget, and low-growth scenarios. We have
not included the high growth with large deficit run, since that is not
a feasible alternative; furthermore, we have already discussed the
difference between the two high-growth runs in the previous
subsection. The principal assumptions and results are presented in
Table 7.
While the decade average figures are useful, they really do not
convey the full flavor of the differences between the runs; this is
best done by examining the differences in the forecast on an annual
basis, which is presented in Table 8. Here we note the great
divergence which occurs in the saving rate, growth in productivity,
and inflation, particularly after 1985. The forecasts are somewhat
76 / S U P P LY - S l D E E C O N O M E T R I C M O D E L
TABLE 7
Growth Rates
(1980 - 1990)
Selected Economic Indicators for Alternative Scenarios
Baseline
Real GNP
Labor Input
Labor Productivity
Labor Force Participation
Real GNP per capita
Relative Price of Energy (PPI)
Growth of Transfer Payments
2.9
2.0
0.9
0.6
2.0
6.6
ll.4
High
Growth
Low
Growth
3.6
1.6
2.0
0.8
2.7
2.1
1.6
1.3
0.3
0.3
0.7
7.2
15.6
0.168
0.20
10%
5.0
10.0
0.284
0.46
0%
10.5
23.0
6.8
Levels in 1985
Personal Income Tax Rate
Corporate Income Tax Rate
Investment Tax Credit
Depreciation Lives, Equipment
Depreciation Lives, Structures
0.227
0.46
10%
8.4
18.4
similar for the first five years but then differ markedly, which
emphasizes the fact that mosl of the effects of changes in supplyside fiscal policies occur only after rhe first five years.
The results in Table 8 point out that the effect of higher
productivity on higher saving and investment on productivity,
growth, and inflation is far from instantaneous. In fact, even if an
optimal fiscal policy were to be implemented immediately, we
would not expect it to have a noticeable effect on slowing inflation
for at least two years. In fact. it is often five years or even more
before the full effect of higher saving is translated into benefits for
the entire economy.
In fact, it is imeresting to note that the initial effect of these tax
cuts is to raise inflation, just as would be the case in a traditional
demand-side model. This occurs because the demand elementshigher consumption and investment-are activated before the
supply elements-higher productivity and lower wage rates-work
TABLE 8
Annual Comparisons of Alternative Scenarios
198119821983 1984 1985 1986 1987 1988 1989 1990
-- -- -- -- -- -- -- -- -- --
Real GNP, % Change
Baseline
High Growth
Low Growth
2.0
2.5
1.7
5.2
5.9
3.8
3.6 0.0 1.5
3.8 0.2 1.6
1.7 -0.9 -0. l
3.0
3.2
I.2
3.6
4.1
2.4
3.6
4.7
2.9
6.3
5.7
8.4
6.2
5.0
8.5
6.3 6.6 6.9
4.5 4.2 3.8
9. 1 10.0 11.6
3.6 3.5
5.2 5.4
3.5 -0.3
Implicit GNP Deflater, % Change
Baseline
High Growth
Low Growth
9.1
9.2
9.8
8.2
8.7
9.2
8.0
8.6
9.2
7.9
8.2
9.2
7 .1
6.9
8.8
Productivity Growth, OJo Change
Baseline
High Growth
Low Growth
1.3
1.3
1.2
1.4
1.7
1.5
1.3 0.3 0.0 0.7 0.8 0.9 0.9 0.9
1.8 1.1 1.2 2.3 2.9 3.4 3.9 4.2
I. I -0.7 -1.4 -1.5 -1.9 -2.4 -2.7 -3.0
Ratio of Fixed Business Investment to GNP
Baseline
High Growth
Low Growth
9.3 9.8 10.8 11.2 10.8 10.8 11.0 I I.I 11.1 11.0
9.4 10.2 11.6 12.1 12.1 12.3 12.5 12.8 12.9 13.0
9.3 9.7 10.4 10.5 10.3 10.1 IO. I 10.2 10.0 9.9
Ratio of Government Spending to GNP
Baseline
36.6 36. l 35.6 36.3 36.8 36.9 36.8 36.6 36.5 36.4
High Growth 36.6 34.0 32.2 31. 9 31.6 3 l. l 30.4 29.9 29.4 29.9
Low Growth 37.2 36.6 36.9 38.I 39.2 39.7 39.7 39.3 38.5 38.9
Personal Saving Rate, 0/o
Baseline
High Growth
Low Growth
4.5
5.0
3.2
5. l
5.1
3.2
6.6
6.1
3.4
6.6
5.7
2.8
6.5
5.5
2.5
7.6
6.5
3.0
8.8
8.0
3.3
9.7 10.3 11.1
9.4 10.7 12.5
3.0 2.2 0.8
78 / SUPPLY-SIDE ECONOMETRIC MODEL
their way through the system. However, by 1985 the situation is
reversed and by 1990 the inflation rate in the higher growth
scenario is almost 3% below the baseline solution.
It is perhaps not very difficult to convince anyone that a higher
rate of growth is preferable to a lower one. However, recently two
groups of lower growth advocates have emerged: those who argue
that we either cannot or should not produce enough resources
necessary to support higher growth, and those who argue that
higher growth would be inflationary and hence ultimately destroy
that which we set out to accomplish.
The resource question is not a trivial one, but can be solved by
an appeal to market economics. The decline in domestic oil
production and the huge increases in the volume of oil imports
during the past decade has been directly related to the decision by
U.S. government officials that we would somehow all be better off
if oil prices were not allowed to rise as fast as increasing costs.
While the problem with energy reserves is the most virulent, similar
problems exist with respect to many other basic industrial
commodities. It is imperative that the higher growth scenario be
accompanied by adequate supply response in terms of profit
margins for those who extract or produce basic materials.
SIMULATIONS AND MULTJPLIER ESTIMATES
One way to approach this subject would be to give the usual
multiplier estimates for small changes in government spending,
personal income tax cuts, corporate tax cuts, and similar measures.
Even these estimates can be quite instructive; we have already used
this model to show that the Carter tax packages are much more
inflationary than the Reagan tax packages. However, the full flavor
of the supply-side model cannot really be savored unless we
introduce massive changes in fiscal policy, and it is these changes
which we report in this section. Specifically, we have prepared three
simulations: a) the baseline case with moderate tax cuts and
essentially a balanced budget after 1982, b) a daring experiment in
which we cut personal and corporate tax rates in half, and c) the
same tax cuts, but combined with enough reductions in transfer
payments to balance the budget.
GENERATlNG A HIGH-GROWTH SCENARIO: THE LARGE DEFICIT CASE
The high-growth run is generated by changing the following tax
parameters:
EVANS/79
1. Gradual reduction in the corporate tax rate from 0.46 to 0.20
by 1985. The actual yearly values are:
1980
0.46
1981
0.40
1982
0.35
1983
0.30
1984
0.25
1985 and later
0.20
2. Depreciation lives for equipment reduced from 10.5 presently
to eight years in 1981 and five years in i 982 and thereafter.
3. Depreciation lives for structures reduced from 23 presently to
18 years in 1981 and lO years in 1982 and thereafter.
4. Gradual reduction in the average marginal federal personal
income tax rate from 24% to 12% in equal increments by 1990.
The revenue losses from these changes are immense, particularly
when calculated in 1990 prices. For example, taxable personal
income is estimated to be $3.4 trillion by 1990. Thus a cut from
24% to 12% in the tax rate would result in a static revenue loss of
some $410 billion.
The changes in the corporate tax rates are not as great, but they
are still substantial. Pre-tax corporate profits are expected to be
about $400 million by 1990; hence, cutting the tax rate from 46% to
20% would reduce tax receipts by $100 billion. In addition,
shortening depredation lives would lower pre-tax corporate income
by $140 billion, although since the tax rate is reduced to 20%, this
only accounts for an additional $30 billion revenue loss per year. In
fact, it should be clear that as the corporate tax rate approaches
0%, the length of depreciation lives is no longer of any importance
for tax purposes.
These figures indicate a static revenue loss of $540 billion per
year. Even when compared with a GNP of almost $7 trillion and a
federal budget of $1. 7 trillion, the amounts are quite large. This
loss amounts to a deficit of 7. 7% of GNP, which is far larger than
the postwar record of 4.6% posted in the recession year of 1975.
It is usually argued that such static revenue loss estimates are
inappropriate, for they fail to consider the higher revenue base
raised by faster growth of the economy, higher employment and
income, and greater profits. However, this leads to one of the
major findings of the supply-side model.
Because lower personal income tax rates generate smaller gains in
wage rates and hence smaller increases in unit labor costs and
prices, current dollar GNP is only slightly larger in the higher
growth scenario than in the baseline case. Real GNP is some 8,6%
higher, since we have defined the high growth alternative to show
real GNP rising approximately I% per year faster for the nine-year
80 / SUPPLY-SJ DE ECONOMETRIC MODEL
period 1981 -1990. However, according to our basic results on the
trade-off between productivity and inflation, every l \lJo increase in
productivity results in a 2% reduction in inflation. Hence in steady
state equilibrium, we would expect current dollar GNP to grow 1%
less per year with this higher productivity growth.
The hypothesis that higher growth leads to more inflation is
effectively defused by the results given in this report. Indeed, the
higher growth scenario is accompanied by lower rather than higher
rates of inf1ation, due to greater productivity and lower wage rate
increases both slowing the rise in unit labor costs. Thus we are able
to generate realistic alternative scenarios which not only provide for
more jobs and greater output, but reduce the rate of inflation as
well by redirecting resources toward saving and investmenL
Finally, it is clear that one of the major contributors of higher
growth in the preferred scenario has been the increase in the
investment ratio, which in turn has been brought about through tax
incentives for increased saving and investment.
The generalized incentives for investment-lowering the
corporate income tax rate and shortening depreciation lives-are
well known and have often been suggested for stimulating
investment. We have not used an increase in the investment tax
credit in this scenario because of our finding that it is not as
efficacious. It increases investment only about half as much as an
equal reduction in the corporate income tax rate and about ¾ as
much as an equal reduction in depreciation allowances. We have
also introduced a net reduction in the capital gains tax by increasing
the exclusion from 60% to 70% of the total gain, a change which
also stimulates investment through raising stock prices and hence
lowering the cost of equity capital.
However, one should not neglect the fact that capital markets are
fungible-that an increase in saving in any major sector of the
economy will result in lower interest rates, greater credit
availability, and hence greater investment and productivity. We can
achieve these gains not only by stimulating investment directly, but
by increasing saving in the personal and governmental sectors. In
particular, we believe that capital formation can be stimulated by
reducing personal as well as corporate income taxes.
Hence in addition to reducing the corporate tax rate to 20% and
restructuring depreciation lives to adjust for inflation, we also favor
broad-based personal income tax cuts accompanied by
commensurate reduction in government transfer payments. It is the
balanced approach~the use of all three legs of the stool-which
we feel is essential for balanced low inflationary growth.
Thoughts on the Laffer Curve
ALAN S. BLINDER
... the ideas of economists and political philosophers, both when they
are right and when they are wrong, are more powerful than is
commonly understood. Indeed the world is ruled by little else.
-J. M. Keynes
The first part of the paper by Canto, Joines, and Laffer, which is
the only part I will discuss, sets up a simple general equilibrium
model with two factors (both taxed proportionately) and one
output, and proceeds to grind out the solutions. The model, while
not entirely unobjectionable, is certainly not outlandish in any
important respect. The authors make no claims that the model tells
us anything about the U.S. economy; nor do they draw any policy
conclusions. They use the model to provide intellectual
underpinnings for the celebrated "Laffer Curve" -the notion that
the function relating tax receipts to tax rates rises to a peak and
then falls. Since, as I will point out shortly, the analytical
foundations of the Laffer curve were in fact established centuries
ago, and require no economic analysis at all, I will devote my
comments to the critical empirical issue: is it possible that taxes in
the U.S. have passed the points at which tax receipts cease rising?
Is the U.S. tax system over the Laffer hill?
Let me note at the outset why this is an important question.
Certainly not because of the implications for the government
deficit. Surely what a tax change does to the budget deficit must be
one of the least important questions to ask. It is important to know
which taxes, if any, have reached the downside of the Laffer hill
because, in an optimal taxation framework, tax rates should be set
to raise whatever revenues are required with minimum deadweight
loss. 1 Since a tax that is past this point causes deadweight loss and
Alan S. Blinder is Professor of Economics, Princeton University, and Research
Associate, National Bureau of Economic Research, Cambridge, Mass.
'The statement assumes that lump sum taxes arc unavailable and ignores
distributional objectives.
81
82 /
THOUGHTS ON LAFFER CUR VE
FIGURE I
G(t)
t*
makes a negative contribution to revenue, it must be irrationally
high, as Canto, Joines and Laffer correctly state. 2
ORIGINS OF THE LAFFER CURVE
Figure 1 is a Laffer curve relating tax receipts, G, to the tax rate,
t. For some types of taxes (example: income taxes), the tax rate has
a natural upper bound at 100%, so we may assume that G(l) = 0.
For others (example: excise taxes) there is no such natural bound at
100%, so we assume instead that G asymptoticaHy approaches zero
as t approaches infinity. The distinction is not terribly important so
long as we keep in mind that taxes greater than 1000/o are indeed
possible in many cases. 3 The Laffer curve reaches its peak at tax
rate t*, which I hereafter call the Laffer point.
'Such a tax might be rational if its avowed purpose was to "distort" behavior
(e.g., an emissions tax to reduce pollution). A purely redistributive objective is also a
potential rationale; but there must be better ways to redistribute income.
'Taxes on such items as cigarettes, liquor, and gasoline have exceeded !00070 of the
producer's price in many times and places.
BLINDER/
83
According to the media, the Laffer curve was born on a napkin
in a Washington restaurant in 1974. This, however, l know to be
wrong. The Laffer curve should perhaps be called the Dupuit
Curve, because Dupuit-a man who was ahead of his times in
many respects-wrote in 1844 that:•
lf a tax is gradually increased from zero up to the point where it becomes
prohibitive, its yield is at first nil, then increases by small stages until it
reaches a maximum, after which it gradually declines until it becomes zero
again.
But Dupuit was just an academic scribbler distilling his frenzy
from a politician of a bygone age. In parliament in I 774, Edmund
Burke used what was perhaps called the Burke Curve by the
journalists of the day to argue against overtaxation of the American
colonists:
Your scheme yields no revenue; it yields nothing but discontent, disorder,
disobedience; and such is the state of America, that af!er wading up to your
eyes in blood, you could only end just where you began; that is, to tax where
no revenue is to be found ...
But, alas, we cannot credit Burke with the idea either, for the
concept goes back even further and is far more basic. One of the
first things that freshmen learn in their first course in cakulus is
RoHe's Theorem. RoUe's Theorem is as follows, Let G(t) be any
continuous and differentiable function with G(a) = 0 and G(b) =
0. Then there must be some point t>t- between a and b such that
G '(t*) = 0. Let a = 0, b be either 1 or infinity, depending on the
type of tax under consideration, add the proviso that G '(0) > 0,
and you have a Laffer curve. The existence of a Laffer curve, in
other words, is not a result of economics at all, but rather a result
of mathematics. We cannot doubt that there is a Laffer hill, i.e.,
there is a tax rate that maximizes tax receipts, so long as the
assumptions of Rolle's Theorem are granted. Are they? I think we
do not want to quibble with continuity or differentiability, and it
must be true that a tax rate of zero yields no revenue. This leaves
only the endpoint condi6on-either G(l) "" 0 or 0( 00 ) "" 0,
depending on the type of tax in question. But I, for one, am willing
to accept that a 100 6/o income tax rate or an infinite sales tax rate
will, to a first approximation, eliminate the taxed activity entirely.
The Laffer curve almost certainly exists.
'This quotation appears in Atkinson and Stern (1980). For other interesting
precursors, see Canlo, Joines and Webb (1979).
84 /
THOUGHTS ON LAFFER CURVE
ARE WE OVER THE LAFFER H1u?
I now turn to the question at hand. Is it plausible that the tax
rates we observe in the real world are greater than t*, so that we
are operating on the down side of the Laffer hill?
First a preliminary point. We all know that the applicability of
the Laffer curve hinges on elasticities being "large" in some sense.
(I will be more precise in a moment.) Thus the possibility of taxing
beyond the Laffer point is much more real for taxes whose bases
are narrowly defined-either in time, or in geographical space, or
in commodity space-than it is for taxes that are broadly based.
Let me iHustrate. A sales tax on pastrami is much more likely to
have a negative marginal revenue yield than a sales tax on all food,
simply because of the much greater substitution possibilities on
both the demand side and the supply side of the market for
pastrami, as compared to the market for all food. Similarly, I
rather doubt that an income tax on earnings between noon and
2 p.m. on Wednesdays would bring in much revenue. As a final
example, I have heard it claimed that if New York City raised its
sales tax, but the surrounding states and counties did not, revenues
would actually decline. The possibility of being over the Laffer hill,
I submit, is a very real one for very narrowly defined taxes. This,
of course, merely strengthens the argument-which economists
have been making for eons, it seems-for using broadly-based
taxes rather than narrow ones.
The important question for current public policy debates, as I
understand it, is: Can it be that some of our broadly-based taxeslike the personal and corporate income taxes-have passed the
Laffer point? This seems to me highly implausible, and let me
explain why.
Tax receipts are the product of the tax rate times the tax base.
For ad va!orem taxes, the latter is itself the product of a price (the
net-of-tax price) and a quantity.' Thus:
(1)
G
=
tpQ.
Since t affects both p and Q, the derivative dG has three terms.
dt
The first term:
pQ
might be called (with some unfairness to the Treasury) the naive
'I assume markets clear so quantity demanded and quantity supplied arc equal.
BLINDER/ 85
Treasury term. It would be a good estimate of marginal tax yield if
there were no behavioral responses. The second term:
tp dQ
dt
is the effect of the celebrated tax "wedge." Normally, we expect a
contraction in the level of any activity whose tax is raised, so this
term makes a negative marginal revenue contribution. The third
term:
tQ~
dt
is the effect that arises from the fact that market prices generally
change when tax rates change. Laffer et al. suggest that some
economists have been led to underestimate the potency of the
Laffer effect by ignoring general equilibrium reactions. Exactly the
reverse seems to be true for many taxes. Consider, for example, a
tax on a factor income where p is the price the firm pays and
p(l- t) is the price the factor supplier receives. Standard tax
incidence theory suggests that normal market reactions would make
p rise and p(l - t) fall when t increases, suggesting that this third
term is positive, not negative. Similarly, if there are possibilities for
factor substitution, the demand curves for competing factors of
production would be expected to shift out; if these factors are
taxed, this will also bring in more revenue. 6
The shape of the Laffer curve depends on the balancing of these
three forces. It is clear that if t* is to occur at an empirically
meaningful level, the "wedge" effect will have to be quite large. To
illustrate the conditions that are necessary, let us work out a
concrete example of a flat rate tax on labor income. Let W be the
wage the firm pays and W(l- t) be the wage the worker receives.
Let S(W(l - t)) be the supply function and D(W) be the demand
function, and assume S(O} = 0 so that a Laffer curve exists. Tax
receipts are:
(2)
G(t) = tWS(W(l- t)),
from which it follows by some simple algebra that marginal tax
yield is:
(3)
dG
dt
=
W · S( · )[ l - ~t_ 17c
1-t
'
+
_!_ dW (1
W dt
+
17c)],
'
'For excise taxes, the argument cuts the other way. If pis the selling firm's price
and p(l + t) is the consumer's price, then p probably falls while p(l + t) rises.
86 /
THOUGHTS ON LAFFER CURVE
where YJ 5 is the elasticity of supply:
YJs
=
W(l - t) S ' (W(l- t))
s
> 0.
The positive Treasury effect, the negative wedge effect, and the
positive price effect mentioned above can be seen clearly here.
Working out the elasticity of W with respect to t, and substituting it
into (3) gives:'
(4)
dG
WS(' )[ 1
dt
+
_t_ riJl + rio)
l-t
rys-Y/D
where r, 0 is the elasticity of demand:
rJo "" WD '(W)
< O.
D
Notice that (4) cannot possibly be negative in the range where
demand is inelastic. The Laffer point, t*, is found by setting (4)
equal to zero:
(5)
t* =
ris - Y/o
-r,D(l +YJs)
Table I shows the values of t* for selected values of the two
elasticities. It is clear that, unless the elasticities are quite high, we
can be over the Laffer hill only when marglnal tax rates are
extremely high. For example, even if each elasticity is as high as 2,
receipts continue to rise until the tax rate reaches two-thirds. In
other words, it is very unlikely (though not totally impossible) that
the peak in the Laffer curve comes at a tax rate that anyone might
seriously entertain.
By exactly the same procedure, it is possible to work out the
formula for the peak of the Laffer hill for the case of an excise tax
at rate t on a commodity with producer price p and consumer price
p(l + t). The answer is:
(6)
t*
=
Y/s -
Y/n
-risO + !'Jo)
and Table 2 provides numerical values for selected elasticities. It is
clear once again that t* is a huge number unless the elasticities are
incredibly high. For example, with elasticities of 2 for both supply
'This is, of course, not a general equilibrium analysis, since I consider only one
market in isolation. l think most economists would be very surprised if a multimarket setting changed things very much. In any case, the next section takes up a
general equilibrium example.
BLJNDER / 87
and demand, tax revenues are maximized at a tax rate of 200%.
Elasticities as high as 5 are necessary to get t* as low as 50%.
I conclude, therefore, that the revenue-maximizing tax rate is
very likely to be so high as to be considered ridiculous for any
broad-based tax. Only very narrowly based taxes, where elasticities
in the neighborhood of 5 start to become at least believable, are
Hkely to encounter the down side of the Laffer hill. For the
important taxes in our economy, the Laffer curve holds no more
interest than Rolle's Theorem.
THE CANTO, JOINES, AND LAFFER (CJL) MODEL
Now the examples just considered were mfoe, not Laffer's. So let
me turn next to the empirical relevance of the Laffer curve in the
model proposed by the authors. The model has perfectly
conventional demands for two factors (called labor and capital,
though both are variable) derived from a Cobb-Douglas production
function. The factor supply equations are somewhat
unconventional, so let me explain them a bit and interpret the
parameters.
Households hold fixed supplies of capital and labor, which they
TABLE 1
Values oft* from Equation (5)
Value of '1s
0
Value
below 1.0
1.00
of
l.O
1.00
-rio
2.0
5.0
1.00
LOO
.25
-
1.00
.90
.84
.50
1.0
2.0
more than 1.00
LOO
1.00
LOO
.75
.67
.83
.73
.60
.47
5.0
...
1.00
.58
.33
TABLE 2
Values oft* from Equation (6)
Value of Y/s
0
Value
of
-rJo
l or below
2.0
5.0
.25
....
.50
LO
2.0
infinity
00
9.0
00
5.25
5.0
2.75
3.0
1.5
5.0
-+
2.0
.88
1.4
.50
88 / THOUGHTS ON LAFFER CURVE
can either supply to the market-at net-of-tax returns R* and W*
respec:tively-or reserve for home production. Laffer et al. view the
factor supply decision in a kind of "utility tree" framework. First,
the household considers the choice of devoting its resources to the
market versus home sectors; this choice depends on the average
level of market returns relative to the average level of home returns
(the latter is, I suppose, always unity). Second, the household
decides on its relative factor supplies to the market by looking at
relative market prices. This analysis suggests supply functions
(assuming constant elasticity functional forms}:
(7)
u =
[(R*)'>(W*)l-<>]' {_WR**)
\
fl
(8)
f
> 0, /3 > 0
.l
>a
where [(R*)"(W*) 1 -"1 is the (geometric) weighted average of market
returns, weighted by the production function weights. The use of
the same "i::" parameter in (7) and (8) reflects the assumption of
CJL that the ratio of L to K depends only on the ratio W* JR*. A
tiny bit of manipulation puts (7) and (8) into the form of equations
(7) and (8) in the CJL paper:
(7 ')
(8 ')
L5 = ( W>j< /-·rn(W*)'
R*
so that the parameters oK and oL that appear in the CJL paper are
seen to have the following interpretations:
Ea.
The authors assume these parameters to be negative, which means
they are assuming a fairly sizable value for E -which is the one
unconventional parameter in this model. The interpretation of£ is
the general price elasticity of supply of factors to the market sector
(from the home sector). That is, if borh W* and R * were to
increase by 1% , then the supplies of both capital and labor to the
market would increase by£%. This is not a parameter for which
much empirical evidence is available.
The authors take pains to make clear that income effects are
ignored in their analysis because marginal tax receipts (positive or
BL l ND ER /
89
negative) are redistributed to the populace in a nondistorting way.
In theory, this is correct. In practice, three caveats must be entered.
First, it seems inconsistent to assume that revenue can be raised
only by distortionary taxes, but can be given away in a
nondistortionary way. Surely, any real way to give back the revenue
(through transfer payments or government gifts of goods and
services) will be just as distortionary as taxes. And isn't reducing
lump sum transfers the same as levying lump sum taxes?
Second, for the argument to hold, it is necessary that the
recipients of the (lump sum?) transfers be the same as the payers of
the additional taxes. If, for example, we consider cutting capital
taxation and making up for the lost revenue by reducing transfers
for the poor, there is no reason to think that income effects are of
second order. In fact, I would be inclined to think that income
effects would be of first order and substitution effects of second
order.
Third, it should be understood that the thought experiments
considered in the paper are balanced-budget alterations in the tax
structure, so we cannot really speak of revenue effects and Laffer
curves at all. The model assumes that lump sum transfers are
available, and what appear to be "Laffer curves" in Figures 2 and
3 represent instead the behavior of aggregate lump sum transfers as
tax rates are increased. If we really care about Laffer curves we
cannof ignore income effects.
Nothing more need be said about the structure of the model. CJL
correctly work out the solutions for prices and quantities and then
compute the revenue-maximizing tax rates on capital and labor
(their equations (27} and (28)). These can be simplified to:
(9)
(10}
(I - a )(1 + £ )(1
+ 1 + [3)
f3 + a
a(l + £ )(1 + 1 + {3)
Let me now pose the $64 question. Is it possible that the tax rates
implied by these formulas could be anywhere near current tax rates,
which I take to be approximately tL = .3 and tK = .4?
There are four parameters in these formulas. The one we know
fairly well is capital's share, a, which I take to be .25. f3 is
approximately the (compensated) wage elasticity of labor supply in
the aggregate. There is much empirical evidence on labor supply.
My reading of the evidence suggests that the lowest and highest
values that can be seriously entertained are O and 0.6 respectively.
90 /
THOUGHTS ON LAFFER CURVE
Values oft{ and
t;
TABLE 3
from Equations (9) and (10)
i::=0
high elasticities
low elasticities
t*L
t*L
= .77
ca,
.91
t*K
t*K
=
t*K
t*K
=
NE
= NE
£=1
high elasticities
low elasticities
t*L
t*L
=
=
.38
.45
.85
.64
i::::c2
high elasticities
low elasticities
NE
t*L
t*L
=
.26
.30
t*K
t*K
.57
.42
Nonexistent (i.e., no tax rate under 100% solves equation
(IO)).
A is a trickier parameter; it is the elasticity of capital supply to the
market (versus to the home sector) with respect to the rate of
return. It is hard to know what to make of this parameter in a
static model. Will I really keep my capital home if the return in the
market is low? Doing what? In a dynamic model, I guess
households supply capital to the market by saving, and the steadystate interest elasticity of capital is the same as the interest elasticity
of saving. I think the absolute limits on reasonable estimates of the
interest elasticity of saving are probably - .05 < A < + .40, with
zero a strong candidate. This leaves the unconventional parameter£.
Since I have no idea of how to "guesstimate" £, I will simply try
three very different values: 0, 1.0, and 2.0.
Table 3 evaluates equations (9) and (10) for a number of
different sets of parameter values. The case denoted "high
elasticities" is (J = .6, A = .4; the case denoted "low elasticities" is
(J = .l, A = 0. The results are unambiguous. If£= 0, revenues keep
on rising right up to the point where the tax rate on capital income
reaches 100% ,8 and the Laffer point for the tax rate on labor
'It might be argued that, because of inflationary distortions in the tax system,
effective rates of taxation of capital under current inflation rates are over 100%
because laxes are being levied on negative real income. If this is the case, however,
the Laffer curve no longer follows from Rolle's Theorem, and may not turn down at
all.
fl LINDER /
91
income far exceeds what we actually observe. If£""' 1, Laffer points
do exist for both capital and labor. But the revenue-maximizing tax
rates still exceed the rates that characterize the U.S. economy
(though perhaps not by much in the case of labor). Only when E
gets as high as 2 does the peak of the Laffer curve come at tax
rates that approximate those actually levied in the U.S.-26-30%
for tabor income and 42-57% for capital income.
Finally, suppose that the elast1cities of supply of capital and
labor are really much greater than I have allowed for here.
Suppose, for example, that both .:I. and {J are unity. Equation (9)
then implies that
will be as low as .30 if E exceeds 1.6; equation
(10) implies that t; is 0.4 when E = 3 .2.
I conclude that, given the CJL model, the only way the
contemporary U.S. economy could find itself on the down side of
the Laffer hill is if the parameter t is quite sizable. Unfortunately,
this is not a parameter we know much about. Pending evidence to
the contrary, I am inclined to think it quite small. But nothing
much hinges on this belief; all that matters is that£ not be huge. As
Table 3 shows, to be anywhere near the top of the Laffer hill with
current tax rates, E wilt have to be about 2. This means that a 100/o
increase in both wage rates and the rate of return on capital must
induce a 20% increase in the quantity of each factor supplied to the
market sector. I find this scenario quite fantastic.
tt
SUMMING
Up
To establish the existence of a Laffer curve in theory, we do not
need to know anything about either economics or the tax system.
Rolle's Theorem will do. But it is a long way from proving the
existence of a Laffer curve to arguing that existing taxes are on its
downhill side. While the down side of the Laffer hill may perhaps
be relevant to very narrowly-based taxes, back-of-the-envelope
calculations such as those presented here make it seem highly
unlikely that broad-based taxes could fall in this range. The specific
model presented in the paper by Canto, Joines, and Laffer does
nothing to dispel this belief unless the tax system (at the margin}
chases huge amounts of capital and labor out of the market system
and into the home production sector (or the underground
economy).
92 /
THOUGHTS ON Li\FFER CURVE
REFERENCES
Atkinson, A.B. and N.H. Stern. "Taxation and Incentives in the
U.K.: A Comment." Lloyds Bank Review, April 1980.
Canto, V.A., D.H. Joines, and R.I. Webb. "Empirical Evidence of
the Effects of Tax Rates on Economic Activity." mimeo,
University of Southern California, September 1979.
Discussion of the Evans Paper
STEVEN BRAUN
Aggregate supply is an old idea. Although discussed by Keynes
and the early Keynsians, most recent econometric models can justly
be criticized for not adequately developing the supply side. It is
therefore exciting to review a supply-side model created by one of
the most prominent model-builders. The Evans model was
commissioned by the Senate Finance Committee as an attempt to
incorporate supply-side effects which were not in existing
econometric models. My remarks are based on a version of the
model furnished to me courtesy of Dr. Evans (Evans, 1980).
Theory suggests a number of channels through which, in the long
run, a reduction in various tax rates might substantially increase
aggregate supply. This would make possible a higher level of real
output without inflationary consequences. Four of these channels
have been built into the Evans model. They are:
l. Because workers bargain for after-tax wages, a reduction in
personal tax rates decreases wage demands;
2. Because income taxes reduce the incentive to work, a
reduction in the personal tax rate increases both the
participation rate and hours worked;
3. Because business taxes reduce the incentives to invest,
reductions in these taxes will increase the stock of business
capital; and
4. Because interest rewards savings behavior, a rise in the aftertax rate of interest will increase savings.
Although theory suggests the possible existence of these channels, it
has little to say about their strength. Earlier model builders have
found substantial empirical support only for the third channelbusiness taxes. Evidence for the others have been mixed at best and
most other models do not contain them.
Steven Braun is Economist, Wage, Prices and Productivity Section, Board of
Governors of the Federal Reserve System. The views contained in this paper do not
necessarily reflect the views of the Board or its staff. The author is grateful for
discussions with Albert Ando and Jared Enzler, and thanks Ron Sege for research
assistance.
93
94 /
EVANS DISCUSSION
Dr. Evans differs from others in claiming to have been able to
measure these channels and he finds their strength to be
considerable. I find this evidence unconvincing.
Let us begin by introducing one of the devils of the supply-side
pantheon. Figure I shows the average and marginal tax rates
computed by Dr. Evans from the IRS tables. Except for the 1964
tax cut these variables show a strong upward trend-a fact which is
important in understanding this model. For comparison purposes I
have computed an average tax rate based on data from the national
income accounts. Since this series allows for the standard and
personal deductions, which are excludable from income used above,
the tax rate level is lower, and its trend is slightly less steep.
Now let us turn to some key equations which incorporate the
various supply-side channels. Let me begin with the wage equation
(which is the first equation in the Appendix). This wage equation is
for the most part a rather standard looking inflation augmented
Phillips curve. The rate of wage change depends on (ignoring the
various dummy variables) the inverse of the unemployment rate, the
rate of change in the CPI, the rate of change of output, and the
level of the average personal tax rate. Presumably, the idea is that
workers bargain for after-tax wages. But if this were true, the
growth of taxes rather than the level of taxes should be included.
The effects of this misspecification produce odd simulation results.
A one time reduction in tax rates will affect the rate of wage
growth not only in the following years, but for eternity. Using the
coefficients of this equation, I have calculated that a reduction in
the tax bill of, say, 3 percent will lower wages also by 3 percent
after 6 years. But after 12 years, wages decline by 6 percent-twice
the reduction in taxes! This equation is going to make the KempRoth tax cut look very good! Notice that the effect of prices on
wages is very low (0.6) implying that workers suffer from money
illusion so that even in the Jong run, a permanent higher level of
inflation could lower unemployment. It appears that the tax rate is
picking up some of the trend in inflation.
Labor force participation rate equations are perhaps the most
visible and the oldest of the supply-side features. Because of
conflicting income and substitution effects, the sign of the wage
variable could go either way. However, an upward sloping supply
curve is plausible. Evans' participation equations (an example of
which appears in the Appendix) does not seem to produce credible
evidence for this proposition. Since one of the independent
variables is the real after-tax wage bill, an increase in employment
FIGURE 1
Average and Marginal Personal Tax Rates
.45
.45
AO
.40
Marginal Rate'
.35
.35
.30
.30
.25
.25
Average Rate {NIA Basis)'
.20
.20
.15
.15
,__.......__...___.,_ _,______......__.__..,___,__ _.____._......__..__.__..___.,_..,j,.._........._.___, . JO
74
76
1978
66
68
1960
62
70
64
72
'Computed by Evans from IRS, Statistics on Income. This series includes state
and local taxes and social security taxes,
'Computed by Evans from IRS, Statistics on Income. This series includes state
and local taxes and social security taxes.
'Average tax rate computed on an NIA basis:
(tax and nontax payments) + (personal contributions for social insurance)
(personal income) + (personal contribmions for social insurance)
96 I
EVANS DISCUSSION
has the same elasticity as an increase in real after-tax wages. We all
know what the trends are in employment and participation. Thus
the coefficient of the wage rate in this equation is guaranteed to
show the correct sign. Notice also that the level of unemployment
does not enter this equation, only its first difference. Will the
participation rate snap back to trend when the unemployment rate
stops growing?
The effect of tax rates on labor supply in this model is only
partially captured in the labor force participation equations.
Claiming that increased taxes reduce hours worked, Evans models a
tax effect in the total manhours equation (shown in the Appendix).
Here taxes are shown to reduce hours worked. This is a curious
equation. If the level of productivity were included, rather than its
growth rate, this equation would be close to an identity. However,
productivity enters only through its growth rate. Because the
omitted variable, the level of productivity, also has an upward
trend just as the tax rate does, it is likely that the negative sign on
the tax rate occurs because it is picking up the trend of the omitted
variable.
Even this negative sign is curious. For a given level of output, a
decrease in the tax rate will decrease manhours worked. Since
output is also in the equation, and therefore held constant, this
means that productivity has fallen. Thus, productivity falls when
the tax rate falls. I seriously doubt that this is the effect that Dr.
Evans wanted to show. I understand that the model presented to
the Senate Finance Committee does not simulate. Surely, this
equation must generate some problems.
Consider how this equation interacts with the participation
equations. When the tax rate rises, manhours fall, causing the wage
bill to fall. This in turn causes the participation rate to fall. So
while it is claimed that the participation equations only captures
part of the effect of higher taxes, we see that in simulations, this
will not be true.
The productivity equation is discussed at length in Evans' paper
in this volume. However, this equation is really superfluous since
productivity is implicitly computed in the total manhours equation.
Besides the growth of productivity appearing in the manhours
equation and the capacity equation, I do not see how else the
productivity variable is utilized. If it were utilized, it would be
inconsistent with the manhours equation. (By the way, why does
the level of secondary workers and the level of government
regulation affect the growth of productivity?)
BRAUN /
97
This model claims the ability to evaluate the effectiveness on
investment of several forms of corporate income taxation. Reducing
the corporate tax rate, for example, is found to be more effective
per Treasury dollar than increasing the investment tax credit. I find
these results to be based on a peculiar structure of the investment
sector (see Appendix). The demands for new orders is separately
influenced by four elements of the cost of capital: an index of
industrial prices, the corporate tax rate, the depreciation allowance,
and the investment tax credit. Then a single cost of capital variable
affects how new orders are translated into investment. This raises
problems of double counting the effects of these taxes. Since
consumer expenditures are also in both equations, there seems to be
double counting here too. These extra terms in the investment
equation raise the possibility that investments may occur without
antecedent new orders. I know of no theoretical explanation for
this peculiar structure, nor has one been offered.
The effect of the interest rate on savings has long been a puzzle.
As Keynes recognized, "Some of the subjective motives towards
saving will be more easily satisfied if the interest rate rises, others
will be weakened." 1 Since Dr. Evans claims a substantial effect, let
us examine his equation (the fourth equation in the Appendix).
Consumption is a function of lagged consumption, current and
lagged income, and the after-tax real rate. However, wealth is
omitted, and this omission is serious in interpreting the effects of
changes in interest rates. Since the savings rate falls when wealth
rises relative to income, and since wealth rises when the interest rate
falls, the interest rate in this equation may be merely picking up the
wealth effect. So after examining this equation, one still does not
know whether the income or substitution effect dominates.
With these remarks in mind, it is time to ask how this model can
help analyze aggregate supply. Reducing the personal income tax to
reduce wage demands is dependent on an equation in which tax
levels influence wage growth. Reducing personal taxes to increase
labor force participation is dependent on an equation that cannot
distinguish an increase in wages from an increase in total
manhours. Reducing personal taxes to add to labor input is
dependent on an equation that omits the level of productivity.
Reducing the corporate tax rate to spur investment seems to be
dependent on an investment sector that counts this parameter twice.
'John M. Keynes, The General Theory of Employmenl, Interest, and Money,
Harcourt, Brace & World Inc., !964, p. 93.
98 /
EVANS DISCUSSION
Reducing taxes on saving to encourage saving seems dependent on
an equation that confuses the wealth effect with the interest rate
effect.
Each of these prescriptions seem to be directly connected with an
error in the model. What then have we learned about the world?
REFERENCES
Evans, Michael K. "Supply-Side Model." Evans Economics, Inc.,
mimeo, 1980.
Evans, Michael K. "An Econometric Model Incorporating the
Supply-Side Effects of Economic Policy." In this volume, 1981.
Keynes, John M. The General Theory of Employment Interest and
Money. Harcourt, Brace, & World Inc., 1964.
BRAUN/99
APPENDIX
EVANS SUPPLY-SIDE MODEL' (selected equations)
Wage Equation (page 8.11)
WRM4
= -
.9 + .004 STRIKES + .008 DWPP + .6 CPI415
(-3.l) (2.3)
(3.3)
(7.1)
+ .3 AVGSUMl8 +. I XIPM4 + .8 UNI18
(3 .5)
(3.0)
(5 .2)
R'= .83
WRM4 = percentage change2 of the average hourly wage in
manufacturing
STRIKES = Dummy variable, auto and steel strikes
DWPP = Dummy variable, wage-price freeze
CPI415 = percentage change' in the CPI, (distributed lag)
AVGSUM 14 = sum of average personal tax rates, (distributed lag)
XIPM4 = percentage change2, index of industrial production
8
UNI 18 = I/( I
UN8), where UN8
i=I
=
unemployment rate if <8
8 if unemployment rate ),8
labor Force Participarion Rare (Females, 25-54), (page 7.53)
LFPF2554 = .335 + .036 WMARG14 - .02 UN13 + .82 CPI41
(14.2) (3.7)
(-4.8)
(3.0)
+ 1.3 CPI45
R'=.85
(6.0)
LFPF2554 = Labor force participation rate, females 25-54
WMARGl4 = (real wage and salary disbursements)(l-marginal tax
rate), (distributed lag)
UNl3 = UN-, - UN-,
CPI41 - percentage change" in the CPI, lagged ( -1)
CPI45 = percentage change' in the CPI, lagged ( - 5)
'Based on Evans, 1980. The page numbers from this document are as indicated.
'These are not simple percentage changes. Rather, they are defined as,
4
X - (1/4) IX_;
I
4
(l/4) IX_,
I
100 / E V A N S D I SC U S S I O N
Manhours (manufacturing), (page 7. 69)
EHMFG40 = 33313 + 101 XIPMS + 64 XIPM14
(26.)
(14.8)
(6.1)
- 41965 AVGSUMI 8 -- 1458 PRODQIS
(--23.4)
(-7.9)
- 9.5 KPPRODIS
(- 2.1)
R'=.97
EHMFG40 = manufacturing manhours
XJPMS = index of industrail production, manufacturing
XIPM14 = distributed lag of XIPMS
AVGSUM I 8 = sum of average personal tax rates, (distributed lag)
PRODQ18 = annual percentage change in private nonfarm business
productivity (distributed lag)
KPROD l 8 = (manufacturing capital stock) - (pollution control
capital stock), (distributed lag)
Consumption, (page 3. 19)
C = constant + .336 C, + .296 Y + .299 Y-, - 2.04 r
(estimated by principal components, long-run MPC = .89), R 2 = .997
C
Y
= total consumption expenditures per capita, 1972$
=
disposable income per capita, 1972$
r = after tax real rate of return
Jnvestmenr Sector
New Orders Equation, (page 4.68)
NOR = 3.4 + .4 PWINOR + .I CDNOR + 3.6 IHSLI
(.7)
(5.0)
(22.8)
(8.0)
+ 45. DCPNOR + .08 XIPDSENO - 35.6 EFFTAX
(9.0)
(5.8)
( - 6.3)
+ 6.4 ZENOR + 6.3 DITC2
(2.5)
(1.4)
R 2 = .994
NOR = New orders, all manufacturing
PWINOR = WPI, industrial commodities, (distributed lag)
BR A UN /
101
CDNOR = consumption expenditures, durables and non-durables,
(distributed lag)
IHSLl = total housing starts, (distributed lag)
DCPNOR = index of capacity utilization (special functional form).
XIPDSENO = industrial production index, defense and space
equipment
EFFTAX = corporate tax rate
ZENOR = tax savings from depreciation allowance
DITC2 = investment tax credit, (distributed lag)
Investment Equation, (page 4.80)
IPE = ·- 12.5 + 1.3 NORL6 - 1.8 CREDLS + .09 CDNL
(4.6)
(17.1)
(-6.1)
(8.4)
- I
a;
RCCPL3~;
(5 .9)
R'"" .992
IPE = business fixed investment, producers durables
NORL6 = new orders, all manufacturing, (distributed lag)
CREDL5 = index of credit rationing, (distributed lag)
CDNL = consumption expenditures, durable and non-durables,
(distributed lag)
RCCPL3 = cost of capital, (distributed lag)
Discussion of the Evans Paper
ALBERT ANDO
While the political discussion in the United States has suddenly
focused on the so-called "supply-side effects," this is not a new
discovery in the literature of economics. I don't believe any one has
denied the theoretical possibility that labor supply may depend on
the real wage rate, and that personal savings may depend on the
real after-tax rate of interest. The question has always been about
the empirical order of magnitudes of these responses. In the case of
savings, there are two further questions: whether or not an increase
in savings will necessarily lead to correspondingly larger investment
in capital goods, and how much the additional investment will
contribute to potential and actual output.
Evans appears to claim in his summary (Evans, 1981) that he has
resolved all these empirical questions, and his new model is now
capable of predicting major effects of macro and micro policies
aimed at supplies of productive factors. A detailed appraisal of his
claims is difficult because they are embedded into a large model,
and the model in question is not laid out for easy understanding.
I therefore propose to look at one critical group of equations in
Evans' model as a representative of the model. Since Evans himself
says that the equation explaining productivity plays the central role
in his model, let us look at this equation as the starter. It is given in
his summary paper (Evans, 1981) and (Evans, 1980, pp. 7.88-7.89).
First of all, we have to presume that Evans, when he defines
PRD as private nonfarm business productivity, means by this
variable output per manhour in this sector. The dependent variable
in this equation is the rate of change of PRD. We may dispute the
choice of variables that Evans introduces to the right-hand side of
this equation. In order to concentrate our attention on less
controversial issues, however, let us accept his choice as
appropriate. There remains the question of the form of this
equation.
Albert Ando is Professor of Economics at the University of Pennsylvania.
103
104 / EV ANS D l SC USS ION
The most curious thing about this equation is the lack of
correspondence of dimensions among variables, and consequent
implausible steady state characteristics associated with it. As I
indicated before, the dependent variable of this equation is the rate
of change of productivity per manhour. Yet some of the
independent variables, notably the ratio of the number of secondary
workers to total employment, and direct federal government
expenditures on regulations in current dollars, are level variables.
To understand clearly the nature of absurd results that follow
from this setup, let us consider the situation in which all
independent variables, including the two variables mentioned above,
remain constant for a while, generating a constant rate of growth
of productivity. Now suppose that the proportion of secondary
workers in total employment increases by some fixed amount, say
l 07o, and remains at the new level thereafter. Then, the rate of
growth of productivity (not the level of productivity) declines by a
fixed amount. (If I believe in the definitions and numerical values
reported in Evans' paper, it does so by .84% per year. However,
this is too large an effect for me to accept for the first year, the
only period in which this equation makes any sense, and I suspect
that there may be some misprint and/ or errors of units in the
definitions.) Consequently, a once-and-for-all increase in the
proportion of secondary workers to total employment will,
according to the Evans equation, lead to a continual decline in the
level of productivity relative to the reference path. Taking Evans'
equation literally, if the ratio of the secondary workers to total
employment rises 1%, say from 400-/o to 4i07o, and remains at the
new level thereafter, the level of productivity will fall by .84% the
first year, 8.7% during the first 10 years, and 18.3% during the
first 20 years, and will continue to decline forever,
The level of federal government expenditure on regulation is even
more absurd. The variable entered is total expenditure in current
dollars. Thus, if the total expenditure in current dollars rises slowly
for whatever reason, perhaps because of inflation, perhaps because
the scale of the economy increases, the rate of increase of
productivity must fall even if the federal government expenditure
on regulation is becoming smaller and smaller relative to total GNP
in current dollars. (As an illustration, suppose that GNP in current
dollars is growing at 7% per year, while the government
expenditure on regulation in current dollars grows at 4% per year
and the inflation rate is 5% per year. The rate of growth of
productivity will still continue to decline, according to Evans'
AN DO
I 105
equation.) Since no other variable on the right hand side of this
equation is an extensive variable that grows with the growth of the
economy as a whole, the presence of this variable, the level of
direct federal government expenditures on regulation in current
dollars, must eventually make the rate of increase of the
productivity negative, even though this expenditure as a proportion
of GNP in current dollars becomes sma11er and smaller.
Even some of the more reasonable-looking variables have their
troubles. The ratio of business fixed investment to gross national
product sounds like a reasonable candidate for influencing the rate
of growth of productivity. But anyone who has worked with models
of growth will soon realize that this is not really a sensible variable.
The variable of this sort that can be fairly readily accommodated in
this context is the rate of growth of capital stock per employee net
of depreciation, not the gross investment-gross output ratio.
His statement that the relevant ratio is investment in constant
dollars to the gross output in constant dollars, and not the ratio of
current dollar values is also a serious suspect. The only theory
bearing on this point in a multi-goods model that I am aware of is
my own (Ando, 1964); the conclusion in that theory was that the
only aggregate ratio that could be interpreted meaningfully was the
ratio of the value of capital goods to the value of output, not the
ratio between implicitly deflated figures in national income
accounts. But that proposition was in the context of a specific,
well-defined model, and here we are dealing with an assertion by
Evans, which is apparently not based on any coherent view of the
world.
On a basis of these observations, I conclude that Evans' equation
explaining the rate of growth of productivity, the equation which,
in Evans' own words, reflects the main thrust of his model (Evans,
1981), is not worthy of our further attention.
Even though Evans imputes great importance to the equation for
the productivity discussed above, the output of this equation feeds
into only two places in Evans' model, and it is probably worth
extending our review of Evans' model to cover these two additional
groups of equations.
The first group of equations in which output of the productivity
equation plays a role is the equation expressing total manhours as a
function, among other things, of total output and productivity. One
typical such equation in Evans' model is given as the third equation
in Braun's discussion (Braun, 1981), also (Evans, 1980, p. 7.69).
Since productivity, PRD, is defined as output per manhour, if all
106 /
E VA N S D JS C U S S I O N
definitions are assured of consistency everywhere, then the
manhours equation must be an identity, namely
EHMFG40 = XIPMS ·
l
PRO
where
EHMFG40: manhours in manufacturing
index of industrial production, manufacturing
XIPMS:
PRO:
productivity per manhour, private, nonfarm, business
sector
The identity does not hold because EHMFG40 and XIPMS refer
to manhours and output in manufacturing industries while PRO
refers to productivity per manhour in private nonfarm business
sector, XIPMS is an index of production rather than total volume
of production, and for a host of other definitional discrepancies.
But I do not see anywhere in Evans' writing or in his handling of
these equations any indication that Lhere are any important
conceptual reasons why the above identity should not hold. Yet, the
manhours equation Evans actually estimates and reports is basically
of the form
EHMFG40
=
constant + cr,XIPMS +
ai
LPRO
PRO + .. · · ·
where dots represent additional terms in the equation which are not
related to output or productivity. In other words, Evans has
substituted for the level of productivity, PRO, in the identity the
rate of change of productivity, linearized the equation, and then
introduced a host of other variables. I see absolutely no
justification for this substitution or for linearization. That it has
disastrous consequences should not come as a surprise to us.
For instance, given a level of output and a rate of growth of
productivity (not the level of productivity), other things equal, the
manhours needed to produce this output remains the same. To put
it another way, if output remained the same from year O to year IO,
while productivity (output per manhour) increased at the constant
rate of 3 % per year, then the man hours required to produce this
same output in year zero and in year ten are nevertheless the same.
If this statement sounds contradictory, it nevertheless accurately
reflects the statement embodied in the equation.
Clearly, such a characteristic of the equation cannot be reconciled
with data, and something else must enter this equation to help
ANDO /
107
reduce the manhour requirement per unit of output over time. The
only variable introduced by Evans into this equation with the type
of time trend for performing this function is, of all things, the sum
of average personal tax rate. (This rate, computed by Evans, has a
strong positive trend over time. Whether or not this is a reasonable
concept is another matter, since one could also compute the average
rate which does not have as much trend). It is therefore not at all
surprising that the average personal tax rate acquires a strong
negative coefficient.
Evans seems to suggest that the definitional identity among
manhours, output and the productivity does not apply here because
manhours and output measures that enter the manhours equation
reflect short-run, cyclical movements of these variables while PRD
reflects the longer-run, secular trend of the productivity. This
excuse does not wash because PRD is simply calculated as the ratio
of output to manhours in each year, and to reflect this fact, the
equation explaining the rate of change of PRD has explanatory
variables that are strictly related to short-run, cyclical variation in
productivity, such as the rate of change of GNP and the index of
capacity utilization.
l must conclude, therefore, that Evans' formulation of the
manhours equation makes no sense, that its fit against data is
purely accidental, and that the large negative coefficient for the
sum of average personal income tax rates estimated in this equation
is at best due to a combination of vagaries of the pattern of time
series data and of serious misspecifications of the equation form.
I would like to repeat here a curious feature of this manhours
equation observed by Steve Braun (Braun, 1981). Since output and
the personal income tax rate enter separately as independent
variables in this equation given the level of output, an increase in
the average personal income tax rate will reduce manhours. That is,
the higher the average personal income tax rate, the higher the
productivity per manhour. I am sure those who are interested in
supply responses to a change in the tax structure are interested in
getting an explanation for this phenomena.
The only other place where the variable PRD plays a role is in
the equation defining the maximum production. It is a definition
rather than an estimated equation, and takes the form (Evans,
1980, p. 11.11)
where
108 //
108
Ev
VAN
ON
N
F
A N 5S 0DJ1 SSC
C U S S5 II 0
XIPC: index of maximum production in the manufacturing
manufacturing sector
= 100.0
1967 =
EM*:
EM: "full
“full employment"
employment” supply of labor in manhours
K:
<lstock"
“stock” of capital goods, somehow measured
We shall not discuss the serious problem of how EM* and K are
measured by Evans, since the focus of our discussion here is how
the measure of productivity is utilized in the model. Evans says
says that
PRODQ is the annual change
change in private nonfarm business
productivity. Evans could not mean what he says, since if we take
him literally, it makes no sense, and
don’t believe that he could
and II don't
have generated the data reported by him preceding the specification
of this definition (Evans, 1980, p. 11.
lO). I therefore assume that
11.10).
PRODQ is something that does make a minimum of sense, say, the
the
accumulated value of the rate of change of productivity starting
starting
from some initial period, with the initial value of it coordinated
with the constant term in the definition so as to fit the data.
Even then, this equation makes no sense. If PRODQ is some
concept such as the one I suggested above, and in any case it is
based on the measure of productivity per manhour, then anyone
who has ever worked with growth models based on homogeneous
production functions, particularly the Cobb-Douglas function, will
know that
that the productivity measure cannot be introduced into the
the
production function unmodified. This is because productivity per
manhour already reflects the contribution of an increase in the
capital-labor ratio, as Evans’
Evans' equation explaining the rate of
of change
of PRD
PRD tries to describe. Therefore its introduction together with
the capital stock into the production function without the proper
restriction is a double counting. One possible, though rather naive
and unrealistic way to handle this problem is to replace the term
00o by e1/,PRooo in the above equation defining XIPC (assuming,
er•
e~°°~
e~RODQ
Evans' PRODQ is basically
always, that my reinterpretation of Evans’
correct). At least, this will make the structure logically
logically consistent.
Even if PRODQ is introduced correctly into aa Cobb-Douglas
Cobb-Douglas
production function, it is most doubtful that such a formulation
will be adequate for estimating the maximum productive capacity of
the U.S. economy. On a year-by-year basis, at least
least some of capital
goods are not malleable. Hence, it is a doubtful procedure to utilize
goods
any production function for the whole economy (or a large segment
of it) incorporating the concept of the aggregate capital stock in
order to describe the production possibilities in the sense that Evans
uses the concept
concept of capacity or maximum output. Moreover, the
AN DO /
/ 109
ANDO
depreciation or abandonment of capital goods may very well
depend on movements
movements of relative prices. But this is really taking us
the most
too far afield away from the
the subject
subject at hand, namely, the
Evans’ model.
obvious defects of Evans'
In this note, I1 have so far limited myself to discussing the
the
Evans’ model and two sets
explanation of productivity in Evans'
sets of his
equations in which the productivity so explained is a critical input. I1
have, however, looked at the large, 850 page document (Evans,
(Evans,
1980), which is an explanation of his model, and II must report that
everywhere I turned, every equation that II have examined, I have
objections rather similar in nature to the ones
ones I have been
discussing. Very few of his equations make "good
“good sense"
sense” as this
convenient term is normally understood by most of us economists,
and most of them imply what I would consider rather absurd
behavior of the dependent variable when
explanatory
when one of its explanatory
all other
variables is changed from one level to another while all
explanatory variables are held constant. That is, most of his
equations have what may be called "unacceptable
“unacceptable steady state
properties.’’
properties."
In his oral discussion, Evans took the position that
that he did not
care what properties individual equations possessed, so long as the
whole system generated dynamic behavior in simulation that
that
appeared reasonable. Although Evans is not alone in taking
taking this
position,
position 1, II for one do not consider this position a tenable one in
building econometric models. Some misspecifications in short-run,
short-run,
of some subsidiary equation might be tolerated,
dynamic behavior of
after a careful examination to make sure that
that such a
misspecification did not affect the overall behavior of the system,
either for good or for bad. The requirement that the whole system
behave in an understandable, reasonable manner under a variety of
shocks is a useful criteria in judging the quality and acceptability of
any econometric model, but is a criteria in addition to, and not in
place of, the traditional
traditional one that each equation in the system be
sensible.
Evans’ two papers (Evans, 1980
My review of Evans'
1980 and 1981), then,
convinces me that the whole model does not make much sense, and
II cannot have any confidence in his model nor in any analysis based
on his model. I have seen many errors and bad judgments in many
econometric studies, including my own. Seldom have I seen,
seen,
1
in
II recollect that Jay Forrester
Forrester tended to take a position somewhat similar to this in
find a specific
reference at
his writings in Industrial Dynamics, but
hut II am unable to find
specific rererence
the present time.
110
110 //
EVANS
EV ANS OtSCUSSION
DISCUSSION
however, a large-scale work such as this one of Evans, undertaken
by a reputable and experienced econometrician,
econometrician, where the pattern
of such major defects have dominated so large a part of the entire
work.
This is really too bad, because the case for rationalizing the tax
tax
This
and transfer payment structure of the United States seems to me to
be quite strong. The shift from the personal income tax to the
the
expenditure tax originally proposed by Kaldor has its appeal,
with adequate taxation of estates. II
provided that is is combined with
the vexing
believe
believe such a shift will make it much easier to handle the
problem of capital gains, to cope with inflation and indexing of the
tax base, and may possibly stimulate savings. A great deal of work
is beginning to be done in this area. I believe the rationalization of
depreciation allowances should be pursued, and the immediate and
complete write-off of capital good purchases as cost should be
considered as one possible alternative, more in the case of
considered
producers’
producers' equipment than in the case
case of structures. Going beyond
that, some form of integration of corporate profit tax, personal
income tax, and the social security tax would
would be worth analyzing.
of taxation by
An even more difficult problem is the coordination
coordination of
the
the federal, state, and local governments. On the transfer side, any
the
movement to make payments less dependent on income of the
recipient is likely to be helpful. The aim is, as it always has been, to
recipient
design the tax and transfer payment system which raises the needed
revenue, approach the desired redistribution of income as closely as
possible while minimizing the distortion of relative prices.
There are many
many careful studies of these possibilities, although
although
they are all quite
quite incomplete, and further research on them as well
as open public discussion of these issues should prove helpful in
formulating our economic policies in the coming decades. A work
such as Evans’
Evans' new model, undertaken at public expense, and well
well
publicized, claiming so much and yet so misleading, is likely to
divert the attention of both economists
economists and the public away from
basic issues and focus it on questionable gimmicks, raising
raising false
will, in the end, retard
expectations in the process. I fear that it will,
rather than advance the cause of fundamental reform of our tax
and transfer payment structure. I hope that I am wrong in this
premonition.
premonition.
ANDO/
ANDO
/ 111
REFERENCES
Ando, Albert. "An
“An Empirical Model of United States Economic
Growth: An Explanatory Study in Applied Capital Theory."
Theory.” In
Models of Income Determination, L. R. Klein, ed., Vol. 18
18 of
Studies in Income and Wealth, National Bureau of Economic
Research, Princeton: Princeton University Press, 1964.
"Comment on Evans’
Evans' Supply-Side Model.”
Model." In this
Braun, Steven. “Comment
volume, 1981.
1981.
Evans, Michael
unpublished,
J.
1. Supply-Side Model. Evans Economics, Inc.,
1980.
"An Econometric Model Incorporating
Evans, Michael J. “An
Incorporating the
Policy." In this volume, 1981.
1981.
Supply-Side Effects of Economic Policy.”
Policy and Corporate Investment
Tax Polky
LAWRENCE H. SUMMERS
1NTROOUCT~ON
INTRODUCTION
The proposition that the level of business fixed investment in the
United States should be increased commands almost universal
support. Increasing the rate of investment is widely seen as a
panacea for a variety
variety of economic problems including inflation,
declining productivity, and the fall of the dollar. While there is
agreement as to the inadequacy of business
business fixed investment, there
shortfall. For example, in
is little agreement as to the causes
causes of the sho1tfall.
a recent proceedings volume of the American Economic Review,
Blinder concludes with Robert Hall that "The
“The principal source
Alan Blinder
of inadequate capital formation has been our failure to do
do anything
about recessions, not our active use of anti-investment stimulative
policies,”
policies," while Martin Feldstein (1980) argues that
that the interaction
of inflation and taxation accounts for much of the decline in
corporate capital accumulation that has taken place over the
last decade.
This paper presents an overview of the issues connected with the
in the
relationship between tax policy
policy and corporate investment,
investment. ln
first section of the paper, post-war trends in capital formation and
corporate sector profitability are examined. While the share of
gross investment in GNP has remained almost constant, the rate of
net productive investment expressed as either a fraction of GNP or
the capital stock has fallen sharply during
during the 1970s.
of the
l970s. This decline
has been associated
associated with a substantial fall in the market price of
capital, and in the after-tax rate of return to investors in
corporate capital,
the corporate sector. The reduction in after-tax returns to corporate
investors, while partially related to a fall in the
the pre-tax rate of
return on capital, is in large part due to the interactions of inflation
and our non-indexed tax system.
The second section presents a cautious view of the social gains
from increased corporate investment. Even a large increase in net
Summers is Assistant Professor of Economics,
Lawrence H. Summers
Economics, Massachusetts
Institute of Technology and Research Associate, National Bureau of Economic
Research, Cambridge, Mass.
115
116 /
TA X
POL l C Y A ND
C O R PO R A T E J N V E ST M E N T
business investment would not be sufficient to offset more rhan a
small part of the productivity slowdown. Given a fixed path of
monetary policy, tax reductions to spur investment are likely to
increase rather than reduce the rate of inflation. The real payoff
from increased investment, it is argued, comes from the very
favorable terms of trade between consumption today and
tomorrow. Foregoing a dollar today leads to an increase in
potential consumption of two dollars only seven years hence. At
these rates, most persons would find more investment attractive.
Traditional econometric studies of the relationship between tax
policies and investment are reviewed in the third section. It is
argued that the type of investment equations embodied in most
large scale econometric models do not offer meaningful guidance as
to the effects of tax policy on investment. Since output is
traditionally held constant, the capacity effects of increased
investment cannot be captured in these formulations. As
fundamental, the usual approach yields results which are very
inconsistent with the assumption that expectations are rational. As
an example of the misleading nature of standard econometric
investment equations, the role of general expansionary policy as a
device for spurring investment is considered. It is argued that as
long as one accepts the view that there is no long run Phillips curve
tradeoff, it is not possible for the level of general stimulus to have
any effect on the long-run growth of the capital stock. The
accelerator does not offer a useful route to increasing corporate
investment.
An alternative methodology for viewing corporate investment
incentives is presented in the fourth section. It is shown that an
asset price approach to evaluating investment incentives avoids the
difficulties inherent in traditional investment equations and avoids
the "Lucas critique" of being unstable across changes in policy
regimes. The effects of various tax policies on investment are
analyzed using this approach. 1t is argued that through judicious
policy choices substantial stimulus to investment can be achieved
without any large revenue cost to the government.
The fifth section examines the general equilibrium effects of a
change in business taxation. ft is argued that business tax incentives
can only spur investment if the supply of savings flowing to the
corporate sector is increased. This can occur in one of two ways.
An increase in the after-tax rate of return may raise the savings
rate. Alternatively, it may lead to an increase in rhe share of wealth
allocated to the corporate sector. Each of these mechanisms is
s UM MER S I 117
examined briefly. The paper concludes by discussing the
appropriate macroeconomic policy mix to accompany business tax
reductions.
INVESTMENT AND THE PERFORMANCE OF THE NON-FINANCIAL
CORPORATE SECTOR
This section examines trends in the rate of non-financial
corporate investment, and profitability during the post-war period.
The focus here is on corporate capital formation because its alleged
deficiencies have received the most attention and it is most plausibly
influenced by tax policies. It is important to recognize, however,
that corporate investment makes up only about 60 percent of the
total. About 25 percent of investment is residential and the
remainder is done by non-corporate business. The trends illustrated
here hold for total business investment as well. There have been
rather divergent movements in the rate of residential investment and
the valuation of housing capital. These are examined in the paper's
final section.
TRENDS IN THE RATE OF CORPORATE INVESTMENT
Various measures of the rate of non-financial corporate capital
investment are displayed in Table 1. The type of measure most
usually relied on, a comparison of gross investment with gross
output, is shown in Table 1. It has been surprisingly constant
throughout the 1951-79 period, and has been close to its long-term
average during the last decade. However, focusing on gross
investment may be very misleading. The key variable for economic
performance is the rate of growth of the capital stock. This depends
on net investment rather than gross investment. The rate of net
investment as a fraction of gross corporate product has declined
quite sharply in the last decade as shown in column 2. 1 While it
averaged 0.036 over the entire 1951-79 period, it averaged only
0.024 during the 1975-79 recovery period. This corresponds to a
33 percent reduction in the rate of net capital formation.
There is a second important issue involved in assessing investment
performance during the 1970s. Regulatory requirements imposed in
order to protect the environment and workers' safety have forced
JThese estimates are based on the assumptions of straight line depreciation and
service lines of .85 Bulletin F. There is a strong argument to be made that both these
assumptions are conservative and so these figures understate depreciation and
overstate net investment.
TABLE I
Alternative Measures of the Rate of Non-Financial
Corporate Investment
Gross I
Net I
Pollution Adjusted Net I
Pollution Adjusted Net I
Year
y
y
y
K
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
0.138
0.134
0.138
0.137
0.136
0.146
0.146
0.131
0.124
0.131
0.128
0.129
0.125
0.130
0.141
0.146
0.139
0.136
0.138
0.133
0.129
0.128
0.135
0.140
0.123
0.120
0.127
0.124
0.127
0.045
0.038
0.042
0.034
0.039
0.047
0.044
0.021
0.022
0.030
0.025
0.033
0.031
0.039
0.053
0.059
0.049
0.047
0.048
0.037
0.032
0.035
0.043
0.040
0.016
0.017
0.027
0.028
0.032
0.045
0.038
0.042
0.034
0.039
0.047
0.044
0.021
0.022
0.030
0.025
0.033
0.031
0.039
0.053
0.059
0.047
0.045
0.046
0.034
0.027
0.029
0.036
0.033
0.009
0.010
0.021
0.022
0.025
0.043
0.036
0.041
0.031
0.038
0.045
0.041
0.018
0.021
0.028
0.024
0.032
0.032
0.041
0.057
0.064
0.050
0.048
0.048
0.034
0.027
0.031
0.039
0.033
0.008
0.010
0.021
0.024
0.027
SUMMERS/
SUMMERS / 119
119
TABLE 1I (continued)
51-54
55-59
60-64
65-69
70-74
75-79
0,137
0.137
0.137
0J29
0.129
0.140
0,133
0.133
0.124
0.040
0.035
0.037
0.051
0.037
0.024
0.040
0.040
0.035
0.031
0.050
0.032
0.017
0.038
0.032
0.031
0.053
0.032
0.018
T 51-79
0.133
0.036
0.034
0.034
0.034
Source:
text.
Source: as described in text.
investment.' This investment does not add
firms to engage in capital investment.’
to the productive (in
(in terms of measured output) capital stock.
Hence,
Hence, it should not be included in assessing changes in capacity
capacity
expanding investment. Data is available from the Department of
Commerce on the share of investment outlays devoted to pollution
control but not for occupational
occupational safety. These outlays have risen
sharply during the 1970s. in
In columns 33 and 4, net productive
output, and
investment is expressed as aa fraction
fraction of gross corporate output,
of the corporate capital stock. They show very pronounced declines
during the 1970s. The rate of growth of the non-financial corporate
sector’s
2.55 percent during
sector's capital stock in column 4 averaged only 2.
the 1970s compared
compared with 3.9 percent during the 1951-1969 period.
A similar pattern is exhibited by the data in column 3. The evidence
suggests that the rate of corporate capital formation has declined
significantly during the 1970s. This
conclusion would be
significantly
This conclusion
strengthened if account were taken of occupational safety
strengthened
investment expenditures, and the more rapid depreciation
depredation of the
the
3
prices.'
capital stock, which has occurred due to rising energy prices.
'H should be emphasized that pollution control expenditures are productive,
productive, irs
in that
9t
that
for clean air and water. These
they provide for
These benefits are real even
even though they do
do not
show sip
up in measured GNP. However, there is no apparent reason why a social
decision to increase environmental quality should lead to a decline in the rate of
“normal” investment. Hence,
"normal"
Hence, the appropriate
appropriate standard
standard of comparison is investment
net of pollution control expenditures,
expenditures.
net
‘The
prices has been
been to reduce substantially the value of
iThe impact of higher energy prlCes
is energy
energy inefficient.
inefficient, If this extra component
existing capital which ls
component were added to
depreciation, estimated net investment would decline even further. If one assumes
obsolete, the
that the energy shock rendered even
even 55 percent of the capital stock obsolete_.
over oneS
oneaverage net investment rate over the
J)07, or over
the last seven years declines by .007.
fourth of its average level.
fourth
TABLE 2
Cyclically Adjusted Rates of Investment
Pollution Adjusted Net I
y
Pollution Adjusted Net I
y
1965
1966
1967
1968
]969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
0.143
0.144
0.136
0.132
0.135
0.134
0.135
0.127
0.130
0.138
0.140
0.131
0.127
0.129
0.127
0.128
0.128
0.131
0.136
0.128
0.130
0.132
0.125
0.125
0.042
0.041
0.038
0.032
0.037
0.041
0.041
0.034
0.037
0.046
0.047
0.035
0.029
0.031
0.029
0.032
0.033
0.036
0.035
0.033
0.032
0.034
0.027
0.028
0.042
0.041
0.038
0.031
0.037
0.041
0.040
0.034
0.037
0.046
0.047
0.033
0.029
0.029
0.026
0.027
0.027
0.028
0.028
0.027
0.025
0.028
0.021
0.021
0.039
0.037
0.039
0.030
0.036
0.041
0.040
0.035
0.038
0.049
0.050
0.034
0.030
0.029
0.026
0.027
0.028
0.030
0.027
0.030
0.027
0.029
0.022
0.022
56-59
60-64
65-69
70-74
75-79
0.139
0.132
0.133
0.130
0.128
0.038
0.038
0.038
0.033
0.031
0.038
0.038
0.037
0.027
0.024
0.036
0.038
0.038
0.028
0.026
T 56-79
0.132
0.036
0.033
0.033
Gross I
Net I
Year
y
1956
1957
1958
1959
1960
1961
1962
1963
1964
Source; as described in text.
K
SU M M E R S /
121
Even casual inspection of Table 1 shows that the state of the
business cycle has a large impact on the rate of corporate
investment. The rate of investment by any of the measures peaks in
the boom years of the mid-60s, and reaches its low in 1975. In
assessing the long-term trends which should guide tax policy, it is
useful to abstract from cyclical factors. This is done by calculating
the cyclically adjusted rates of investment shown in Table 2. The
cyclical adjustments are based on regression equations of the form:
R,
= ao + a,RUMMt + a2RUMMt-1 + lit
where R1 is the rate of investment, and RUMM 1 is the married-male
unemployment rate which is used as a cyclical indicator.• The
" is calculated as:
cyclically adjusted investment rate Rt
A
Rr
= R1
-
a, (RUMMt -
RUMM) - a2(RUMMt-1 -
RUMM)
.It corresponds to the rate of investment which would have taken
place if the unemployment rate had been at its mean level.
The results show that the decline in net productive investment in
the 1970s is not a cyclical artifact. The share of corporate product
(column 3) going to this source on a cyclically adjusted basis has
declined from 3.8 percent during the 1956-1959 period to 2.5
percent during the l 970s. Thus, the decline in investment is almost
as great on a cyclically adjusted basis as on a cyclically unadjusted
basis. This conclusion also holds for the other measures of the
investment rate. The conclusion that the 1970s have witnessed a
large reduction in investment, inexplicable on the basis of cyclical
factors, appears almost inescapable. Below we examine some
possible underlying causes including the rate of profit and the
extent of capital taxation.
TRENDS IN CORPORATE PROFIT ABILITY
The data in Tables l and 2 illustrate the declines in investment.
Table 3 shows how various indicators of corporate profitability
have evolved over the last 25 years. The first column shows the pretax rate of profit of the corporate sector. While the rate of profit
has declined somewhat in the 1970s, it appears to have been fairly
constant at about 11 percent over the entire period. The second
column shows the total tax rate on corporate capital arising from
'Similar results were obtained using other indicators of the cyclical conditions such
as the unemployment rate of all men 25 and over, the GNP gap, and the rate of
capacity utilization.
TABLE 33
Corporate Sector Profitability
Year
aTotal
'Total Rate
of Return
bTotal
hTotal
Effective
Tax Rate
1955
1955
1956
1957
1957
1958
1959
1960
1960
1961
1962
1962
1963
1964
1964
1965
1966
1967
1967
1968
1968
1969
1969
1970
1971
1971
1972
1972
1973
1974
1974
1975
1975
1976
1976
1977
1978
1979
13.2
11.4
11.4
10.5
9.0
11.2
10.4
10.4
10.3
11.7
12.4
12.4
13.4
13.4
14.5
14.5
13.0
13.0
13.0
13.0
11.7
9.6
10.0
10.8
10.5
8.2
8.6
9.5
9.7
9.7
9.1
66.5
72.4
71.7
70.7
67.3
66.5
66.4
61.5
60.6
56.2
55.1
56.0
56.4
62.6
67.3
70.5
67.7
62.5
70.1
70.l
90.1
72.4
68.1
68.l
68.3
72.2
74.5
hReal
bReal Net
Rate of
Return
cRatio of
of
'Ratio
Market Value to
Cost
Replacement Cost
of Net Assets
4.4
3.2
3.0
2.6
3.6
3.5
3.5
4.5
4.9
5.9
6.5
6.4
5.7
4.9
3.8
2.8
3.2
4.1
3.1
0.8
2.4
3.0
3.1
2.7
2.3
0.92
0.92
0.85
0.87
1.04
1.02
1.14
1.09
1.20
1.29
1.35
1.35
1.20
1.21
1.25
1.25
1.12
0.91
1.00
1.07
1.01
0.75
0.71
0.80
0.73
0.68
0.65
Sources:
“State and Local
Local Taxes and the Rate of Return on
on NonaFeldstein and Poterba, "State
Financial Corporate Capital,”
Capital," NBER Working Paper #508R, p. 10.
10.
p. 23
bbJjjj~
Ibid., p.
23
CEconomic Report
of the
the President.
1980, Table
B-85.
'"Economic
Report of
President, 1980,
Table B-85.
SUMMERS/
123
SUMMERS
/ 123
the combination of federal and state taxes at both the corporate
and individual levels. A fuller discussion of the calculation of these
effective tax rates is contained in Feldstein and Summers (1979) and
Feldstein and Poterba (1980).
(1980), These data clearly show a very
pronounced increase in the
the taxation of corporate capital during the
55.1 percent in 1965
1970s. The tax rate has risen from 55.l
1965 to 74.5
percent in 1979.
This increase in taxes has largely been the result of inflation.
Inflation increases the taxation of corporate capital in three ways.
The two most important are historical depreciation, which added
over $25 billion
billion to corporate tax liabilities in 1979, and the taxation
tax liabilities
of nominal inventory profits which raised corporate tax
In addition, the taxation of
of nominal
by over $30 billion in 1979.'
1979.’ In
JO
capital gains is estimated to have imposed a tax burden of over $$10
billion. It is frequently argued that
that these effects are offset by the
the
fact
fact that corporations can deduct nominal interest payments for tax
purposes. This gain to corporations, however, is itself almost
completely offset by the increase in individual taxes on nominal
interest. Feldstein and Summers (1979) show that in assessing the
burden on
total tax hurden
on corporate capital, the taxation of nominal
interest nets out and can be neglected.
The after-tax
after-tax rate of return on corporate capital is displayed in
the third column. In the late 1970s it fell to only about one-half
one-half of
its level during the late 1960s.
seen
l 960s. From columns 1I and 22 it can be seen
that over half of this fall can be attributed to increased taxes rather
than to a decline in the pre-tax rate of return. This suggests that it
the return to capital
may be taxation more than any decline in the
which has accounted for the 1970s investment slowdown.
The values of Tobin's
Tobin’s q ratio of the
value of the capital
the market valne
stock to its replacement cost
are
shown
in
column
cost
column 4. The large
decline in the value of q during
the
1970s
of
course stands
during
stands out. It is
noteworthy that the 50
50 percent fall in q from the late 1960s almost
fall in the
exactly parallels the fall
the net return to corporate capital shown
in Table 3. It appears that a significant portion of
of the
the fall in the
total market valuation of corporate capital can be attributed to the
extra tax burdens imposed by inflation. If one accepts a "q"
“q” theory
extra
of investment
of the
the type discussed in the fourth section, this
investment of
provides further support
the hypothesis that increased taxation
support for the
5
Ttñ; extra
extra tax burden is in some sense voluntary since firms
firms could avoid it by
'This
switching to LiED
LIFO inventory accounting,
accounting, This does not make it less real. Firms
FlED because, rationally or irrationally, they perceive
presumably stay with FIFO
perceive some
so. Nonetheless inflation
inflalion does penalize
intramarginal economic gain from doing so,
burdens.
by raising their tax burdens,
them by
124 /
TAX POLICY AND CORPORATE INVESTMENT
has been an important cause of the decline in investment which has
taken place during the 1970s. Before examining the data bearing on
this question, we turn in the next section to an analysis of the
potential gains from increasing the rate of investment.
THE GAINS FROM INCREASED INVESTMENT
This section examines the potential social gains from tax policies
designed to increase corporate investment. The arguments which
have received the most popular attention, those linking investment
to productivity, inflation and unemployment, are examined first. It
is shown that none of these considerations provide a strong case for
investment tax incentives. A case for reducing the tax burden on
corporate capital is then developed in terms of micro- and macrointertemporal economic efficiency.
INVESTMENT, PRODUCTIVITY AND GROWTH
The poor performance of productivity in recent years has often
been attributed to the low rate of growth of the capital stock. It is
argued that increasing the rate of investment could have a large
effect on the rate of growth over the next decade. This prospect
seems unlikely. Prominent studies of the productivity slowdown,
Denison (1979), Norsworthy, Harper and Kunze (1979), show that
even after full account is taken of the decline in capital
accumulation, most of the productivity slowdown cannot be
explained. The limited potency of increased investment in spurring
productivity growth can be illustrated by a simple calculation.
Consider an economy which evolves according to the following
model:
(la)
yt "" KfL;·
(1 b)
Kt
(le)
It-l °"' dKt-1
(ld)
Li "" (l + g)L[ -
0
(l-6)KH
+
+
11~1
yYt
l
Equation (la) is a standard Cobb-Douglas aggregate production
function. Since the variable Y is to be interpreted as net output, it
is plausible to take a = .15 in using the model to interpret U.S.
economic performance. 6 The second equation (1 b} describes the
'The standard assumption !hat a ~ .25 is simply wrong in an analysis of this type.
The figure of interest is the share of net return to capital in net output. For the
corporate sector, this has averaged .15 over the !ast quarter century.
SU MME RS /
125
TABLE 4
The Rate of Growth of Output
Under Alternative Investment Policies
Years
y = .045
y=.060
y = .075
y = .090
0-5
6-10
l l-20
21-30
3.00
3.00
3.00
3.00
3.10
3.11
3.09
3.07
3.20
3.22
3.17
3. 13
J.30
3.3 I
3.24
3.17
accumulation of capital in the standard way. In the calculations
reported below, it is assumed that 6 = .08. Equation (le) specifies
that net investment is a constant fraction (y) of net output. This
figure has averaged about 4.5 percent1 over the last two decades for
the U.S. non-financial corporate sector. The final equation specifies
that the effective labor force grows at rate g. In the calculations
below g is taken to equal .03.
It is apparent the model has a steady state with a capital output
ratio of 1.5, and a rate of return on capital of .10. This is quite
realistic. As shown in Table 3, the pre-tax rate of return on
corporate capital averaged 9.6 percent over the last decade.
The 1979 capital-output ratio was 1.48. By simulating the model it
is possible to examine the effects of an increase in the share of
output devoted to net investment. This is done in Table 4 which
shows the rate of growth of output under alternative investment
policies.
The limited potency of increasing investment to spur growth
emerges clearly. Even a doubling of the share of output devoted to
net investment would increase the economy's rate of growth by only
0.3 percent per year over the next decade. The long-run gains are
even smaller. In steady state the rate of growth is independent of
the investment rate. The effects of more feasible increases in the
rate of investment are much smaller. Increasing the share of net
investment by one-third would only raise the growth rate of
productivity by about 0.1 percent per year over the next decade.
This calculation has assumed that all technical change is
disembodied-that is, independent of the accumulation of capital.
It might be argued that instead technical progress is embodied in
'This figure is greater than those in Table I, because ii takes account of growth in
land and inventories.
126 / TAX POLICY AND CORPORATE INVESTMENT
new capital goods, so that an increase in the rate of investment
raises productivity by speeding the introduction of new technology.
The model can easily be modified to take account of this possibility
by allowing technical change to affect the growth of the effective
capital stock rather than the effective labor force. That is, the
model becomes:
=
(2a)
Y,
(2b)
KEFF( = (l+g)t-lfl-1 +
(l-d)KEFFt-l
(2c)
Kt
(2d)
It
(2e)
Li = (l+n)L1 -i
KEFFfL[-"
= It-I
=
+ (l-d)K1 _1
dKt-1 + yY1-1
where g is now to be taken as the rate of embodied technical
change and n the rate of population growth. For the U.S. economy
it seems reasonable to take n = g = .015.
The results of simulating this model for alternative values of y
are displayed in Table 5. They indicate that assuming that technical
change is embodied does somewhat increase the estimated potency
of increased investment. Even so, a doubling of the share of output
devoted to net investment only raises the productivity growth rate
by .6 percent over the first decade. This calculation surely is an
overstatement since at least some technical change is disembodied.
The conclusion of this analysis, that even a large increase in the
rate of investment will have only a minor effect on productivity,
may at first seem surprising. However, it is in line with most
previous research. One of the striking discoveries of the "growth
accounting" literature dating from Solow (1958) has been the
unimportance of capital accumulation as a factor accounting for
increasing affluence. Estimates of the sources of inter-temporal and
international differences in productivity, Denison (1979), have
consistently found that capital intensity plays only a minor role.
The major factors appear to be human capital and technological
progress. It is little wonder, therefore, that increasing capital
accumulation is not likely to have major effects on productivity
growth.
Proponents of the view that increased investment would yield
large output gains frequently point to the apparently high
correlation across countries between capital formation and growth.
It is possible that this is because high rates of capital formation
spur research, or give rise to "learning by doing" effects. If so,
SUMMERS/
127
TABLE 5
The Rate of Growth of Output
Under Alternative Investment Policies
with Embodied Technological Change
Years
y= .045
Y"" .060
ye::: .075
y :::c .090
0-5
6-10
11-20
21-30
3.00
3.00
3.00
3.00
3.21
3.14
3.10
3.06
3.40
3.25
3.16
3 .11
3.59
3.36
3.23
3.15
convent1onal analyses may underestimate the gains from increased
investment. However, it seems more pfausible that causality runs
the other way and high savings rates are caused by rapid
technological progress. This implication flows naturally from the
standard Life.Cycle Hypothesis.~
INVESTMENT AND lNFLATlON
It is difficult to know how to frame the question of the effects of
policies to encourage investment on the rate of inflation. The
outcome of such policies obviously depends on what other
concurrent policy choices are made. We begin by considering the
effects of measures to encourage investment holding the rate of
growth of money constant.
Unless there is a change in the velocity of money, the effect of
increased investment on the rate of inflation is just the negative of
its impact on the growth rate of real output. The calculations in the
preceding section suggest that this is likely to be only a small effect
on the order of several tenths of a percentage point per year.
An investment oriented tax cut is likely to raise the returns
available on stocks and bonds. This will reduce the demand for
money, thereby increasing velocity and tending to raise the price
level. Suppose, for example, that an investment stimulus raised the
yield to bond holders by one percentage point. Assuming an initial
'Two other qualifications to the analysis in this subsection should be
acknowledged. First, an increase in the rate of capital accumulation will tend to
increase real wages, which may spur some labor supply response giving rise to extra
growth. lt is easy lo show that this effect is likely to be negligible even if a very high
labor supply elasticity is assumed. Second, the gains from additional investment may
be slightly underestimated because no account is taken of the advantage from
replacing energy intensive with energy conserving capita!. Preliminary analysis
suggests that this effect could not possibly raise the estimates reported above by
more than . l percent.
128 / T A X P O L l C Y A N D C O R P O R AT E ! N V E Si M E N T
interest rate of 10 percent, and an interest elasticity of money
demand of only .25, the price level would have to rise by 2.5
percent beyond normal inflation to restore asset market equilibrium.
This inflationary pressure is much greater than the deflationary
force from increased productivity growth. Hence, the net effect of
an investment oriented tax cut is likely to be an increase in the rate
of inflation unless the rate of money growth is reduced at the same
time.
Depending on the exact formulation of wage-price dynamics it is
possible to argue that increases in productivity may make it possible
to bring down the rate of money growth and inflation without
causing unemployment. Essentially the argument is that
productivity growth is like a favorable supply shock. A one-time
shock, by reducing past inflation, may moderate wage demands
leading to further reductions in inflation. This argument depends
on the implausible premise that workers are not able to obtain
higher real wages when increased capital intensity raises their
productivity. It also suggests that any measure (e.g., cutting sales
taxes) which reduces prices wiU reduce long-run inflation. Hence, it
does not single out increased investment incentives as the way to
fight inflation.
In sum, it does not appear that tax policies to spur investment
are likely to reduce the rate of inflation. This proposition is true
a fortiori if account is taken of their effects on aggregate demand
and the government defic5t.
INVESTMENT AND EMPLOYMENT
There is no reason to favor investment oriented policies as a
vehicle for encouraging employment. As long as labor and capital
are substitmable, either within individual production activities or
through shifts in the mix of production activities, it will be possible
to achieve full employment with any level of capital intensity. Fears
that insufficient capital accumulation must cause unemployment are
as groundless as earlier concern about unemployment due to
automation. Indeed, since capital and labor are substitutes in
production, unless output also expands increased capital
accumulation will actually reduce the level of employment.
lNVESTMENT AND !NTERTEMPORAL ECONOMIC EFFICIENCY
The justification for measures to increase the rate of economic
growth, if such a justification exists, must lie in the area of
intertemporal economic efficiency. There are two types of issues
SUMMERS/ 129
involved here which I will refer to as macro- and microintertemporal efficiency. Macro-efficiency here refers to society's
decision about the allocation of consumption between those alive
today and future generations. The huge literature on the Ramsey
optimal economic growth problem is concerned with this issue.
Micro-efficiency here refers to the distortion of individual
consumption plans by capital income taxation. This is the subject
addressed by traditional welfare analyses of the effects of capital
income taxes.
INVESTMENT AND MACRO-EFFICIENCY
The allocation of consumption between current and future
generations inherently involves ethical choices. Even a policy of
consuming the entire capital stock and leaving nothing to future
generations is Pareto optimal. Hence traditional welfare economics
can offer little guidance. The problem -is normally formulated on
choosing a growth path to maximize the discounted value of utility
subject to the constraints imposed by the production technology.
That is:
co
(3)
Max
J
U(ct)e-<0 ·0 nJtdt
s. t.
0
C
ko
= f(k)
(n+ g)k - k
= k
where c is consumption, d the discount rate, n the rate of
population growth, and g is the rate of Harrod-neutral technical
change. It is not difficult to show (see Solow (1970) for an intuitive
exposition) that an economy which is moving along a path which
solves the maximization problem given in (3) approaches a steady
state path with the property that:
(4)
f'(k) = d+rg
where £ is the elasticity of the marginal utility function. A value of
= - I implies that as consumption doubles, the value of a small
increase in its rate halves. With£ = - 2, the value falls by 75
percent and so forth.
Equation (4) can be used to make a judgment about the
efficiency of the path currently followed by the U.S. economy. The
data in Table 1 suggest that the marginal product of corporate
capital, f '(k), approximately equals .10. The value of g is very
optimistically assumed to be .02. The parameters £ and d describing
£
130 / TAX POLICY AND CORPORATE INVESTMENT
how the social marginal utility of consumption changes with the
level of consumption and time cannot be estimated empirically. A
value of £ = - 2 implying that society is willing to take a dollar
from someone with a $30,000 income in order to transfer 12 cents
to someone with an income of $10,000 seems very egalitarian. This
implies that current levels of investment are insufficient unless
d > .06.
There is little that an economist can say about the value of d. 9
However, it is difficult to see a rationale for discounting the utility
of future generations at a rate nearly as high as six percent. Ramsey
himself saw no argument for any discounting at all. Thus, there is
an ethical argument pointing to the desirability of more capital
accumulation.
It might be argued that this hardly provides a warrant for
government policies to spur investment. The future will be provided
for by bequests from parents to their children. The level of capital
intensity ground out by the free market is almost bound to be the
optimal rate. Careful consideration of this line of argument suggests
that there is a presumption that private capital formation will be
insufficient. First, the private return to capital is far less than the
social return to investment. The data in Table 2 indicate the average
return to corporate capital was about 10 percent during the 1970s.
The after-tax return to investors is only about one-fourth as great,
creating a presumption that insufficient provision will be made for
investment. Second, as long as individuals' concern for posterity
extends to the children of others, there is a benefit externality from
increased capital formation. Third, there is no more reason to rely
on private provision for the future than there is to rely on private
charity to meet current social needs. The existence of a transfer
motive is hardly sufficient to establish the sufficiency of the
resulting transfers.
While no definitive statement can be made, the foregoing
arguments suggest that macro-efficiency considerations dictate the
desirability of increased corporate investment. The amount of the
increase is of course more difficult to judge.
INVESTMENT AND MICRO-EFFICIENCY
Even if taxation has no effect on the amount of capital
accumulation, it may lead to substantial welfare costs due to the
distortion of individual consumption profiles. This will be true even
'Note the term gin (4) already takes account of the fact that future generations
will be richer than those alive today.
SUMMERS/
131
if the overall level of capital intensity is constant at its optimal
level. Feldstein (1978), Boskin (1978) and Summers (1980) all
estimate annual welfare costs of capital income taxes at current
levels which exceed $ 100 billion annually. Below, I illustrate how
capital taxes can give rise to large welfare costs, without having an
effect on capital intensity.
Consider the following model. Consumers live two periods
supplying labor inelastically in the first period and consuming in
both periods. That is, consumers maximize:
(5)
a=
wT
where C 1 and C refer to first and second period consumption, t is
the tax rate on capital income, and WT is first period income. If
the utility function is Cobb-Douglas, U ::c:: QC\~«, it is easy to
show that Ci = a-WT independent of the capital income tax rate.
Thus the tax has no effect on the level of capital formation which is
given by:
(6}
K
= WL
~
c,
The welfare cost of the tax can easily be measured. Solving the
maximization problem (5) it can be shown that the indirect function
is given by:
(7)
V(t,r,WL)
= WLa-"(l-a)"(l +(l--t)r)(h,J
This expression can be solved to find the change in labor income
necessary to compensate the representative consumer for any given
change in his tax rate on capital income. The revenue yield of the
tax can then be subtracted from this expression to calculate the
deadweight loss.
This model is highly stylized. Nonetheless, it can prov1de some
insight into the orders of magnitude of the welfare losses from
capital income taxation. It is assumed that each period in the model
corresponds to a generation, or 25 years. Hence, the value of a is
taken to equal .5, and the pre-tax rate of return is taken to be
e· 10(25 l = 12.18.
These parameters imply that relative to lump sum taxation, the
welfare loss from a 75 percent tax rate on capital income is 8
percent of labor income, compared to 4 percent of labor income for
a 50 percent capital tax rate, and 1 percent with a 25 percent tax
rate. These welfare losses are very large-a 50 percent capital
income tax has a welfare loss of over $50 billion annuaHy at current
132 / T A X P O L l C Y AN D
C O R P O R A TE
l N V E ST M E N T
levels of national income. As is to be expected, the welfare loss
rises much more than proportionally with the tax rate. Cutting the
tax rate by one-third from 75 percent to 50 percent reduces the
deadweight loss by one-half. A further halving of the tax rate to 20
percent reduces the loss by three-quarters. Thus the marginal gains
in intertemporal efficiency from cutting high capital tax rates are
large. The reduction in deadweight loss equals half the revenue loss
in the case of reduction in the tax rate from 75 to 50 percent.
This calculation omits two important features of reality. The
result may be overstated because of the assumption that lump sum
taxes are available. If the alternative is the taxation of labor
income, then deadweight losses may also result from this source.
However, it is not at all dear that consideration of variable labor
supply would reduce rather than increase the estimated welfare
losses from capital taxation. Capiral taxes, by raising the price of
future consumption, reduce real wages as defined by an appropriate
intertempora1 cost of living index. ' 0 Hence, they also distort the
labor-leisure choice. Moreover, they distort the intertemporal
allocation of labor, which is not affected by a labor income tax."
Feldstein (1978), without considering the latter effect, found that
there are substantial net gains which can be realized from a shift
towards labor taxes. Considering the intertemporat labor supply
effects would strengthen this conclusion.
The calculation also is carried on as if all capital were located in
the corporate sector. This means the final losses from the
misallocation of capital are not included. Available evidence,
Fullerton, et al. (1976), suggests that these losses may not be too
great.
Any reduction in the tax burden on corporate capital wou1d tend
to reduce the wedge between the social return to capital and
investors' private return, and so would reduce the deadweight loss.
The calculation presented here suggests that even if the policy did
not increase capital formation there would be substantial gains in
intertemporal economic efficiency. If parameter values consistent
''This crucial point is overlooked by many authors who hold that with variable
labor supply, optimal tax rules are compktely indeterminable. In the plausible case
of separable utility, it is optimal to place no raxes on labor income regardle,s of the
elasticity of labor supply, It is easy to construct e:-:ampks in which a subsidy to
capital income is optimal.
"A long tradition in labor ec(rnomics dating from the work of Mincer has
recognized that the intertemporal elasticity of labor supply far exceeds tbe static
elasticity.
S Li M M E R S /
133
with a positive effect of investment incentives on saving had been
assumed the estimated welfare gains would have been much greater.
These results imply that there is a substantial scope for improving
economic welfare through increased incentives for investment. The
next sections discuss the empirical estimation of the extent to which
tax policy can increase investment.
TRADITIONAL APPROACHES TO EVALUATING CORPORATE
INVESTMENT INCENTIVES
This section examines previous empirical evidence on the
relationship between corporate investment and tax policy. The large
literature on this subject is based almost entirely on single equation
econometric models of the demand for equipment and structures. A
detailed survey and criticism of some prominent models may be
found in Chirinko and Eisner (1980). There have been relatively few
efforts to examine the effects of investment stimuli within plausible
general equilibrium frameworks. The efforts of this type which
have taken place have been carried out using large scale
econometric models which are ill-suited to questions of long-run
capacity growth.
The standard method of evaluating the effects of tax policy on
investment follows the seminal work of Hall and Jorgenson (] 967).
They begin by postulating that the desired capital stock, K*,
depends on the level of output, Y, and the cost of capital, c. The
cost of capital is a complex function of the interest rate and tax
parameters. A general expression for it is given by
q [ (l - u)
(8)
C
=
Q -
!L +
6 ]
q
(I - u)
[I -
k -
uz]
where q is the supply price of capital goods, u is the corporate
income tax rate, Q is the opportunity cost of capital, 6 is the rate of
economic depreciation, k is the investment tax credit, and z is the
present value of the tax depreciation expected from a dollar of
investment.
From this point, empirical implementations differ across studies.
It is usually assumed that the rate of investment depends on some
distributed lag on K. * The distributed lag is usually justified as
deriving from lags in the delivery of investment goods or in the
formation of expectations. The equation is then estimated
econometrically.
134 / TA X P O L I C Y A N D C O R PO R A TE I N V E S T M E NT
Changes in tax policy are studied by examining the effects of a
tax change on the cost of capital and then of the cost of capital on
investment. Chirinko and Eisner (1980) present a detailed
description of how this is done in the major large scale econometric
models.
While there is room for substantial disagreement about the
proper way to carry out this procedure, these issues are ignored
here. There are several fundamental problems which make this
approach an undesirable way of evaluating investment incentives.
First, by holding the level of output fixed, the investment
equation approach makes it impossible to capture the effects which
are at the root of the case for tax policies to encourage investment.
If one believed that the level of output was in fact independent of
the path of investment, it is difficult to see why investment stimuli
should be advocated. The essence of the way in which investment
stimuli are supposed to work is by reducing the cost of capital and
encouraging firms to increase investment in order to supply more
output.
The second fundamental difficulty with these investment
functions is that they are susceptible to the "Lucas critique." There
is no reason to suppose that their parameters would remain
constant if policy rules were changed. Hence they cannot provide
useful policy guidance. A trivial example is provided by considering
the difference between a variable and a permanent tax credit. It is
easy to see that a temporary credit will provoke a much greater
investment response since firms will all schedule their investment to
coincide with it. Hence the estimated effect of the investment tax
credit (ITC) will depend on what policy rule has been followed. A
related point is that conventional investment equations offer no way
of considering the effects of policy announcements. Taken literally,
the investment equations in all the major macro.econometric
models would imply that an announcement today that six months
hence the corporate income tax would be abolished would have no
effect at all on current investment decisions. Nor does anything in
the equations suggest how they might be modified to meet this
objection.
The third difficulty with traditional investment equations is that
they are really adjustment equations without a theory of
adjustment. The question of ultimate interest is the effect of
changes in tax policy on the long run capital stock. This question
can be answered simply from the production function requirement,
FK = c, holding that the marginal product of capital is equated to its
SUMMERS /
135
rental cost. The investment equation is essentially irrelevant. Seen in
this light, it is clear that the focus of efforts to examine the effects
of tax policy should be on the aggregate production function rather
than the investment equation. Worse, the production functions
which are implied by the results of fitting investment equations are
typically wildly implausible.
The only role for an investment equation is in explaining the
economy's adjustment path in response to a policy shock. Yet
existing econometric investment equations proxy adjustment
without any explicit treatment of adjustment costs. They can hardly
be interpreted as offering useful guidance on the process of
convergence to equilibrium because the equilibria they imply are
typically so far wide of the mark.
THE ROLE OF DEMAND
Previous studies all suggest that the state of business activity is a
prime determinant of the level of investment. It is this evidence that
has led many observers to conclude that more vigorous antirecession policies offer the greatest hope for raising the level of
investment. This conclusion typically emerges from both single
equation studies (e.g., Clark (1979)) and full model simulations.
This finding can be traced directly to the flaws in these studies
noted above. In fact, economic theories which command almost
universal support among Keynesians as well as classical
macroeconomists indicate that reliance on the accelerator offers no
route to increased capital formation in the long run.
The high correlation between output and investment which is
observed in the data does not imply that a permanent increase in
the level of output will permanently increase the rate of investment.
As emphasized above, output and investment are simultaneously
determined and in the past have moved in tandem because of
common causes. Indeed the apparent potency of the accelerator
reflects, in large part, the impact of investment on total output. It
does not follow that the correlation would be the same if general
expansionary policy was regularly used to spur investment.
There is a second important argument supporting this conclusion.
Many, though not all, previous investment studies fail to impose
the restriction that investment depends only on the growth in
output not its level. Since high output has in the past been
correlated with high output growth it appears that expansion is a
potent policy to stimulate investment. A policy of permanent
expansion would eliminate this correlation and so would be much
136 / TA X P O L I C Y A N D CO R P O R A T E I N V E S T M E N T
less effective than conventional econometric specifications suggest.
The analysis so far has been partial equilibrium in character. It
has suggested that there is reason to doubt that a permanent
increase in GNP would have a large impact on investment. There is,
however, a much more fundamental flaw in the argument for
expansionary policy to spur investment. Stated baldly, the natural
rate hypothesis implies that there is no such thing as "permanent
expansionary policy." Any attempt to keep the level of economic
output performance above some "natural" level, will lead to
accelerating inflation. If we rule out policy rules which will lead to
steadily increasing rates of inflation, we are confined to policies
which on average keep the economy at its natural rate. Permanent
expansion or contraction is not possible.
What about a policy of systematically more vigorous response to
recessions than has been observed in the past? While this would
increase investment, it would also lead to permanently accelerating
inflation, unless an equal offset was applied in boom times. Such
an offset would negate any gains which might be realized in terms
of investment.
EVALUATING INVESTMENT INCENTIVES
This section summarizes the methodology for evaluating
investment incentives developed in Summers (1980), and presents
some estimates of the effects of alternative tax policies on
investment. The method described here is an application of Tobin's
q theory of investment. It yields estimates of the effects of tax
policies on the valuation of the stock market as well as on rate of
investment. Below I present a heuristic account of the method. For
a fuller treatment, the reader is referred to my earlier paper.
METHODOLOGY
For simplicity, the dynamics of investment and market valuation
are examined in a simplified model where all investment is financed
through retained earnings and the only tax is a proportional levy on
corporate income. In this setting it is reasonable to assume that
investment depends on the ratio of the market value of existing
capital to its replacement cost. Unless the market value of the firm
will be increased by more than one dollar by a one dollar
investment, there is no reason for it to be undertaken. Given costs
of adjustments and lags in recognition and implementation, there is
no reason to expect that all investments which will raise market
value by more than their cost will be made immediately. As Tobin
SU M M E R S / 137
(1969) has argued, these considerations lead to an investment
equation of the form:' 2
(9)
I(Y)K
K
1(1)
=
I'> 0
0
where I represents gross investment and V /K is the "q" ratio of
market value to replacement cost. The assumption that it is 1/K
which depends on q insures that the growth rate of the capital stock
does not depend upon the scale of the economy.
It is assumed that equity owners require a fixed real rate of
return to induce them to hold the existing stock of equity. This
return comes in the form of dividends, equal to after-tax profits
less retentions for new investment, and capital gains. Hence we
have the condition:
(10)
which implies:
V
(11)
= eV -
(l-r) F'(K)K
+ I(y
K
)K- dK
where T is the corporate tax rate, and F(K) is the production
function for net output.
It will be most convenient to examine the dynamics in terms of K
and q = Y. Equations (9) and (11) imply that the system's
K
equations of motion are:
(12)
(13)
K
q
=
Qq
=
I(q)K - dK
l(q)q +d q + I(q) - (1-T)F'(K) -d
where cl is the rate of depreciation.
The steady state properties of the model are easily found by
imposing the conditions K = 0 and q = 0. These imply:
(14)
(15)
q = 1- 1(d)
(I - r)F '(K)
=
eq
"A rigorous foundation for an investment equation of this type is provided in
Abel (1979) and Hayashi (!980). An important implicit assumption of this approach
is the homogeneity of capital. If capita! is heterogeneous, shocks may reduce the
market value of existing capital but raise the return on new investment. The recent
energy shock illustrates this phenomenon.
138 / TAX POLICY AND CORPORATE INVESTMENT
The former equation indicates that the steady state value of q must
be greater than 1 by an amount just large enough to induce
sufficient investment to cover depreciation. The latter equation
holds that firms equate their net marginal product of capital to the
cost of capital. Inspection of (14) and (15) makes it clear that a
change in the corporate tax rate affects the steady state capital
stock but has no effect on steady state q. This is a consequence of
the assumption that it is investment relative to the capital stock
which varies with q.
The phase diagram of the system (12) and (13) is displayed in
Figure I. It is readily verified that the pair of equations is saddle
point stable". The arrows indicate the direction of motion and the
heavy line represents the saddle point path along which the system
will converge. A change in the corporate tax rate is depicted in
Figure 2' 4 • If the expectations about pre-tax profits were static, the
value of q would jump from E to A when the tax change took
place. This expectations assumption has been used in previous
works on the effects of taxation on the stock market, e.g., Feldstein
(1979), Hendershott (1979). It neglects the effect of the induced
changes in investment on the present value of future profits. With
perfect foresight, as assumed here, the value of q will jump only to
B. The magnitude of the jump will depend upon the speed of
adjustment of the capital stock to the shock.
The system of equations (12) and (13) can be solved numerically
to estimate the impact of any type of shock on the path of q and
the capital stock. The effect of tax changes on the level of the stock
market can be easily calculated. This can then provide a basis for
estimating the effects of tax changes. The model actually used to
calculate the effects of tax changes is considerably more complex. It
takes account of the complexities of the tax code and of the fact
that investment is partially financed through the issuance of debt.
The results reported below are based on empirically estimated
production functions and investment relations for the corporate
sector.
RESULTS
We begin by considering the impact of the investment tax credit,
since this issue has been a focus of previous work. Standard single
equation approaches to the investment function have yielded
"This is a common feature of models with asset prices.
"It is assumed that the market selects the unique stable perfect foresight path.
SU M M ER S /
139
TABLE 6
Permanent and Temporary Removal of the
Investment Tax Credita
bTemporary
Permanent
Year
l
2
3
4
5
10
15
20
50
Steady
State
V
I
K
V
l
K
-2.8%
-3.0IIJo
-3.0%
-3.3%
-3.5%
-4.0%
-4.4%
-4.7%
-5.6%
-6.0%
-4.80/o
-4.9%
-6.1%
-6.2%
-6.4%
-7.9%
-8.1%
-8.8%
OOJo
-0.4%
-0.9%
-1.30/o
-3.5%
-4.8%
-6.0%
-8.9%
-2.0%
-0.5%
-0.5%
-0.6%
-0.6%
-0.3%
-0.3%
0%
0%
00/o
0%
-4.9%
-3.7%
00/o
0%
0%
0%
-0.lOJo
-0.1%
-0.1%
-0.4%
-0.9%
-0.7%
-0.6%
0%
0%
-0.1%
-5.6%
-9.6%
-9.6%
0%
0%
0%
-L7%
Notes_. •The numbers shown in the table are the changes relative ro the 8 percent
inflation path in the absence of tax reform.
bThe temporary investment tax credit is imposed in year 4 for three years.
divergent results. In perhaps the most widely cited study, Hall and
Jorgenson (1971) conclude that the investment tax credit has a
potent impact, which reaches its peak after about three years. They
estimated that the 7 percent credit on equipment enacted in 1962
raised the 1970 capital stock by about 4 percent above the level it
would have reached in the absence of the credit. Other estimates
typically suggest much smaller estimates of the effect of the credit.
None of the estimates takes explicit account of the possibly
temporary nature of changes in the level of the credit.
In Table 6 the effects of alternative tax credit policies are
considered, The first column considers the effects of a correctly
perceived permanent removal of the credit. The results indicate that
the credit has potent effects on investment, even though it has only
a small impact on market valuation in the short run. Its immediate
effect is to reduce investment by about 6 percent, and it decreases
the capital stock by 8.9 percent in the long run. The estimated
response is much more gradual than that predicted by standard
140 / TAX POL [CY AND CORPORATE l NV EST MEN T
investment equations. The effect on investment declines between the
first and second years and then rises steadily as the reduced capital
stock requires less replacement investment. Since the change
considered here is the removal of a 9 percent investment credit,
these results indicate a slightly larger effect than those of Hall and
Jorgenson, and a much larger effect than that found in most other
studies.
The right half of the table considers the impact of a temporary
removal of the ITC. Such a measure leads to a sharp decrease in
investment during the suspension period. This leads to an increase
in net investment after the suspension is removed. Gross investment
does not increase because the lower capital stock requires less
replacement investment. Note that the catch-up following the
restoration of the credit is very slow. Two-thirds of the gap caused
by the suspension in the capital stock remains 15 years later. These
results show the importance of the adjustment costs, which explain
investment's sluggish response to q. In the absence of any
adjustment costs, one would expect to see substantial disinvestment
during the period of the suspension. Because the adjustment costs
of returning to the steady state capital stock would be high, this
does not take place. These findings illustrate the importance of
considering expected future policy. If the credit suspension were
permanent its effects on net investment in the short run would be
far less pronounced.
The effects of reductions in the corporate tax rate are examined
in Table 7. An immediate rate reduction from .48 to .40 is
constrasted with an announcement that in year 4, such a tax cut
will take place. Both measures are equivalent in the long run, and
raise the steady state capital stock by 15. 7 percent. They increase
the long-run value of the stock market significantly more because
the reduced corporate tax raises the effective price of new capital
goods by diminishing the value of accelerated depreciation and the
expanding of adjustment costs.
The simulations show that the announcement policy has a
significantly greater short-run impact on investment than the
immediate implementation policy. The former raises the capital
stock by 3 percent after three years compared with 2 percent for the
latter. This occurs even though the immediate implementation
policy has a greater immediate impact on the capital stock. The
reason again is the effects of accelerated depreciation and the
expanding of adjustment costs. Firms find it optimal to accelerate
their investment plans to take account of the lower effective price
SU M M E R S /
141
TABLE 7
Unanticipated and Anticipated Permanent
Corporate Tax Cut a
hAnticipated
Unanticipated
Year
V
I
K
V
1
2
3
+ 18.6%
+ 19.4%
+20.0%
+20.4%
+20.7%
+22.3%
+23.2%
+24.1%
+25.9%
+ 7.1%
+ 7.2%
+ 8.5%
+ 7.3%
+ 8.6%
+ 9.0%
+ 10.5%
+ 10.8%
+ 14.7%
0%
+ 0.5%
+ 1.1%
+ 1.6%
+ 2.0%
+ 4.5%
+ 6.5%
+ 8.1%
+ 13.5%
+15.1%
+ 16.9%
+ 19.0%
+20.9%
+21.2%
+22.7%
+23.5%
+24.3%
+25.9%
4
K
+ 9.5%
+ 10.8%
+ 12.2%
+ 8.5%
+ 8.6%
+ 10.3%
+ 10.5%
+ 10.8%
+ 14.7%
+
+
+
+
+
+
+
+
0%
0.8%
1.6%
2.5%
3.0%
5.1%
7.0%
8.6%
13.8%
5
10
15
20
50
Steady
+26.7% + 15.3% + 15.3% +26.9% + 15.3% + 15.3%
State
Notes: asee footnote (a) in Table 6
bTax cut takes place in year 4
of capital goods which prevails before the tax reduction actually
takes place. This implies that if the goal of the corporate rate
reduction is to increase capital formation, the measure should be
announced well in advance of its enactment. Similar considerations
suggest that a temporary increase in the corporate tax rate would
actually spur investment.
These findings have important policy implications. They indicate
that a policy of announcing a future reduction in corporate taxes
will spur investment with no current revenue loss. Indeed, the effect
on investment would actually be enhanced if corporate taxes were
raised immediately and then cut. By combining temporary
corporate rate increases with temporary increases in the investment
tax credit or accelerated depreciation it would be possible to
provide substantial investment stimulus at no budgetary cost.
Most previous analyses of the effects of investment incentives
have neglected the role of individual tax measures. The effects of
reforms in the individual tax system are considered in Table 8.
Eliminating capital gains taxes would raise the stock market by 7 .3
142 I
TAX
POL f CY
AND
CORPORATE
INVESTMENT
TABLE 8
Reforms in Individual Taxesa
b Anticipated
Dividend Relief
Capital Gains
Tax Eliminated
Year
2
3
4
5
10
15
20
50
Steady
State
V
I
K
V
I
K
+ 7.3%
+ 8.1%
+ 8.5%
+ 8 .9%
+ 9.3%
+ 10.8%
+12.1%
+ 13.2%
+ 16. 1%
+ ll.9%
+ 12.0%
+ 13.4%
+ 12.2%
+ 13.6%
+ 16.7%
+17.1%
+20.30/o
+26.5%
00/o
+ 0.9%
+ 1.8%
+ 2.7%
+ 3.6%
+ 7.5%
+11.1%
+ 14.0%
+24.0%
+60.3%
+68.50/o
+77.3%
+86.3%
+85.7%
+83.7%
+82.5%
+82.0%
+79.3%
+40.50/o
+47.0%
+53.70/o
+ 6.1 OJo
+ 6.2%
+ 5.1%
+ 4.0%
+ 2.7%
+ 1.5%
0%
+ 3.2%
+ 6.7%
+10.7%
+ 10.2%
+ 8.5%
+ 7.0%
+ 5.7%
+ 1.7%
0%
0%
+ 17.3% +27.7% +27.7% +78.6%
Notes: asee footnote (a) in Table 6
bExpected abolition of the dividend tax in year 4
percent in the short run. Because it would increase the advamages
to the firm of retaining earnings, the impact on investment is
substantially greater. Its long-run effect would be to raise the
capital stock by 29.5 percent. The transition is however very
gradual with only half the adjustment occurring within the first
decade.
The second reform considered is an announcement that in year 4,
the dividend tax will be eliminated. This corresponds to an extreme
form of partial integration of the corporate income tax. As
explained in Summers (I 980), changes in the dividend tax rate have
no effect on steady state capital intensity. The announcement that a
dividend tax reduction will occur however gives firms a very large
incentive to defer paying of dividends. This is done by accelerating
investment. The simulations suggest that the announcement effect
raises investment by 40.5 percent.
The estimates of the potential gains from reductions in taxes on
capital income described here are quite robust. As explained in the
previous section, the long-run results depend almost entirely on the
production function. The Cobb-Douglas form which provides the
SUM MER S /
143
basis for the estimates reported here is widely accepted as a
reasonable aggregate approximation, The propositions that the
stock market's level reflects the present value of future profits, or
that investment responds positively to q are also uncontroversial.
This is all that is necessary to accept these results.
Taken together the results indicate the large scope for tax policy
to affect capital accumulation in the long run. Politically
conceivable measures, such as the abolition of capital gains taxes or
the allowing of replacement cost depreciation would have a very
substantial impact on long-run capital intensity. Measures can be
designed which have a large impact on investment with a relatively
tow cost in foregone government revenue. A final conclusion which
emerges from these simulations is the dangers of indiscriminate tax
cutting. The incentive effects of announced and unannounced cuts
vary greatly across tax measures so that careful policy design can
increase the investment stimulus per dollar of lost government
revenue.
THE SUPPLY OF FUNDS FOR CORPORATE INVESTMENT
The analysis in this paper so far has assumed that the rate of
return required by investors in the corporate sector is fixed,
independent of tax policy or the level of corporate investment. As
Figure 2 illustrates, this is equivalent to assuming that the supply of
funds to the corporate sector is perfectly elastic, Unless this
condition is met, investment incentives wiH lead to increases in the
rate of return required by corporate investors. In the limiting case
where the supply of funds to the corporate sector is completely
inelastic, and the KS curve in Figure 1 is vertical, investment stimuli
will have no effect on capital accumulation.
It is therefore crucial to assess the elasticity of the supply of
capital to the corporate sector. A full discussion of this issue is
outside the scope of this paper, but a few remarks are sufficient to
establish that the elasticity is likely to be quite high. The elasticity
of the supply of savings to the corporate sector depends on both
the elasticity of total savings with respect. to the rate of return and
the substitutability of corporate and non-corporate assets in wealth
portfolios. These issues are considered in turn.
Until recently, it was widely believed that the rate of saving was
largely independent of the rate of return. This notion was
supported by verbal reference to conflicting income and substitution
effects, and to the near constancy of the saving rate. Recently, both
144 / TAX POL l CY AND CORPORATE l NV EST MEN T
FIGURE l
q
L
7
q
=0
------------- ------------- -K
theoretical and empirical evidence have accumulated suggesting that
the elasticity is quite high. The "infinite horizon" model of
intertemporal consumption decisions implies that saving is perfectly
elastic with respect to the interest rate. Summers (1980) shows that
plausible life cycle formulations almost inevitably imply a high
interest elasticity of saving. It also demonstrates that the two period
model which provided the basis for most previous theoretical
studies of the interest elasticity of saving is likely to be very
misleading.
At the same time, recent empirical evidence tends to support a
positive interest elasticity of saving. Boskin {1978) was the first
study to use a measure of the proper variable, the real after-tax
interest rate, in a study of the interest elasticity of saving. His study
found an interest elasticity of about A. There are strong reasons to
believe that this is an underestimate of the elasticity of response to
a permanent change in tax policy. The variations in real after-tax
interest rates during Boskin's sample period are almost all
transitory. As Summers (1980) shows, the response of policy to a
SU M ME RS /
145
FIGURE 2
q
L
7
0
q
=
0
'-----------------------------K
transitory shock in interest rates is likely to be much Jess than the
response to a permanent shock. Of greater importance, Baskin, in
calculating the interest elasticity of saving, takes no account of the
wealth effects of interest rate changes. Part of the saving response
to increases in interest rates occurs because of induced changes in
wealth. Taking account of these effects can easily raise the
estimated elasticity from .4 to 2.
These considerations suggest that there are strong reasons to
believe that the supply of capital to the corporate sector is highly
elastic. This conclusion is strengthened by considering the allocation
of capital between sectors. The U.S. corporate sector accounts for
only about one-fifth of American physical wealth and a much
smaller fraction of world capital. Hence even if the total supply of
capital were fixed, the supply of capital to the corporate sector
might be quite elastic. There is no direct evidence bearing on the
extent of these effects. Summers (1981) shows how the relative
valuation and accumulation of corporate and housing capital over
the last decade has been affected by increased taxation.
146 / T A X P O L I C Y A N D C O R PO R A T E l N V E S T M E N T
In Feldstein and Summers (1978) an attempt is made to gauge the
elasticity of the supply of capital to the corporate sector. This is
done by examining the effects of changes in the MPIR-the
Maximum Potential Interest Rates firms can afford to pay on a
given investment project-on actual interest rates. The results
indicate that a one percentage point increase in the MPIR raises
interest rates by .25 points. Loosely speaking, this means that 25
percent of the stimulus afforded by investment tax incentives is
offset by rising asset prices. This is further evidence that investment
incentives are unlikely to be crowded out by rising costs of capital.
If crowding out due to a limited supply of capital appeared to be
a significant factor impeding corporate investment, government
policy could easily increase the supply of funds to the corporate
sector. This could be done through measures to encourage saving or
more plausibly through increased public saving. The latter action
could be achieved by reducing budget deficits and limiting
commitments to future expenditures.
The analysis here of the supply of funds to the corporate sector
has important implications for policy towards investment. In
particular it implies that measures directed at increasing national
saving will have little effect on investment. In the limiting case
where saving is infinitely elastic, such measures would have no
effect at all. Policies to spur investment, if they are to be effective,
must be specifically directed at corporate capital. Our analysis
suggests that such measures are likely to have potent effects.
SUM M ER S /
147
REFERENCES
Abel, Andrew. "Investment Theory: An Integrative Approach."
mimeo, 1979.
Blinder, Alan. Discussion of Martin Feldstein. "Tax Rules and the
Mismanagement of Monetary Polley." American Economic
Review, 70 (May 1980), 189-190.
Boskin, MichaeL "Taxation, Saving and the Rate of Interest."
Journal of Political Economy, (April, 1978).
Chirinko, Robert and Robert Eisner. "The Effects of Tax Policies
on Investment in Large Scale Econometric Models." Paper
presented at the 4th World Congress of The Econometric
Society, August 28, 1980.
Denison, Edward. Accounting for Stower Economic Growth: The
United States in the 1970s. Brookings Institution, 1979.
Feldstein, Martin. "Investment, Inflation, and Taxes." mimeo,
1980.
_ _ _ _ _ and James Poterba. "State and Local Taxes and the
Rate of Return on Nonfinancial Corporate Capital." NBER
Working Paper #508R, July 1980.
_ _ _ _ _ and Lawrence H. Summers. "Inflation, Tax Rules,
and the Long Term Interest Rate." Brookings Papers on
Economic Activity, 1978:L
_ _ _ _ _ _ _ _ _ _ _ , "Inflation and the Taxation of Capital
Income in the Corporate Sector." National Tax Journal, 32
(December 1979), 445-470.
Fullerton, Don, et al. "Static and Dynamic Resource Allocation
Effects of Corporate and Personal Tax Integration in the U.S.:
A General Equilibrium Approach." NBER Working Paper #337,
AprH 1979.
Hall, Robert and Dale Jorgenson. "Tax Policy and Investment
Behavior." American Economic Review, 57 {June 1967),
391-414.
Hayashi, Fumio. "The q Theory of Investment: A Neo-Classical
Interpretation.'' Econometrica.
148 / TAX POL l CY AN[) CORPORATE INV [ST MEN T
Norsworthy, J.R., Michael Harper, and Kent Kunze. "The
Slowdown in Productivity Growth: Analysis of Some
Contributing Factors." Brookings Papers on Economic Activity,
1979:2, 387-421.
Solow, R.M. Growth Theory, Oxford University Press, 1970.
Summers, Lawrence H. "Taxation and Capital Accumulation in
a Life Cycle Growth Model." American Economic Review,
forthcoming. (1981).
______ . "Inflation, Taxation and Corporate Investment."
mimeo, 1980.
Tobin, James. "A General Equilibrium Approach to Monetary
Theory." Journal of Money Credit and Banking, I (1969), 15-29.
Estimates of Investment Functions
and Some Implications for
Productivity Growth
PATRICH. HENDERSHOTT
My original assignment was first to evaluate Larry Summers'
paper as a description of the current state of the art regarding
investment behavior and second to determine the adequacy of the
investment sector of Michael Evans' econometric model (Evans,
1980) in light of Summers' paper. The late arrival of Larry's paper
forced me to alter my strategy, and it is just as well. Summers'
investment function is a very long-run relationship that does not
purport to explain cyclical movements in business investment
outlays, while Evans' relationship is a more traditional analysis of
quarterly expenditures. 1• Moreover, Summers is concerned with only
corporate investment, while Evans deals with all of domestic fixed
investment. My revised strategy was to employ two papers recently
presented at Brookings Conferences (Hendershott, 1980, and
Hendershott and Hu, 1981) as the standard with which to contrast
Evans' work.
The first two sections of the present paper are concerned with
nonresidential and residential fixed capital outlays, respectively. In
each of these I first summarize my earlier work and then critique
Evans' treatment of the same investment component. A general
discussion of the relationship between the form of investment and
productivity growth is the subject of the third section, and a
Patric H. Hendershott is Professor of Economics and Finance, Purdue University.
The author gratefully acknowledges support from the National Science Foundation
under grant DAR-8016064 and the National Bureau of Economic Research for his
work in the broad area of capital formation.
;Summers' equations explaining the annual ratio of gross real investment to the
beginning period capital ,wck over the 1932-78 period have R' that range from 0.05
(no autocorrelation correction) to 0.75 (Summers, 1980, Table 2, p. 34). Of course,
investment equations must have plausible long-run properties if they are to be useful
in examining the long-run impacts of tax changes, but this does not rule out
relationships that also explain cyclical behavior.
149
150 / l N VE S T M E N T F U N CT I O N S
summary concludes the paper. Summers' imaginative work is
referred to periodically when it bears on the issue at hand, but time
and space constraints prevent me from discussing his analysis at
length.
NONRESIDENTIAL INVESTMENT
GENERAL DETERMINANTS
Investment outlays (or orders) can be thought of as the sum of
four components: Those due to normal growth, to disequilibrium,
to replacement, and to mandates of governments. The general
determinants of each of these parts are the following:;
Normal Growth (In): Normal growth in the economy requires
greater production capacity. How capital intensive this is should
depend on the real user cost of capital (c). Thus one can write
where y represents any of a variety of variables that proxy for the
expected growth rate in real output, and the expected signs of the
partial derivatives are indicated above the arguments in the
function. I emphasize here that the relationship is between net
investment and the rate of change in output, not the level of
output. As Summers (1981) and others have noted, the latter is a
major misspecification of an investment function and has
nonsensical macroeconomic policy implications.
Disequilibrium (Id): Disequilibrium investment (positive or negative)
arises when factor prices or aggregate demand change unexpectedly.
Proxies often employed to represent disequilibrium are deviations
between current and long run or "normal" values of Tobin's Q
(the ratio of the market value of corporate debt and equity to the
replacement cost of nonfinancial assets) and capacity utilization
(CU). Thus
Id
=
+
+
Id(Q-Q*, CU-CU*),
where * denotes normal or long-run values (assumed to be
constant).
'This analysis assumes a CES production function. The use of a variable elasticity
function, such as the translog (see Berndt and Christensen, 1973), requires inclusion
of either the user costs or quantities of other factors in the estimation equation.
HEN D E RS H OTT
I 151
Replacement (1,): In a pure putty-putty world where changes in the
capital/labor ratio can occur both before and after the installation
of capital, replacement investment is reasonably approximated by
the product of the depreciation rate and the existing capital stock.
But in a putty-clay world, where variable factor portions exist only
for net investment and upon replacement of old capital,
replacement investment also depends on changes in the real user
cost since the capital being replaced was initially instaHed. More
specifically, one can write
00
Ir
L Yr(c_.Jc)
= K-,6
T=0
where Yr equals LO for r "" 0 and 0.0 otherwise, if technology is
putty-putty, or equals the fraction of each vintage of capital in the
total existing stock, if technology is putty-day, and the symbol or
denotes the optimal feasible replacement investment fraction/
Mandated investment Om): This investment is mandated by law and
is thus reasonably treated as exogeneous.
Combining the four investment (orders) components into a single
function,
+ - + +
(J)
I ""
HY,
C'
Q, CU)
+ ofK-, + Im.
Our empirical results suggest the following. First, the user cost
variable, which affects both t and Jf, is a fundamental factor
affecting investment.' Second, the accelerator variable, y, works as
expected. And third, the capacity utilization rate, but not Q, is an
important determinant of disequilibrium investment.
REAL USER COST OF CAPITAL
Consider the following assumptions/ definitions:
i) aU prices are expected to rise at rate rr forever,
ii) the productivity of an investment declines at rated over an
infinite holding period,
1
Putty~clay technology is a possible source of long lags in investment functions,
bu! it is still diffkult to explain Summers' 16 year adjustment period to obiain half
of the impact of an inflation shock (1980, Tab[e 4, p. 45).
'With d = 0.!3, Jf varies from a low ofO.! !5 in 1957:1 to a high of0.156 in
1971:4. Replacement of JK .. , with dfK -, in the estimated equation significantly
raised the explanatory power.
152 / l N V E ST M E NT F U N CT I O N S
iii) the statuatory income tax rate isµ,
iv) the rate of investment tax credit is k,
v) the present value of depreciation allowed for tax purposes
on a dollar of capital is z,
vi) pollution control outlays of 4' dollars are required for every
dollar of capital investment,
vii) the ratio of inventories based on FIFO accounting to the
stock of capital is 11, and
viii) the real after-tax financing rate is r.
With these assumptions, one can derive the real user cost of capital
as
(2)
C
(1 +tp)q
(1-µ)p[(l-k-µz) (r+cl)
+
µvn},
00
where z ""
L
t""' l
the
tax
q "" the
p ""' the
dxr
(l+r+rr)l
fraction of the capital price allowed to be treated as
depreciation in period t,
price of capital goods, and
general price of output.
This equation is identical, in appearance, to equation (4.2), p. 4,15
of Evans except for the addition of the inventory term to allow for
the taxation of FIFO-based inventory profits. Assuming that a
portion a of investment [(1-k)q] is debt financed, the debt and
equity portions are expected to remain constant forever, and debt
finance charges are deductible from the income tax base,
(3)
where i is the nominal debt yield and ea is the nominal after-tax
cost of equity funds. A plausible proxy for e8 is the sum of the
after-tax earnings-price ratio (E/P) and n/(l - a). The division by
l - a reflects the fact that all inflation gains accrue to shareholders
(except those indirectly built into i). Substitution into (3), 5
(3) I
r "" a(l -µ)i
+
(l -a)E/P.
'This equation looks like an analogue to the Modigliani-Cohn stock market error:
it appears that a nominal debt yield is being averaged with a real equity yield. ln
faet, the expression is an average of two real yields (I -µ)i- rr and E/P +
[a/{l -a)]n. The rr terms cancel when the expression is simplified.
H E N D E R S H O TT / 153
EVANS' ANALYSIS
The aggregate investment equations reported account for the
normal growth and disequilibrium investment components in a
reasonable fashion. A variety of sectoral income variables drive
investment; the user cost variable generally performs as expected
(more on this below); and the capacity utilization rate, the
unemployment rate and stock prices all appear as disequilibrium
proxies. The putty-clay optimal feasible replacement investment
fraction does not appear, but the establishment of its relevance is of
recent "vintage." However, I cannot even find the lagged capital
stock in the equations, although it is referred to in the text. Even
more disconcerting is the absence of mandated investment outlays.
The importance of these outlays is emphasized by Evans and these
outlays are incorporated in the calculation of the user cost, but the
actual outlays are ignored in the estimation. To put these outlays in
perspective, during the I 972-78 period they were roughly 4 percent
of total new orders for equipment and 12 percent of net new orders
(roughly t~o-thirds of orders were for replacement).
One final point on these equations. An undefined index of credit
rationing appears in the equipment equation with rationing
(supposedly a slowdown in deposit flows) reducing equipment
outlays. While outlays on trucks and autos (p. 4.67) may be
reduced, as are housing starts (see below), it would seem to me that
outlays somewhere in the economy should be stimulated. That is, if
accelerated flows into open market paper, defined broadly to
include large CDs and money market funds, are detrimental to
outlays financed by regular deposits, then these flows ought to be
favorable to the outlays financed by open market paper; rationing
ought to have an allocative, zero-sum impact rather than a
cumulative negative impact. Finally, if rationing matters for
business investment, then business cash flows obviously matter to
investment, a fact Evans denies on p. 4.10.
Evans spends a great deal of time and effort in the construction
of user costs of capital for business investments. For this he is to be
commended. Unfortunately, there appears to be a number of errors
in the calculations. First, consider the measurement of the real
after-tax financing rate. In the aggregate investment equations
(pp. 4.70 and 4.76), the yield is
r
=
0.4 i + 0.6 E/P.
Note that the interest rate is before-tax when it should be after-tax."
'In the industry studies (pp. 4.19 and 4.42), the dividend-price ratio replaces E/P,
and the 0.4 and 0.6 weights may have been switched.
154 /
l N V E ST M E N T
FUN CT ION S
Also, there does not appear to have been any attempt to adjust
earnings for the overstatement due to historic cost depreciation.
Thus the real after-tax financing rate is clearly overstated by a
significant amount. Second, depreciation rates of 0.095 (structures)
and 0.181 (equipment) have been employed. These, too, are far too
high (by about 0.05). Third, the effective (average) rather than
statuatory (marginal) corporate tax rate is utilized. To the extent
that the vagaries of the tax code are already accounted for-the
investment tax credit, tax depreciation, and FIFO accounting-the
statuatory rate is clearly the appropriate variable. Just as
important, the average tax rate moves cyclically, being high when
profits are great and low when profits are small, but the expected
tax rate over the life of the investment, the relevant rate in the user
cost calculation, is unlikely to move in this manner. This illustrates
an important point about the user cost expression (2). All values in
it denote expected values over the life of the investment asset. If
these values are expected to change in the short run, then such
expectations could have a large impact on the timing of orders or
investments, even if only the long-run expected values affect Jongrun capital accumulation. To illustrate, a temporary increase in the
investment tax credit would have a far larger short-run stimulative
impact on investment (Lucas, 1976, pp. 30-35) than would a
"permanent" increase. 7 Further, as Summers illustrates ( 1981)
anticipations of tax changes can have major, and even surprising,
effects.
In summary, the Evans model has not advanced econometric
modeling of nonresidential investment. Replacement and mandated
investment are not accounted for, and there are significant errors in
the calculation of the important user cost variable. Moreover, the
measurement and inclusion of z in the user cost is hardly
innovative, as is suggested on p. 4.16. This variable was included in
early Jorgensonian formulations and has been part of the data bank
for the various versions of the Federal Reserve econometric model
for at least a decade.
RESIDENTIAL INVESTMENT
AN OVERVIEW OF THE HOUSING MARKET
My view of the determination of increments to the real housing
stock is depicted in Figure 1. The major financial variables are
circled: the mortgage payment constraint (roughly the product of
the nominal after-tax mortgage rate and the real price of
"Evans sta,es the opposite on p. 4.13.
FIGURE lI
Detennination of Chang~s
Changes in the Housing Stock
Determination
Stock
Mortgage Payment
Constraint
Interest Rates
Tax Law
Productivity of the
Construction Sector
Finanda! Structure
Number of
Real User Costs
Credit
Availability
Tenure Choice:
Buy vs. Rent
Starts;
Single vs.
Multiple
X
Average
Quality
Per
S!art
Rea! value
Value
Real
of Starts
Real
Change in Real
Housing Stock
Housing
156 /
l N VE S T M E N T F U N C T I O NS
structures), the user costs of capital for owner-occupied and rental
housing (the former is approximately the product of the real aftertax mortgage rate and the real price of structures), and credit
availability. (An inflation-induced increase in the mortgage payment
constraint will limit the size of house purchases if imperfections in
the capita! market prevent households from borrowing against
future housing capital gains.) These variables depend on those in
the box on the left: the level of interest rates, tax law, the financial
structure, and the relative productivity of the construction sector
(which determines the real price of structures). The three doublelined boxes represent the important economic decisions. Tenure
choice depends on the rental price (user cost) of housing services
generated by an owner-occupied dwelling versus that of services
produced by a rental unit. This choice, along with total household
formations and credit availability, determines the numbers of single
and multifamily starts. The average quality (square feet, number of
fireplaces, etc., valued in constant dollars) per start, in turn, is a
function of real income per household and "prices," both the real
user cost (user cost divided by the price of non-housing goods) and
the real mortgage payment constraint. The product of the number
of starts and their average quality is the real value of starts, and
this is converted to real housing outlays or the change in the real
housing stock with a short production lag,
Implicit equations for single (SST) and multifamily (MST) starts
and explicit equations for the average real qualities of single (SQ)
and multifamily (MQ) starts are
+
+
SST
=
4s(hHH, b(c/r), AVAIL)
MST
=
4rn(AHH, A(c/r), AVAIL)
SQ
MQ
= iµs(
+
+
+
+ - y, c, m)
+
= t/!in( y,
r),
where HH is the number of households, c and r are the real user
costs for owning and renting, AVAIL represents credit availability,
y is real income per household, m is the mortgage payment
variable, and the signs of the partial derivatives are indicated above
the variables. Significant lags exist, particularly with regard to the
tenure decision.
H E N DE R S H O TT /
157
The above starts equations are consistent with a world in which
prices of new housing units are a mark-up on costs and builders
determine starts so as to equate the expected future supply and
demand for incremental units. An alternative view, which I label
the pure supply view, has the price of new units determined by the
supply and demand for existing units and has builders responding
to profit opportunities, as well as the availability of credit:
ST
=
+
+
fp{Ph/p, Cost/p, AVAIL),
where Phip is the real price of housing and Cost/p is the real cost
of production.
THE EVANS MODEL
Starts equations appear in the Evans investment chapter, but
average quality equations do not. Multiplication of starts by a
housing price translates starts into nominal dollars, and a
production lag converts these into nominal outlays on housing.
How or whether real outlays are determined is unclear. Thus, my
discussion relates only to the behavior of starts.•
It is difficult to fit the Evans starts equations into either of the
above frameworks. The equations are of the forms
+ ~ +
+
SST = 4c,(Y, m, rt1i, AVAIL)
+
+
+
MST = f, (y, RENT INT + RENT 'AVAIL, OVER),
cm
COST ,
WAGES
where the signs over the variables are the signs of the estimated
coefficients, rrh is the housing inflation rate, RENT /COST is a
profitability measure, (INT + RENT)/WAGES is the ratio of NIA
interest and rent income to wage income, and OVER is a measure
of overbuilding (the cumulated difference since 1970 between 600
thousand and actual annual starts). The first equation has no cost
variables and looks more like an average quality rather than
number of starts equation. The inflation and mortgage payment
'One exception. It is stated that "most recent estimates indicate that the (income)
elasticity (of housing) is now closer to L5 [than unity]" (p. 4.94). My estimate is
0.68 and those of the micro studies I have seen are only slightly higher. Possibly the
studies referred to (not cited} intermingle the income and price effects. The price
(user cost} is lower for households with higher incomes {in higher tax brackets}. If
the income variable captures this price effect, then a higher elasticity would be
estimated.
158 / 1 N V E S T M E N T F U N C T I O N S
variables could be reflecting tenure choice, and the rise in income
over time, too, likely reflects the shift towards home ownership,
although there is no reason why higher income per se should
increase ownership. Unfortunately, this would suggest that income
should enter the multifamily equation with a negative sign. The
multifamily equation also includes a profit variable, a factor share
variable, and a measure of overbuilding consistent with desired
starts over time being a constant 600 thousand and thus
independent of any economic considerations. That is, the equation
appears to include most any variable that "worked."
Because Evans' credit availability index is undefined, discussion
of its plausibility is impossible. However, the impact of the change
in FHLB advances, another availability proxy employed, is subject
to interpretation. This variable reflects what appears to be a
common problem with econometric models of housing: availability
of funds variables work far too well. During the 1976- 79 period,
only 23 percent of savings and loan loans closed, net of
refinancings, were used to finance new construction of dwelling
units. Yet the coefficients on the advances variables in the starts
equations suggest that a billion dollars of advances would generate
$2 billion in nnv construction.• A quite careful specification of
starts equations is needed to prevent a vast overstatement of
availability effects. My own estimates are that a billion dollars of
deposits generates only $0.3 l billion of 1-4 family housing, and
even this seems to be too large an effect.
Regrettably, the residential investment sector of the Evans model
is no improvement on poorly formulated existing models.
SUPPLY-SIDE ECONOM[CS AND THE PRODUCTIVITY OF CAPITAL
Supply-side economics is concerned with increasing economic
growth and thus the size of the economic pie. This can be achieved
by increasing either the level of effort (more manhours ,vorked) or
the quality of a given level (more output per manhour). One way of
increasing productivity is to increase capital per worker, and this is
most directly achieved by raising the saving rate. Thus the most
important supply-side economic issues are the sensitivities of labor
supply to real after-tax wage rates and of saving to real after-tax
interest rates. Because neither of these topics relates to investment,
it is fortunate that other means of raising productivity exist, In
''Somewhat similarly, Jaffee and Rosen, 1979. and Poterba, 1980, report that an
additional billion dollars of thrift d~ro,it, would kad to $1.5 billion in construction,
H EN D E R S H O TT /
159
order to focus on such means, I assume in what follows that labor
supply and saving, respectively, are independent of wage and
interest rates.
Economic policy can affect economic growth in such a world via
two routes. First, an increase in government saving that is not
accompanied by an equal decrease in government investment or
private saving will increase the capital stock. A reduction in
government "consumption" outlays would reduce government
borrowing and thus real interest rates, thereby stimulating
investment. Alternatively, an increase in taxes on private
consumption outlays would accomplish the same objective. Second,
a reallocation of investment from less to more productive uses will
raise the productivity of a given total stock of capital.
There are two general means of channeling investment into more
productive uses. There has been a surge in explicitly mandated
investments in the last decade, some of which have been of
questionable value. The massive retrofitting of transportation
networks to allow access of the handicapped comes to mind.
Similarly, government regulations implicitly require overinvestment
in some areas. For example, our trucking fleet is larger than it need
be owing to "gateway" requirements whereby trucks are forced to
make empty return trips on suboptimal routes. A reduction in
explicit and implicit mandated investments would free resources for
more productive uses.
A second means of improving the productivity of capital is to
reduce the relative subsidy extended to owner-occupied housing.
The user cost of capital for owner occupied housing tends to be low
because neither the implicit rents from the unit nor the capital gain
earned is taxed. Moreover, this user cost has declined in response to
increases in anticipated inflation because real after-tax debt yields
have fallen. Estimates of real user costs for owner-occupied housing
and corporate structures in 1964 and 1978 are listed in Table L The
1964 data illustrate the relationships among user costs in a
noninflationary period. The costs for housing are lower because of
its preferred tax treatment, and the costs are lowest for those in the
highest tax brackets. The 1978 data reflect the decline in real aftertax debt yields; the decline is largest for those in the highest tax
brackets. The fall in the user costs for owner-occupied housing
would have been greater but for a sharp rise in the real price of
structures. Referring back to equation (2), the near doubling of the
user cost for corporate structures reflects: l) a decline in z, the
present value of tax depreciation, owing to the use of historic cost
160 / I N V E ST M E N T F U N CT l O N S
TABLE l
Real User Costs of Capital, 1964 and 1978
(Percent)
1964
Owner-Occupied Housing:
15 Percent Tax Bracket
30 Percent Tax Bracket
45 Percent Tax Bracket
Corporate Structures
1978
9
5
8
7
15
2
0
27
Sources: Owner-occupied Housing, Hendershott and Hu, 1981a. Corporate Structures, Hendershott and Hu, 1981b.
depreciation, 2) an increase in taxes paid on inventory profits, and
3) a rise in the real price of structures (q/p). Also, the real aftertax financing rate for structures has not fallen because the heavilyweighted equity yield component has risen by enough to offset the
decline in the real after-tax debt yield. Given this movement in user
costs, the surge in the levels of sales and production of singlefamily housing in the second half of the 1970s and the sluggishness
of investment in nonresidential structures are hardly surprising.
America is now investing resources in housing that has a net (or
depreciation) marginal product of near zero and foregoing the
construction of corporate structures that have a net marginal
product of over 20 percent.
The relative subsidy for owner-occupied housing and the resultant
misallocation of capital can be reduced through a variety of
methods. Most obviously, implicit rents and housing capital gains
could be taxed. Not only does this appear politically infeasible, but
the taxation of largely nominal capital gains has little appeal on
equity grounds. Alternatively, a wide range of business tax cuts
could be employed to offset the subsidy to owner-occupied housing:
these include a switch to replacement cost depreciation, expanded
investment tax credits, a reduction in the double taxation of
corporate dividends and a general cut in the corporate income tax
rate. The investment stimulated by these cuts would drive up real
interest rates, thereby rechannelling resources from housing to
nonresidential investments.'° Feldstein, 1980, has generalized this
argument by calling for a switch from an easy-money /tight-fiscal
"'See Hendershott and Hu, !980, for an analysis of the impact of these tax cuts on
,he user costs for business investments and owner-occupied housing.
H EN DE RS H OTT /
161
policy mix, in which real after-tax mortgage rates are negative and
the taxation of capital income is great, to a tight-monetary I easyfiscal policy mix, in which the reverse is true.
My own favorite method of reducing overinvestment in owneroccupied housing is a large, say $12,000, exemption of interest and
dividends from taxation, subject to the netting of personal (largely
mortgage) interest expense. To illustrate, consider two households,
each with $12,000 in interest income but one with a mortgage
entailing an annual interest expense of $9,000 and the other with no
mortgage and thus no interest expense. The former household
would pay taxes on $9,000 of interest income (only $3,000
= $12,000 - $9,000 would be exempt), while the latter would pay
no tax on interest income. This would reduce both the relative tax
advantage to owner-occupied housing and the inequitable current
taxation of largely nominal interest income. In effect, a tax break
(cessation of taxation of nominal interest) would be extended to
those not leveraging investment in owner-occupied housing. Finally,
we should discourage any further subsidies to housing such as the
use of tax-exempt financing (mortgage revenue bonds).
SUMMARY
It is not clear that the new emphasis on supply-side economics
has implications for major revisions in the form of business
investment equations. There is, of course, a need to account
carefully for the interaction between inflation and taxes and to
incorporate mandated investment outlays into the analysis. But
existing models either already do this or can be easily adapted.
Possibly as a result, the equations in Evans' model do not appear
to be particularly innovative. Moreover, there seem to be some
errors in the calculation of user costs, and replacement and
mandated investment outlays are overlooked.
The residential construction equations in existing models are not
in as good shape as the nonresidential investment equations. The
major problems are a failure to measure user costs as carefully as is
done for the business sector, and a tendency to attribute far greater
impact to credit availability than is remotely plausible.
Unfortunately, the equations in Evans' model do not appear to
address these problems in a useful manner.
While the new supply-side emphasis should not be expected to
alter greatly the form of investment equations, hopefully its
emphasis on supply constraints will alter the type of policy
162 /
I N V E S T M E N T F U N CT I O N S
simulations run with the models. Too often in the past, simulations
of policy actions or legislation designed to encourage a specific type
of capital outlay have been run in the context of a world with
unlimited resources or infinite supplies. The result implied by such
simulations is, not surprisingly, an increase in not only the targeted
capital good but in all capital and consumption goods. In such a
world, any capital-specific policy should be pursued. In the real
world, resources are limited. Even in the intermediate run, the only
policies that should be analyzed are those designed to have zero
aggregate demand impact. For example, a specific tax cut should be
accompanied by other tax increases, expenditure cuts, or higher real
interest rates (a more restrictive monetary policy). 11 Of course, if
the policies are well-designed, then productivity and thus the total
size of the economic pie, will increase. As Summers emphasizes
(1981), however, significant quantities of these aggregate benefits
will not be achieved quickly.
"For discussion of the issue, raised in this paragraph, sec Hendershott, !980.
H E N D E R S H OTT /
163
REFERENCES
Berndt, Ernst R. and Laurits R. Christensen. "The Translog
Function and the Substitution of Equipment, Structures, and
Labor in U.S. Manufacturing, 1929-68." Journal of
Econometrics, 1 (1973), pp. 81-114.
Evans, Michael K. "Supply-Side Model." Evans Economics, Inc.,
mimeo, 1980.
Feldstein, Martin. "Tax Rules and the Mismanagement of Monetary
Policy." American Economic Re1Jiew, 70 (May 1980), 182-86.
Hendershott, Patric H. "Analysis of the Impact of Capital Specific
Policies or Legislation: Application to Reforms of the TaxExempt Market." Journal of Money Credit and Banking, (May
1980), 377-99.
~~--· "Real User Costs and the Demand for Single Family
Housing." Brookings Papers on Economic Activity, 2: I 980.
Hendershott, Patric H. and Sheng Hu. "Inflation and Extraordinary
Returns on Owner-Occupied Housing: Some Implications for
Capital Allocation and Productivity Growth." Journal of
Macroeconomics, forthcoming (Spring 1981).
~~~~·
"Investment in Producer's Equipment." ln Aaron and
Pechman, eds., How Taxes Affect Economic Behavior.
Washington: Brookings Institution, 1981.
____ . "The Relative Impacts of Various Proposals to Stimulate
Business Investment." In von Furstenberg, ed., The Government
and Capital Formation, Ballinger Publishing Co., 1980, 321-36.
Jaffee, Dwight M. and Kenneth T. Rosen. "Mortgage Credit
Availability and Residential Construction." Brookings Papers on
Economic Acfivity, 2: 1979, 333- 76.
Lucas, Robert E., Jr. "Econometric Policy Evaluation: A
Critique." The Phillips Curve and Labor Markets, North
Holland, 1976.
Poterba, J. "Inflation, Income Taxes, and Owner-Occupied
Housing." NBER Working Paper No. 553, September 1980.
Summers, Lawrence H. "Inflation, Taxation and Corporate
Investment." mimeo, I 980.
____ . "Tax Policy and Corporate Investment." In this volume.
Discussion of the Summers Paper
NORMAN B. TURE
I believe that the term "supply-side economics" is a misnomer.
The analytical system going under this name really consists of
nothing new or fancy but merely the application of price theory to
public policies concerned with major economic aggregates. This
analytical approach and the public policies developed therewith do
not focus particularly on supply conditions to the exclusion of
effects of policy on aggregate demand. The distinguishing attribute
of "supply-side" economics, and the principal issue it casts up,
rather, is that it identifies the initial impact of public policies and
actions in terms of alterations in (implicit or explicit) relative prices
instead of changes in income.
One of the principal consequences of this distinction is that if one
wants to model economic responses to public policy actions in the
supply-side context, one must make very certain that the behavioral
functions in one's model preclude identification of first-order
income effects of government actions. The mere addition of supply
equations to a standard "aggregate demand" model does not
convert that model into a supply-side model.
The implications for policy of assigning first-order price effects
to government actions and of rejecting the possibility of first-order
income effects of such actions are enormous, but not because
public policies guided by supply-side economics focus exclusively or
primarily on aggregate supply conditions or because such policies
primarily affect supply conditions. Rather, it is because supply-side
economics dictates different policy strategies and tactics from those
which have long been pursued and looks to results which differ in
character and magnitude from those urged by the Keynesian
aggregate demand approach.
While Summers does not provide an explicit supply-side context
Norman B. Ture was President, Institute for Research on the Economics of
Taxation, Washington, D.C., when this speech was presented. He is currently Under
Secretary of the Treasury for tax and economic affairs.
166 /
S t.: M M E R S D I S C U S S I O N
for his discussion, his paper is very much in that spirit.
Summers' provocative paper presents a wide-ranging discussion,
each of the topics of which itself deserves and would make an
interesting paper. I shall comment briefly on several of these,
reserving more extended comments for two of his topics.
Summers first turns his attention to the postwar trends in net
capital formation in the non financial corporate sector. He shows
that the decline during the last half of the 1970s in the rate of net
investment (other than for pollution control facilities) and in such
investment in relation to gross corporate product is associated with
a decline in the real net rate of return. This, in turn, more reflects
increases in the effective rate of tax on corporate earnings than
decreases in the pre-tax rate of return. The increase in the tax rate,
in turn, is attributable to inflation. Accordingly, Summers
concludes that the interaction of the tax system and inflation
accounts for the l 970s investment showdown.
I take no issue with this conclusion or more generally with the
proposition that tax factors materially influence the pace and
volume of capital formation.
The question is why the acceleration of capital formation is
important. Summers properly identifies the popular concern with
the adequacy of investment in terms of effects on productivity,
inflation, and unemployment. He finds, however, that changing the
rate of investment is unlikely to have a significant effect on the rate
of growth over the next decade, that increasing investment is likely
to accentuate inflation, and that there is no reason to seek to
promote investment as a means of encouraging employment. With
each of these conclusions and Summers' means of arriving at them,
strong issue is to be taken.
First, Summers' finding that increasing investment has an
extremely limited potential for increasing growth in output is
derived from a model the specification inadequacies of which
include a labor supply function unrelated to anything but the
passage of time and a capital supply function devoid of any
behavioral arguments. Associated with this is an investment
function specifying net investment as a constant function of net
output. Summers' model is not useful for dealing with the question
whether increasing investment implies significant gains in output
and employment and decreases in the inflation rate. Nor can the
model be treated as representing reality. Indeed, as specified, it
serves no purpose other than to illustrate a proposition which needs
no illustration, viz., if the elasticity of output with respect to a
TUR E /
167
production input is very small, large increases in the amount of that
input will result in relatively small increases in output. By the way,
even in this unrealistically limited context, the effect on the growth
rate of increasing the share of output allocated to investment is
substantially more impressive than Summers' exposition would lead
one to believe. He finds, for example, that doubling the share of
output allocated to investment would increase the growth rate "by
only 0.3 percent per year over the next decade." But this is more
correctly read "0.3 percentage points" and amounts to a JO percent
gain in the growth rate.
A model correctly specified to analyze the effects of a change
in the rate of capital formation on growth of output will show how
the initial change in the capital: labor ratio increases the marginal
value productivity, i.e., real wage rate, of labor, and the
consequent increase in both the demand for and supply of labor
services. These increases in labor inputs, along with the initial gain
in capital inputs, result in gains in output of significantly larger
magnitude than Summers estimates. Moreover, the second-order
income effects of the output gains also generate an increase in the
optimum stock of capital, hence a further expansion of capital
inputs.
Summers' line of analysis leads him to conclude that "Fears that
insufficient capital accumulation must cause unemployment are as
groundless as earlier concern about unemployment due to
automation." This conclusion is, of course, dead wrong. It is
arrived at by way of a mechanistic observation that since
production inputs are substitutable it is possible to have some
given amount of labor employed with virtually any given amount of
capital. All this statement amounts to is that one can conceive
production functions with any combination of exponent values one
wishes. It is this analytically useless observation that leads to
Summers' next assertion that increasing capital will decrease labor
unless there is an increase in output. This is, of course, precisely the
fear about the consequences of automation which Summers
dismisses as groundless. Aside from being inconsistent, Summers is
wrong. Other things equal (i.e., the pertinent demographics, the
state of technology, the basic conditions of factor supplies, etc.),
the only way to increase employment is by increasing labor's
productivity which requires, unless the laws of production have
been repealed, an increase in the capital: labor ratio. Indeed, the
basic criterion for assessing the sufficiency or insufficiency of
capital accumulation is whether it affords an increase in the capital:
168 /
SUMME RS D l SC U SSIO N
labor ratio sufficient to maintain an acceptable rate of gain in
productivity, real wage rates, and employment.
One of Summer's most startling conclusions is that if the rate of
growth of the money stock is held constant, investment-oriented tax
changes which increase investment, hence, one must presume,
increase total output above levels otherwise attained, will result in
an increase in the inflation rate. This conclusion derives from
misspecification of the direct effects of the tax change and of the
responses thereto. The correct specification is that the tax change
reduces the real supply price for any given amount of capital, the
response to which is a shift in the use of current income from
consumption toward saving. Insofar as the reduction in real capital
supply price is reflected instantaneously in an increase in the returns
on stocks and bonds, this entails no shift from money to securities,
as Summers claims, but from purchase of consumption goods and
services to purchases of claims on capital assets. Nothing in this
response mechanism necessarily pertains to any change in velocity.
All that is left as a source of effect on the price level, therefore, is
the effect of larger stocks of capital and the consequent increases in
labor inputs on total output. As Summers correctly notes-but
denies-" ... the effect of increased investment on the rate of
inflation is just the negative of its impact on the growth rate of real
output.''
To summarize to this point, on the score of the effects of
increasing the stock of capital on output, employment, and the
price level, Summers negative conclusions are derived from
misspecification. While certainly not dismissing the welfare gains
which Summers believes are the real payoff from increased
investment, I think he grossly underestimates the gains in output,
hence employment, which would result from increased investment
in response to reducing the existing tax bias against saving and
capital formation.
Summers' discussion of how tax "incentives" affect investment
behavior-the last three sections of his paper-are more useful. He
is quite right in criticizing the treatment embodied in the standard
large-scale econometric models. For the most part, these models
depend on a capital stock adjustment formulation but take a nothink approach to the adjustment process. Yet as Summers himself
points out, the lack of theory to explain the pace of adjustment
from one optimum stock of capital to another is not, itself, a fatal
flaw in analyzing the effects of tax changes on the economic
TURE /
169
aggregates. To be sure, it impairs the usefulness of these models for
forecasting purposes but the social welfare is little diminished by
any such model imperfections. More to the point is whether these
or any other models are so specified as to capture correctly the
effects of tax "incentives" on the desired stock of capital.
The relevant formulation for this purpose proceeds, as Summers
notes, from the specification of the production function, from
which the schedule of the marginal product of capital is derived.
This is the capital "demand" function, obviously unaffected
initially by any tax change, since it is not a behavioral function.
The capital supply function is the schedule showing the amounts of
capital individuals wish to hold at varying net, real rates of return,
given the level of total income. With taxes of the character the
U. S. relies upon, market or pre-tax rates of return required for
each quantity of capital must, obviously, exceed the net or after-tax
rates. It is the intersection of the downward sloping marginal
product and upward sloping supply schedules which determine the
optimum stock of capital. Clearly, changes in tax provisions affect
this optimum by altering the capital supply schedule in pre-tax
terms. A tax change per se can have no initial effect on the
marginal product of capital. Nor has it any initial first-order
income effect to alter the supply of capital. It affects only the pretax returns required to obtain the after-tax return at which a given
amount of capital will be held.
I belabor you with this simple exposition only to emphasize that
the effect of a tax change on investment derives solely from the way
in which taxes affect the supply of capital, hence saving behavior.
With no change in the tax regime and other things given (i.e., the
rate of technical progress, the condition of labor supply, etc.),
saving= investment will increase with the increase in total income,
hence the increase in the desired stock of capital, through time.
Given the level of income, however, a change in taxes affecting the
rental cost of capital generates a new optimum stock of capital at
that total income level. It consequently impels a change in the
amount of saving out of that total income, hence a change in
consumption, as people seek to shift to the new desired stock of
capital. It is, therefore, only through its effects on saving that tax
changes can alter the stock of capital.
For purpose of analyzing the ultimate effect of tax changes on
the stock of capital, nothing more is needed. For purposes of
estimating the effects of tax changes on saving= investing, i.e., the
170 / SUMMERS DISCUSS l ON
adjustment from one optimum stock to another, far more is
needed, specifically theory and data to explain the pace of the
adjustment.
The search for this explanation is complicated by virtue of the
fact that few, if any, feasible tax changes will affect the desired
stock of each component of the total stock of capital in the same
proportion. Virtually all such tax changes will result in some change
in the composition of the capital stock. The time required to
effectuate that change will differ from one type of capital to
another; it takes a good deal longer, ordinarily, to build a
petroleum refinery than to manufacture a new machine tool.
Searching the data for stable saving functions, therefore, is chasing
a will o' the wisp.
But instability in the saving function does not imply instability or
shifting parametric values in the desired stock of a capital function.
Accordingly, there is no real problem rising from changes in policy
rules, of the sort Summers suggests, in the use of a properly
specified cost of capital formulation. Set in the correct model
context, this specification entails no difficulty whatever in
differentiating the effects of temporary or permanent investment
tax credit changes. Moreover, it generates the carefully
differentiated, with respect to both magnitude and timing, estimates
of the effects of different types of tax changes of the sort Summers
illustrates without resort to the exotic sort of explanation Summers
offers.
I find myself mostly in agreement with Summers' conclusions
about the relative magnitude of the effects of capital-favoring tax
changes, despite the fact that I largely disagree with the way he
arrives thereat. What this proves is that even when marching to
different drummers, people can arrive at the same destination. It is
heartening to discover that despite quite different perceptions of
what supply-side economics is about, it is possible to come quite
close together on tax policy prescriptions aimed at regeneration of
economic progress.
Income and Payroll Tax Policy
and Labor Supply
JERRY HAUSMAN
INTRODUCTION
Income and payroll taxes account for about 75 percent of federal
revenues. The proportion of federal tax revenue raised by these two
taxes has gone up markedly in the past decade with the amounts
growing faster than the underlying inflation rate. The rise in the
income tax collections occurs because of its progressive rate
structure and insufficient indexing of tax brackets to account for
inflation. The rise in the payroll tax has occurred because of
legislative actions to fund social security payments. Both the tax
rate of the payroll tax and the maximum earnings limit have
increased significantly. In Table 1 we indicate the effects of the
income and payroll taxes over the last two decades. Note that the
combined percentage of the two taxes has risen from 56% of
government revenues in 1960 to 76% of government revenues in
1978. This increasing trend is likely to continue in the future.
The current social security law calls for further tax rate increases
up through 1990 and beyond, and earnings limit increases up to
1982. While the income and payroll taxes have certainly received
adequate attention from economists, it is probably fair to say that
most economists accepted their structure as reasonably good. Most
economists liked the distributional consequences and believed that
the economic cost in terms of economic efficiency was small. This
latter conclusion was based on limited empirical work and survey
responses that the income tax caused little reduction in labor
supply. Some evidence existed which indicated that wives labor
supply might be affected by taxation, but the general view was that
prime age males' behavior was hardly affected at all.
Jerry Hausman is Professor of Economics, Massachusetts Institute of Technology
and Research Associate, National Bureau of Economic Research, Cambridge, Mass.
Peter Diamond and Nan Friedlaender have provided helpful comments. Paul Ruud
and Ken West were research assistants for this project. The NSF provided research
support.
173
TABLE l
Revenues from Income and Payroll Taxes (billions)
Earnings Limit
for Payroll
Tax
Year
Income Tax
Revenues
Payroll Tax
Revenues
Income Tax%
of Federal
Revenues
Payroll Tax %
of Federal
Revenues
1960
$ 40.7
$ 10.6
44%
12%
3.0%
$ 4800
1965
48.8
16.7
42
15
3.625
4800
1970
90.4
38.4
47
22
4.8
7800
1975
122.4
75.7
45
29
5.85
14100
1978
198.5
106.1
46
30
6.05
17700
Tax Rate for
Payroll Tax
HAUSMAN / 175
Two mistakes arose from this common interpretation of the
income tax. First, even if we grant the hypothesis that the income
tax has little overall effect on labor supply, its economic cost might
still be substantial. Income taxes have two effects on labor supply.
Taxes lower the net wage and reduce labor supply by the
compensated substitution effect. But taxes also have an income
effect, which causes individuals to work more since they have been
made worse off by the tax. The two effects have opposite signs and
might well approximately cancel causing only a small net effect on
labor supply from income taxation. But, the economic cost of the
tax arises from the first effect alone. Thus, the conclusion by many
economists that the cost of raising revenue by the income tax is
very small is not supported by economic theory if, in fact, the
income effect and substitution effect are cancelling each other out.
The second problem occurs because virtually all empirical work on
labor supply disregarded taxes. The market wage rather than the
after-tax wage was used in the labor supply functions. Or
alternatively, the tax system was treated as a proportional tax
system rather than a progressive tax system. 1 In a recent paper,
Hausman (1979c), I have built on previous research and conducted
a study of the effect of tax policy on the labor supply behavior of
prime age males, wives of the prime age males, and females who
head households. When progressive taxes are entered into a model
of labor supply we see a significant effect. The findings indicate
that labor supply of the husbands is reduced by about 8% because
of the income and payroll taxation while labor supply of wives is
reduced by about 30%. Thus, income taxes do affect labor supply
in an important way.
But as I argue in the next section of the paper, economists should
focus on the economic cost of income taxation more than on labor
supply effects, My findings indicate that the economic cost of
raising a dollar of government revenue by the income tax is about
25¢ on average in terms of lost welfare. The marginal cost of
raising an additional $1 government revenue by this means is
approximately 40¢. Thus, the economic cost of the income tax is
substantial. At least three possible policy recommendations may
follow from these conclusions. First, government expenditure might
well be reduced given the cost of raising the necessary revenue. To
recommend this policy we would need to study the benefits created
'Hal! (!973). Hamman and Wise (!976), Burtless and Hausman (1978), and Wales
and Woodland (1979) provide the major exceptions for analyzing U,S. tax policy,
176 / INCOME TAXES AND LABOR SUPPLY
by marginal government expenditure. Here and earlier, questions of
income distribution become important. Income distribution
considerations are discussed in this paper, but we have very little
grasp of what constitutes marginal government expenditure or the
benefits which arise from it. A further narrowing of policy options
would be required to analyze the expenditure option more deeply.
The second policy option is to consider raising a greater proportion
of tax revenue from other federal taxes. To recommend this option,
we need to know the economic cost of other taxes, such as the
corporation tax, in terms of their effect on economic efficiency. We
do not have adequate knowledge of the cost of other taxes to
explore this option. Lastly, we could consider altering the income
tax structure to raise the same amount of revenue but at lower
economic cost. In the paper, we investigate progressive linear
income taxes which seem to have favorable effects both with respect
to economic cost and labor supply.
Policy options one and three are investigated in this paper. Policy
option one is similar to Kemp-Roth type proposals for a decrease in
income tax rates. Since our model is partial equilibrium, we look at
the effect on tax revenue and the economic cost of taxation holding
other factors constant. Our findings indicate that income tax
revenues in our sample would decrease by about 6.1 0/o for a IOOJo
tax cut and by about 20.3% for a 300/o tax cut. Labor supply
effects and the effects on economic cost are discussed in this paper
as well as distributional effects of the tax cut. It is certainly possible
that general equilibrium effects would eliminate the estimated
reduction in tax revenues, but my results lead me to doubt this
possibility, especially in the short run. The third policy option
appears much more favorable. The progressive tax considered there
is basically as progressive as the current tax system for low incomes
but decreases the high marginal rates for high incomes. When
raising the same amount of revenue as the current system, the
economic cost is decreased by more than one half on average with
even a greater decrease at the margin. On the usual efficiency
grounds this policy option looks extremely good. But as we
discuss in the last section of the paper, objections might well be
raised to it because it worsens the income distribution. Questions of
the tradeoff between the economic cost (efficiency) and income
distribution (equity) are very difficult to treat without making
judgments on unobservable preferences. Yet, the investigation of
this paper is useful because it indicates the size of the potential
tradeoff in terms of a marked reform of our income tax system.
H AUSM AN /
177
LABOR SUPPLY, TAXES, AND DEADWEIGHT Loss
In this section we first consider a model of individual labor
supply of the type which has been used in most empirical analysis.
The model is based on individual decision makers rather than some
larger unit like a family decision process. In fact, in the empirical
estimates which we present we consider only husbands and wives.
Thus, our model has the husband's labor supply decision
independent of the wife's labor supply decision. The wife makes her
decision conditional upon her husband's choice. While this model
set-up has been traditionally followed in empirical research in labor
supply, I expect research in the near future to be more general in its
approach. A more symmetrical treatment of family labor supply
decisions would be helpful. A second limitation to the model is that
it is both static and partial equilibrium. Intertemporal decisions
such as the amount of education that a person receives which may
well be affected by taxes are omitted.' Also, the model does not
consider demand factors for labor in terms of types of jobs offered
with respect to wage and hour packages. Again, a more complete
model which incorporates these factors would be desirable.
Once we outline the model of labor supply we will then consider
the effect of taxes on labor supply. Labor supply has been the
focus of much attention in recent discussions of supply-side
economics. As a theoretical proposition, it is well known that the
effect of taxes can either be to decrease or increase labor supply.
However, the accepted hypothesis among supply-side economists
has been that the effect of the current U.S. income tax system has
been to decrease the labor supply. The labor supply model helps us
to consider this question which is answered in the next section with
the empirical estimates. But it needs to be emphasized that the
labor supply cannot be the sole focus of discussion of the effect of
taxes. Instead, measures of individual welfare need to be
considered. Therefore, we introduce the appropriate measures of
individual welfare, the equivalent or compensating variation. From
the equivalent variation and tax revenue raised we then develop the
notion of deadweight loss (often also ca1led excess burden). From
an economists viewpoint, deadweight loss is the correct measure of
the effect of taxation. While deadweight loss is a somewhat difficult
concept, I believe it, rather than labor supply, should be the focus
of informed discussion of the effects of taxation. If we accept the
'Other institutional factors such as pension and social security benefits are not
treated due to lack of appropriate data.
178 / INCOME TAXES AND LABOR SUPPLY
FIGURE I
y
y
notion that the purpose of the income tax is redistributive as well as
a means to raise tax revenue, then deadweight loss defines the
correct way to measure the economic cost of the income tax. The
error in considering labor supply only is that we can easily design
feasible tax policies which raise a given amount of tax revenue
while increasing labor supply from the no tax position even though
the individual is made worse off by the tax. In this situation it
would be incorrect to conclude that the tax is desirable due to its
effect on labor supply when the individual's utility has decreased.
Furthermore, the redistributive aspect of the income tax would be
eliminated by this type of tax so that the change from the current
type of system would not be acceptable.
THE MODEL OF INDIVIDUAL LABOR SUPPLY
The typical model of labor supply used in empirical work has a
very simple structure. The individual is assumed to maximize a
utility function over hours of work H and net of tax income Y,
U(H,Y). 3 Thus, all consumption goods, except leisure, have been
'Some treatments replace hours of work H by leisure, T-H, where Tis total time
available. However, since T is an unobservable variable this approach often leads to
unnecessary empirical problems.
HAUSMAN /
179
FIGURE 2
y
-H
H'
0
aggregated into a composite good which is represented by the
expenditure variable Y. Note that since H is a supply variable,
rather than a demand variable, the derivative of the utility function
has a negative sign with respect to it. The budget constraint then
becomes Y = y + w H where y is nonlabor income and w is the net
after-tax wage rate. 4 In Figure 1 we present the two-good diagram
which corresponds to this model of labor supply. The tangency of
the indifference curve which arises from the utility function U(H, Y)
with the budget line determined by non-labor income and the wage
then leads to desired hours of work H*.
In Figure 2 we then consider the effect of a wage change from w
to w '. This change could occur if the government levied a wage tax
and exempted nonlabor income, e.g., income from savings. In our
subsequent analysis we also allow for taxation of non-labor income,
but here look at the simpler case.
'In this formulation the wage and income variables are given in terms of the price
of the composite good.
180 /
I N CO M E TA X ES
A N D
L A BO R SU PP L Y
Note in the diagram that after-tax hours of work H ' exceed pretax hours H*. Nothing pathological exists in Figure 2. We merely
have the counteracting influences of the income and substitution
effects which have opposite signs under normal assumptions.' The
income effect along with the assumption that leisure is a normal
good implies that labor supply increases when non-labor income
decreases holding the wage constant. In Figure 2, the movement
from point A to point B arises from the income effect. The dashed
line which is tangent to the lower indifference curve at point B
represents the income effect since it is drawn parallel to the original
budget line and represents the same wage. The movement along the
lower indifference curve from point B to point C, then represents
the (compensated) substitution effect. It holds utility constant but
lowers the wage from w to w '. Economic theory states that the
substitution effect when the net wage falls will decrease labor
supply. Thus, even in the most simple case of a wage tax, the
income and substitution effects are of opposite sign. Econometric
estimates are necessary to measure the total response and
magnitudes of the two separate effects. In terms of the Slutsky
equation we have the formula
(I)
JH
Jw
= JH
ow
IU
+ H JH
Jy
where the first term on the right-hand side is the substitution effect
and the second term is the income effect. It is important to consider
both the income and substitution effects when considering taxation
and labor supply. As we will see shortly, it is the substitution effect
alone which measures the amount of economic cost of a tax. But
the income effect cannot be lost sight of because it normally serves
to increase labor supply when a tax is levied and determines how
much worse off an individual is made by the imposition of a tax.
THE EFFECT OF PROGRESSIVE TAXATION
We now consider the effects of two types of progressive income
taxes. The first type is a linear income tax with a constant marginal
tax rate while the second type of progressive tax has increasing
marginal rates and is closer to the current U.S. tax system. The
linear income tax has many favorable aspects. Since it has only one
'This example should not be confused with the textbook case of a Giffen good
which may never have existed in practice. Given many empirical estimates of labor
supply response, we might expect this behavior over a certain range of w and w '.
H AUSM A N /
181
FIGURE 3
y
y
A
marginal rate it would decrease socially unproductive behavior
which individuals currently engage in to reduce their tax liability.
The linear tax would lower top marginal tax rates decreasing the
incentives for certain types of tax shelters. It can also be made very
progressive at the low end through the use of a lump sum grant
amount G or an exemption level E.~ In Figure 3 we consider the
case of a linear tax with a given exemption level. For income up to
point E the individual is not taxed so that he recovers his gross
market wage w. Depending on his wage the exemption level E
defines labor supply H beyond which the individual receives a net
wage rate, w' = w(l- t) where t is the constant marginal tax rate.
Note that while the marginal tax rate is constant beyond H the
average tax rate is increasing, hence the progressive feature of the
tax. And the tax can be made extremely progressive for low Y by
adjusting E. However, a disadvantage occurs at the high end
because the progression declines as the average tax rate increases
toward the marginal tax rate t.
"The lump sum grant makes the tax similar in part to the negative income tax
proposals. For a model of individual behavior and empirical estimates under a
negative income tax see Butt!ess and Hausman (!978) and Spigelman et al. (1978).
0
182 /
l N CO M E TA X f_ S A N D L A B O R S U PP L Y
FIGURE 4
-------- --- ----- _____ _
......
y,
H,
0
The general progressive tax case is similar to Figure 3 except with
more linear segments.7 However, it differs from the previous
diagram in that no exemption is present so that each budget
segment is determined by a net after tax wage rate of wi = w(l - t,)
and the income brackets over which t holds. After-tax non-labor
income is given by y,. In Figure 4 we indicate such a budget set
with 3 tax segments although the reader should note that the actual
U.S. tax code currently has about 15 brackets.
We now address the question of how to use our labor supply model
when the budget set is no longer linear as in Figure 1. There we
assumed that the individual chose H to maximize U(H, Y) subject to
Y "' y + wH. Here we have a multiplicity of wage rates instead of
just w. The appropriate technique to use is to define the "virtual"
incomes Y; which correspond to the wages wi on a particular budget
'It is sometimes not recognized that the U.S. tax system is not progressive over its
entire range because of the effects of the earned income tax credit, social security
contributions, and the standard deduction. These tax provisions make the
appropriate budget sets nonconvex instead of convex as in Figures 3 and 4. We do
not treat this additional complication here but instead refer the reader to Hausman
(1979c).
H AUSMAN /
183
segment. Then along each budget segment the individual maximizes
utility subject to Yi + wiHi. The resulting choice is constrained by
the bracket limits which determine H, and Hi in Figure 4. That is,
the chosen hours of labor supply must be feasible in the sense of
being on the budget line in Figure 4. However, a more
straightforward approach is to use a labor supply function (which
may be determined from the original utility function) of the form
(2)
where Z is a vector of individual socio-economic variables and (3 is
a vector of parameters to be estimated. We enter each set of net
wages w and virtual income y and at most one tangency with the
feasible budget set is found. The tangency then determines labor
supply. This result follows because indifference curves for which
g( ·) is derived are concave and the budget set is convex. If no
feasible tangency is found then we will have bracketed one kink
point, e.g., Hand it will be the optimum labor supply.s Thus, in
the case of progressive taxes the situation becomes somewhat more
complex, but the usual economic theory applies. Also, the notion of
virtual income plays a crucial role in the measurement of the
welfare costs of taxation which we now turn to.
DEADWEIGHT LOSS FROM TAXATION
It is incorrect to measure the economic cost of a tax by its total
effect on labor supply. As we see in Figure 2 the wage tax served to
increase labor supply so on labor supply grounds the tax might be
deemed favorable. Yet the individual has been made worse off by
the tax since his post-tax indifference curve lies below his pre-tax
indifference curve. Furthermore, even if the government returned
the amount of tax revenue they raised, which is given by the line
segment CD, in the form of the consumption good, the individual
has still been made worse off by the tax. Thus, in our simple
example the ''size of the pie'' has increased because the tax has
brought forth more labor supply. But still the individual's utility
decreases because of the tax. It seems clear that an appropriate
welfare measure, rather than labor supply alone, is needed to
measure the effect of taxation.
The first component of a welfare measure is the effect of the tax
on individual utility. Here the measure long used by economists has
'This approach is put forward by Hausman (1979b). Other approaches have been
used by Ashworth and Ulph (1977) and Wales and Woodland (1979). See al.~o
Burtless and Hausman ( 1978).
184 / J N C O
M E
TA X ES A N D
L A B O R SU P P L Y
been some form of consumers' surplus. Consumers' surplus
corresponds to the concept of how much money each individual
would need to be given, after imposition of the tax, to be made as
well off as he was in the no tax situation. Measurement of
consumers' surplus often is done by the size of a trapezoid under
the individual's demand curve or here it would be the labor supply
curve. But Hausman (1979a) has demonstrated that in the case of
labor supply this method is very inaccurate. Instead the
theoretically correct notion of either the compensating variation or
equivalent variation must be used. 9 These measures, set forth by Sir
John Hicks, are probably best defined in terms of the expenditure
function. The expenditure function determines the minimum
amount of money an individual needs to attain a given level of
utility at given levels of wages and prices. ' 0 Its form is determined
by either the direct utility function U(H, Y) or the labor supply
function, equation (2). In our simple example of the wage tax of
Figure 3 the compensating variation equals
(3)
C.V. (w, w', U) = e(w', U) - e(w,U)
Equation (3) states that the welfare loss to the individual, measured
in dollars of the consumption good, equals the minimum amount of
non-labor income needed to keep the individual at his original
utility level U minus his non-labor income in the no tax situation,
y. Since utility is kept at the pre-tax level U, the compensating
variation arises solely from the substitution effect in the Slutsky
equation (I). The income effect is eliminated because the individual
is kept on his initial indifference curve. In the more complicated
case of progressive taxes, the only difference is that we use virtual
non-labor incomes in equation (3) rather than actual non-labor
income. 11
We need one more ingredient to complete the measure of the
welfare loss from taxation. The government has raised tax revenue,
and we need to measure the contribution to individual welfare
which arises from the government spending the tax revenue. The
assumption commonly used is that the government returns the tax
'These measures correspond to the area under the compensated demand curve
which is determined by the substitution effect in the Slutsky equation (!}. For
further discussion see Hausman (1979a) or Varian (1978).
"For a more formal treatment see Varian (1978) or Dicwcn (1979).
''The alternative measure of the equivalent variation uses post-tax utility U' as the
basis for measuring welfare loss. For labor supply in the two good set-up the
equivalent variation typically gives a higher measure of welfare loss than docs the
compensating variation.
H A U S M A N / 185
FIGURE 5
y
...
' ' ...
y
-H
H*
H'
0
revenue to the individual via an income transfer. Here it would
correspond to increasing the individual's non.labor income by the
amount of tax revenue raised. Then the total economic cost of the
tax is given by the deadweight loss (or excess burden) as
(4)
OWL (w, w', U)
=
=
C.V.(w, w', U) - T(w, w', U)
e(w', U) - e(w, U) - T(w, w', U)
Equation (4) states that the deadweight loss of a tax equals the
amount the individual needs to be given to be as well off after the
tax as he was before the tax minus the tax revenue raised
T(w, w ',U)." Dead weight loss is greater than or equal to zero
which makes sense given that we expect taxation always to have an
economic cost. Thus, even if an individual chooses to work more
after the imposition of a tax as in Figure 2, he still has not been
made better off by the tax. And the economic cost of the tax to
him is given by the deadweight loss formula of equation (4). Of
course, if no tax revenue is returned the compensating variation
gives the welfare loss to the individual. In Figure 5 the
compensating variation and deadweight loss are shown in terms of
our simple wage tax example of Figure 2.
"Here we follow Diamond and McFadden (1974) and use taxes raised at the
compensated point. Kay (1980) has recently argued in favor of using the
uncompensated point. As with C.V. and E.V. measures the problem is essentially
one of which is the better index number basis.
186 / I N C O M E TA X E S A N D L A B O R SU P P L Y
Here the effect of the tax is to reduce labor supply from H* to H '.
The compensating variation is measured by the line segment yy 1 •
We then decompose the compensating variation into its two parts.
The line segment CD measures tax revenue collected while the line
CE measures the deadweight loss of the tax. Since the taxpayer has
been made worse off but no one has benefited from the amount of
the deadweight loss, it represents the economic cost of raising the
tax revenue.
DEADWEIGHT LOSS AND TAX POLICY
Much of public finance theory is concerned with the question of
raising a given amount of tax revenue while minimizing the
economic cost as measured by the deadweight loss.' 1 But in
considering tax policy redistribution must be accounted for or
otherwise we certainly would have no need for a progressive income
tax.
Suppose the government wanted to raise tax revenue equal to R
dollars. The deadweight loss minimizing tax is a lump sum or poll
tax of amount T = R/N where N is the number of taxpayers.
Figure 6 portrays such a tax. The deadweight loss is zero because in
comparison to Figure 2 or Figure 5 note that only an income effect
is present in the movement from point A to point B. No
substitution effect is present since the pre-tax wage and post-tax
wage are identical. The compensating variation from equation (3)
equals T, the amount of tax revenue raised. Thus, the first term of
the Slutsky equation (1) is zero and the change in hours of labor
supply comes totally from the income effect. No distortion in
relative prices occurs and so no deadweight loss occurs. In equation
(4) the compensating variation term is exactly cancelled out by the
tax revenue term. Deadweight loss is zero. Furthermore, note that
labor supply increases because of the income effect. The result of
the lump sum tax is to increase labor supply while not creating any
deadweight loss. On economic efficiency grounds it is an ideal tax
and also would satisfy supply-side economists goals. 1• But it is
doubtful such a tax would ever be acceptable on political grounds
since the redistributive aspect of the current income tax has been
lost. In fact, the lump sum tax is extremely regressive since the
''For an exposition and references see Chapters 12-14 of Atkinson and Stiglitz
(1980). Mirrlees (1971) wrote the seminal paper on optimal income tax theory. See
also Mirrlees (1979).
"I do not claim to know what the exact goals or supply-side economics are.
However, an increase in the national product certainly seems high on 1he list.
HAUSMAN /
187
FlGURE 6
average tax rate decreases with labor income. Even with its
favorable supply-side effects, it is doubtful that such a tax would be
politically acceptable.
The simple example of a lump sum tax raises a number of
important issues. Taxes take away income from people. Taxes,
therefore, make people worse off, even if they are nondistortionary.
In Figure 6 the individual is on a lower indifference curve after the
tax is levied. We measure the economic cost of the tax with the
deadweight loss measure of equation (4). But if the tax revenue is
not returned to the individual who paid it, he is still worse off. The
question of individual losses from the income tax and individual
gains to the recipients of tax revenue expenditures involves
questions of redistribution. These questions cannot be avoided in
discussions of tax policy. Taxes also effect individual behavior
again even if they are nondistortionary. Along the lines of Figure 6
we can demonstrate that a lump sum tax which raises revenue T
always involves greater labor supply than a linear income tax like
Figure 3 or a completely progressive tax like Figure 4 so long as
188 / I N C O M E TAX E S A N D L A BO R S U P P L Y
leisure is a normal good. Therefore, a tradeoff exists between the
degree of progressivity that society wants in the income tax and the
economic cost measured by the deadweight loss. Thus neither
deadweight loss nor labor supply are sufficient measures alone in
evaluation of the income tax. Deadweight loss gives the economic
cost of the tax, but the "benefit" of the tax which arises due to its
redistributive aspect must also be accounted for. Unfortunately, the
correct degree of redistribution is difficult to reach agreement on,
which makes consideration of income tax policy changes a difficult
subject.
AN EMPIRICAL LABOR SUPPLY MODEL AND THE
EFFECT OF TAX REFORM PROPOSALS
In this section we first briefly discuss an empirical labor supply
model estimated by Hausman (1979c). The estimates from this
model are used to evaluate the effects of income taxation. We then
evaluate the effects of the current income tax via both deadweight
loss and labor supply effects. Following the analysis of the current
tax system, we consider two types of tax reform proposals. The
first proposal is referred to as the Kemp-Roth proposal and here we
consider reductions in the income tax rates of 10-30%. Besides
deadweight loss and labor supply effects we are also interested in
the effect on tax revenue. The change in tax revenue depends on the
labor supply response when taxes are changed. If the labor supply
response is not uniform across individuals, the change in tax
revenue will be sensitive to whether the response is concentrated
among high income or low income earners. The other type of tax
reform proposal we consider is an equal yield progressive linear
income tax like that in Figure 3. That is, we consider income taxes
with constant marginal rates which raise the same amount of
revenue as the current income tax. The overall tax will still be
progressive by Jetting the exemption level vary across tax reform
proposals. The linear tax systems that we consider are similar in
progressivity at the low income levels but display much less
progressivity at high income levels than the current tax system does.
A linear income tax is attractive because it has the potential of
sharply decreasing deadweight loss by decreasing high marginal tax
rates. But how far it can do so while raising equal tax revenues
depends on the labor supply response which we also consider. For
each of the tax reform proposals we attempt to account for
distributional effects by considering effects among population
quintiles. It is important to emphasize that all our results are partial
H A U S M A N / 189
equilibrium in nature. Potentially important general equilibrium
results are not captured by the econometric model.
AN EMPIRICAL MODEL OF LABOR SUPPLY
The essential feature that distinguishes econometric models of
labor supply with taxes from traditional demand models is the nonconstancy of the net, after-tax wage. As we saw in the previous
section, the marginal net wage and the virtual income depend on
the specific budget segment that the individual's indifference curve
is tangent to. Econometric techniques have been devised which can
treat the nonlinearity of the budget set. An econometric model
takes the exogenous nonlinear budget set and explains the
individual choice of desired hours of work. Our model is based on
the linear labor supply specification
(5)
where w is the net after-tax wage, and y is the virtual income on
budget segment i. The vector Z represents socioeconomic
characteristics of the individual. The unknown parameters a, f3, and
y are estimated using econometric techniques. Now actual hours h
may differ from desired hours h* because of stochastic reasons.
Another source of stochastic variation enters the model by allowing
for a distribution of preferences in the population via random {3.
The specific way in which these enter the model is described in
Hausman (1979c). Also a zero constraint for hours as well as fixed
costs to working enter the model. The model is estimated first for a
sample of husbands who are between 25-55 years old for the year
1975.' 5 We then estimate the model over a sample of women who
are wives of the husbands' sample. The husbands' earnings are
treated as non-labor income for the wives. Thus, wives labor supply
is conditioned on husbands labor supply. Wives also face initial
marginal tax rates given by the last tax bracket which contains their
husbands earnings.
The federal income tax is represented in the model by 12 tax
brackets. The first bracket is $1,000 wide with succeeding brackets
falling at intervals of $4,000. Since we are interested in the taxes on
labor supply, we consider only taxes on earned income. Because we
do not have access to actual tax returns, a number of assumptions
';It is important to note that neither the model nor the simulations treat the young
or old segments of the working population. We would expect a labor supply model
to differ markedly for such individuals. Nor do we treat non-married individuals.
190 /
I N C O M E TA X E S A N D L A BO R S U P P L Y
are required. We assumed that all married couples filed jointly. In
forming the taxable income we took account of personal
exemptions and assumed that individuals used the standard
deduction up to the (1975) limit of $16,250. The standard deduction
was used on approximately 2/3 of all tax returns in 1975. Beyond
$20,000 we used the average of itemized deductions for joint
returns for each tax bracket found in Statistics of Income. We also
take account of the earned income credit and social security
contributions which were 5.85% up to a limit of $14,000 for 1975.
Lastly, we take account of state income taxes by putting the tax
laws of the 41 states who taxed earned income into the budget set
calculations. Thus we had a reasonably complete characterization of
taxes which individuals faced on their earned income.'"
We briefly discuss the results from the model for the average
individual in the sample. A more complete discussion is contained
in Hausman (1979c). For husbands we found the uncompensated
wage elasticity to be very near zero. This result is similar to the
findings of previous research. However, by taking account of the
tax system via the virtual incomes we find an income elasticity at
the mean hours of work to be approximately - .177 for the mean
wage in the sample. Thus, the presence of a non-zero income
elasticity implies that husbands' labor supply decisions are affected
by the income tax. Also the deadweight loss may be significant
because the substitution effect of the Slutsky equation (I) will be
non-zero given our estimates. For wives we find the uncompensated
wage elasticity to be .906. The income elasticity for the mean
woman who works full time is approximately - .504. 17 Thus, both
the uncompensated wage elasticity and income elasticity are nonzero which indicates that taxes have an important effect on both
labor supply and deadweight loss.
Given the model specification and estimates, we can now apply it
to evaluate the effect of income taxation. Suppose we want to
evaluate a tax reform proposal. The estimated change in labor
supply can be found from equation (6) by entering the new tax plan
via the marginal tax rates w; and virtual incomes Y;- A micro
simulation is done on the sample of husbands and wives, and the
"City income or wage taxes could not be included due to lack of specific job
location data. Minor problems may also be created because of the tax treatment by
states or earnings of non-residents.
"It is important to note that this elasticity is calculated at a mean virtual income
of approximately $8200. The reason for the high virtual income is that husbands'
earnings are included in the non-labor earnings of the wife.
H AUSMA N /
191
change in labor supply is calculated. The specific manner in which
stochastic elements of the model are treated in the simulations is
given in Hausman (1980). To do deadweight loss calculations we
need the expenditure function for equation (3). Hausman (1979a)
derives the expenditure function which corresponds to the labor
supply function, equation (5), to be
(6)
-/1Wi
U +-a
{1
W·
I
+
a {12
Zy
-
{1
We take the marginal wage wi from the budget set and then
calculated the deadweight loss from equation (4) using taxes raised
at the compensated labor supply point. We then have our welfare
measure of the cost of the income taxation. Two possible objections
to our welfare measure are that we aggregate across individuals,
giving each individual the same weight in the implicit social welfare
function. Also different individuals are allowed different
coefficients in their expenditure functions. The problems created for
analysis of vertical equity considerations for these choices are
discussed in Atkinson and Stiglitz (1976). But we attempt to
indicate the importance of these considerations by looking at
distribution measure across different income categories.
CURRENT TAX POLICY AND KEMP-ROTH REDUCTIONS
We begin our analysis of the current tax policy by considering the
effect of the current tax system on the labor supply of husbands.
First, we consider the mean individual in the sample. His before tax
wage is $6.18 per hour and his non-labor income is $1266. Without
taxes the labor supply model predicts he would work 2367 hours
per year, but the effect of the current tax system is to lower his
labor supply to 2181 hours per year. Thus, the effect of taxes is to
decrease his desired labor supply by 8.2%. To calculate the welfare
loss for these husbands we look at the deadweight loss (DWL) based
on the compensating variation measure of deadweight loss from
equation (3). For the mean individual we calculate the deadweight
loss to be $235 which is 21.8% of the total tax revenue collected
from him. It is 2.40Jo of his net, after-tax income. Thus, we see that
taxes on earned income have an important effect on both labor
supply and on deadweight loss. These results differ markedly from
the received knowledge in the field, e.g., Pechman (1976), which is
that taxation has almost no effect on the labor supply of prime age
males. Also, the deadweight loss calculation indicates that the
192 / ( N CO M E TA X E S A N D L A BO R S U P P L Y
TABLE 2
Mean Tax Results for Husbands
Market
Wage
$ 3.15
4.72
5.87
7.06
10.01
DWL
$
66
204
387
633
1749
OWL/Tax
Revenue
OWL/Net
Income
Change in
Labor Supply
9.4%
14.4
19.0
23.7
39.5
0.8%
2.0
3.1
4.5
9.9
4.5%
6.5
8.5
-10.1
-12.8
income tax is a relatively high cost means of raising tax revenues.'"
If less expensive means to raise federal tax revenue do not exist, the
large amount of redistributive expenditure by the federal
government is being done at relatively high economic cost.
Now the mean individual calculation leaves out two potentially
important factors. First, because of the nonlinearity of the tax
system, it may provide a poor guide to population averages. It can
be shown that deadweight loss is proportional to the square of the
marginal tax rate so that deadweight loss will grow quickly as
marginal rates rise. Second, distributional considerations are
neglected. We have emphasized that an important objective of the
income tax system, in addition to raising tax revenue, is to
redistribute income. We attempt to investigate distributional
considerations by looking at quintiles based on the market wage.
The market wage seems a better measure than income to base
distributional categories on, because it is closer to the notion of the
opportunity set of the individual. In an optimal tax calculation, the
tax is based on the opportunities facing the individual instead of
post-tax behavior.
In Table 2 we look at the effect of the current tax system for five
categories defined by the market wage. Overall, we find that the tax
system decreases labor supply by 8.5% and the mean deadweight
loss as a proportion of tax revenue raised is 28.7%. Thus, the
results are not too different from the results for the mean
individual. However we note important differences among the five
categories.
"Of course, the economic cost of raising revenue from other federal taxes would
need to be investigated before an informal choice could be made. Federal taxes on
labor income currently raise about 75% of federal revenues.
H AUSMA N /
193
First, we see that deadweight loss rises rapidly with the market
wage as we expected. In terms of the welfare cost of the tax we see
that the ratio of deadweight loss to tax revenue raised starts at
9.4% and rises to 39.5% by the time we reach the highest wage
category. Again we see that the cost of raising revenue via the
income and payroll taxes is not negligible. In terms of a
distributional measure we see that the ratio of deadweight loss to
net income also rises rapidly. In fact, this measure indicates that
individuals in the highest wage category bear a cost about 10 times
the lowest category while individuals in the second highest category
bear a cost 5 times as high. Without specific social welfare measure,
we cannot decide whether the current tax system has too much, too
little, or about the right amount of progressiveness. But the
measures of Table 2 seem an important step in thinking about the
problem. Lastly, note that the change in labor supply from the no
tax situation again rise with the wage category. The high marginal
tax brackets have a significantly greater effect on labor supply than
do the low tax brackets.
We now do a similar set of calculations for our sample of wives.
While we found both significant deadweight loss and an important
effect on labor supply for husbands compared to the no tax
situation, the situation is more complicated for wives. First, about
half of all wives do not work. In the absence of an income tax, the
net wage would rise causing some of them to decide to work and
others to increase their labor supply. But, at the same time their
husbands' after-tax earnings would also rise which has the opposite
effect on labor force participation. Thus, both effects must be
accounted for in considering the effects of the income tax.
TABLE 3
Mean Tax Results for Wives
Market
Wage
$2.11
2.50
3.03
3.63
5.79
OWL
DWL/Tax
Revenue
OWL/Net
Income
Change in
Labor Supply
23
119
142
184
1283
4.6%
15.3
15.9
16.5
35.7
.3%
1.3
1.5
L7
8.6
+3L20Jo
~ 14.2
-20.3
-23.8
-22.9
$
194 / INCOME TAXES AND LABOR SUPPLY
Overall for wives, we find the ratio of deadweight loss to tax
revenue to be 18.4%. But it should be remembered that this ratio
understates the effect on labor force participants alone. For labor
supply, we find that taxes serve to increase labor supply in the
lowest wage category, but decrease labor supply as the wage rises.
Overall, they decrease labor supply by 18.2%. Thus, again for
wives we see that the current income tax system has both an
important labor supply effect and imposes a significant cost in
welfare terms for raising tax revenue.
We now turn to a consideration of Kemp-Roth type tax
proposals. We will consider two levels of tax cuts, 10% and 30%.
The question which has been focussed on most is what effect these
tax cuts would have on tax revenues. Our results are partial
equilibrium so that general equilibrium effects are not accounted
for. The main effect here arises from the change in labor supply.
But increased labor also moves some individuals into higher tax
brackets. Both effects need to be accounted for. In Table 4 we
present the two Kemp-Roth simulation results. For the 10% tax
deduction mean hours of labor supply for husbands rise 22.5 hours
or 1.1%. Tax revenues fall by 7.4%. Even given the fact that our
model is partial equilibrium, rudimentary calculations demonstrate
that general equilibrium effects are very unlikely to be large enough
to cause tax revenues from decreasing significantly in the short run
as our results show. In terms of the welfare cost of the tax we see
that the DWL falls significantly. The ratio of mean deadweight loss
to tax revenue falls from 22.1 % under the current system to 19 .0%
under the 10% tax cut plan. ' 9 For the 30% tax cut labor supply
increases by 2.7% while tax revenue falls by 22.6%. Again we see
that deadweight loss decreases significantly with the ratio of
deadweight loss to tax revenues raised decreasing to 15.4%. Thus
Kemp-Roth type tax cuts have large effects both in terms of
decreasing deadweight loss and in decreasing government revenue.
Without knowledge of marginal government expenditure, it is
difficult to evaluate the tradeoff. But we cannot recommend KempRoth on welfare grounds alone given the substantial fall in
government revenue.
"A problem arises here because we are doing welfare calculations with different
indifference curves because of the tax changes. But we are using a common basis of
comparison, the no tax situation.
TABLE 4
Kemp-Roth Tax Cut Proposals for Husbands
300/o Tax Cut
10% Tax Cut
Market
Wage
$
3.15
DWL/Tax
Revenue
DWL/Net
Income
Change in
Labor Supply
DWL/Tax
Revenue
DWL/Net
Income
Change in
Labor Supply
8.50/o
.7%
+.4%
6.8%
.4%
+ 1.3%
4.72
13.3
1. 7
+.5
10.9
1.1
+ 1.6
5.87
17.4
2.6
+.9
14.5
1.8
+2.7
7.06
21.8
3.8
+1.1
17.9
2.5
+ 3.1
10.01
36. l
8.2
+ 1.4
29.5
5.3
+4.6
196 I I N CO M E TA X E S A N D L A B O R S U P P L Y
For wives we do not present detailed quintile results because the
overall pattern is similar to husbands. The mean results are given in
Table 5.
TABLE 5
Overall Kemp-Roth Tax Cut for Wives
Tax Cut
Change in
Tax Revenue
Change in DWL
Change in
Supply (Hours)
10%
30
- 3.8%
-16.2
-10.6%
- 17.4
+ 50.2
+ 117 .o
Overall, we see that the labor supply response to a tax cut is greater
for wives than for husbands. We expect this since the wage
elasticity is about twice the income elasticity so we should have a
net increase in labor supply. Furthermore the difference in the
elasticities is about four times that of husbands, and we do observe
a significantly larger response. For the 10% tax cut case labor
supply increases by 4.1% and tax revenues fall by 3.8%. For the
30% tax cut case labor supply increases by 9.4% and tax revenues
fall by 16.2%.
Our overall evaluation of the Kemp-Roth tax proposals is that
while tax revenues will decrease by significantly less than the tax
cut, overall government revenue from the income and payroll tax
will decline. An argument might be made that general equilibrium
results may be large enough to reverse this conclusion, but I doubt
that it is a valid argument, especially in the short run. Thus, unless
a strong argument can be made for reducing government
expenditures with little welfare loss from the recipients, the KempRoth tax cut proposals cannot be supported on the basis of our
results. They certainly do not have the "free lunch" properties
claimed by some of their supporters.
A LINEAR INCOME TAX
We now consider an equal yield change from the current tax
system to investigate whether the welfare cost in terms of
deadweight loss can be significantly decreased. The type of tax
system which we consider are linear income taxes with initial
H A USM A N /
197
exemptions like the tax system drawn in Figure 3. Thus, we specify
an initial exemption E and then search our marginal tax rates until
we find the minimum tax rate which raises the same amount of tax
revenue as the current tax system. We might expect such a linear
income tax to do well in two respects. io
First, in Table 2 we saw that deadweight loss increases rapidly as
marginal tax rates increase. Since the linear income tax will not
have such high marginal rates, deadweight loss should be decreased.
Second, we would expect a significant labor supply response given a
decrease in the marginal tax rates. Thus, the tax rate should not
have to be too high to raise equal revenues to the current tax
system. Yet a potential problem still exists. Even if total deadweight
loss decrease, some individuals may still be made worse off by a
change from the current tax system to a linear income tax.
Although overall deadweight loss will decrease, we have the
problem of potential versus actual compensation which was the
basis of the Kaldor-Hicks-Scitovsky-Samuelson debate of the l 940s.
However, we will see that the linear income tax does so well that
the problem may be overcome in some cases.
In Table 6 we consider the equal yield linear income tax for
husbands. Note first that the tax rate begins at 14.6% with an
exemption level of zero and rises to 20.7% with an exemption of
$4000. Each tax measure gives a substantial welfare gain. Since tax
revenues remain the same the change in deadweight loss gives the
welfare improvement. Note that even with the highest exemption
level of $4000 the deadweight loss falls by 49% from the current
system. The labor supply also increases substantially from the
current system. My conjecture is that except for a lump sum tax,
we have done about as well as possible because labor supply is now
only approximately 1.5% below the no tax case. Lastly, we look at
the question of distribution. By considering the average tax rate for
various exemption levels, we see that either the $2000 or $4000
exemption is superior to the current tax system since the average (as
well as the marginal) tax rate is lower at every tax bracket. The
results are sensitive to various deductions and credits an individual
taxpayer declares but yield the conclusion that approximately all
taxpayers are made better off by this type of linear income tax
system. 21
"Mirrlees (197]}, when he considered the optimal nonlinear income tax, found
that the optimal tax was nearly linear for the particular labor supply function he
considered.
"The earned income tax credit is taken into account in these calculations.
TABLE 6
Equal Yield Linear Income Tax
With Initial Exemption for Husbands
Average Tax Rate at:
4000 8000 16000 24000
Exemption
Level
Tax Rate
Change in
Deadweight Loss
Deadweight Loss/
Tax Revenue
Change in Hours
0
14.6%
- 825.75
.071
+ 170.0
.146
.146
.146
.146
$1000
15 .4
- 798.82
.083
+ 169.3
.116
.135
.144
.148
2000
16.9
-765.31
.098
+ 167.6
.085
.127
.148
.155
4000
20.7
-659.18
.145
+ 163.0
0
.104
.155
.172
.119
.147
.173
.188
Current
Tax Code
IRS Code
.287
H A U S M AN
I 199
TABLE 7
Linear Income Tax for Wives
Exemption
Level
Tax Rate
0
$1000
2000
4000
14.6%
15.4
16.9
20.7
Change
In Taxes
5.1%
.3
+ 4.6
+ 11.2
Dead weight
Loss/
Tax Revenue
Change
In Hours
.104
.110
.114
.143
+372.6
+ 345.l
+302.2
+232.8
We briefly consider what effect this type of tax system would
have on wives. We assume here that each family gets only one
exemption and faces the same marginal tax rates as her husband.
We use the tax rates from Table 6 so that tax revenue for wives is
not held constant. The results are presented in Table 7. As we
expect, labor supply increases for women with the linear income tax
because the marginal tax rate has decreased. Because of the increase
in labor supply, the revenue changes are not that large. Tax
revenues fall by 5 .1 % for a 14.6% tax rate but rise by 11.2% for
the case of a 20.7% tax rate. The ratio of deadweight loss to tax
revenues falls markedly from the current tax system. Thus, for
wives as well as husbands, the linear income tax has favorable
implications from an economic cost viewpoint.
Our example bears out to some extent the lessons from the
optimal tax literature. The crucial parameters there are the weighted
(compensated) substitution response and the net revenue raised
from each individual. We use the same weights for each individual
in our deadweight loss calculations. Our results indicate the
importance of the net revenue consideration. Because of the labor
supply response, Tables 6 and 7 demonstrate that lower income
groups can gain from lowering the top marginal income tax rates.
Can anyone then object to the case for a linear income tax? The
answer is unfortunately yes, if it is relative rather than absolute
income or utility that matters for society's choices on distribution
matters. 22 Economists used to the Pareto principle typically think of
each individual's or family's welfare apart from the rest of the
"Such cases are analyzed by Fair (1971) and Boskin and Sheshinski (1978).
200 / INCOME TAXES AND LABOR SUPPLY
population. Since the linear income tax has the possibility of
making everyone better off, most economists would favor it on
these grounds. But by sharply decreasing the top marginal rates
from say 50% to 20.7%, the highest paid individuals have a greater
increase in welfare than do the lowest paid. Therefore, on a relative
basis or by some income distribution measures, the linear income
tax might not be an improvement from the current tax system.
These arguments would need to be considered in tax reform
discussions. I favor such a change in our tax system because I do
not give great weight to the relative welfare argument. Favorable
economic effects could occur with less progression in the tax system
at higher income levels. This type of proposal emphasizes the
economic efficiency aspects of the tax system. Thus, it seems that a
more linear type of tax system is to be favored over the current
system. The Kemp.Roth tax cuts do not do nearly as well by
comparison.
HA USMA N /
201
REFERENCES
Ashworth, J. and D. T. Ulph. "On the Structure of Family Labor
Supply Decisions." mimeo, 1977.
Atkinson, A. B. and J. E. Stiglitz. "The Design of Tax Structure:
Direct Versus Indirect Taxation." Journal of Public Economics,
6 (1976).
_____ . Lectures on Public Economics, New York: McGrawHill, 1980.
Auerbach, A. J. and H. S. Rosen. "Will the Real Excess Burden
Please Stand Up? (Or Seven Measures in Search of a Concept)."
mimeo, 1980.
Boskin, J. J. and E. Sheskinski. "Optimal Income Distribution
When Individual Welfare Depends on Relative Income."
Quarterly Journal of Economics, 92 (1978).
Burtless, G. and J. A. Hausman. "The Effect of Taxation on
Labor Supply." Journal of Political Economy, 86 (1978).
Diamond, P. and D. McFadden. "Some Uses of the Expenditure
Function in Public Finance." Journal of Public Economics, 3
(1974).
Diewert, W. E. "Duality Approaches to Microeconomic Theory."
mimeo, 1979.
Fair, R. C. "The Optimal Distribution of Income." Quarterly
Journal of Economics, 85 (1971).
Hall, R. E. "Wages, Income and Hours of Work in the U. S.
Labor Force." in G. G. Cain and H. W. Watts, eds., Income
Maintenance and Supply, Chicago: Academic, 1973.
Hausman, J. A. "Exact Consumers' Surplus and Deadweight
Loss." American Economic Review, forthcoming, 1979a.
_____ . "Labor Supply with Convex Budget Sets." Economic
Letters, 3 (1979b).
______ "The Effect of Taxes on Labor Supply." In H. Aaron
and J. Pechman, eds., How Taxes Affect Economic Behavior,
Washington: Brookings Institution, 1981.
_____ . "Stochastic Problems in the Simulation of Labor
Supply." prepared for NBER conference, October 1980.
202 / lNCOME TAXES AND LABOR SUPPLY
Hausman, J. A. and D. Wise. "Evaluating the Results from
Truncated Samples." Annals of Economic and Social
Measurement, 5 (1976).
Kay, J. A. "The Deadweight Loss from a Tax System." Journal of
Public Economics, 10 (1980).
Mirrlees, J. A. "An Exploration in the Theory of Optimum Income
Taxation." Review of Economic Studies, 38.
~~~-~· "The Theory of Optimal Taxation." mimeo, 1979.
Pechman, J. Federal Tax Policy (3rd ed.) Washington: Brookings
Institution, 1976.
Varian, H. Microeconomic Analysis. New York: Norton, 1978.
Transfers, Taxes and the NAIRU
DANIEL S. HAMERMESH
Just as war is too important to be left to the generals, the impact
of taxes and transfers on the aggregate unemployment rate is too
important to be left to the macroeconomists. I therefore subject the
issue of how tax and transfer policy affects unemployment and
aggregate supply to a detailed, microeconomic examination of the
effects of individual tax and transfer program structures. This
inductive approach is, I believe, likely to provide a far better guide
to discovering how changes in these policies have worked through
the economy than would a macroeconomic approach that ignored
the programs' complexities.
Throughout the discussion we need to distinguish the programs'
effects on two different aspects of economic performance. First,
they may affect the measured nonaccelerating-inflation rate of
unemployment (NAIRU). Such effects would be important for
planning macroeconomic policy, though it is not clear how
informative knowledge of any effects on the NAIRU is for learning
about aggregate supply. Second, each tax and transfer policy may
change the amount of employment observed at the NAIRU;
assuming productive efficiency, this means that these policies will
affect the amount of output, and thus per-capita incomes observed
in the economy. It is this second set of effects that is more in the
spirit of the supply-side discussions of recent years. Unlike the first
effect, it is more than just an issue of measurement.
Before proceeding to present first a macro approach to the issue,
then a detailed micro approach, it is worth considering some wellknown (to labor economists) aspects of labor force change over the
past twenty years. For selected years of roughly comparable
aggregate demand pressures (though 1969 was probably somewhat
tighter than the other two years), we present the aggregate
unemployment and participation rates, and unemployment rates,
Daniel S. Hamermesh is Professor of Economics, Michigan State University,
and Research Associate, National Bureau of Economic Research, Cambridge, Mass.
Helpful comments on an earlier version of this paper were provided by Alan Blinder.
203
204 / TRANSFERS, TAXES AND NAIRU
participation rates and labor force shares of five demographic
groups. Several features, in decreasing order of my estimate of their
importance in the history of the U.S. labor market over the past 20
years, stand out: 1) The adult female participation rate has
skyrocketed, causing that group's representation in the civilian
labor force to jump from 30 to 38 percent; 2) As a result of the
post-war baby boom, the teen-age share of the labor force has also
increased, a rise that has been accentuated by the simultaneous rise
in (mostly part-time) labor-market participation in this group; 3)
The participation rates of older males have decreased drastically,
substantially lowering their representation in the labor force. (This
change is a major focus of my discussion in the fourth section
below.); and 4) Partly as a result of the first two changes and their
interaction (see Grant and Hamermesh, 1981), the unemployment
rate of teenagers has increased sharply. Teenagers are indeed one of
only two groups among the five whose pattern of unemployment
rates across the three years departs obviously from the aggregate
rate. (The other is older men, whose unemployment rate is lower in
1979 than in 1957.)
A MACRO APPROACH TO THE EFFECTS OF TRANSFERS AND TAXES
If you are an unreformed macroeconomist, and you believe that
taxes and transfers have affected the NAIRU, your initial
inclination should be to specify a time-series equation to estimate
the direction and magnitude of their effects. In the case of
unemployment insurance benefits, such a time-series model has
been estimated by Grubel and Maki (1976). Postulating that the net
effect will be positive, they find, in a regression of the logarithm of
the aggregate unemployment rate on the gross replacement rate of
UI benefits and other variables, that this effect is observed in the
data. Unfortunately for believers in such models, the size of the
effect is so large as to imply that unemployment would be reduced
nearly to zero if the UI program were abolished. 1
Taking this simplistic approach to its logical conclusion, we
estimate in this section an equation explaining variations in
aggregate unemployment. The dependent variable is log (U* / 100-U*),
a transform of the adjusted unemployment rate. Rather than using
the published aggregate unemployment rate, we use a constantweight average of unemployment rates of teenagers, women 20 +,
'The implied effect of a . l increase in gross replacement by Ul in the Grubel-Maki
study is an extra 6.31 percentage points of unemp!oymem !
HAMERMESH I 205
TABLE 1
Selected Labor Force Data, 1957, 1969, l 979
1957
1969
1979
Aggregate
Unemployment Rate
Participation Rate
4.3
59.6
3.5
60.1
5.8
63.7
Teens
Unemployment Rate
Participation Rate
Fraction of Labor Force
8.8
49.7
.064
8.8
49.4
.086
16.1
58. l
.092
Women 20+
Unemployment Rate
Participation Rate
Fraction of Labor Force
4.1
36.5
.297
3.7
42.7
.340
5.7
50.6
.378
Men 20-24
Unemployment Rate
Participation Rate
Fraction of Labor Force
7.8
87.0
.054
5.1
82.8
.065
8.6
86.6
.080
Men 25-54
Unemployment Rate
Participation Rate
Fraction of Labor Force
3.1
97 .1
.455
1.6
96.1
.395
3.4
94.4
.362
Men 55+
Unemployment Rate
Participation Rate
Fraction of Labor Force
3.5
63.4
.130
1.9
56.1
.114
2.9
46.7
.088
men 25-54, and other men, where the weights are their shares in the
civilian labor force in 1957:1. This refinement circumvents the
problem that growing replacement rates of transfer programs are
observed to be positively correlated with an aggregate
unemployment rate that is rising because of the very substantial
changes in the demographic mix of the labor force that have
occurred since 1957.
206 /
TRANSFERS, TAXES
AND NAJRU
To represent transfer and tax policy, two variables are used, in
each case with lags to avoid part of any problem that may be
caused by simultaneity. These are: I) NRR, the net replacement rate
of transfer payments in aggregate. This is computed as personal
transfer payments, divided by wages and salaries minus personal
contributions for social insurance minus a prorated (by wages'
share in personal income) share of personal income taxes; and 2)
TAX, the sum of personal income taxes on wages and salaries, and
individual and employer contributions for social insurance, all
divided by the sum of wages and salaries and employer social
insurance contributions.' This is designed to measure any
disincentive effects that taxes on wages and salaries may have
beyond their effects through the financing of transfer payments.
Also included in the model are a time trend variable and the
change in the rate of growth of per-capita real GNP. 3 This
acceleration term seems more appropriate than the growth rate
itself, as it is hard to argue that the NAIRU will vary with the
steady-state growth rate of an economy. The model is estimated
over U.S. data from 1954:II through 1978:IV. Both simple lag
terms in NRR and TAX are included, and variants that include
polynomial distributed lags in these variables are also estimated.•
All of the equations are estimated using the Cochrane-Orcutt
technique to account for first-order autocorrelation in the residuals.
The results of estimating four versions of the equation relating a
logarithmic transformation of the adjusted unemployment rate to
the variables defined above arc presented in Table 2. The change in
the rate of per-capita real GNP growth has the expected negative
sign. Interestingly, the trend coefficient is negative. (Remember, we
have removed any trend effects produced by demographic changes
in the labor force.) Including all lagged terms (in both NRR and
TAX) significantly increases the explanatory power of the
'A TAX variable that excluded employer contributions from both nnmeralor and
denominator was also used in place of the variable discussed in the text. While the
results were qnalitativcly similar, the coefficient of detcrminaiion was in every case
sllghlly lower.
'The model was also estimated with the theoretically improper variable, percent
change in GNP. Though the R' exceeded those reported for comparable equations in
Table 2, and though the implications of NRR and TAX were the same as in the
table, the lack of a good justification for this variable suggests the discussion should
be based on the model including its rate of change.
'The polynomial lags were estimated with the far end-point coefficients
constrained to equal zero. A test of the validity of these constraints in the equation
in column (4) yielded F(J,87} ~ .49. (The 95 percent significance level with these
degrees of freedom is 2. 7 I.)
TABLE 2
Effects on log (U* /100-U*)
1954:II-1978:IV
Constant
GNP-GNP._,
(sum of four
lagged terms)
Time
NRR_,
(1)
(2)
(3)
(4)
-3.39
(-17.93)
-3.34
(-12.79)
-3.45
(-19.01)
-4.26
( - 9.91)
- .036
- .037
-.047
-.045
(- l.64)
(- 1.66)
(-2.11}
( -- 2.04)
-.011
-.Oll
(-3.0])
(-2.71)
-.014
(- 3.04)
-.022
(-3.74)
6.71
(4.80)
6.22
(4.79)
6.15
(5.00)
5.91
(4.87)
2.21
(3.98)
(4.44)
NRR_,
-.13
NRR_,
NRR"",
TAX._,
2.42
(- .17)
.27
(.36)
- .86
( - 1.38)
( - .85)
-.54
.15
(.13)
- .361
(- .29)
L56
(2.04)
TAX_z
TAX_,
2.01
(2.27)
1.49
(2.16)
TAX_.
Ri
D-W
Q
.9320
l.31
.912
.9320
1.31
.911
.9348
L31
.902
.9384
1.29
.900
208 / TRANSFERS, TAXES AND NAIRU
equation. s We thus base our discussion of these variables' effects on
the results in column (4) of Table 2. Both the terms in the net
replacement rate and those in the tax rate are significant, and the
sum of each set of four coefficients is positive.
Since NRR grew from .095 in 1954:II to .265 in 1978:IV
(reaching a high of .290 during the 1973~75 recession), we may infer
that the growth of transfer payments relative to net wages and
salaries has induced an increase in the unemployment rate. A
similar inference may be drawn from the positive coefficients on
TAX and the increase in TAX from .167 to .301 (its highest value)
during this period. However, lest this be reported in tomorrow's
Wall Street Journal as proof positive of the deleterious effects of
transfers and taxes on labor income, two considerations are in
order. First, the coefficients imply incredibly large effects of taxes
and transfers on the adjusted unemployment rate. For example, a
one standard deviation increase in NRR from its mean is seen co
induce an increase in U* from its mean, 5.00, to 7.85. Similarly, an
increase in TAX of one standard deviation from its mean of .231
induces an increase of U* from its mean to 6.08. • Both of these are
ridiculously large, suggesting other things are going on that we have
not accounted for. Second, it may be the skepticism of one who has
seen too much simple-minded macroeconometric "evidence," but l
tend to disbelieve studies whose bold conclusions are based solely
on time-series results. Accordingly, I would give little weight to the
results in this section, and would instead base my conclusions on
careful thought about the programs' effects and on cross-section
evidence about their impact.
SOME THEORETICAL CONSIDERATIONS
Given my skepticism about using macro estimates of the effects
of taxes and transfers on unemployment to deduce their effects on
the NAIRU, it is incumbent upon me to propose some alternative
method of answering this question. Help is provided by the
approach of Perloff and Wachter (1979) and others who use
aggregate production and pricing models to deduce what aggregate
unemployment rate, adjusted for demographic change, is consistent
with nonaccelerating inflation. This method is dearly the correct
'In an equation like that in column (4) from which TIME was excluded, the sum
of the coefficients on NRR was 5.35, and that on TAX was 3.00.
"NRR has a mettn of .171 and a standard deviation of .060; TAX has a mean of
.Z3! and a standard deviation of .040. Their correlation is .933.
HAMERMESH / 209
one for macro policy planning; it does not, though, as its users
would readily admit, indicate whether changes in tax and transfer
policy are responsible for changes in the NAIRU. (This approach
really says little about the causes of changes in the NAIRU.) Thus,
while it may be helpful for other purposes, it provides no evidence
on the positive issues under consideration here.
A second approach is simply to make grandiose statements about
how the NAIRU has increased tremendously, or, depending upon
one's political views, how unemployment much above four percent
is evidence of a recession. In the former camp we have statements
from at least one ex-Chairman of the Council of Economic
Advisors; sympathetic to the latter, a recent annual report of the
Council of Economic Advisors made the bold admission that, "A
number of forces have been at work ... to raise the overall
unemployment rate at which inflationary pressures begin to appear
above the neighborhood of 4 percent. .. ? 1 Neither statement has
the least bit of scientific basis, and neither should therefore receive
any serious attention. Nonetheless, because of the political
importance of the issue, and because of the attention those making
such statements command, they have infected the public debate.
They do not, though, tell us anything about how or to what extent
transfers and taxes have affected the labor market.
A third approach is inductive; it tries to construct, from available
estimates of the effects of individual tax and transfer programs, the
likely impact on the NAIRU of the sum of such programs. The
problem with this approach is that, unless one examines the
underlying estimates carefully before basing one's conclusions upon
them, one quickly comes to outlandish results. For example, taking
Feldstein's (1973) estimate that unemployment insurance (UI)
benefits and taxes induce a 1.25 percentage point increase in the
NAIRU, and combining it with Clarkson and Meiners' (1977)
estimate that AFDC and Food Stamps work registration
requirements have raised measured unemployment by two
percentage points, the absurdity of the exercise becomes apparent.
It is impossible to believe that without these two fairly small
programs, the unemployment rate in 1979 would have been reduced
to below 3 percent. Either these effects are not additive, or the
'Herbert Stein noted, "I am not in a position to insist that it [l"ul! employment] is
7 percent unemployment. But it is a possibility that must be given weight. Suppose
we accepted the idea that there is a 50-50 chance that we are now at full
employment." ( Wall Street Journal, September 14, 1977, p. 22) The CEA statement
is from the Report, 1978, p. 171.
210 / TRANSFERS , TA X ES A N D N A l RU
underlying estimates are grossly overstated. (The former criticism
may be correct, though I present no evidence on it; the latter does,
as I show below, have substantial support.) Given these difficulties,
this third approach is also not one that is likely to produce precise
estimates unless great care is given to the interpretation of the
underlying studies.
What I do here is recognize that the NAIRU has increased since
the 1950s, probably by the slightly more than 2 percent implied for
1977 by the Perloff and Wachter study. Of this increase a bit more
than one percentage point has been attributed by Wachter (1976) to
changes in the demographic mix of the labor force. Using the four
groups underlying the calculation of U* in the estimates in the
previous section, I find that the unemployment rate would have
been .85 percentage points lower in 1978:IV had the labor-force
weights of 1957:l prevailed. (I am somewhat uncomfortable with
the assumption implicit in this approach that the relative
unemployment rates of the various demographic groups must
remain unchanged from 1957. In any case, those who loved the
implications of this approach for the 1970s' labor market may be
less enthralled with its implications for the late 1980s!) The task,
then, is to consider on a program-by-program basis whether the
remaining one percentage point increase could have been produced
by changes in transfer policy. In conjunction with this we consider
whether the slowdown in the growth of real output per capita may
also have been in part induced by these policy changes.
Although it is impossible to summarize in a succinct way the
massive amount of theoretical work on the incentive effects of
various transfer programs, I believe that there are sufficient general
similarities among the programs' effects to make a general
discussion of their likely economic impact worthwhile. The purpose
of doing so is to point out some aspects of these effects that have
been ignored by research that has been concentrated narrowly; to
demonstrate the similarities among various strands of research; and
to provide a focus for the discussion of specific programs' effects in
the next section. Throughout this analysis we assume that leisure
and unemployment are synonymous-both are voluntary. We also
recognize that any attempt to synthesize a general model will surely
ignore some important programmatic details within individual
transfer schemes.
We examine the likely effects of transfers under the assumption
that each member of the adult population faces two separate
situations vis-a-vis these programs. In the first the individual is
H AM E R M E S H / 211
FIGURE l
Budget Constraints Before Eligibility for Benefits
Income
B
H
E
A
'----------------------'---Leisure
0
ineligible for benefits under the program. Nonetheless, the program
affects his behavior because of the incentives it provides to establish
eligibility for benefits later on. This represents the entitlement effect
discussed for Ul in Hamermesh (1979b), part of the effect of OASI
on hours of work before age 62 implicit in Burkhauser and Turner
(1978), and the work incentive effect of OASI through automatic
benefit recomputation noted in Blinder et al. (1980). As Figure 1
shows, the budget line in the absence of the transfer scheme (and
the taxes that finance it) is OAB. With the transfer program and its
concomitant tax structure the line shifts to OACFGH. As compared
to the budget line OADE, describing the choice set available to the
worker who sees only the wage net of taxes, the constraint
OACFGH induces substantial changes in behavior. (See Moffitt and
Kehrer, 1980; Burtless and Hausman, 1978; and Hamermesh, I 980.)
Some persons who would have been at the corner solution at A, or
who would have found an internal maximum along AC, are
induced by the entitlement aspect of the transfer program to
increase their supply of labor and move to point F. (In addition to
its effects in UI and OASDI, it may also be operative in affecting
military enlistments, as the post-service educational and other
benefits are an added bonus to enlistees.) Though this entitlement
effect has no immediate impact upon unemployment rates, it may
212 / TRAN SF E RS , TA XE S AN D NA l RU
FIGURE 2
Budget Constraints When Eligible for Benefits
Income
C
A
,.___ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __.__Leisure
0
change the aggregate rate insofar as it increases labor force
participation among persons whose probability of being
unemployed differs from the average. So too, it will clearly increase
market employment and thus measured real GNP.
Once eligibility for the transfer is established, the individual faces
a different set of constraints. Under UI and OASDI these can
mutatis mutandis be described as resulting from a lump sum benefit
paid if no work, or only a small amount of work, is undertaken; as
reflecting the sum of the wage rate and a steadily reduced benefit as
hours increase, until the point at which no more benefits are paid.
The budget line OACFGHJ in Figure 2 describes this choice set. As
compared to the case in which the only perceived effect is through
the tax (along OADE), the impact of the program is to induce those
who otherwise would have supplied labor along FC to reduce their
supply (assuming leisure is a normal good). This effect likely occurs
beneath the ceiling on OASI benefits (currently $5000 per year),
though this does not appear to have been analyzed empirically; and
the same effect is expected beneath the $280/month at which an
individual no longer is eligible for Disability Insurance.
In addition to the possible effect in shifting persons rightward
from F in Figure 2, transfer programs also shift them from points
to the left of F toward point F. These are the disincentive effects
that have received so much attention in the literature (see Feldstein,
H AM E R MES H /
213
1973, and Hamermesh, 1977, on regular UI; Munts, 1970, on
partial UI benefits; Quinn, 1977, and Baskin, 1977, on OASI; and
Parsons, 1980, Leonard, 1979, and Haveman and Burkhauser, 1980,
on DI.) In each program there is some, occasionally nearly infinite,
tax rate on additional earnings beyond point F such that labor
supply is reduced. It is this effect that has been viewed as the
culprit in reducing market employment and, in the case of leisure
that is measured as unemployment, in increasing the unemployment
rate.
Throughout the discussion we have glided over the effect of taxes
that finance the transfer payments. Since the concentration of this
paper (and most of the literature) has been on the effects of
transfers, that seemed appropriate. Nonetheless, some attention to
this difficult issue is in order at this point. The following
considerations seem relevant. 1) At least for transfer programs, the
issue of what the financing method does to labor supply is
unusually murky because of the extreme difficulty of extricating the
effects of taxes that are, for some programs, experience rated (see
Hamermesh, 1977, and Ehrenberg et al, 1978). 2) Assuming that
the financing is through a payroll tax, a very complicated
simultaneity problem seems to be operating. Without knowing the
incidence of the combined employer-employee tax that finances
OASI and DI, we cannot know the true shape of the budget
constraint facing the worker-consumer. But, without knowing the
shape of the constraint, we cannot deduce the labor supply
elasticity that partly determines the incidence of the tax. This means
that any consumer-theoretic analysis of the effect of a combined
tax-transfer program rests on shaky ground. 3) Despite these
problems, we do know that the payroll taxes are at least partly
borne by workers, so that it makes sense to represent the slope of
the budget lines OADE in Figures I and 2 as -w(l- st), where w is
the wage rate, t is the (total) tax rate, and s is the fraction of the
tax borne by workers. 4) Because of the ceiling on payroll taxes,
there is a convexity in the budget constraint facing the workerconsumer over some range. This will affect labor supply and thus
market output in that range. (Clearly, though, if one modelled the
entire structure of taxes on earnings, one would find that the
appropriate constraint is concave to the left of some point.)
The net effect of taxes and transfers on aggregate supply
combines all of these separate impacts implied by this general
model. Entitlement effects, induced unemployment, bunching at
notches in benefit structures, and behavior induced by taxes, either
214 / T RA NS FE R S , TA X ES A N D N A I R U
general income taxes or earmarked taxes that finance a particular
program, must be considered as we discuss how each specific
transfer program affects the labor market,
While our discussion abstracts from changes in the demographic
mix that have affected the NAIRU, we should recognize that there
are other changes in the composition of the labor force that are
induced by transfer schemes and that will have an impact on the
NAIRU. Within each demographic group, for example, those
persons with the lowest market productivity (relative to their
productivity at home) will be induced to leave by any given increase
in transfer payments. So long as relative market-household
productivity is positively (negatively) correlated with the individual's
probability of being employed when in the labor force, this will
induce a decrease (increase) in the measured unemployment rate
within the particular demographic group. Though this is a change
induced by transfers, it is also a measurement problem of a sort
similar in quality to that which we have circumvented by assuming
constant labor-force weights.
EFFECTS OF SPECfflC TRANSFER AND TAX PROGRAMS
That transfer payments have formed an increasing fraction of
disposable income was made clear in our discussion in the second
section, and it is underscored by the totals in the bottom two lines
of Table 3. The growth of transfer payments has been very uneven,
however; it is interesting to note that the phrase "welfare mess" is
hardly apropos, as "welfare" -usually thought of as AFDC-has
grown more slowly than disposable income. Disability Insurance
payments have been the most rapidly growing among programs that
were ongoing in 1966, and we have seen the birth and explosive
growth of payments under SSI and Food Stamps. The data clearly
suggest that transfers could, by virtue of their increased generosity
and coverage, have induced substantial changes in the labor market
since the mid-1960s. Whether this is in fact the case can be seen by
a program-by-program consideration of the transfers' effects.
Prompted by Feldstein's (1973) seminal work, there was a
resurgence of research on the effects of UI on the labor market.
Unfortunately the bulk of this work is on only one of the potential
impacts of UI, namely on the duration of spells of unemployment.
The twelve studies summarized in Hamermesh (1977, Chapter 3)
show a substantial consensus that higher UI benefits do induce
people to remain unemployed longer (as our discussion in the
previous section suggested). Further work (e.g., Kiefer and
H A M E R M E S H / 215
TABLE 3
Income Maintenance Programs
1966 and 1978
(billions of dollars)
Program
Old Age and Survivors'
Insurance
1966
1978
Growth Rate
(% per year)
$ 18.07P
$ 78.524a
12.2
Unemployment Insurance
(state and railroad)
1.891
9.233
13.2
Workers' Compensation
(state laws and federal
programs)
1.320
6.760
13.6
General Assistance (AFDC)
4.306
10.700
7.6
4.595a
Food Stamps (value of
federal contributions)
Disability Insurance
(under OASDHI)
1.72P
Supplemental Security
Income
All Transfer Programs
Disposable Income
12.2143
16.3
6.551
44.7
224.1
13.4
510.4
1458.4
8.7
• fiscal year bas is
Neumann, 1979, and Katz and Ochs, 1980) has done nothing to
dispel this consensus, and even my synthesis "best-guess" impact.5 extra weeks of unemployment for each . I increase in the net
replacement rate-seems supported by more recent studies. 8 There
should be no doubt whatsoever that Ul benefits in the U.S. do
induce longer spells of unemployment.
Feldstein (1976) and Baily (1977) have shown how the partly
experience-rated tax that finances UI can induce increases in
'The weak evidence available suggests that this effect is smaller in looser labor
markets (Hamermesh, 1977, Chapter 3).
216 / T R A N S F E R S , TA X E S A N D N A l R U
employment fluctuations and thus increases in the number of spells
of unemployment. This is postulated to occur because the marginal
tax cost to employers of another layoff is zero. Many employers'
UI taxes already exceed the benefits paid to prior employees
because of nonzero minima on state UI taxes, and some others'
taxes are limited by maxima on state tax rates. (Elsewhere,
Hamermesh, 1977, I have shown that roughly only 2/ 3 of Ul taxes
are experience rated.) Recently, there has been some effort to
quantify the impact of the tax structure on the labor market.
Brechling (1981) has carefully parameterized state UI tax laws and
shown that they appear to have a substantial effect in raising
manufacturing layoff rates across states and over time. Halpin
(I 979) has presented similar evidence for seasonal fluctuations in
employment in several industries. I find this evidence, and the
theoretical structure underlying it, to be nearly as convincing as that
on unemployment duration.
The provision of UI benefits represents a safety net under
workers' participation in the labor market. As such, it induces the
potential worker to choose to participate where she otherwise might
not. This entitlement effect (Hamermesh, I 979b) is especially likely
to be important among demographic groups whose attachment to
the labor market is fairly loose. It will affect the composition of the
labor force by increasing the weight accorded to such groups, and
will raise (lower) the aggregate unemployment rate if these groups'
unemployment is greater (less) than average. I have shown for adult
women that this effect does appear important in increasing
participation, and one might assume that it affects the behavior of
teenagers and older workers too. Since these groups generally have
higher-than-average unemployment, we may infer that it adds to the
positive effect of UI on aggregate unemployment. However, by
inducing persons marginally attached to the labor market to spend
more time in the work force, it also increases market employment
in these groups.
The net effects of an expanded UI program-higher benefit
amounts, longer potential duration and wider coverage-have been
clearly demonstrated empirically: Unemployment duration is raised;
employment variability is increased, and the composition of the
labor force is tilted toward groups having higher-than-average
unemployment. There is no question that UI raises the NAIRU, by
an amount that I elsewhere (Hamermesh, 1977) have
"guesstimated" to be .7 percentage points. Part of this effect has
been added since the mid-I 960s, due to expansion of coverage of
HAMERMESH / 217
this program and to recession-triggered extensions of the potential
duration of benefits. The program also induces declines in
employment (as unemployment duration is increased, and
additional layoffs occur when product demand decreases), but may
also increase market employment among secondary workers. The
net effect on aggregate employment, and thus per-capita GNP, is
an empirical question; however, as I have shown elsewhere
(Hamermesh, 1979b) that even among adult women the net effect is
negative, we may conclude it is negative in aggregate as well.
As Table 3 shows, retirement benefits under Social Security
represent the largest component of the transfer panoply. While our
discussion in the previous section hinted at the program's major
effects, there is one other effect that deserves mention first. Not
only does OASI raise the cost of working for those eligible; the
structure of benefits is also such that the cost is especially raised for
younger eligibles. This occurs because: 1) at age 72 the earnings
ceiling is removed, whereas it applies before then; 2) the increase in
monthly benefits if a man (woman) postpones filing beyond age 65
(age 62) is far less than would be actuarially fair; 9 and 3) the
ceiling on earnings is a more important constraint among younger
eligibles, because their market wage rates are greater. These last two
considerations coalesce to induce those eligible for benefits to file as
soon as eligibility for full benefits is achieved. The removal of the
ceiling at age 72 likely comes too late to have much impact on
persons who have been out of the labor force, and whose skills
have deteriorated.
Far more important than the induced switches among eligibles,
the system has provided increasing incentives for early retirement
through expanded support levels. (In terms of an ultra-rational lifecycle model, though, the opposite is true: The ratio of expected
benefits to OASI contributions has been falling since the 1940s. In
such a model the income effect works toward greater lifetime labor
supply. I doubt people are that rational, and the participation data
for older males in Table 1 suggest they are not.) As Munnell (1977)
showed, these rose sharply between the late 1960s and 1976, both
because of ad hoc statutory increases and the now-repealed double
indexing of benefits. Even though the 1977 Amendments will
prevent further increases in gross replacement, the projected rises in
'Each month beyond age 65 in which benefits are not claimed raises ihc monthly
benefit eventua!ly claimed by J/4 of one percent; each month before age 65 in which
benefits are claimed reduces ihc monthly benefit by 5/9 of one percenl. (Department
of Health, Education and Welfare, Social Security Handbook, 1978)
218 / T RA N S FE R S , TA X E S
A N D NA I R U
payroll tax rates, and a continuation of current trends in taxes on
earnings, indicate that net replacement may continue rising. This
suggests that the incentive that benefits give for early retirement will
continue to increase unless further amendments to the Social
Security Act are passed.
The magnitude of the increases in net replacement is large enough
to have had substantial impacts on the labor market. Quinn (1977)
and Boskin (1977) provide some weak evidence for the empirical
importance of these effects in cross-section data, and Pellechio
(1979) has provided a very convincing demonstration that it is
higher Social Security benefits particularly that are responsible for
the earlier findings. However, Blinder and Gordon's (1980)
estimates show only slight effects. One might infer that the data on
labor-force participation rates for older men in Table I reflect the
time-series analog of this cross-section evidence. This effect has
served to decrease employment; it says nothing per se about effects
on the NAIRU. Indeed, our arguments on composition in the
previous section; the observation that the unemployment rate
among older males decreased between 1957 and 1979; and the
evidence that early retirement is more likely among less educated,
lower skilled workers, precisely those for whom incidence of
unemployment is greater, all imply that the increased generosity of
OASI benefits may have reduced measured unemployment by
inducing nonparticipation by older workers with the poorest labormarket prospects.
We showed in the previous section that an entitlement effect can
also exist in OASI payments, as workers seek to establish greater
monthly retirement benefits later on through work before age 62.
This effect is compounded by the incentive the system provides to
shift hours of work away from periods of eligibility for OASI,
when the implicit marginal tax rate on effort is 50 percent.
Burkhauser and Turner (1978) use aggregate time series to "show"
that inclusion of Social Security wealth explains much of the
sudden halt in the decline in the workweek after World War IL
I am skeptical about attributing so much of this important
phenomenon to what appears to be so far-removed an incentive,
and I refuse to be convinced by time-series evidence alone. Some
cross-section evidence seems to be required. Even without this,
though, we should note that this effect implies an increase in labor
input and market output, and probably no effect on the NAIRU, as
hours are increased among prime-age workers whose participation
rates are already high.
HA MERMESH /
'.219
Because the shared payroll tax finances OASI benefits, one
cannot assess the program's effects without knowing the burden of
the tax. While some aggregate evidence implies the burden is
entirely on workers (Brittain, l 971), other macro evidence (Vroman,
1974} and micro studies (Hamermesh, 1979a) imply that it is shared
by workers and capitalists through higher product prices. It is likely
that the tax reduces effort. (J believe that substitution effects
outweigh income effects for some groups, and that they are roughly
equal for others.) However, though this does imply a reduction in
total labor inputs into production, it may also imply a reduced
NAIRU, since the greatest labor supply elasticities are among
groups with a high incidence of unemployment (compare Borjas
and Heckman, 1978, and Cain and Watts, 1973).
AH these considerations suggest that OASI retirement benefits
change labor-force participation in such a way as to reduce the
NAIRU: The composition of the labor force is induced to shift
toward groups with a low incidence of unemployment. With the
exception of the (to me) secondary effect on the distribution of
hours of work over the lifetime, the theoretical arguments and
empirical evidence suggest the major impact of OASI retirement
benefits is to decrease employment. Because of increased net
replacement and earlier eligibility, this effect has moreover likely
increased since the 1950s, and has increased since the late 1960s for
the first of these reasons.
Federal Disability Insurance has since 1960 provided benefits to
disabled workers of all ages. As Table 3 showed, the program has
received increasing attention from potential eligibles, drawn by
increased replacement rates and a not overly harsh interpretation of
eligibility rules. While there is a five-month waiting period during
which the person is not to be involved in substantial gainful
activity, an initial denial of benefits still leaves the applicant four
appeals levels; and the evidence (Haveman and Burkhauser, 1980)
suggests that claimants are increasingly aware of this and
increasingly successful in their appeals.
Like OASI under Social Security, Disability Insurance provides
incentives that affect the NAIRU and aggregate employment.
Workers with low market productivity, either because of severe
impairments or because of minor lmpairmems coupled with a lack
of marketable skills, have a substantial incentive to apply for and
continue to seek Dl benefits. (This is not, though, a decision to be
made lightly: Once eligibility is established, the individual cannot
earn more than $280 per month and then reapply successfully for
220 /
TRANSFERS, TAXES
AND NAIRU
benefits.) We should thus expect low-wage workers, minority
workers, older persons, etc., to be represented disproportionately
among DI recipients. Indeed, one might view DI partly as a
retirement program for those in their fifties.
These predicted effects are exactly what we observe: Leonard
(1979) shows that among males 45-54 nonwhites have twice the
representation among DI recipients as they do in the labor force.
He also shows that the probability of filing for DI is negatively
related to one's past wage rate. ' 0 Haveman and Burkhauser (1980)
show that the "overwhelming majority of DI benefits are initially
made [sic] to workers age 50-64."
The most clearly demonstrated impact of the program's increased
legal and administrative attractiveness to potential eligibles is on the
labor-force participation of older men. Among nonwhites, for
example, Siskind (197 5) has shown using time-series data that much
of the decline in participation can be attributed to the changes in
the DI program. In a more complex model Leonard (1979) confirms
Siskind's results. Parsons (1980) finds similar results for the
participation of males ages 48-62 using cross-section data for 1969.
He also finds that the effect of higher DI benefits in 1969 is greater
among persons who died within the next few years and who
presumably were in poor health when they filed for benefits. The
results suggest strongly that the growth of DI has induced a decline
in the NAIRU. AH the groups which the program data and
empirical work demonstrate are induced to leave the labor force are
composed disproportionately of persons with an above average
incidence of unemployment. This means that the composition of the
labor force is shifted by DI benefits away from persons with higher
unemployment rates, and thus that measured unemployment is
lower at a given level of labor market tightness.
The effects of DI on the labor-market issues of interest-the
NAIRlJ and the size of the work force-are the same as those of
OASI: Market employment is reduced, as is the NAIRU. This
rapidly growing program may well have contributed to reducing the
rate of GNP growth, but it has also disguised some of the
unemployment that would otherwise have been observed.
While the Food Stamp program is relatively new and has grown
rapidly, AFDC payments were established under the Social Security
Ac:t and have grown relatively slowly in the last decade.
'"Because of the problem of specifying full-capacity earnings to hold consrnnt rar
the effects of health on the probability of filing, Leonard's results should be viewed
as quite tentative.
H A M E R M E S H / 221
Analytically, though, they can be lumped together for our
purposes. The first consideration for each program is the work
registration requirement each entails: Recipients of benefits must
register with the state Employment Service and accept suitable work
if such is found for them. Clarkson and Meiners (1977) have argued
that this has induced a 2 percentage point increase in measured
unemployment. The calculation is based on the assumption that no
registrants would have been in the CPS labor force before the work
registration requirement was imposed, and that all report
themselves as unemployed in the CPS. Both assumptions seem
highly questionable, and Cagan (1977) and Devens (1978) have
argued that the Clarkson-Meiners number is greatly overstated.
Without econometric evidence based on observation of the effect of
Food Stamp or AFDC on labor force status, little credence appears
owed to this finding. One would need longitudinal data to test
the issue properly; though such are available, the test has not been
undertaken. Perhaps the best conclusion on the issue, based upon
consideration of the enforcement of the work-seeking requirements,
is that there may have been some one-shot effect on the NAIRU in
the early l 970s, but it was likely tiny.
If one believes the registration effect on the NAIRU was
important, one must also believe that the requirement has induced
an increase in employment and thus in aggregate supply: Some of
these induced to register presumably did find work when they
otherwise would not have. Since I do not believe the effect on the
NAIRU is large, I do not believe this positive effect on employment
is large either. Far more important is likely to be the effect of the
benefit structure under both programs. Saks (1975), for example,
has shown that the implicit tax rate on AFDC mothers in New York
in 1967 was .6, and that there was a substantial guarantee. (Casual
evidence suggests the implicit tax rate is somewhat lower today.)
Similarly, Food Stamps have increasingly substituted for the
negative income tax that was never enacted: There is no longer a
purchase requirement; a certain amount of Food Stamps is
guaranteed, and the allotment is reduced by less than 100 percent as
other income increases. This implies that both programs will induce
the usual negative effects on labor supply that we know are
associated with negative income taxes, assuming, as seems likely,
that recipients' supply elasticities are positive (see Saks, 1975, for
strong evidence on this).
How much have the induced changes in labor supply resulting
from AFDC and Food Stamps changed the NAIRU and aggregate
employment in the past 15 years? Since AFDC has not expanded
222 /
T R A N SF E R S , T A X E S A N D
N A I RU
relatively, it is hard to argue its effect has changed, so that one
must conclude it has not contributed to higher unemployment or a
changed employment rate. (Though, clearly, reducing the guarantee
or the tax rate would increase supply.) Food Stamps are new since
the mid-1960s, though; it is thus likely that they have affected
unemployment and employment. However, as with the other
programs that have reduced labor supply, one can reasonably argue
that the reduction has been disproportionately among persons with
the highest incidence of unemployment. Thus, if anything, the
benefit structure of Food Stamps has reduced the NAIRU slightly.
Without careful econometric evidence (and there is currently none),
this conclusion is based only on a little logic and on an analogy to
the demonstrated effect of other programs whose benefits can be
modelled similarly to those of Food Stamps.
There are numerous other transfer programs that one could
examine, and some, such as Workers' Compensation or
Supplemental Security Income, are fairly important. However, there
has been little or no work studying the effects of these other
programs on the NAIRU or on employment; since the discussion
above has given the flavor of the likely directions of the impacts of
most programs, there is little point repeating the analysis absent
specific empirical results. Suffice it to say that these other programs
most likely accentuate the effects we have already discussed.
I have avoided analyzing the effect of income taxes on the
NAIRU and on aggregate supply. While the latter issue has received
tremendous popular attention (and far too little scientific analysis),
the former has received none. There is no obvious direct effect of
the progressive income tax on the NAIRU, though there may be
some compositional effect of the sort we have stressed throughout
this section. Whatever the impact of the income tax on the labor
supply of high-wage earners, it is unlikely to have induced them to
withdraw from the labor force. A reduction in weekly hours seems
far more likely. Thus if anyone is induced to reduce market work
to zero, it is probably those whose market opportunities are least
attractive. To the extent that the income tax does affect supplyand, I stress, this has not been demonstrated directly-it has likely
done so among persons with the greatest probability of being
unemployed. Thus, if anything, the progressive income tax reduces
the NAIRU by changing the composition of the labor force.
The effect of the progressive income tax on hours of employment
cannot be answered here. (Hausman's paper covers this in more
detail.) Nonetheless, we should note that the induced reduction in
H A MER MESH /
223
output (assuming wage rates reflect marginal productivity) is
Z: t;rJ;W;N;, where t is the marginal tax rate on the i'th group of
i
potential workers; 11 is their labor supply elasticity; w is their
market wage, and N is the number of persons in the group. Across
different groups of workers both a higher marginal tax rate and a
higher supply elasticity will induce a greater reduction in effort (and
thus presumably in market output and real GNP). Among high,vage groups the marginal income tax rate on effort is fairly high;
however, all the available evidence suggests I'/ is quite low (Borjas
and Heckman, 1978). Thus it is unlikely that income taxes are
inducing much shortfall of output from this group and, conversely,
laughable to think that tax reductions will induce a sharp rise in
workhours and total earnings.
For low-wage groups the evjdence is much less clear. While it is
true that most studies find fairly high values of Y/ for these groups
(see Cain and Watts, 1973), some recent evidence suggests that, at
least for women with children, these findings are due to fixed costs
of entering the labor market {see Cogan, 1980). This suggests that
the effect of increases in the marginal tax rate on hours of effort
will be small. Also, the marginal tax rates on low-wage workers are
not very high.
Taken together, the evidence says that it is unlikely that the
progressive income tax has reduced employment much. Moreover, it
has, if anything, reduced the NAIRU. There may be difficulties
with the current income tax structure in this country; taxes may be
"too high''; but these statements should not be based on fears
about any huge detrimental effects on the labor market.
CONCLUSIONS
I would like to give one grand number indicating the effect of
income transfer programs on the NAIRU. I cannot. AH I can do is
note that UI does raise unemployment, but that the other, often
larger-scale programs have the opposite impact through their effects
on the composition of the labor force. Since I have not been able to
quantify these, I cannot weigh them against the effect of UI that I
have previously "guesstimated." Nonetheless, if forced to pick one
number to summarize the entire impact of transfers and taxes on
the NAIRU, zero would appear to be a good choice. At the very
least, it is a far better choice than that implied in the regressions in
the second section or in much of the popular discussion.
224 / T R A N S F E R S , T A X E S
A. N D
N A 1 R lJ
Zero would be a very bad estimate of the effect of taxes on
aggregate employment. Every program we have discussed likely
reduces labor supply on net, While we have not quantified this
reduction for all the programs and taxes discussed, the studies that
have done so for particular programs suggest the decline is
substantial. That transfers induce such a reduction should be
especially disturbing, as the tax structure in the U.S. economy
already contains a (probably increasing) bias against market work.
(Though, as we saw above, its effects may not be very large.) While
guessing the size of the induced drop in employment is not possible,
it is worth noting that, if even one-half of the decline in
participation of men 55 + has been caused by changes in OASI and
DI benefits and regulations, that alone would have induced a .8
percent reduction in aggregate employment since the mid-1950s.
The effect for the entire labor force is likely somewhat larger than
this. This guess, though, creates a conundrum: Why has aggregate
labor force participation risen by 3.6 percentage points since 1969,
at the same time we estimate that taxes and transfers have induced
a decline? Have nonmarket substitutes for women's time in the
home experienced such huge relative price reductions? Has the
structure of tastes changed (a thought that is repugnant to me as an
economist)? Perhaps the rea! issue we should be addressing is: Why
has the aggregate participation rate grown so much, departing from
its long-term near constancy just below 60 percent'?
While this is not a policy paper, a few conclusions for policy
seem clear, The evidence is abundant that we cannot ease program
eligibility and pay higher benefits without inducing changes in
behavior. This raises program costs, and thus the taxes that finance
the programs, and it targets benefits toward persons who were not
(at least apparently) meant to be targeted. At a time when the older
population is becoming healthier, DI has induced substantial
decreases in participation of men 55-62. OASI benefits have done
the same for persons 62 + and caught them in what Maggie Kuhn
of the Gray Panthers has called the "retirement trap": They are
induced to leave the labor force early, find they cannot maintain
their financial status during an unexpectedly long retirement, and
discover it is difficult to reenter the labor force at the same rate of
earnings." Clearly, unless we wish to see the growth rate of real
per-capita income decline further, steps such as raising the
, 'Case histmies and a discussion of this problem are presented in Waif Street
Journal, November 5, 1979, p. l er. seq.
H A M E R M E S H / 225
minimum age of eligibility back to 65 for men, and 62 for women,
seem perfectly reasonable and consistent with a healthier and
longer-lived population. Similarly, DI cannot be allowed to grow
further into a retirement program, as that will reduce the benefits
that the politics of the program will allow to be paid to the
seriously disabled who do need them. In short, we risk hurting
those persons for whom all these programs were designed by letting
them expand far beyond their original purposes with no thought to
the tax burdens they impose or their induced effects on production.
226 / T R A N S F E R S , T A X E S A N D N A I R U
REFERENCES
Baily, Martin. "On the Theory of Layoffs and Unemployment."
Econometrica, 45 (July 1977), 1043-1063.
Blinder, Alan and Roger Gordon. "Market Wages, Reservation
Wages and Retirement." NBER Working Paper No. 513, July
1980.
_____ and Donald Wise. "Reconsidering the
Work Disincentive Effects of Social Security." National Tax
Journal (December 1980).
Borjas, George and James Heckman. "Labor Supply Estimates for
Public Policy Evaluation." Industrial Relations Research
Association, Proceedings, 31 (1979), 320-331.
Baskin, Michael. "Social Security and Retirement Decisions."
Economic Inquiry, IS (Jan. 1977), 1-25.
Brechling, Frank. "Layoffs and Unemployment Insurance." In
Sherwin Rosen, ed., Studies in Labor Markets, Chicago:
University of Chicago Press, 1981.
Brittain, John. The Payroll Tax for Social Security. Washington:
The Brookings Institution, 1971.
Burkhauser, Richard and John Turner. "A Time-Series Analysis on
Social Security and Its Effect on the Market Work of Men at
Younger Ages." Journal of Political Economy, 86 (Aug. 1978),
701-716,
Burtless, Gary and Jerry Hausman. "The Effect of Taxation on
Labor Supply; Evaluating the Gany Negative Income Tax
Experiment." Journal of Political Economy, 86 (Dec. 1978),
1103-1130.
Cagan, Phillip. "The Reduction of Inflation and the Magnitude of
Unemployment." In W. Fellner, ed., Contemporary Economic
Problems, Washington: American Enterprise Institute, 1977.
Cain, Glen and Harold Watts. Income Maintenance and Labor
Supply. Chicago: Rand McNally, 1973.
Clarkson, Kenneth and Roger Meiners. "Government Statistics as
a Guide to Economic Policy: Food Stamps and the Spurious
Increase in Unemployment Rates." Policy Review, I (Summer
1977), 27-54.
HAMER MESH /
227
Cogan, John. "Labor Supply with Costs of Labor Market Entry."
In James Smith, ed., Female Labor Supply: Theory and
Estimation, Princeton: Prineeton University Press, 1980.
Devens, Richard. "Food Stamps and the Spurious Rise in the
Unemployment Rate Reexamined." Policy Review, 3 (Winter
1978), 77-83.
Ehrenberg, Ronald, Robert Hutchens and Robert Smith. ''The
Distribution of Unemployment Insurance Benefits and Costs."
U.S. Department of Labor ASPER Technkal Analysis Paper,
No. 58, 1978.
Feldstein, Martin. Lowering the Permanent Rare of Unemployment.
U.S. Congress, Joint Economic Committee, 1973.
- - - ~ ~ · "Temporary Layoffs in the Theory of Unemployment."
Journal of Political Economy, 84 (October 1976), 937-958.
Grant, James and Daniel Hamermesh. "Labor Market Competition
Among Youths, White Women and Others.'' Review of
Economics and Statistics, 63 {August 1981).
Grubel, Herbert and Dennis Maki. "The Effect of Unemployment
Benefits on U.S. Unemployment Rates.'' Weltwirtschaftliches
Archiv, 112 (1976).
Halpin, Terrence. "The Effect of Unemployment insurance on
Seasonal Fluctuations in Employment." Industrial and Labor
Relations Review, 32 (April 1979), 353-362.
Hamermesh, Daniel S. "Entitlement Effects, Unemployment
Insurance and Employment Decisions." Economic Inquiry, 17
(July 1979b) 317•332.
- - - - - · Jobless Pay and the Economy. Baltimore: Johns
Hopkins University Press, 1977.
______ "New Estimates of the Incidence of the Payrolt Tax.''
Southern Economic Journal, 45 (April 1979a), 1208-1219.
______ "Unemployment Insurance and Labor Supply.''
International Economic Review, (October 1980), 517-527.
Haveman, Robert and Richard Burkhauser. "Economic Issues
Regarding Public Policy Toward the Disabled." Unpublished
Paper, University of Wisconsin, 1980.
228 / TRANSFERS, TAXES AND NAJRU
Katz, Arnold and Jack Ochs. "Implications of Potential Duration
Policies in Unemployment Duration." Unpublished Paper,
University of Pittsburgh, 1980.
Kiefer, Nick and George Neumann. "An Empirical Job-Search
Model, with a Test of the Constant Reservation-Wage
Hypothesis." Journal of Political Economy, 87 (Feb. 1979),
89-108.
Leonard, Jonathan. "The Social Security Disability Program and
Labor Force Participation." NBER Working Paper No. 392,
August 1979.
Moffitt, Robert and Kenneth Kehrer. "The Effect of Tax and
Transfer Programs on Labor Supply.'' Research in Labor
Economics, 4 (1980).
Munnell, Alicia. The Future of Social Security. Washington: The
Brookings Institution, 1977.
Munts, Raymund. "Partial Benefit Schedules in Unemployment
Insurance: Their Effect on Work Incentive." Journal of Human
Resources, 5 (Spring 1970), 160-176.
Parsons, Donald. "The Decline in Male Labor Force Participation."
Journal of Political Economy, 88 (Feb. 1980), 117-134.
PeUechio, Anthony. "Social Security Financing and Retirement
Behavior." American Economic Review, 69 (May 1979), 284-287.
Perloff, Jeffrey and Michael Wachter, "A ProductionNonaccelerating Inflation Approach to Potential Output; Is
Measured Potential Output Too High?" In Karl Brunner and
Allan Meltzer, eds., Three Aspects of Policy and Policymaking,
Amsterdam: North Holland, 1979.
Quinn, Joseph. "Microeconomic Determinants of Early
Retirement." Journal of Human Resources, 12 (Summer 1977),
329•346.
Saks, Daniel. Public Assistance for Mothers in an Urban Labor
Market. Princeton: Industrial Relations Section, 1975.
Siskind, Frederic. "Labor Force Participation of Men 25-54, by
Race." Monthly Labor Review, 98 (July 1975), 40-42.
H AM E R M E S H /
229
Vroman, Wayne. "Employer Payroll Taxes and Money Wage
Behavior." Applied Economics, 6 (June 1974), 189-204.
Wachter, Michael. "The Demographic Impact on Unemployment."
In National Commission for Manpower Policy, Demographic
Trends and Full Employment, Special Report No. 12, 1976.
Discussion of the Hausman Paper
JEFFREY M. PERLOFF
Jerry Hausman's paper makes major contributions to both the
labor supply and taxation literatures. His paper provides the most
reliable labor supply estimates to date since he takes account of the
nonlinear budget constraint created by federal and state income
taxes. His work also helps rectify the misleading approach taken by
politicians, the popular press, and many economists which stresses
the revenue effects of tax cuts: the relevant question is the welfare
effect of tax cuts. Hausman is able (amazingly enough!) to
rigorously cakulate the deadweight loss imposed by a tax.
Of great policy importance is his conclusion that an across-theboard tax cut of the Kemp-Roth variety would lower welfare (and
tax revenues), while a reduction in the progressivity of the tax could
raise welfare. As Head argued in 1966, a progressive tax will have
greater disincentive effects than a proportional tax so long as the
economy is not in the prohibitive range where a reduction in the
proportional tax rate would raise revenues.'
If (as Hausman defines it) the progressive tax differs from the
proportional tax in that some level of income is exempted from the
tax, then revenues collected under the progressive tax system will be
less than under the proportional system for any marginal tax rate,
as shown in Figure l. Holding revenues fixed at R, so long as the
economy is not in the prohibitive range (as Hausman 's estimates
show), the marginal tax rate which corresponds to the proportional
tax, t*, will be less than that under the progressive tax, t**. As a
result, the proportional tax ·will have less of a disincentive effect, as
shown in Hausman's estimates.
While Hausman's research is destined to become one of the
classics of applied econometrics, I have a few minor quibbles. First,
Jeffrey M. Perloff is Assistam Professor of Economics, University of
Pennsylvania,
·Head, J. G., "A Note on Progression and Leisure: Comment," American
EconomicReview, V. 56, l966, pp. 172-179,
231
232 / H A U SM A N D I S C U S S l O N
FIGURE I
Revenue
R
revenues from a progressive
tax
..__ _.._._ _ _.___ _ _ _ _ _ _ _ _~ ' - - - ' - - - - - marginal iax rate
t•
t**
the effects of taxation on the amount of education people
undertake may be pronounced. This effect, however, is likely to
reinforce the distortions Hausman estimates. 2 Second, the
estimation process used assumes that the income effect is always
normal, which seems unreasonable in genera!.
Third, these estimates presume individuals know their marginal
tax rates. There is some justification for this approach, however,
according to Harvey Rosen and some of Hausman's other papers,
so this potential problem is probably not serious. 1 Fourth,
Hausman assumes that women are the secondary workers in a
family, while it would have been more reasonable to assume that
the lower wage family member was the secondary worker. Hausman
is currently working on a model where the family makes a joint
decision so that this problem will be eliminated in the future. In
any case, in his sample, few if any households had women earning
more than their spouses.
'Perhaps some handle on this effecr can be obtained by examining people who
made their educa1ion decision~ before WWII when income taxation was relatively
unimportant.
'Rosen, Harvey S., <;Taxes in a Labor Supply Model with Joint Wage-Hours
Determination," Econometrica, V. 44, N. 3, May, !976, pp. 485-508.
P E R L O F F / 233
One should show care in interpreting some of Hausman's results
(though he is fairly careful about pointing out these limitations).
Because utility levels are different across experiments, one cannot
compare dead weight losses directly. Moreover, his implicit social
welfare function, which is not very egalitarian, favors the policy
prescription which he favors. Finally, his experiments where he
compares progressive and proportional taxes are (necessarily)
relatively arbitrary. A more reasonable experiment might be to
reduce the number of kinks in the progressive tax constraint rather
than eliminating all but one kink. That is, an intermediate policy
might be even more favorable to Hausman's argument.
Hausman also argues that his results, while partial equilibrium in
nature, are likely to be dose to the general equilibrium effects.
Since this proposition was not immediately obvious to me, I tried a
few "back of the envelope" calculations to confirm this conjecture.
A tax on labor income will have complicated general equilibrium
effects. While the taxes are likely to influence capital, energy,
output prices, and wages, Hausman's partial equilibrium analysis
implicitly treats these variables as constants. The calculations
reported here are actually less partial equilibrium than Hausman's
rather than fully general equilibrium results, since capital and other
variables are still treated as constants: only wages are allowed to
adjust. In some sense, these results may be viewed as "short-run"
general equilibrium ones, where the labor market has time to
adjust, but the other markets have not yet adjusted.
Suppose, for simplicity, the labor supply equation is written as
(I)
i = ( (1 - ti)w/iti,
where i represents the hours worked by the ith group, t; is their
marginal tax rate, W; is their wage, di their after-tax wage elasticity,
and I is the nonearned income. The tax-supply elasticity is
(2)
ri· =
1
aiii
8t/t;
d1-t; ,.
= -=..!i_
The demand for each demographic group is derived from an
aggregate translog production function (assuming competition):
(3)
wl
= _Q_ (a· + L
•
l
1
j
0
y--lj In J
)
=Q
~
M 1'
l
where Q is aggregate output and M; is the factor share of the i1h
group (Mi = wii/total cost).
234 /
HAUSMAN
IJJSCUSSJON
Combining (2) and (3) and differentiating, we obtain
(4)
£ii
= ...ill.LL
ot;ft;
i
d;
l
( (Mi+Y;/M;)di - (d,+ 1))
I-ti
(5)
YulM,
+ M-
and, if t, = ti = t,
(6)
Ej
= aui =
_t_
at1t
1-t
cti
x
I - (yii / M; + Mi)/Mi + Yii /Mi - (di+ 1)/ di)
((Mi+ Yi/Mi)d; ~ (d; +I})
If the production function uses a single labor index, then only
equation (6) is relevant. Using an aggregate production function
with aggregate labor, capital, and energy factors, then in 1977
fourth quarter;•
ct,
e
I
0.1
0.9748
0.2
0.9508
0.3
0.9279
0.5
0.8854
1.0
0.7944
2.0
0.6589
'This production function, the estimated coefficient\, and a description or the data
is contained in Jeffrey M. Perloff and Michael L. Wachter, "A Produciion
Function-Nonaccdcraiing Inllation Approad1 to Potential Output: h Mea,ured
Potcmial Output Too High?" Carnegie-Rochester Conl"ercnce Series Vol. IO, 1979,
Journal of Afoneiary Economics, pp. 113 - 163.
PERLOFF/ 235
That is, the supply elasticity Y/i is only likely to deviate substantially
from the equilibrium elasticity, Ej, if di is relatively large. For
example, if
di = .I, then Ei = 0.9748r,;; while if d; = 1.0, Ei = 0.7944r,;.
There is substantial evidence, however, that it is inappropriate to
aggregate labor into a single index. Grant and Hamermesh, using
1969 cross-sectional manufacturing data in a translog production
function, have shown that it is reasonable to aggregate youths and
white females, but that it is not reasonable to aggregate all of
labor. Using time series data, Michael L. Wachter and I have
estimated a comparable production function for the private
economy using inputs of capital, energy, prime age males (M), and
all other labor (0).' Using our estimated coefficients, the following
adjustment factors can be calculated using equation (6):
do
d:vi
80
B:v1
0.1
0.1
0.1
1.0
l.O
0.1
0.5
1.0
0.1
0.5
l.O
LO
0.982
0.991
0.999
0.832
0.839
0.846
1.000
0.907
0.813
1.192
1.082
0.969
Thus, if dM is approximately 0. I and d 0 is approximately 1.0,
then EM ~ 1. 192r,M and r 0 ;:: 0.832r, 0 . That is, the equilibrium
elasticity for prime age males would be almost 20 percent higher
than the supply elasticity, while the supply elasticity would be
almost 20 percent higher than the equilibrium elasticity. Of course,
even if Hausman's estimates were off by as much as 20 percent, it
would make no difference to most of his conclusions.
Hausman's analysis is very useful in determining the costs of our
income tax system. This cost must be balanced against the benefits
of government goods and services and transfer programs. It should
be noted, however, that a substantial part of funds collected at
'A similar model i, described in "Productivity Slowdown: A Labor Problem?" in
or Boston Conference
Series No. 22. June, 1980, pp. 115-142. The only difference in that model is !ha! one
labor series consists of young people (under 25 years) and 1he other of older
workers. The coefficients are: M 0 °· ,23465, M~1 = .49218, Yoo = .13152, Y\1\l •=
.12096, Yo~1 = - .10972
The Decline in Produclivirr Growrh, Federal Reserve Bank
236 / HAUSMAN DISCUSSION
some levels of government go to collecting taxes. Small U. S.
counties (populations under 100,000) spent 7 .4% of their tax
revenues, on average, on financial administration; while the federal
government spent only about 0.7%. These figures are low, since
they include only central fiscal operations (which reached $1,798
million for the federal government in 1976). The U. S. government
spent 6.22% of tax revenues on general administration (which
includes the cost of tax collection and all administration costs not
directly attributable to specific programs). 6
'These statistics are discussed in Dick Netzer, "State-Local Finance and
Intergovernmental Fiscal Relations," in Economics of Public Finance, (Washington,
D. C.: The Brookings Institution, 1974) and Jeffrey M. Perloff "Economies of Scale
in Tax Collecting: Evidence for the U.S. and Abroad," Working Paper.
Discussion of the Hamermesh Paper
FREDRIC RAINES
Daniel Hamermesh has undertaken an extensive survey of what
we know about the impact of income maintenance programs on
employment, unemployment and labor force participation.
Reflected in this paper is an awesome amount of research, both
that of others and his own. And, on balance, he has done an
excellent job of synthesizing this literature. He is, certainly, the
resident expert in this area among us. If this conference is a supplyside harvest, we may note that Hamermesh has been busy tilling the
fields, and gathering the crops.
However, there are problems. The first problem Hamermesh has
is where to look for evidence of supply-side effects. He starts by
looking at macro time series data, regressing log ~ (where U*
100-U*
is the unemployment rate adjusted for shifts in demographic
composition) on lagged values of two policy variables:
(I) NRR-net replacement rate of aggregate transfers payments;
(2) TAX-the overall tax rate on earnings.
Unfortunately the results seem not to be to his liking, though
they would warm the heart of a Lafferite. A one standard deviation
increase in NRR from its mean raises U* from 5% to 7.85%, and a
similar increase in tax raises U* from 5 % to 6. I 9%.
Hamermesh then decides that truth may only be revealed by an
examination of the effect of individual programs. But not
everybody's examination. For instance, the 1973 study of benefits
by Feldstein, which finds that Unemployment Insurance benefits
and taxes have raised NAIRU by 1.25 percentage points, and a
1977 study by Clarkson and Meiners, which finds that AFDC (Aid
to Families with Dependent Children) plus Food Stamps have raised
the measured unemployment rate by 2 percentage points, are
rejected as patently too large.
Finally Hamermesh hits upon a solution. He takes the Perloff
and Wachter (1979) finding that NAIRU has increased since the
Fredric Raines is Associate Professor of Economics, Washington University in
St. Louis.
237
238 / HAMERMESH DISCUSSION
mid- l 950s by about 2 percentage points (of which slightly less than
1 percentage point is due to demographic shifts), and sees if, by an
examination of individual programs, he can work up to that modest
total. He also investigates what appear to be the more important
effects of income maintenance programs-those on employment.
A point about what it is we are trying to measure the effect on is
in order here. Hamermesh makes it quite clear that a given program
may have distinctly different effects on employment and
unemployment. But the unemployment concept that he chooses,
and the one commonly used in these studies-NAIRU-is, I would
argue, incorrect.
NAIRU refers to that rate of unemployment associated with
balance in the product market. But the relevant concept for labor
market studies is that unemployment rate which is consistent with a
balanced-the number of job vacancies equal to the number of
unemployed workers, say-labor market. Unless you are
sufficiently neo-classical to deny or ignore differing adjustment
speeds, these two concepts need not yield the same number. Indeed,
if I define the latter concept as a "full employment" benchmark
adjusted over time for demographic shifts-call it the natural rate
of unemployment (NRU)-then I can cite the above Perloff and
Wachter study as giving evidence that NRU and NAIRU have been
diverging over time. But the point is that NAIRU might be
consistent with a 5 percent unemployment rate at one point in time,
and an 8 percent rate at another, without there being any
implication or deducible inference for the impact of supply-side
programs on unemployment.
Putting this consideration aside, what does Hamermesh find?
Examining research on four different programs: Unemployment
Insurance (UI), Social Security, Disability Insurance (DI), and Aid
to Families with Dependent Children plus Food Stamps
(AFDC/FS), the consensus he finds is that the employment effects
(and labor force participation effects) are negative in each case.
However, the unemployment effects are mixed, implying reductions
in NAIRU for Social Security and DI, and increases in NAIRU for
UI and AFDC/FS. For the overall net effect on NAIRU of these
programs, Hamermesh likes the number "zero."
It should be pointed out that Hamermesh gets his reductions in
NAIRU entirely through changes in the composition of the labor
force. Those induced to leave the labor force due to the benefit
structure of Social Security and DI, for instance, are assumed to be
those with below average marketable skills and above average
RAINES/ 239
unemployment rates. This is a testable proposition, and while
Hamermesh does present some evidence, the full implications do
not appear to have been explored. One implication is that average
worker productivity should have been increasing as a result of these
programs. If so, it was much more than wiped out by other factors.
Another implication, which does seem borne out by overall
participation data, is that the composition of the labor force is
tilted toward younger workers.
One may ask, is the Hamermesh approach of counting the trees
to measure the forest a reasonable one? I strongly agree with him
that the foundations of imputing a supply-side effect must come
from observing micro behavior. There are just too many
complexities that get washed out in aggregate data-and our policy
proposals must deal with these complexities. At the same time, the
effect of these individual programs may not be additive as
Hamermesh is inclined to assume. For example, Hamermesh
concludes that the net effect of AFDC/FS on labor force
participation, employment, and NAIRU is slight. This conclusion is
based in part on the gradual reduction in the AFDC implicit tax
rate over time. But the AFDC implicit tax rate compounds with the
food stamp implicit tax rate, and these compound with the implicit
tax rate for Medicaid, Housing Assistance, Child Nutrition, and a
few other programs. This is known as the "stacking" problem, and
it implies overall effective tax rates in many cases in excess of 100
percent with numerous notches and kinks. If a 100 percent tax rate
doesn't have any effect on labor supply then Laffer is really
barking up the wrong tree.
I have a final comment to make on "where to look" for supplyside effects. I think that, methodologically, we may want to
examine the trees, but conceptually we should be thinking about the
forest. The subsidies and implicit taxes of welfare programs, the tax
system, and the pattern of government spending are imbedded in
our institutions and our culture. I don't know what it means to say
that if you abolish UI, the unemployment rate will decline by 3
percentage points. What is being held constant and what is
changed? There are important trends to be explained-slow
economic growth, virtually stagnant productivity, chronic inflation,
dramatic shifts in labor force composition-and the causes may be
bigger (or at least more subtle) than our independent variables.
But the tax/transfer system, in toto, does make a difference:
consider the following data on income distribution prepared by
Watts & Skidmore (1977):
240 /
HAMERMESH DISCUSSION
INCOME SHARES OF HOUSEHOLDS
lowest 40 percent
highest 40 percent
Before Taxes
and All Transfers
aAfter Taxes
and Transfers
7.5%
76.2%
17.8%
64.7%
aprograms include insured cash transfers, cash assistance, in-kind transfers, and
income and payroll taxes.
One way of looking at these numbers is to say that government
programs currently move one third of the way toward instituting a
completely egalitarian income distribution. I have no idea what a
redistribution of this order of magnitude-and the policies that
brought it about-entails for the economic behavior of individuals.
But I would venture the guess that those seriously concerned about
supply-side economics have bought themselves a rather large and
complex research agenda.
RAINES/ 241
REFERENCES
Perloff, Jeffrey and Michael Wachter. "A ProductionNonaccclerating Inflation Approach to Potential Output: Is
Measured Potential Too High." In Karl Brunner and Allan
Meltzer, eds., Three Aspects of Policymaking. Amsterdam:
North Holland, 1979.
Watts, Harold and Felicity Skidmore. "An Update of the Poverty
Picture Plus a New Look at Relative Tax Burdens." Focus,
Institute for Research on Poverty Newsletter, Fall 1977.
The Power of
Negative Thinking:
Government Regulation and
Economic Performance
MURRAY L WEIDENBAUM
Let me start off with a proposition duly overstated-which
should fit comfortably with the remarks of other contributors to
this conference on supply-side economics: it is futile to focus so
heavily on tax incentives to encourage economic activity at a time
when the governmental regulatory apparatus is imposing such a vast
and rapidly expanding array of obstacles to economic activity.
The lack of parallelism in my language is deliberate. It is not just
a matter of the disincentives of regulation offsetting some of the
incentives which can be provided by tax reform. Rather, it is a case
of insurmountable government-imposed barriers which any
increases in the normal, after-tax rate of return can do little
to hurdle.
For example, the most generous of tax credits will not help a
company to market a product that has been banned by the
government. The most liberal depreciation allowance will not assist
a firm in obtaining the numerous permits which are essential to the
operation of a new power plant. Indexing income tax rates will not
encourage the job applicant who is turned aside by companies
administering government-imposed quotas in their hiring. Nor will
massive reductions in personal income taxes help the teenager who
is priced out of the labor market by the latest increase in the
compulsory minimum wage.
Of course, this is not truly a matter of either/ or. We need not
and should not choose between tax reform and regulatory reform.
Rather, we should understand that the two go together. In practice,
supply-side tax cuts and reductions of regulatory burdens are
Murray L. Weidenbaum was Mallinckrodt Distinguished University Professor and
Director of the Center for the Study of American Business, Washington University
in St. Louis, when this speech was presented. He is currently Chairman of the U.S.
Council of Economic Advisers.
245
246 /
GOVERNMENT REGULATION/ECONOMIC PERFORMANCE
mutually reinforcing. Both can increase the capacity of the
economy to produce goods and services, the willingness of investors
to take risks, of management to innovate, and of workers to
produce.
To put it less dramatically, but more specifically than I did in my
opening statement, tax reform is a necessary but not sufficient
condition for substantially improving the performance of the
American economy. We must simultaneously deal with what I call
the power of negative thinking-the ability of, or at least the
tendency for, the regulatory apparatus (in truth I cannot call it a
system) to make economic activity difficult to perform. So many
government regulatory agencies have the power to say no to new
economic undertakings; few, if any, have definite authority to say
yes. To the typical entrepreneur, government is not a source of
help, but the possessor of the power to stop or at least to delay and
confuse. As a federal judge recently declared, "The federal
bureaucracy is legally permitted to execute the Congressional
mandate with a high degree of befuddlement as long as it acts no
more befuddled than the Congress must reasonably have
anticipated.''
It is fascinating to consider the attitudes of the proponents of
that increased regulation: they view the modern corporation
simultaneously as venal and omnipotent. That is, they implicitly
assume that society can impose an endless variety of so-called social
responsibilities on the business firm without affecting its basic
ability to carry on its economic function, that of meeting consumer
needs for goods and services.
To bolster my point, let me cite high authority, a recent issue of
the magazine Mother Jones. The editor was reporting on a
conference of business executives that he had recently attended. He
explained his surprise at the attitude that he had encountered. As he
put it, "We had come to view executives as the sort of men who
blithely market fire-trap cars, fill the Love Canal with lethal
chemicals, dump hazardous products on Third World countries and
conceal the dangers of asbestos from their workers ... To have
perpetrated so much, unscathed-surely they must be a strong,
confident breed, boldly planning new drives for profits."
That is not satire, but journalism, I keep reminding myself. But
the Mother Jones editor, to his surprise, found a different spirit
among the executives, who "saw themselves as innocent, aggrieved
producers-unfairly assaulted by environmentalists [and] regulatory
agencies ... " He went on to point out, "The corporate sector, we
WEIDENBAUM /
247
discovered, felt besieged. Barry Commoner, Ralph Nader, Leonid
Brezhnev, Teddy Kennedy and Jane Fonda were all out to get
them."
It is not my purpose today to evaluate the innocence or the guilt
of American business executives (whatever that would mean), but to
point out the economic consequences that result from the massive
range of government intervention in economic activity-which, in
turn, has resulted from the pressures of the self-styled public
interest groups. Subsequently, I will try to show how any effective,
supply-oriented approach to public policy can take account of this
phenomenon.
THE MANY COSTS OF GtWERNMENT REGULATION
Most public and professional attention to the costs of
government regulation has focused on the direct burdens of
complying with government directions. You may recall my estimate
that, at the federal level, these costs were in the neighborhood of
$ l 00 billion in 1979 and rising rapidly. Granted the imperfections of
my rudimentary techniques-I note that nobody else has attempted
to take on such a task -1 now acknowledge the important costs of
regulation that I neglected to take into account in my computations.
Let me enumerate some of these costs. It will become clear soon
enough why I did not include them in my numbers. I am referring to
the induced effects of regulation, the most diffuse and elusive aspect
of measuring the impacts of regulation. But for the policymaker,
what is truly important is not the precise dollar quantities but the
direction of the impacts. Clearly, most of these induced effects of
regulation impair the basic ability of the American economic system
to perform. Let me enumerate the key types of induced regulatory
costs.
1. The innovative product and process research and development
that is not undertaken because corporate research and development
budgets increasingly are being devoted to what is termed ''defensive
research. " Many companies report that they devote large and
growing shares of their scientific resources-from one fifth to one
half -to meeting regulatory requirements or avoiding running afoul
of regulatory restrictions. Surely, the longer it takes for a new
product to be approved by a government agency and the more
costly and uncertain the approval process, the more likely that
innovation will be delayed and the rate of innovation reduced.
Invariably, it is discouraging to the innovative instincts of
business firms to undergo experiences like the one recently had by
248 / GOVERNMENT REGULATION/ECONOMIC PERFORMANCE
Monsanto, the chemical company, with its recyclable plastic bottle
for soft drinks. The Food and Drug Administration banned this
new product because it was made with acrylonitrile. The regulators
say that if the bottles were filled with acetic acid (and not soda
pop) and stored for six months at 120°F., an infinitesimal amount
of the acrylonitrile could leak into the solution-and that would
constitute a carcinogenic (and hence unlawful) food additive. On
the basis of this less than brilliant experiment, Monsanto dosed
down aH the factories making the product and laid off several
thousand workers.
But these problems are not just a matter of large companies or of
one obstinate government agency. A small R&D oriented company,
Nutrilite Products, reported similar negative experiences. After
repeated efforts to obtain approval for a new "biological" form of
insect control (instead of the more environmentally hazardous but
traditional "chemical" approaches), the company concluded,
"We're going back to making vitamin supplements and trying to
stay as far away as possible from the Environmental Protection
Agency." In effect, government is building what Lee Loevinger,
former chairman of the Federal Trade Commission, calls " 'legal
envelope' around existing technology.''
2. The new investments in plant and equipment that are not made
because of regulatory barriers and the diversion of investment funds
to meeting government-mandated social requirements. The cost of
potential new investments is raised by the uncertainties generated in
the permit-approval process and by the cloudy future of new
rounds of regulation. Delays surely have become the order of the
day. In 1975, it took Deere and Company, the agricultural
equipment manufacturer, only three months to receive a complete
environmental permit review for constructing a new plant.
Currently, Deere estimates the lag at two years. Although the
company has received most of the permits it has requested, it
reports that EPA has insisted that these permits contain reopener
clauses in case the agency adopts more restrictive standards in the
future. In another instance, after noting that 42 different federal,
state, regional, county, and municipal agencies regulate his new
aquaculture company, George Lockwood, president of Monterey
Abalone Farms, stated in a paper to the AAAS that the major
problem ls not the direct costs of compliance but "the great
uncertainty" about whether any new activity wiH meet rapidly
changing regulatory standards.
Professor Ossar Lindbeck of the University of Stockholm has
commented on this phenomenon which apparently is not unique to
W E IDE NBA UM /
249
the United States. He points out that if laws and regulations change
"violently" all the time, the returns accruing from correct
speculation about the next moves of the regulatory authority often
become higher than the returns from careful investment in skills,
product development, choice of production technique, and
marketing. Professor Lindbeck contends, and I tend to share his
concern, that the sluggish behavior of investment activity in most
Western economies during recent years is derived not only from low
short-term profits, but also from increased uncertainty about future
government policies and the future rules of the game.
The problems facing firms which introduce new technology arc
especially great. Here is the assessment of a task force of the U.S.
Energy Resources Council on the overall impact of regulatory
activity on the establishment of a new energy industry: "In
summary, some of these [regulatory] requirements could easily hold
up or permanently postpone any attempt to build and operate a
synthetic fuels plant." The recent cancellation of the SOHIO
pipeline project provides striking evidence that the regulatory
uncertainties are not limited in their adverse impacts to new
technologies or even controversial ones.
Where government approvals are forthcoming, we find that a
rising share of company investment is being devoted to meeting
governmentally imposed social requirements. In recent years,
outlays mandated by EPA and OSHA have come to about 10
percent of new capital formation in American industry. In a
pioneering study, Edward Denison estimated that the diversion of
this amount of new capital resulted in business productivity in 1975
being I .4 percent lower than it otherwise would have been. One
percent may not seem like much but, in recent years, that would
have been the difference between a rise and a fall in the overall
productivity of the economy.
Moreover, we cannot always assume that the loss of private
productivity is offset by an improvement in some area of social
concern. For example, Armco Steel Corporation was required to
install special scrubbing equipment at one of its plants to reduce the
emission of visible iron oxide dust. The scrubber does succeed in
capturing 21.2 pounds per hour of the pollutant. However, it is run
by a 1,020-horsepower electric motor. In producing the power for
that motor, the electric utility's plant spews out 23 .0 pounds per
hour of sulfur and nitrogen oxides and other gaseous pollutants.
Thus, even though Armco is meeting government regulations on
visible emissions, the air is actually 1. 8 pounds per hour dirtier
because of the government's regulatory requirements.
250 / GOVERNMENT REGULATION/ECONOMIC PERFORMANCE
The Armco case is no isolated example. Scrubbers are
increasingly becoming required equipment for electric utilities that
are attempting to comply with EPA regulations. The federal
agencies, by being unable or unwilling to consider the adverse but
indirect effects of their actions, are likely to produce more instances
in which unintended but undesirable side effects swamp the
benefits. Consider the sad story of the Pennsylvania Power
Company. That utility has a new 825-megawatt complex that
utilizes scrubbers. In extracting the pollutants from coal, it
produces 18,000 tons of sludge a day. To dispose of the sludge, the
company has been forced to build a 350-foot-high dam, the largest
earth and rock enbankment east of the Mississippi River. Behind
the dam, there is now a lake of sludge, which already covers 900
acres in a picturesque valley of Western Pennsylvania!
Moreover, the regulations issued under the 1977 Clean Air Act
Amendments will slow down, if not halt, industrial expansion in
many parts of this nation. If and when the rulings are fully
enforced, failure of a state to win EPA approval of its detailed
clean air plan will result in an absolute prohibition of any new
industrial construction in that state.
3. The workers that are not hired because federal regulations have
priced them out of labor markets. A variety of serious academic
studies has shown that the steady increases in the statutory
minimum wage have reduced teenage employment significantly
below what it otherwise would have been-without a comparable
offsetting increase in adult employment. The DavisBacon Act yields similar results in government-financed construction
-lower employment and higher inflation rates.
4. The immeasurable effects of government regulation on the basic
entrepreneurial nature of the private enterprise system. To the
extent that management's attention is diverted from traditional
business concerns to meeting government requirements, a significant
bureaucratization of corporate activity results. Many chief
executives now report that one third or more of their time is
devoted to governmental and public policy matters.
Donald Rumsfeld, chief executive of a major drug company and
former Congressman and Secretary of Defense, has described very
personally the pervasiveness of government involvement in business:
When I get up in the morning as a businessman, l think a lot more about
government than I do about our competition, because government is that much
involved-whether it's HEW, IRS, SEC, FTC, or FDA. l always understood the
problem intellectually, but the specific inefficiencies that result from the
government injecting itself into practically every aspect of our business-that is
something one can feel only by being here.
W E 1 D E N B AU M /
251
This bureaucratization of entrepreneurial activity, albeit
undramatic, is not of modest dimensions. Professor Douglas North
of the University of Washington contends that the key marg1n of
decision making in our society today is access to government
influence. As he describes the matter, the predictable result is "to
shift the focus of the investment of resources into attempts to
favorably influence the strategic government official or to prevent
the enactment of government policies that will adversely affect the
interest of groups." The point may be overstated. There are still
many more opportunities for private undertakings. Moreover, the
adverse public reaction to massive use of business resources in
politics would, under present circumstances at least, be
overwhelming. Nevertheless, North is indicating an important
emerging development, especially in the case of the larger business
organizations.
Furthermore, Professor Lindbeck, from his different vantage
point, has made a similar observation, As he puts it, "there will be
great temptations," particularly for large firms, to bargain with
politicians over the rules and to seek various "deals" with
governmental authorities. Lindbeck notes the risk of businesses
entering into "zero-sum games" where they concentrate on
bargaining with governments rather than trying to increase output.
APPROACHES TO POLJCY CHANGES
It may, however, be easier to identify the regulatory problem
areas than to develop effective strategies for change. At the outset,
we must recognize the source of many of the pressures for
regulation-the self-appointed, self-styled public interest groups.
Large segments of the media, as well as many legislators, view these
groups automatically as both "representatives" and as underdogs.
This simpleminded attitude results in the characterization of people
who disagree with them as the "heavies." But just because I may
disagree with Ralph Nader or Jane Fonda should not inevitably be
taken as my representing some special interest opposed to the public
welfare. Why not think the unthinkable? It just may happen that,
on occasion, Ralph (or Jane} may be wrong.
Many-but not all-representatives of the public interest groups
confuse their personal prejudices with the national well-being.
Surely, I do not claim to represent the public interest. In all of my
years in government, I never met a mortal man or woman who
truly represented the public interest. As someone who was
intimately involved in government policymaking, I know that
252 / G O V E R N M E N T R E G U L AT l O N / E CO N O M l C P E R F O R M AN CE
making good policy is far more difficult than merely choosing, in a
simpleminded fashion, between ''public" or "consumer" interests
(which are presumably good and to be supported) and business
interests (which are presumably evil and to be opposed), Effective
policymaking consists not of dramatic confrontation, but of
carefully balancing and reconciling a variety of legitimate interests
-such as clean air and low inflation, safe products and high
employment, healthy working conditions and rising productivity.
In addition, the one thing this new breed of interest groups lack
is a sense of humor. For example, they attacked OSHA for
stopping the distribution of one of its pamphlets. OSHA had issued
a pamphlet on farm safety which treated farmers like dummies.
One of the newspapers in the nation's farm belt answered with the
following editorial in the form of a Dick and Jane book, the kind
you read in the first grade. Let me read it so you can decide for
yourselves.
DICK AND JANE V[SIT THE FARM
See the book.
See the little book.
See the little OSHA book.
What is OSHA?
OSHA is your government.
OSHA is the Occupational Safety and Health Administration,
OSHA helps people.
OSHA helps people to be safe.
OSHA made the little book for farmers.
What does the little book say?
This is what it says:
"Be careful around the farm . .. hazards are one of the main causes
of accidents. A hazard is anything that is dangerous.
"Be careful when you are handling animals. Tired or hungry or
frightened cattle can bolt and trample you. Be patient, talk softly
around the cows. Don't talk fast or be loud around them. ff they
are upset, don't go into the pen with them.
"Be careful thal you do not fall info the manure pits. Pu/ up
signs and fences to keep people away. These pits are ve1y
dangerous. ''
WETDENBAUM /
253
See the farmer.
See the farmer go to the mail box.
See the farmer get the little book.
The farmer can read.
The farmer can read big words.
The farmer can read long sentences.
The farmer knows about fences.
The farmer knows about manure pits.
Now the farmer knows about OSHA.
See the farmer kick the mail box.
Hear the farmer say bad words.
See the farmer throw the little book.
See the farmer throw the little book into the manure pit.
See OSHA.
See OSHA write.
See OSHA throw money into the manure pit.
Say bad words about OSHA.
Basically, we have to realize that the variety of regulatory activity
requires a variety of reform approaches. Eliminating regulation
makes good sense in those areas where the consumer is better
served by market competition. Energy is a prime example.
Eliminating the entire apparatus of energy price restrictions,
allocation controls, entitlements, and reporting requirements would
result in more domestic production, more conservation, and
reduced imports of foreign oil. Deregulation of airlines, trucking,
and railroads are also good examples of regulatory reform oriented
to supply-side concerns.
For the social regulations, there is no good alternative to revising
the basic statutes under which the regulations are promulgated. The
zero-risk approach of the Delaney Amendment to the Food, Drug,
and Cosmetic Act is a cogent example of unrealistic and
unreasonable social regulation which can be effectively curtailed
only by rewriting the law. Given the multiplication of regulatory
statutes, what would truly help is, yes, yet another statute, one
requiring compulsory benefit/ cost tests. Each agency should be
required to demonstrate in advance that its rulings will generate
more benefits to the nation than costs-hopefully, that the
marginal benefits equal the marginal costs and that it has chosen
the most cost-effective approach.
The promulgation of rules, of course, is not rhe only means of
accomplishing public objectives. As economists have been trying to
254 / GOVERNMENT REGULATlON/ECONOMlC PERFORMANCE
explain to government decision makers, pollution taxes could
constitute a far less costly method of achieving water quality
objectives. Interestingly enough, the business community, which
shows little enthusiasm for regulation, is adamantly opposed to this
use of the price system. Not that it is necessarily relevant, but I
note that environmental standards, unlike pollution taxes, tend to
be rougher on new industries than on established facilities. But as
we have learned over the years, the most adamant foe of
government intervention eventually learns how to convert a
government rule to a barrier to entry. As Lee Loevinger has noted,
"Thus small enterprises are slowly squeezed out and barriers to
entry are established by government fiat that would make an oldfashioned monopolist either envious or embarrassed."
In many other areas of government intervention, notably
consumer product safety, an information strategy is an alternative
to compulsory standards or product bans. Interestingly enough, this
approach often is favored in consumer surveys, although not by the
more vocal consumer organizations.
A word of caution: any realistic appraisal of government
regulation must acknowledge that important and positive benefits
have resulted from some of the regulatory activities-less pollution,
fewer product hazards, a reduction in job discrimination, and other
desirable goals of our society. But the "externalities" generated by
federal regulation do not justify government attempting to regulate
every facet of private behavior.
CONCLUSION
To sum up: the response of the economy to supply-oriented tax
policy will be greatly enhanced by reducing the numerous regulatory
obstacles to economic activity. Failure to eliminate or at least
substantially cut back the regulatory inhibitions to work, invest,
and produce will result in disappointing returns from tax policy
changes.
Government policymakers must come to realize the lack of
symmetry in the two different policy mechanisms: tax changes can
provide strong incentives to undertake private economic activity,
but regulation can provide a simple but effective veto. Too many of
the debates on supply-side economic policy have ignored or at least
deemphasized the crucial power of negative thinking on the part of
the regulatory apparatus.
The Politics of Supply-Side
Economics
ORRIN G. HATCH
We have, in the Congress, a thing called a "budget process."
You may not have noticed it, but it's there. It was the subject of
heated debate when it was established in the middle 1970s, when
many legislators who were worried about spending voted for it on
the grounds that it would force us all to think about the financial
consequences of our various programs, to reconcile them, and to
set priorities.
It hasn't done that. In fact, deficits have gotten worse since the
budget process began, and government spending is now approaching
proportions of the GNP previously reached only in wartime. What
the budget process did achieve was an infallible method of
providing rationales for increased spending, usually in terms of an
alleged rise in unemployment should government spending be
reduced in the economy, but sometimes by means of specialized
studies on technical issues. I might add here that the one area where
the budget process did act to inhibit spending was defense, where it
tended to challenge the specific requests made by the Pentagon and
its friends in Congress. By coincidence, this reflected the political
priorities of the party controlling Congress at the time and the
predilections of the staff members coming onto the Hill during the
Vietnam era.
All this happened in spite of the fact that the Congressional
Budget Act of 1974 set up a body called the Congressional Budget
Office, which was supposed to provide politicians in both Houses
with dispassionate, objective, and professional assessments of policy
proposals. As it turned out, it was the CBO that provided the
arguments for increased spending, and it backed them up with an
imposing array of evidence from a variety of econometric models,
much of it written in Greek and emanating from computerswhich, as you know, never lie. For that matter, since economics is a
Orrin G. Hatch is U.S. Senator (R.-Utah)
255
256 /SUPPLY-SIDE POLITICS
science, many legislators, although puzzled, concluded that
economists couldn't lie either, and that if they said deficits were
OK, they must be.
There are in reality value judgments at the heart of the Keynesian
orthodoxy, and particularly at the heart of the Keynesian
proponents. This is not just a matter of Alice Rivlin (the supposedly
impartial head of the CBO and one of what Newsweek magazine
called the "half dozen leading liberal economists") dining with
Senator Kennedy to prepare him for his challenge to Mr. Carter last
year. (Another CBO projection bites the dust!) It isn't even just a
matter of the faulty underlying assumptions contained in the CBO
projections, although these are often rather odd. The CBO, as you
all know from reading the literature, has for years systematically
favored spending increases over tax reductions as a means of
stimulating the economy, and, at one time, it was even using a
model which assumed that a decrease in corporate taxation would
reduce GNP.
Where the element of faith in the Keynesian orthodoxy really
comes into its own is in the CBO's steady resistance to any sort of
analytical or empirical debate about its assumptions. We had a
particularly graphic example of this in the spring of 1980. There is
abroad in the Western world at the moment, a particularly lethal
weapon that has totally altered the balance of power between
employers and the employed. This weapon is called the Xerox
machine, and some anonymous dissident on the Budget Committee
staff used it to send us a copy of a memo (written to Ed Muskie,
then Chairman of the Budget Committee, from his staff director)
discussing detailed collaboration between the CBO head and the
Democrats on the Committee to suppress Republican efforts for a
hearing on the econometric models CBO uses. These models, of
course, are under severe attack for ignoring the incentive effect,
and we were hoping to get CBO to consider some of the supply-side
thinking now going on, of which this conference is a symptom.
The memo told Muskie: "Alice [Rivlin] doesn't really want to
have hearings and would like to put Hatch off somehow. She says
-and Susan Lepper (the Majority Economist) supports her in this
- that the critics of the models CBO uses for forecasting are an
extreme right wing claque who should not be given an audience, lest
it legitimize their views and give Hatch a forum which should be
denied him if we could. If we are to hold hearings, Alice believes
they should involve noted economists telling the Committee that
Hatch's witnesses are wrong .... "
HATCH / 257
Later on in the memo, the staff director told Muskie: "I am
tempted to have him [me] off on this tangent, which few people
know or care about outside the economics profession, rather than
leave him with time to become involved with something that might
be more serious .... "
None of this looks particularly objective or dispassionate, or for
that matter even scientific, to me. Of course, I'm just a lawyer. I
think the sad thing about all of this is that the people involved,
whether political appointees like the Democratic staffers or civil
servants like the CBO functionaries, ar; not in themselves dishonest
or conniving people. The nature of the system causes them to act in
this way because their own short-term interests are so very clearly
involved.
Although bureaucrats and politicians-at least certain politicians
-do benefit from continual deficits and pervasive inflation, the
system is unstable. Inflation is only a temporary answer to the
problem of separating the taxpayers of this world from their
earnings. For one thing, the dislocation it causes annoys and
distresses them. For another, the combination of inflated incomes
carrying more individuals into higher tax brackets, and government
expenditures which are steadily mounting, means that the
underlying resistance to taxes is steadily increasing. More and more
people are being pushed into the fiscal free-fire zone. They are
reacting by digging fox-holes, constructing tax shelters, and
generally refusing to obey orders.
This is a particularly acute problem for the economists of our
"ruling class" -because that's what the Keynesians, in effect, are.
Their system is entirely set up to suppress insurrections from people
who believe in balanced budgets-and there are still a lot of them
about, incidentally. All they have to do is show that balancing the
budget will cause economic disruption, besides requiring either tax
increases or spending reductions. But they don't have any way of
dealing with the negative incentives of their system, except more
government intervention to divide up the pie or to treat the
symptoms of rising prices and wages. This is why we hear so much
now about "zero-sum societies," lowered expectations, spiritual
malaises, and so on. Tacitus said the Romans made a desert in
Germany, and called it peace. We can say of the economic
establishment that it has made a stagnant pond, and called it the
Great Society. Still, it has been good for real estate prices in
Georgetown. And it is causing us to rejoin the human race-with a
command economy, with welfare for people and corporations.
258 / SUPPLY~ SIDE POLITICS
In other words, supply-side economics has arrived in exactly the
situation the late Harry Johnson diagnosed as existing at the time
of the advent of Keynes, at the onset of the Depression:
... On the one hand, the existence of an important social and political
problem with which the prevailing orthodoxy was unable to cope; on the
other hand, [a new theory with] a variety of characteristics that appealed to
the younger generation of that period-notably the claim of the new theory
to superior social relevance and intellectual distinction, its incorporation in a
novel and confusing fashion of the valid elements of traditional theory, the
opportunity it offered to bypass the system of academic seniority by
challenging senior colleagues ... [and] the advancement of a new empirical
relationship callenging for econometricians to estimate.
This may sound cynical, but it isn't really. As we have seen,
economic policy is an area where even the most qualified
professionals seem to have trouble keeping their minds open to new
and inconvenient ideas. In that respect, it's unlike academic lifeI hope. If any theory is to flourish in this environment, it must be
protected by its political mentors. Keynes, incidentally, was fully
aware of this and used every trick he could think of to advance his
views. He had an extremely active mind, so he thought of a lot of
tricks.
The best way of thinking about economic policy is by comparing
it to a dog fight between World War II fighters. You have to aim at
some point other than at the target itself in order to hit it, given
your relative motion and so on. This is something that Keynes
understood. He told Friedrich von Hayek that he realized his policy
prescription would be inherently inflationary, but that when the
moment came he would step in and turn public opinion around in
six weeks. When Hayek tells this story, he always adds, with an
ironic grin, that six weeks later Keynes was dead. But the point is
that Keynes wanted to solve certain problems and he wanted to
change policymakers' thinking about them, and the importance they
put on them. In a sense, you could argue that there's an element of
myth about all economic policy proposals-as defined by the
French historian Sorel, who said many years ago that myths in
human society were not factual statements, but were instead
expressions of intentions to act.
Keynes was successful in getting all of us-not merely liberalsto accept his values. And I believe that those who have developed
the supply-side theories will be successful in shifting our attention
once again to incentives and production and the economic
applications of liberty. As I say, this isn't merely an academic
achievement. It is a political achievement of no small merit. What
HATCH / 259
the supply-siders have done is to point out that the war between the
proponents of incentives and the federal government's spending
constituencies is not necessary. It is possible to attack at another
point: to get tax rates down and stimulate growth sufficiently to
pay for the current rate of social services, hence bypassing the
question of whether social spending is too high.
Now, will these services be paid for out of tax revenues that have
increased absolutely, while decreasing in terms of rates levied on
individuals? Or will they be financed out of additional savings
generated by increased production? Or will we in fact find further
deficits, albeit in the context of a policy that promises to get the
country moving again rather than sinking under taxes and
regulation? There are various answers to these questions, but in a
broader sense, these questions are upstaged by the new awareness in
the public debate of incentives-that there is supply as well as
demand.
An example of this new awareness came in Mr. Carter's recently
proposed tax package, which seems as if it were designed to catch
attention as an alternative to Mr. Reagan's tax proposal. No one
can deny the White House's exquisite sensitivity to currents abroad
in the land-to style, if not to substance. When you look at
President Carter's proposals in detail, you can see the extraordinary
gains the supply~side offensive has made in the last two years-and
also the stubborn and ferocious determination of the economic
establishment to maintain and expand its power and that of the
government, come what may. A recent H. C. Wainwright study by
Paul Craig Roberts shows how President Carter's tax cut is really
aimed at objectives other than tax reduction.
In the matter of a few short months over the summer, President
Carter went from teBing the American people that the $36 billion
tax cut proposed by Governor Reagan would cause "fierce
inflation" to proposing a $27 .6 billion tax cut of his own, which he
said would be "anti-inflationary." Following on the heels of the
Senate Finance Committee's proposal for a $39 billion tax cut, it
put to rest the argument that the Reagan-Kemp-Roth tax cut was
bad politics. So we can now move to the merits of the proposals
and determine which would provide the most incentives to increase
production.
By comparison, the Kemp-Roth tax cut bill proposed by
Governor Reagan is clearly a supply-side proposal, since it
concentrates solely on reducing marginal tax rates, Measured by
static revenue losses, it is more heavily weighted toward
260 /SUPPLY-SIDE POLITICS
"individual" rather than "business" tax reductions. The Senate
Finance Committee bill, although it wastes about $7 billion on
enlarging the zero bracket amount, personal exemption, and earned
income tax credit, is largely an application of incentive-oriented
supply-side economics. It gives 56 percent of its cut to individuals
and 44 percent to business. President Carter's proposal is more
heavily weighted toward business, giving it 55 percent of the cut.
But, although the Carter proposal is cloaked in supply-side rhetoric,
a closer look shows that it is designed to achieve ends quite
different from lowering marginal tax rates or increasing production
incentives.
One example is the refundable investment tax credit. The purpose
of the investment tax credit is to boost the incentive for investment
in new equipment; there is no economic sense to excluding firms
with no tax liability. It is often new and rapidly growing firms that
have no tax liability against which to apply a non-refundable credit.
But the main problem with the refundable investment tax credit is
the precedent it establishes. How could we hope to avoid making,
say, the child care tax credit refundable for poor people if big
business has it? The child care tax credit is expensive-up to $800
per eligible return-and making it refundable would be a big step
toward expansion of the federal welfare system.
The refundable investment tax credit would also expand the
federal welfare concept to business. It would establish the concept
of extending the dole to businesses that lose money. It would result
in an institutionalized bail-out scheme instead of making the
Congress consider it on a case-by-case basis. This is hardly the way
to "make careful investments in American productivity" -Carter's
way of differentiating his tax cut from Reagan's.
Another part of the President's proposal that will contribute to
the growth of government intervention in the economy is the
additional 10 percent refundable investment tax credit targeted to
revitalize depressed areas. Firms that want to qualify must obtain
certificates of necessity from the Commerce Department, but the
criteria for determining eligible areas are not defined. This would
give the government the ability to reward its friends and withhold
the credit from the uncooperative. Even if the system could be kept
free of political corruption, government allocation of resources will
certainly reduce efficiency in the economy.
We should also note that President Carter is also suggesting that
the Treasury Secretary be given the power to adjust depreciation
rates at will. This is another expansion of the government's
HATCH /
261
discretionary power. And it's likely that the accumulated effect of
his proposed substitution of open-end for vintage accounting will
tend to reduce the present value of the depreciation allowance for
technical reasons. So the pro-business aspects of Mr. Carter's plan
can be-and have been-exaggerated.
On the individual side of the Carter tax package, an income tax
credit is used to partially offset the scheduled increase in the social
security tax out of general revenue funds. Instead of reducing
marginal tax rates, it is a scheme to redistribute income and turn
social security into a welfare program by taking the first step into
general revenue financing. If the President were really interested in
avoiding the economic damage that will result from the social
security tax increases, he could just postpone or repeal the
scheduled increase. The only reason for the income tax credit
approach is to attack the contributory nature of social security and
plunge into general revenue financing. This type of tax cut is likely
to guarantee continuing revenue losses and deficits. Although it has
the smallest static revenue loss, it would probably be the most
expensive, net of feedback, because of the negative supply-side
effects.
On the whole, the Carter tax cut encompasses the welfare rather
than the incentive approach to tax policy. Most of its provisions
increase the discretionary power of the government to control the
economy. It would divert resources from economic to political uses,
and would lead to deficits and revenue losses that would prevent us
from getting the incentive tax cuts the economy needs to grow.
Furthermore, the Democratic Platform contains 70 separate items
that will result in federal government spending. Over the next five
years, the platform would cost $608 billion in budget authority and
$431 billion in outlays. In comparison to the Senate Budget
Committee's second budget resolution for FY 1981, the Democratic
Platform would add $74 billion in budget authority and $30 billion
in outlays in FY I 981, and $566 billion in budget authority and
$389 billion in outlays over the FY 1981 to 1985 period. If enacted
into law, the Democratic Platform would cause federal outlays to
increase to 24. 7 percent of GNP in 1982, and this includes no
additional outlays for interest on the public debt due to the higher
deficits. Coupled with President Carter's tax cut, it would create a
$261 billion deficit over the next five years as opposed to the $75
billion surplus Governor Reagan's plan would create.
I want to conclude tonight by commenting on the checkered
fortunes of the tax revolt since it first materialized in California in
262 / S U P P L Y - S l D E P O L I T I C S
1978. Since then, it has been periodically proclaimed to have run
out of steam. Certainly the lobbying groups arrayed on the side of
increased spending still seem to be alive and dangerous; victory has
been by no means as automatic as it first appeared it might be. But
it might be remembered that we are fighting a momentum that has
built up over a period of decades. The proponents of income
redistribution, deficits and government intervention took years to
perfect their appeal to the broad electorate, and to overcome the
doubts, scruples and skepticism of the American people about
charity, the expropriation of property, and the surrender of
independence that the welfare state entails. It will take us years,
too-although the success we have had in forcing our opponents to
steal our rhetoric is evidence of some sort of progress. And in the
end, our task will be easier. It is the processes of liberty that we are
fighting for, and they are intrinsic to the American tradition. After
all, it was a dispute over taxation that triggered the American
Revolution. It is not surprising-it is, indeed, highly appropriatethat we should have gathered here to think about tax policy in the
consciousness that what we have been doing in reality is to
contemplate at least the success and perhaps, ultimately, the
survival of liberty itself.
@FedFRASER