..\.~.1p.:
....rqo.«ouaa.e
1;.

v

to.
I. 2:
tr: u!

t 1 rue-(-
:ustfviillloilivzlrrwrvn
Stifrf-Ivvlvfllip .?
1.}. irrtvualrarvl

av.rl. rl.lr

o..A$..:€v

. . '
..:.....t.:.
(..

.la::Vr. 1..
roll Itlvevo a:
turf!
a 5r.

4
C 2
1:4... a...

v y

vv.!-. 1

{rioov
.(lra.t t.

r Ilualll
.u .c.-lvl,l..01lloll
a... 1. r. ..xao(1:le
111.. l r.r.r

t; . .t.?\..;

 

 

 

 

 

V..:
1.9:...

I. .
n:.:..vr:ror:

 

o......:c4
4:... :1,

o.
a." if! n’llFIIa (’7. it
(I Irizuufflla I

(5...!-
!!(uoullrip .5. .u.
0!!

1- .V
Clo r!!!
1.

trrrvftv
lu0¢.vll.ldtfl'!v:ll<. (tr
..

a

It. £11171);
1 21.6

1...

I

...>x
(I 1.!Il‘é

9.1;». Il.llluvl a1:!..’l tvlf.
Elf:o..!¢av..v.!.avil;1€:.5!

a) 1!!!! 1! Ir}...
1:: I'll.1:41:4(5)..srl2rlvl.vlvrnl
I'Itllltlfirr.

 

 

 

ruFfils

Illl’ll’l‘lllllllllllll‘lllllllllllfill

3 1293 00791 7580

This is to certify that the
dissertation entitled
AN ASSESSMENT OF TAX POLICIES IN MODELS WITH ENDOGENOUS
FINANCIAL BEHAVIOR
presented by

John Charles Navin

has been accepted towards fulfillment
of the requirements for

Ph . D . degree in Economics

 

 

Major professor

Date [2 afibe” /9 9 2

M5 U is an Affirmative Action/Equal Opportunity Institution 0-12771

 

LIBRARY
Mlchlgan State
I University

1—.

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

I DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU is An Affirmative Action/Equal Opportunity Institution
cmmux

AN ASSESSMENT OF TAX POLICIES IN MODELS WITH ENDOGENOUS
FINANCIAL BEHAVIOR

BY

John Charles Navin

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Economics

1992

ABSTRACT

AN ASSESSMENT OF TAX POLICIES IN MODELS WITH ENDOGENOUS
FINANCIAL BEHAVIOR

BY

John Charles Navin

The purpose of my dissertation is to study the effects of
corporate tax policies on the financial and investment
decisions of firms in the U.S. economy. In the first section,
I construct a model of firm investment and financial behavior
which can be used to study the effects of various tanpolicies
in a computational general equilibrium (CGE) model.

In the second section, I use a panel data set of 781
firms to examine the impact of changes in statutory tax rates
on firms' debt/capital ratios. Data on firms' debt/capital
ratios were taken from the COMPUSTAT data tape, covering the
years 1969-1988. A fixed-effects model was used to measure
the impact of tax rate changes on firm's debt/capital ratios.
The results of the empirical estimation indicate that the
response of firms to changes in the tax rates was very small.
The elasticity estimates, after correcting for
autocorrelation, ranged. between -0.11 and -0.73 for' the
debt/capital ratios with respect to changes in the personal
income tax rate.

In the final section, I use a nine-sector CGE model to
examine the effects of three different tax policy proposals:

full corporate tax integration, a reduction in the corporate

income tax rate, and a dividend exclusion from the personal
income tax. The model consists of nine production sectors, a
single consumer, and also a government sector. The model
incorporates the empirical elasticities of firm financial
behavior, in.order to allow firms to adjust.their debt/capital
ratios in response to changes in tax policy.

The tax policy simulations resulted in welfare gains
under each policy proposal. The welfare gains ranged between
$463 billion, under full corporate tax integration, and $340

billion, for the reduction in the corporate income tax rate.

This dissertation is dedicated to my brother, Jim.

iv

ACKNOWLEDGEMENTS

I would like to thank all of the members of my committee,
John Goddeeris, Larry Martin, Ron Fisher, and Charles Ballard,
for all of their time and patience in seeing this work through
to its completion. I would especially like to thank my
advisor, Charley Ballard, for all of the long hours spent
reading earlier drafts of my dissertation and for his
unwavering support.

My wife Lynn, as well as my parents provided phenomenal
emotional as well as financial support. Without them I could
have never gotten this far. I would especially like to say
thank you to my dad, without his constant pep talks I would
have given up long ago.

Finally, I would like to thank all of my friends at
Michigan State for all of their help, especially Kari

Foreback, Marneta Griffin, Gil Davis, and Kel Utendorf.

TABLE OF CONTENTS

1 INTRODUCTION . . . . . . . . . . . .
Introduction . . . . . . . . . . . . .
Overview of Some Corporate Income

Tax Distortions . . . . . . . . . . .
Review of the Literature . . . . . . .
Corporate Tax Integraton and

CGE Models . . . . . . . . . . . . .
Theories of Corporate Financial

Leverage . . . . . . . . . . . . . .

2 MODEL OF FIRM BEHAVIOR . . . . . . .

3 THE EFFECT OF PERSONAL AND CORPORATE
TAXES ON FIRMS' DEBT/CAPITAL

UUUMQ

UNNNN
0..
UMP

3.

NH

hhbhéhfihfi
\lO‘O‘O‘UIéwNH

RATIOS . . . . . . . . . . .
Empirical Implications of the

Bankruptcy Models .
Estimation . . . .
Model . . . . . .
Data . . . . . . .
Results . . . . .
Conclusions . . .

4 DESCRIPTION OF THE SIMULATION
MODEL . . . . . . . . . . . .
Introduction . . . . . . . . .
The Consumer's Problem . . . .
Producer's Problem . . . . . .
The Government Sector . . . .
General Solution to the Model

Calibration . . . . .
Production Sector . .
Consumption Parameters
Data . . . . . . . . .

5 SIMULATIONS . . . . . . . .
Introduction . . . . . . . . .
Full Integration Results . . .

Reduction of the Corporate Tax

Dividend Deduction . . . . . .
Sensitivity Analysis . . . . .

vi

39

39
4O
40
41
47
54

56
56
56
60
73
74
76
77
79
82

84
84
87
92
93
95

TABLE OF CONTENTS (cont'd)

5.5.1 Adjustment Costs . . . . . . . . . . . . . . . 96
5.5.2 Depreciation . . . . . . . . . . . . . . . . . 97
5.5.3 Debt/Capital Elasticities . . . . . . . . . . 98
5.5.4 Classification of Corporate and

Noncorporate Sectors . . . . . . . . . . . . 102
5.5.5 Required Consumption Ratio . . . . . . . . . . 104
5.6 Comparison with Other Studies . . . . . . . . 107
5.7 Conclusions . . . . . . . . . . . . . . . . . 109

REFERENCES 0 O O O O O O O O O O O O O O O O O O O O O 1 12

vii

LIST OF TABLES

TABLE 1-

Summary of Features of Computational General

Equilibrium Models . . . . . . . . . .

TABLE 2-
Listing of Corporate Sectors Used

in Estimation O O I O O O O O O O O 0
TABLE 3-

Summary of Debt/Capital Ratios
by seCtor 1969-1988 e e o e o o o o o 0

TABLE 4-
Initial Regression Results . . . . . . .

TABLE 5-
Average Autocorrelation Coefficients
by Sector . . . . . . . . . . . . . . .

TABLE 6-
Regression Results After Correcting

for Autocorrelation . . . . . . . . . .
TABLE 7-

Production Sectors Used in the
Simulation MOdel O I O O O O O O O 0

TABLE 8-

Breakdown of Capital Stock Classification
by sector 0 O O O I O O O O O O O O 0

TABLE 9-

Assets and Their Tax Parameters . . . .

TABLE 10-

Base-Case Data . . . . . . . . . . . . .

TABLE 11-

Summary of Initial Simulation Results . .

viii

15

44

45

48

51

52

63

65

70

83

95

Chapter 1

INTRODUCTION

1.1. Introduction

The corporate income tax may have long-term effects on
savings, investment, and firm financial decisions. The
purpose of my dissertation is to study the effects of
corporate tax policies on the financial and investment
decisions of firms in the U.S. economy. I begin with a brief
survey of earlier work. I then construct a model which can be
used to study the effects of various tax policies in a
computational general equilibrium (CGE) model. The model
yields implications for corporate debt/capital ratios that may
be tested econometrically. In Chapter 3, I report on my
investigation into the effects of changes in the personal and
corporate tax rates on firms' financial decisions, using a
panel data set of annual data from 1969 through 1988. Then in
Chapter 4, I describe the CGE model and in Chapter 5, report

on the results of the policy simulations.

1.2. Overview of Some Corporate Income Tax Distortions
The corporate income ‘tax: is an important (although
relatively’ declining) source of :revenue for the federal

government, ranking third behind the payroll tax and the

2

1 It has been criticized on the grounds

personal income tax.
that it distorts individual and firm behavior.2 The
following are some decisions which are believed to be
distorted by the corporate income tax:

(1) The firm's decision on whether to finance new
capital with new issues of equity or new issues of debt.
Under the current tax code, interest payments on debt are
deductible from the corporate tax base, while disbursements to
equity holders are not. This asymmetry in the tax code
heavily favors debt finance. See, for example, Stiglitz
(1973) or Gordon and Malkiel (1981).

(2) The firm's decision on whether to undertake new
investment. A key factor in determining whether to undertake
new investment is the after-tax rate of return received by the
firm. As the corporate income tax rate changes, the rate of

return required on investments will fluctuate. This

distortion has been discussed by Auerbach and Hassett (1989) ,

 

1 According to Pechman (1987) the corporate income tax
accounted for approximately 8 percent of federal revenues
compared with 45 percent from the personal income tax and 37
percent from payroll taxes in 1986. Pechman points out that
the percentage of total federal tax revenue raised by the
corporate income tax has been declining steadily over the past
30 years, falling from 27.3% in 1955, to 8% in 1986.

2Although the percentage of total federal tax revenue
received form the corporate income tax is declining, this does
not necessarily imply that the size of the welfare losses
incurred due to the existence of the corporate income tax are
declining. In fact, it may imply that the size of the
distortions caused per dollar of tax revenue raised may be
increasing.

3
Feldstein and Frisch (1977), Fullerton and Henderson (1989),
and Hall and Jorgenson (1967).

(3) The allocation of capital between the corporate and
noncorporate sectors. Different tax rates between sectors
will cause capital to move out of the higher-tax sector into
the lower-tax sector. The seminal work on this distortion was
done by Harberger (1962, 1966), using a two-sector model.

Harberger's most famous result is that a tax on capital
in the corporate sector is borne by all owners of capital,
even if their capital is not incorporated. Harberger also
estimated the efficiency loss from the taxation of capital.
He estimated that the deadweight loss from the taxation of
capital between 1953 and 1959 was between 2.4 and 7.0 percent
of corporate income tax revenue.

The development of CGE techniques has allowed the
construction of multiple-sector models. See, for example,
ShoveniandHWhalley'(1972), Shoven (1976), and Fullerton, King,
Shoven, and Whalley (1981).3 Gravelle and Kotlikoff (1989)
have explored the misallocation of capital between corporate
and noncorporate establishments which produce in the same
sector. Their results indicate that, if one takes into

account the presence of corporate and noncorporate capital in

 

3 Shoven (1976) has twelve production sectors in his
model. In addition to disaggregation, Shoven also corrects a
mathematical error by Harberger. Harberger understates the
true amount of capital income in his model by almost four
million dollars. Both the higher level of disaggregation and
the correction of Harberger's error lead to welfare gains
which were, on average, 40 percent larger than Harberger's
results.

4

the same sector, the distortion is much larger than previously
believed. They examine this distortion at a high level of
industry aggregation. It may be the case that, if one were to
take a more disaggregated view of the economy, this distortion
may be much smaller than their results indicate. Another
criticism is the accuracy of their model in capturing the
decision on whether to incorporate. Their model provides a
high elasticity of incorporation with respect to tax parameter
changes within the model. Gordon and MacKie-Mason (1990) find
that, empirically, there appears to be little elasticity in
the decision to incorporate with respect to changes in the tax
system. In any event, I choose to focus on the more
traditional question of the misallocation of capital between
the corporate and noncorporate sector.

(4) The individual savings decision. If one accepts
Harberger's result that the burden of the corporate tax is
shifted to all owners of capital, then the corporate income
tax lowers the after-tax rate of return for all savers. This
biases their savings decision in favor of current consumption
instead of future consumption. The effect of taxation on
intertemporal decisions has been explored by a number of
authors, including Atkinson (1971, 1980), Auerbach and
Kotlikoff (1987), Ballard (1990), Feldstein (1978), Feldstein
and Frisch (1977), and Periera (1988).

(5) The dividend payout policy of firms. Under the
current tax structure, dividends are taxed twice, once as

corporate income, and once as personal income. This increases

5
the cost to shareholders of receiving dividends, and biases
the firm toward a low payout ratio. The effect of the tax
system on corporate dividend policy is explored by Poterba and
Summers (1985) and Feldstein and Green (1983).

The simplest method for removing these distortions would
be the elimination of the corporate income tax. The outright
elimination of the corporate tax is unlikely for two major
reasons. The first is that, despite its relative decline, the
corporate income tax is a major source of federal revenue,
raising approximately $83.9 billion in 1987.4 Secondly, the
elimination of the corporate income tax also may lead to some
distortions of its own. One such distortion would be the
incentive for individuals to shelter income in corporations in
order to avoid paying personal income taxes.

Short of simple elimination of the corporate income tax,
some economists have suggested the integration of the
corporate and personal income taxes. One of the most thorough
discussions of corporate tax integration is found in McLure
(1979). He discusses the administrative implications of tax
integration as well as various other proposals. The possible
proposals for integration range from complete integration to
partial dividend relief} 'Under the partnership method of full
corporate integration, the corporate tax is eliminated and all

corporate income is attributed to equity holders of the

 

‘ U.S. Department of Commerce, Bureau of the Census,

Statistical Absttact of the United States, Washington, D.C.,
January 1990, P. 274.

6
corporation, based on the number of shares held, and is taxed
as personal income.

Another proposal is to allow a deduction for dividends
within the corporate tax. If firms choose to pay out all
earnings as dividends, a 100% dividend deduction would have
the same effect as full integration. However, even with a
100% deduction, this would probably not have as great an
effect as full integration, because firms may have other
reasons for retaining earnings.

Dividend relief, within the personal income tax, reduces
or eliminates the double taxation of dividends. One method
would be to issue a credit to taxpayers for the corporate
income tax jpaid on income disbursed. to shareholders as
dividends. Besides full dividend relief, it is also possible
to give individuals a partial dividend credit. Although this
would not eliminate the double taxation of dividends, it would

lessen the distortion.

1.3. Review of the Literature

1.3.1. gotpotate Tax Integration and CGE models

This section briefly reviews earlier work on tax
integration, and CGE modeling. Readers familiar with this
literature may want to proceed to section [1.3.2.]

A number of studies have examined the effect of various

corporate tax integration plans. Using tax return data from

7

the U.S. Department of Treasury's Tax File of 1973, Feldstein
and Frisch (1977) examine the distributional and revenue
implications of full corporate integration and a revenue-
neutral partial integration plan. They report that full
corporate integration may result in annual welfare gains as
large as $7 billion (approximately 19.25% of federal corporate
income tax revenue in 1973) by reducing the distortion on the
individual savings decision alone.

Ballentine and McLure (1981) examine the effects of full
integration of the personal and corporate tax systems, while
taking into account the financial behavior of firms. They
modify a two-sector Harberger general equilibrium model to
incorporate a model of firm financial decision making. They
find that, with the incorporation of the corporate financial
structure, the implementation of corporate tax integration
will lead to greater welfare gains to the economy, than in an
otherwise identical two-sector model which abstracts from firm
behavior. This is due to a reduction in the difference
between the rate of return on corporate and noncorporate
capital, and a reduction in the risk premiums required on
corporate securities. The reduction in the risk premiums will
lead to relatively larger increases in the income of investors
in the noncorporate sector, because previously the corporate

income tax burden was shifted onto owners of noncorporate

8

capital. This will also result in an increase in the
progressivity of the income tax.5

These studies by Feldstein and Frisch and by Ballentine
and McLure made important contributions to the examination of
integration. However, these studies are lacking in that they
are not able to model more completely the complexity of the
tax laws and the ability of individuals to change their
behavior in response to changes in these laws. Advances in
computer technology have allowed researchers to use CGE models
to'simulate the effects of changes in tanpolicy, using large-
scale models of the United States economy. Computational
models allow the modeling of multiple production sectors as
well as many different types of consumers. These models also
allow for a complex modeling of the current tax system. These
models have also been used to study the effects of corporate
tax integration. Fullerton, Shoven, King, and Whalley [FKSW
(1981)] use a CGE model to examine the welfare effects of four
corporate tax integration plans, ranging from full integration
to partial dividend relief. Their model consists of 19
producing sectors which produce 16 different consumer goods.
The model also includes 12 differing consumer classes, which
are separated based on income levels. Their model's data set
consists of information based on the 1973 tax rates, capital

stocks, and income distribution. of the United States.

 

5 This result is based on the fact that the ownership of
corporate capital tends to be more highly concentrated in the
upper-income classes than the ownership of noncorporate
capital.

9
However, FKSW do not explicitly model the effects of changes
in the tax system on corporate financial behavior.

The authors run simulations under revenue-neutral
conditions, adjusting personal tax rates so that there is no
loss in total tax revenue. The static gains from total
integration are very large (on the order of $6 billion per
year, approximately 16.5% of total federal corporate income
tax revenue, in 1973 dollars). The plans for dividend relief
resulted in efficiency gains of about half as much as the
total integration plan. FKSW also look at dynamic gains,
i.e., gains which accrue in the future due to increases in
the accumulation of capital. These changes are very large,
with a present value of at least $100 billion in every case.

Fullerton (1983) examines the effects of full corporate
tax integration in a model with partially mobile industry-
specific capital. Fullerton uses the nineteen-sector FSKW
model, but restricts the flow of capital out of some sectors
to the amount of depreciation that occurs in that sector's
capital stock. In his model, a large portion of the economy's
capital stock is freely mobile. A portion of the capital
stock is, however, restricted to be sector-specific. The
sector-specific capital is assumed to remain in the original
production sector. The sector-specific capital will
depreciate away over the course of simulations, and after ten
years, all capital is assumed to be perfectly mobile. In his
most restrictive case, Fullerton assumes that five production

sectors consist of sector-specific capital. Full integration

lO
simulations, under' his :most. restrictive case, result in
dynamic welfare gains that are $5 billion dollars lower than
those found by FSKW.

Fullerton and Gordon (1983) also use a CGE model to
examine the effects of changes in tax policy on the United
States economy. They modify the FKSW model to incorporate the
effects of changes in the tax system on the selection of
corporate debt/capital ratios. The authors rely on the
bankruptcy theory of corporate finance, which is discussed in
Gordon and.Malkiel (1981). Under this theory, firms trade off
the benefits of financing new investment with debt against the
increased costs of bankruptcy associated with a greater amount
of leverage. The theories of financial leverage will be
discussed in greater detail in section [1.3.2].

Fullerton and Gordon examine the effects of full
integration via the partnership method. Uhder full
integration, there is no longer any tax advantage to using
debt, so firms have an optimal debt/capital ratio of zero.
In every revenue-neutral simulation in which personal tax
rates were increased to compensate for lost revenue, full
integration resulted in welfare losses. There were welfare
gains when the loss in tax revenue was replaced by a lump-sum
tax, but they were very small in comparison to those found by
FKSW (1981). The changes in the FKSW model resulted in a
dramatic change in the effects of corporate tax integration.

Although the Fullerton-Gordon paper would seem to

indicate that there are few benefits to corporate tax

11

integration, one must be cautious in evaluating their results.
The reason for this caution is the type of risk that they
incorporate into their model. In their model, Fullerton and
Gordon focus on the income risk of firm ownership. The total
return on a capital asset is composed of two parts. These are
the current return on the asset and changes in the value of
that asset. Income risk is associated with changes in the
current return.to the asset» Fullerton and.Gordon assume that
the government assumes full income risk sharing.

Bulow and Summers (1986) point out that fluctuations in
the value of the asset, referred to as capital risk, are much
larger than assumed by FUllerton and Gordon. Variations in
stock prices are generally much larger than fluctuations in
current income. The findings of Bulow and Summers call into
question the assumption of a large amount of income-risk
sharing by the government" The fact that Fullerton and Gordon
incorporate endogenous financial behavior into their model is
clearly an important step forward in the modeling of the U.S.
economy. However, although the financial decisions of the
firms are endogenized, the investment decisions of the firm
are not. If changes in the financial decisions of the firm
affect the cost of investment to the firm, the welfare
effects of corporate tax integration could be very different
from those found by FKSW or Fullerton and Gordon.

Slemrod (1983) uses a CGE model to examine the effect of
indexing the federal tax system for inflation. One of the

major features of his model is the modeling of financial

12

behavior of the corporate sector. In modeling the choice of
a firm's debt/equity ratio, Slemrod also relies on the
bankruptcy tradeoff theory. As the tax parameters of the
model change, the optimal debt/equity levels will change.
Slemrod relies on econometric estimates, his own and those of
King (1978), of the elasticity of firm's debt levels and
dividend payout ratios with respect to changes in the personal
and corporate tax rates. Slemrod finds that indexation of the
federal income tax system will result in a "significant"
reallocation of financial resources. He also finds that
indexation results in large welfare gains for upper-income
individuals. Although this paper is a major contribution
toward CGE modeling, it is lacking in that it groups all
corporations into one sector and does not allow for
intersectoral variation.

Galper, Lucke, and Toder (1988) also use a CGE model to
examine the effects of changes in the tax system on a model
which incorporates the debt/equity decision of the firm.
Their general equilibrium model is used to examine the effects
of the 1986 Tax Reform Act. Their modeling of the corporate
debt/equity choice differs from Slemrod (1983). Galper,
Lucke, and Toder use theoretical considerations to derive
explicit demand functions for debt, instead of relying on
econometric estimates of firm responses to changes in the tax
system. Firms in their model select an optimal debt/equity
ratio to minimize the cost of capital, given the interest

rate, the corporate tax parameters, and the parameters of

13

individual demand functions for equity. Galper, Lucke, and
Toder find that the Tax Reform Act of 1986 resulted in large
welfare gains to the upper-income households, because upper-
income households tend to hold a disproportionate amount of
capital assets. They also find the there should be a slight
decline in corporate debt/equity ratios. This is the result
of two offsetting effects: a reduction in the corporate
income tax, and an increase in the tax rate on capital gains.

Although this model proves very useful, it suffers from
the same criticisms as Slemrod (1983): the aggregation of the
corporate sector and the lack of dynamics in the model.
Galper, Lucke, and Toder make an important contribution,
however, by using an explicit model of firm debt behavior.

Jorgenson and Yun (1986) use an econometrically-estimated
general equilibrium model to examine the efficiency of capital
allocation among corporate and noncorporate assets. Their
model uses empirically estimated production and consumption
parameters based on time series data from 1955-1980. A
business sector and a household sector are included in their
model. The business sector produces consumption and
investment goods from labor and corporate and noncorporate
capital services. 'The capital services in the business sector
are divided among short-lived and long-lived assets. The
household sector, in their model, is an aggregate of
consumption goods in the economy as well as household capital

services, and leisure time. A government sector is also

14
included in their'model. Jorgenson.and Yun examine the impact
of the ERTA of 1981.

They find that the ERTA of 1981 had potential welfare
gains of $542 billion, if a lump-sum tax was used to replace
the revenue loss. Using an increase in the tax rate on labor
income to recoup the revenue shortfall ‘would result in
potential welfare gains of $328 billion, while a sales tax
replacement could result in potential welfare gains of $295
billion dollars.

The Jorgenson and Yun paper is important, because unlike
many other CGE models, it relies on actual empirical estimates
of consumer and producer behavior. But, it is does not include
endogenous financial behavior.

The papers reviewed up to this point have dealt with two
major issues, corporate tax integration and CGE :models.
Although the modeling of corporate tax integration has
improved greatly over the past twenty years, there is still a
great amount of work to be done. Table 1 presents a listing
of the major features of the CGE models which are discussed

here.

1.3.2. Theories of Corporate Financial Leverage

This section ‘will give a brief examination of the
theories of corporate financial leverage. The corporate
leverage decision plays a major role in the development of my

model in Chapter 2.

15

The seminal research on corporate financial policy was
done by Modigliani and Miller (1958). Their major result was
that corporate financial policy is irrelevant in determining
the value of the firm, i.e., ceteris paribus, investors will
be indifferent to the debt/equity policy or the dividend
payout ratio of the firm. Modigliani and Miller's findings
are reached under some very restrictive assumptions. They
assume: (1) There are perfect capital markets. (2) All bonds
are riskless assets. (3) There are no bankruptcy or leverage
costs. (4) There are no taxes. The Modigliani-Miller Theorem
relies on the ability of investors to borrow and lend to

offset any changes in firms' borrowing or dividend policy.

 

Table 1
Summary of Features of Computational General Equilibrium Models
Number Endogenous
of D/E Endogenous
Agthogs Sectors Ratio Dividends Dynamtcg
Fullerton, King,
Shoven e Whalley
(1981) 19 No No Yes
Fullerton (1983) 19 No No Yes
Fullerton & Gordon
(1983) 19 Yes No Yes
Slemrod (1983) 5 Yes Yes No
Galper, Lucke,
and Toder (1988) 5 Yes No No
Jorgenson & Yun
(1986) 1 No No Yes

When corporate and personal taxation are introduced, the
Modigliani-Miller Theorem no longer holds. Under the current

tax system, interest payments to bondholders are deductible

16
from corporate income, while dividend payments of a firm are
not. This makes it cheaper for a firm to use debt than
equity. Because debt is the cheaper means of financing an
investment, a profit-maximizing firm would finance all of its
investments out of debt.

Casual observation of the corporate sector shows that
this is not the case in the United States. Firms only finance
a portion of their capital stock with debt. There are a
number of theories on why firms do not finance all of their
capital with debt. The existence of agency costs, clientele
effects, and bankruptcy costs will be discussed here.

The use of large amounts of debt finance may lead to
agency costs in the form of distorted investment incentives
for firms. One way to reduce these agency costs is to force
managers of firms to finance:a portion of their new investment
with equity, to preserve some future investment opportunities.
The theory of agency costs of this type is discussed by Meyers
(1977).

Although the current corporate tax law favors the use of
debt finance, the existence of personal income taxation may
work to reduce the benefits of using debt. The existence of
clientele effects, the interaction of personal and corporate
tax rates, and the relative costs of debt and equity may also
effect corporate debt/equity positions. The effects of
personal taxation on corporate financial decisions have been
explored by Stiglitz (1973), Miller (1977), and Auerbach and

King (1983).

17

Stiglitz (1973) and Miller (1977) focus on the relative
costs of borrowing based on the relationship between the
corporate income tax rate and personal income tax rate. They
point out that certain features of the personal income tax
system work to minimize the benefits of using leverage. When
a firm uses debt finance, a lower portion of the investment
will be taxed as capital gains. Until the Tax Reform Act of
1986, realized capital gains were taxed at a lower rate than
interest or dividend income. In addition, capital gains are
taxed only on.a:realization basis instead of an accrual basis.
This lowers the effective tax rate on equity even further.
The existence of a progressive income tax system leads to
differing advantages of debt and equity finance for
individuals. Because individuals have differing tax rates on
bond and equity income, they will want to invest in firms
with differing degrees of leverage. This will lead to a
market equilibrium with some firms which are very leveraged
and some firms that are not very leveraged.

Auerbach and King ( 1983) explore the interaction of
personal and corporate taxation in a general equilibrium
model. Investors in this model select themselves into one of
two categories based on their income tax bracket and their
personal level of risk aversion. Investors prefer all-debt
finance or all-equity finance. Depending on an individual's

income class, income from debt may be taxed at a lower rate

18

6 Another major factor in the

than equity, or vice versa.
selection of the debt/capital ratio is that some investors may
be risk averse. If risk aversion is taken into account,
investors may put themselves in an all-equity position, even
when debt is the tax-preferred form of financing. The
existence of differing tax: brackets and levels of risk
aversion in the economy will lead, in the aggregate, to some
capital being financed with debt and some capital being
financed with equity. It is important to note, however, that
investors end up in either of two extremes, all-debt or all-
equity. This model does not provide for mixed portfolios.
The clientele models do not provide for individual firm
interior debt/equity positions. This is a major drawback. In
the real world, firms do not seek an all-debt or an.all-equity
position. In these models, the advantages and disadvantages
of using debt stem only from the interaction of the personal
and corporate tax systems, and varying levels of risk
aversion. The introduction of real costs in the use of debt
will lead firms to select an optimal interior debt/capital
ratio. The possibility of bankruptcy imposes leverage costs
on the firm for using debt. Firms will trade off the tax

advantages of using debt against the increase in the

 

6 Auerbach and King assume that the tax rate on equity is
an effective tax rate which combines the lower tax rate on
capital gains with a slightly higher rate on dividend income.
The tax rate on personal interest income is assumed to be
greater than the effective tax rate on equity. However, it is
the interaction of the personal and corporate tax rates which
leads to a consumer preferring all debt or all equity.

19

probability of bankruptcy (which is assumed to impose real
costs on the firm). As mentioned earlier, this theory is
explored in Gordon and Malkiel (1981), and adapted to a CGE
model in Fullerton and Gordon (1983). The addition of
bankruptcy costs may have varying effects across sectors and
firms. This will lead to differing leverage positions of
firms. I believe that this is an important advantage of the
bankruptcy model. Consequently, bankruptcy costs are used in
this paper to model firms' leverage decisions.

The bankruptcy theory takes into account the asymmetry of
the corporate tax code with respect to its treatment of the
use of debt and equity by corporations. It also takes into
account the effects of the personal income tax system on
reducing the tax advantages to investors of firms relying on
debt finance, as explored in Stiglitz (1973), Miller (1977),
and Auerbach and King (1983). The following is a brief
description of a standard version of the bankruptcy theory.

Interest income received by bondholders is taxed at their
marginal tax rate, 6. The returns to equity holders are taxed
in two ways. The portion of the return that is received in
the form of dividends, a, is taxed at the marginal tax rate 6.
The remaining portion, (1-a), is taxed at the capital gains
rate c. Dividend payments are taxed at the corporate level,
as corporate income, at the corporate tax rate 1. Interest
payments on bonds are deductible from the corporate income tax
base, so they are not taxed at the corporate level. The

after-tax return on a corporate bond is (1-9)r, where r

20

represents the interest payment on the bond. The after-tax
return on equity is {a(1-0)+(1-a) (1-c)}(1-r)r. Note that this
model does not take into account any risk, and that the
parameter a is assumed to be fixed.

Following Fullerton and Gordon (1983), if a firm issues
a dollar of debt and then uses that dollar of debt to
repurchase a dollar of equity, the gain to investors is (1-
0)r, the new interest payments on the debt, minus {a(1-6)+(1-
a)(1-c)}(1-r)r, the reduction in the return to equity. This

expression reduces to

r{(1-6)-(1-1:)[a(1—e)+(1-a)(1—c)]}. (1)

As long as expression (1) is positive, the investors in
the firm will gain because the use of debt saves them money.
Gordon and Malkiel (1981) explore the conditions under which
expression (1) would be positive, or equal zero. As noted in
Fullerton-Gordon (1983) , they find that, for reasonable values
of the tax rates, expression (1) will be positive.

If there were no leverage costs and expression (1) were
positive, one would reach the all-debt-finance conclusion.
However, with leverage costs, the managers of the firm will
realize that they can reduce the cost of acquiring capital by
using debt finance, but that this increase in leverage results
in real costs to the firm. The bankruptcy costs can be
summarized by a function t(o), where o represents the firm's

debt/capital ratio. It is assumed that ¢'>0 and t">0.

21
Setting the costs and benefits of debt equal to each other
yields the equilibrium condition

We» = r{(1-0)-(1-1:)[a(1-8)+(1-a)(1—c)]}. (2)

Expression (1) implies that changes in the personal and
corporate tax rates will affect the incentive for firms to use
debt. The derivative of expression (1) with respect to the

corporate tax rate yields7

We’rgeastsjm 1’ =r{a(1-0) +(1-a) (1-c)}. (3)

 

which indicates that an increase in the corporate tax rate
will increase the benefits of using debt. It increases the
value of the interest deduction on debt, and should lead to an
increase in the use of leverage.

A change in the personal tax rate (6) will change the
taxes that must be paid on interest income and change the

taxes that must be paid on dividend income.

6(exprigfion l) = [{(41)+(1-c)(a)} (4)

The derivative is clearly negative, indicating that the

overall impact of an increase in the personal tax rate is a

 

7 Note these are partial equilibrium derivatives. In a
general equilibrium setting, it will generally not be possible
to alter one tax rate without also altering the other tax
rates to achieve revenue neutrality.

22
decrease in the benefits of relying on debt finance. This
will cause firms to reduce their leverage position. This is
because the increase affects the total return to debt, while
affecting only a portion of the return on equity.

Finally, an increase in the capital gains tax rate should
result in an increase in the use of debt because it lowers the
after-tax rate of return on equity, but has no effect on the
after-tax rate of return on debt. This is shown by equation
(5), which is positive, indicating an increase in the benefits

of using leverage.

6(1)

= r{(1-t) (1-a)} (5)
ac

 

Because equation (5) is positive, it indicates that an
increase in leverage should result from an increase in the
capital gains tax rate. Interest income received by
bondholders is taxed at their marginal tax rate. The
bankruptcy theory yields empirically testable results with
respect to the effect of the personal and corporate tax rates.
These results will be tested in Chapter 3.

The purpose of this essay is to take the modeling of CGE
models and. corporate tax: policies one step further, by
presenting a model of firm behavior which can be used in a CGE
model. This model will then be used to examine corporate tax
policies in a setting where the effects of policies on the

financial decisions of the firm, as well as the effects on the

23
investment decisions, are taken into account.8 It will also
allow for examination of the dynamic effects of tax
integration proposals. The model of firm behavior is
presented next. Econometric estimates of the impact of
changes in the personal and corporate tax rates on a firm's
debt/capital ratio will be obtained from a panel data set of
U.S. non-financial corporations covering the period from 1969

until 1988 in Chapter 3.

 

3 My model does not attempt to explain the dividend
behavior of firms, but instead focuses on firms' choices of
debt-to-capital ratios. However, the effects of changes in
the dividend behavior of the firm may be explored in
subsequent work.

Chapter 2

MODEL OF FIRM BEHAVIOR

This section presents a more complex model of firm
behavior. The model encompasses the debt/capital choice of
the firm as well as its investment decision. It relies
heavily on earlier work by Goulder and Summers (1989), which
in turn, is based on Summers (1981). The behavior of firms is
based on the premise that managers of firms seek to maximize
the value of the firm for its shareholders.

I begin. by specifying' the fundamental asset. market
equilibrium condition which requires that the after-tax rate
of return from ownership in a firm (risk adjusted) must be
equal to the rate of return that could be achieved.by' holding
a riskless asset, such as a municipal bond. The equilibrium

is characterized by

(1-c) {V-SN} + (1-6)[Div] -'qV = <1(1-6))v, (6)

where c is the capital gains tax rate, V is the change in the
value of the firm over time, SN is book-value of new share
issues, 6 is the personal tax rate on dividend and interest
income, Div are dividends paid by the firm, n is the equity
risk premium, which is assumed to be exogenous, V'is the value
of the firm, and i is the nominal interest rate. Equation (6)

indicates that the after-tax return on capital gains plus the

24

25
after-tax dividend income from the firm (which equals the
total net return on the firm), risk adjusted, must equal the
after-tax return on the safe asset. Equation ( 6) can be

rearranged to yield the differential equation

 

‘_ I = y_ (1‘6) .

V ——(1-c) V S (1_C)D1v, (7)
where r = i(1-6) + 7). If one imposes the transversality
condition

1' i ' 1

1m -r

s-ooo Vs [emf W dUJ = O (8)

t

to rule out explosive growth in the value of the firm,

equation (7) can be solved to yield

- 58"] {expf -I"

1 9
r (1-C) leJ dS. ( )

 

 

This expression implies that the value of the firm in time t
is equal to the present discounted value of the perpetual
stream of after-tax dividends less new share issues. Managers
will seek to maximize (9) each period.

Firms in this model are assumed to maintain a 100 percent
dividend-payout ratio. All current after-tax earnings are
returned to shareholders in the form of dividend payments.
Investment expenditures of the firm are entirely financed out

of new debt and new share issues. There are no retained

26
earnings in this model.9 The assumptions of a 100 percent
dividend-payout ratio and only new-debt and new-share

investment finance place two constraints on the firm:

DiV= EARN (1°)

and

IEXP = B" + s". (11)

EARN represents the firm's after-tax earnings, BN represents
new bond issues and IEXP represents new investment
expenditures.

The after-tax earnings of the firm are

(PF(K,L,M) -WL-P,,M-1‘D) (1-r) +rDep, (12)

where: P = the price of output, F = the production function
of the firm, K = capital inputs, L = labor inputs, M = the
vector of intermediate inputsw W’= the wage rate, Pk:=‘vector
of intermediate input prices, D = nominal debt, r = corporate
tax rate, Dep = the value of current depreciation allowances.

Investment expenditures are specified as

 

9The assumption of no retained earnings in this model is
important. By not allowing firms to use retained earnings,
often thought of the least expensive form of investment
finance, one may prevent the firm from undertaking some
investment. This may introduce some distortion into the firm's
investment decision.

27
IEXP = (1-ITC)PK1'+ (1-r)P¢(TI()I, (13)

where ITC is the investment tax credit and ¢( I /K) measures the
investment costs per unit of investment. It is assumed that
the adjustment costs are homogeneous of degree one in.I and.K.
Following Goulder and Summers, I model adjustment costs as
being internal to the firm. This implies that, holding
constant the amount of investment, an increase in adjustment
costs results in a reduction in output. This is due to more
inputs being diverted from production to the installation of
new capital. I also assume, as do Goulder and Summers, that
output, Y, is separable between inputs and adjustment costs,

i.e.,

Y = F<K,L,m — «phi-gr. (14)

I deviate from Goulder and Summers in that I allow the
firm to select its level of debt as well as the issue of new
shares. The use of more debt will impose real costs on the
firm in the form of an increased probability of bankruptcy.10

I posit the following bankruptcy function

 

1° The wa;; Street Journal, October 23, 1990, reports
that the number of corporate bankruptcies has risen
substantially over the past five years. The number of
bankruptcies rose to nearly 680,000, which they report is
almost double the number of bankruptcies filed in 1985. This
coincides with the increased use of debt finance in leveraged
buyouts during the 1980's.

28
= .2.
Ir “ms" (15)

where 8 represents the book-value of outstanding shares. Note
that I assume Y'>0 and Y">0. This function will cause the
firm to incur real costs as it increases the share of debt
relative to the total value of its capital stock. The
bankruptcy costs are modeled as a continuous function of the
debt/capital ratio. There is no minimum level of leverage
that must be reached before there is a positive probability of
bankruptcy. Any positive leverage will result in real costs
to the firm.11

Introducing explicit bankruptcy costs also plays an
important role in determining how investment is financed. In
a similar’model, Goulder and Summers indicate that, due to the
tax disadvantages of equity, firms would never issue new
shares. This is not the case in this model. Although new
shares are disadvantageous from a tax standpoint, they work to
the reduce the probability of bankruptcy to the firm. In this
model, it is feasible for the reduction in bankruptcy costs,
and hence an increase in the value of the firm, to offset the
tax disadvantages of issuing new shares.

The introduction of bankruptcy costs allows one to

rewrite equation (9) as

 

11 The use of an explicit bankruptcy function makes the
model tractable, as well as providing a realistic result:
firms achieve an optimal interior debt/capital ratio.

29

[(1—0) . l
Vt f ‘1'“ l (1-.» Dw- ' Selle”: W

"I" u} ds. (16)

 

Vt now represents the value of the firm when it is not
bankrupt. The expression (1-!) represents the probability
that the firm will not go bankrupt. It is assumed that if the
firm is bankrupt, the value of the finm is equal to zero.
Equation (16) indicates that the bankruptcy costs will be
proportional to the value of the firm. As the value of the
firm.increases, the loss in income that the shareholders would
incur if the firm went bankrupt would increase. Therefore,
the higher the value of the firm the greater the bankruptcy
costs incurred by the firm.

In order to specify the value of the firm over time, it
is important to represent explicitly the value of allowable
depreciation allowances. The depreciation allowances are
composed of depreciation allowances on current capital and the
value of depreciation allowances on future capital
acquisitions. The value of total depreciation allowances in

time t is equal to12

 

12 At this point, I am not concerned with the actual
calculation of the value of depreciation allowances. However,
when simulations are run, this will become necessary. In
order to calculate the present value of depreciation
allowances, I‘will appeal to the earlier work of Fullerton.and
Henderson (1989), and Hulten and Wykoff (1981).

30

5
Dept = fonmIJeprMt-ul) }du. (17)
0

The parameter 6 represents the exponential depreciation rate
under the tax code.13 PR is the price of capital, and I
represents the quantity of investment. Equation (17) says
that the value of depreciation allowances at time t is equal
to sum of the present value of depreciation allowances at each
point in time, from time t=0 until current time t. In
examining the value of the firm at each point in time,
depreciation allowances can be broken down further into
depreciation allowances on current capital (Bt) and the value
of depreciation on a dollar of capital acquired in the future
(2,) . This is done because the value of depreciation expenses
on existing capital may be taken as a fixed parameter, because
future investment decisions are independent of Bt. This
allows the firm to ignore Bt during its value maximization.
The value of depreciation expenses on current capital is

expressed as

Bc=f r86 expl-Ms—tH PK: [(9po (1'—_rc) du} ds. (18)
C

 

13 Note that the tax code does not actually give
exponential depreciation rates. The exponential depreciation.
rate is used for analytical convenience. See Fullerton and
Henderson (1989) for an example of the methodology of
transforming depreciation rates.

31
The value of depreciation allowances on future capital
acquisitions will affect the cost of future investment, by
providing future tax deductions. The expression for the value
of depreciation allowances on a dollar of capital acquired in

the future is

= _ _ ‘I 19
Z, ftuoexp( Mn 3)) [£9po (1_C)da] du. ( )

 

This expression indicates that the present value of
depreciation allowances on a dollar of new investment is equal
to the total of the present values at each time period in the

future. The earnings of the firm may now be expressed as

EARN = (PF(K,L,M) -WL-p,,M-iD) (1-1) +z,p,,1+3t. (20)

Substituting earnings into equation (10) implies that

dividends are equal to

Div = (PF(K,L,M) -WL-P,,M—iD) (1-1) +B+ZPKI. (21)

Equation (21) can then be substituted into equation (16) to
yield an expression for the value of the firm.at time t. After
substituting and rearranging terms, the value of the firm at

time t is

The managers of the firm will want to maximize equation

(22), by selecting L, M, B", I, and SN subject to four

32

 

vc = fu-wfl 8"” (norm, L, m- WL- -p,,M-1D] (1-1)
I:

(22)
+ zprr) — S" J 1{epo--(—1—_———- I ddu}ds+B.
constraints:
I6 = I-bRK (23)
15=B”-1:D. (24)
3' = S" (25)
(1-ITC') prr+(1-r)P¢(-TI() = B” + s" (26)

The first constraint, equation (23), is the capital
accumulation equation. This indicates that the change in the
capital stock is equal to new investment less the depreciated
capital stock. Note that 6R is the economic rate of
depreciation, which may differ from the depreciation rate
allowed under the tax code. Equation (24) gives the change in
the value of debt outstanding. It is equal to new debt issues
less the decrease in the value of outstanding debt due to
inflation, n. The third constraint, equation (25), indicates
that the change in value of shares outstanding is equal to the
value of new share issues. The final constraint insures that
investment expenditures will be equal to the value of new'debt

plus new shares.

33

The maximization problem of the firm can be solved by
applying Pontryagin's maximum principle. I set up a current-
value Hamiltonian and introduce three multipliers, 11 12, and
13, which are interpreted as the shadow prices of capital,
shares, and debt respectively. I also introduce a static
multiplier, p, to account for the fourth constraint. Because
the depreciation allowances on the current capital stock (Be)
are independent of future investment choices, they can be
ignored in the firm's maximization problem.

The firm then selects L, M, I, S", and BN to maximize

f -r
H g {3- We“ duh {1-T}{:—::—2—;—<[PF(K,L,M) -WL-P,,M—iD] (1-1) +ZPKI+>-

+ 11(1-6“) + 12(3") + 13(BN-1CD)
+ B{(1-ITC)P,.I+(1-r)P¢(7I()I-S"-B”ll.

(27)

Maximization yields the following first-order conditions:

5H.

 

E. FL=%’ ‘28)
65:1”: 13 = (3 (30)

 

11: -(1-T) iii); ZPX - B{(1—ITC)PK+(1—I)P {¢+(-I€,)o’}}

Bias

(31)

34

 

3H. . R _ (1-6)
32.1 111(—(——fc)+6) (1- T)cK-—TPF(1(1_1:-)+
2 (32)
I
- (ml-mu?) «V1
6H _ I (1- 6) 1
-a—D.:1.313(——(1_c) +1!) --(1 ‘1’) (1_ C)(-1')(1- 'c) + ‘P(—( S)2)X
(33)
8H
—63N:B = 12-(1-1’) (34)
OH I / D
6? ‘2‘ “(Tr-75$" * "' ‘W’X‘ "5’

In first-order conditions (33) and (35), Xt is equal to
current earnings less the book-value of new share issues.
First-order conditions (28) and (29) imply that labor and
intermediate outputs should be used up to the point where
their marginal products equal their respective input prices.
Using the first-order conditions for new debt and new shares,
one can solve equations (33) and (35) for the value of H, the
static multiplier. Equations (33) and (35) can then be
combined to achieve the condition

v’<—S——)x,-T’(—£—
(D+S)2 (D +S S)2

r + _(1— 0) _
(1-c) ‘n -(1_ C)(i)(1 1)}

)Jsr,{1———’t ”’0’ }=
I

(36)
(1-T){

 

This expression represents the marginal costs and benefits of

transferring a dollar of equity to a dollar of debt. The

35

left-hand side represents the increase in bankruptcy cost.
The right-hand side represents the gain from using debt. The
first term represents the required after-tax rate of return
that would have been paid on a new share. The second term
represents the benefits of inflation on debt. During periods
of inflation, the real value of outstanding debt decreases.
This causes the benefit of using debt to increase. The final
term represents the value of the interest payment on the
dollar of debt. This expression implies that a firm should
select its level of debt such that the marginal increase in
leverage costs on the left-hand side of the equation are just
equal to the marginal advantages of using debt on the right-
hand side of the equation. ‘This equation implies that changes
in the personal and corporate tax rates will affect the firm's
debt/capital ratios. This effect will be measured empirically
in Chapter 3.

Next, I solve for the investment equation of the firm.
Note that equation (22), the value of the firm, is homogeneous
of degree one in all inputs K, L, M, and I. This is the
result of the constant returns to scale production function
and the homogeneous adjustment cost function. This implies
that ‘the ‘value of 'the firm. minus the ‘value of current
depreciation is proportional to the value of the capital

stock, or

(Vt-BC) =nyK, (37)

36
where y is the factor of proportionality, or average q. The
maximum principle and the definition of the shadow price imply

that

6Vt_
am: "A" t

(38)

Equations (37) and (38) imply that marginal and average q are
equivalent.’4 Because average and marginal q are equal in
this model, equations (37) and (38) can be combined to yield
__ (Vt-BC)

lint PK

(39)

Included in equation (39) is the value of the firm, after
taxes, relative to the value of the capital stock. This is
known as Tobin's marginal q adjusted for taxes, or what
Summers (1981) refers to as "tax-adjusted Q."

This expression can then be substituted into the first-
order condition for investment, along with the value of 8
derived from equations (34) and (35), to yield an investment
equation of the form

In order to specify the investment equation completely,
I must specify the adjustment cost function completely. I
adopt the adjustment.cost function first introduced by Summers

(1981). It is assumed that adjustment costs are zero until

 

1‘ In this model, average and marginal q will always be
equivalent. However, it is important to note that empirical
evidence has shown that marginal and average:q often.differ in
the real world.

37

 

 

l 1 IQ -'Bt (1’3)
WW3) + «M = { + (1-Y)——z1
K l (14) - w't—D >X. P22" (1'9)
(ms)2
+ ( 1 + ITC)? ]———l
' ‘JP(1-t)

(40)
some level of investment and then rise linearly; The
adjustment cost function is assumed to be

m I
—((—)-¢vz)2
¢(TI{)= 2 K (41)

NIH

Because adjustment costs have been specified,the investment
function can be derived by substituting for ¢(I/K) in equation
(40). .After substituting and.rearranging'terms, equation (40)

reduces to the investment function

-I£(=a+ Q , (42’

eIH

where Q represents the right-hand side of equation (40) and o
and a are estimated adjustment cost parameters. This
investment function can then be used to model the response of
firms to changes in tax policy. Econometric estimates of the
change in a firm's leverage position with respect to changes
in personal and corporate tax parameters can also be used to
fully specify the value of the firm.ambedded in the investment

function.

38
The next section examines empirically the implications of
the effect of corporate and personal taxation on firms'

leverage decisions implied by equation (36).

Chapter 3
THE EFFECT OF PERSONAL AND CORPORATE TAXES ON FIRMB'
DEBT/CAPITAL RATIOS

3.1. Empirical Implications of the Bankruptcy Models

Equation (36) implies that changes in the personal and
corporate tax rates will affect the benefits that firms
receive from the use of leverage. The effects of changes in
the tax rates generated from my model do not differ from those
that are predicted by expression (1). This can be seen by
examining the effects of changes in the personal and corporate
tax rates on the benefits of using leverage. Define P as the
benefits of using leverage. Gamma is equal the right-hand

side of equation (36),

_ _ r _ (1-6) - _
I‘ - (1 ‘1’) {_M + a: (1_C) (1) (1 1)}. (43)

 

Differentiation of equation (43) with respect to the marginal

personal tax rate 6 yields15

3(P) _ _ _ i r
3767‘ (”WT—7:7 “"

Equation (44) indicates that an increase in the personal tax
rate will decrease in the advantages of using'debt. This Will

result.in.a decrease in a firm's debt/capital ratio. iBoth.the

 

15 This uses the fact that r = (1-6)i + n.

39

40

simple bankruptcy model and.my model of firm behavior predict
that an increase in the capital gains tax rate and the
corporate tax rate will result in an increase in the use of
leverage. It may appear that.the construction of my model was
simply a theoretical excise which could have been avoided,
because it reaches the same empirical implications as the
simple bankruptcy model. However, although the theoretical
predictions of both models are similar, my model provides a
framework for the tax policy simulations which are conducted
in Chapter 5.

In this chapter, I will examine empirically the role of

taxation in the firm's choice of debt/capital ratios.

3.2. Estimation

3.2.1. Mgggl

A fixed-effects model was used to measure the
responsiveness of the corporations to changes in the tax

rates. The model takes the form:

Yit=pXic+Ai+aT+€ic a (‘5’

Yit is a measure of the ith firm's debt/capital ratio in year
t. lit is a vector containing information on firm sales,
corporate and personal tax rates, and a measure of inflation.
A time trend, T, is also included. The cit denotes the firm-

and year-specific random error. These errors are assumed to

41

be distributed 'N(0,02). Lambda (1.1) represents a firm-
specific effect that is assumed to be time invariant. This
effect allows the estimation.to»account for the fact that some
firms may have consistently high or low debt/capital ratios,
over time, due to a firm-specific characteristic.

The fixed-effect model is estimated by differencing each
firm's observations around the individual means in the

following way:16

(46)

This differencing will remove the fixed effect because it is
time invariant. The resulting model is then estimated using
ordinary least squares“ The natural logarithms of all
variables are used, with the exception of the time trend, so

that the resulting coefficients are in elasticity form.

3.2.2. Data

Estimation was performed on a panel data set consisting
of annual data from 1969 to 1988. Firm-specific data were
obtained from Standard and Poor's 1989 COMPUSTAT data tape.

Information on a firm's outstanding debt as well as its

 

1‘ Note that this differencing is done for all variables,
not just the X and Y vectors.

42
current capital stock and sales was collected. The firms were
separated into twelve sectors, according to their Standard
Industrial Classification (SIC). The separation into sectors
is to control for sector-specific effects. In Table 2, I
provide a listing of the sectors, the number of firms in each
sector, and the SIC codes for each sector.

The debt variable for each firm includes both long- and
short-term debt, and is based on the book value of the
company's debt. The value of the current capital stock is
equal to the value of outstanding debt, plus equity, plus any
minority interest of the firm. Data on annual sales were
obtained for each firm, to control for firm size as well as
for fluctuations in business activity. Because the sales
variable is used to control for firm size, it was adjusted
using the Consumer Price Index (CPI), as reported in LYLE
fitatiatigal Absttact of the Qaitea States. In addition, the
annual percentage change in the CPI was included to control
for the effects of inflation on debt/capital ratios. During
periods of inflation, the real value of a firm's outstanding
debt will decrease. If firms are concerned with maintaining
a debt/capital ratio in real terms, they may issue more debt
during inflationary periods.

Table 3 contains time—series information on the
debt/capital ratios of each sector. The data indicate the
differences in the use of debt between sectors. The average
debt/capital ratios over the sample period range from .29 in

the chemical, rubber, and plastics sector, to .52 in the

43

transportation, communications, and utilities sector. .Almost
all sectors show growth in the use of leverage during the
sample period, especially in the late 1980's. One may note,
however, that even given the growth in debt/capital ratios,
the variation over time appears limited. Data on corporate
tax rates were obtained from Pechman (1987). The top marginal
corporate tax rate was used for each year. unfortunately,
this variable shows little variation over the twenty-year
period, changing only four times. The statutory rate is the
variable of choice. In my model, the statutory rate is the
rate which.will affect the value of the interest deduction for
the firm. Note that, in my model, changes in the depreciation
schedules and investment tax credit have no effect on firms'
leverage positions. There may be second-order effects, but
those effects are not investigated here.

Data on personal marginal tax rates were obtained by
updating the series of average marginal tax rates compiled by
Barro and Sahasakul (1983). They build a time series of

average marginal tax rates weighted by the amount of adjusted

44

 

Table 2
Listing of Corporate Sectors Used in Estimation
figctor .. .. .flgsoge__ W
Food and Tobacco 2000 - 2141 33
(Food)
Textiles, Apparel, and 2200 - 2399 & 31
Leather 3100 - 3199
(Text)
Paper and Printing 2600 - 2796 46
(Paper)
Petroleum Refining 2910 - 2911 & 23
(Refine) 2992 - 2999
Chemicals, Rubber, and 2800 - 2899 71
Plastics 3000 - 3089
(Chemical)
Lumber, Furniture, Stone, 2400 - 2599 & 22
Clay, and Glass 3200 - 3299
(Lumber)
Metals, Machinery, 3300 — 3699 8 209
Instruments, and Misc. 3800 - 3999
Manufacturing
(Metal)
Transportation Equipment 3720 - 3799 23
(Transportation)
Motor Vehicles 3710 - 3716 19
(Motor)
Transportation, 4000 - 4971 187
Communications, and Utilities
(Communications)
Wholesale and Retail Trade 5000 - 5999 81
(Trade)
Services 7000 - 8999 35
(Services)

 

TOTAL: 780

1969
1970
1971
1972
1973
1974
1975
1986
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
Avg:

Std.
0".

KEY:

£1 1! EB 28
.35 .42 .31 .36
.39 .38 .32 .39
.36 .37 .33 .37
.37 .41 .31 .36
.36 .44 .30 .35
.41 .44 .31 .33
.35 .35 .30 .34
.33 .40 .30 .35
.33 .42 .30 .36
.36 .43 .29 .36
.39 .43 .27 .34

Food and Tobacco
Textiles

Paper and Printing
Motor Vehicles
Services
Communications
Petroleum Refining
Chemicals

Lumber

Metals Manufacturing
Trade

Table 3 .
Summary of Debt/Capital Ratios by Sector 1969-1988

45

51
34
56
33
32
35
so
35
32
52

15

a!
33

.33
.32

£9
.56
.56
.56
.55
.55
.58
.55
.52
.50
.50
.51
.50
.51
.50

.03

std.

46
gross income of each marginal bracket. They construct tax
rates for 1916 through 1980. I extend their data through
1988. Data on the highest marginal tax rate paid by gross
income group are obtained from the U.S. Department of

Treasury's W. The average marginal
rate is obtained by using the formula

i-n

72; (AGIi/AGI) Ti, (47)
-1

where the subscript i refers to the adjusted gross income
class, AGIi refers to the adjusted gross income of class i,
AGI refers to total adjusted gross income, T1 refers to the
highest marginal tax rate paid by that class, and T is the
average marginal tax rate. This equation corresponds to Barro
and Sahasakul's equation (11).

The capital gains tax rate was based on.the effective tax
rate on capital gains after allowing for the capital gains
exclusion. Between 1969 and 1978, the rate was 50% of the
personal tax rate, due to the 50% exclusion of capital gains
income. Between 1979 and 1986, the rate was 40% of the
personal rate because of the 60% exclusion. In 1987 and 1988,
the capital gains and.personal tax rates were identicall Data
on the highest marginal rate paid by adjusted gross income
class were not available for 1987 and 1988. These amounts had

to be imputed based on the average taxable income of each

47
adjusted gross income class. The average marginal tax rate

was then calculated using equation (47).

3-2-3- W

The results of the initial regressions are presented in
Table 4. They show that, in four of the twelve sectors
(metal, communications, transportation equipment, and lumber),
the coefficients on the personal income tax rate were
significant and negative.

The coefficients on the inflation measure, CPI, ranged
from 0.02 to 0.20. They indicate that, during times of
inflation, firms slightly increase their debt loads. The
trend variable does not indicate any overall trend in the
debt/capital ratios, but instead emphasizes the differences
between sectors“ In the sectors of refining and lumber, there
was a positive trend in the debt/capital ratios over the
sample period. In the communications and transportation
sectors, there was a downward trend in the debt/capital
ratios.

The results from the basic model were not overwhelmingly
strong. In order to improve my estimates, I attempted a
number of different approaches, including correcting for
autocorrelation, aggregation, and estimating lag models.

One possible reason for the insignificance of many of the
coefficients may be the presence of autocorrelation in the

data. In order to test for autocorrelation, the correlation

118

.80... dance. ¢.¢L_e a a modem uuceau as.» aooc__ n
ucoa» uxouc_ ou.ca Loesacou use c_ vac-go acoocoa u _au unc_ou .au_eeo co one; xou socoaaoa u chug “one; xeu ouecoeaou u «bu "Queue. ucoo_s_o

 

a acetone. co one; xau seepage: a «pa .aooogucoaoa c. uumuo_uouu.p u.o>o. moo we ucoo_*_cuwn aoueu_nc_ c u.o>os xmo us acao_v_copa noueo_ucm ac
nmN.~v ANN.v A—a.nv arm.v now.v Aem.pv
mo. oo.w aeno.. no. team. or. «P. 0¢.- moum>eom
an6.nv AOm.wv nop.ov aqw.pv Aop.~v A¢~.Pv
no. nh.P cano.. no. on. eOM.- eeo.P- nQ. gonna
Aoo.wv Ao¢.~v Acm.nv new.v Anh.v Anv.wv
oo. aeoo.¢ No. stow. sann.. no. no.- 95v.- modumgo> Laue:
“on.~v nm¢.nv «mo.~v Am¢.Nv Aop.~v Amm.~v
mo. et0.N «ewe. temp. acne. acne. can.« ae~o.u 500534
8.}... 2%: 3%.: at: $3 5.8 8:955“
ON. eeon.~— eemo.. cp.u sam¢. mm. kn.- sea—.9- compoueonocoL»
ah.~v Acm.Pv Aoo.nv Ao~.v Aco.Pv App.v
no. eNF.N echo. oo. seen.. 00. an. no. acycmwox
mop.v a¢0.pv “No.v Aeo.v ANM.FV am.pv
«o. 00." ~06. so. no. n~.- ¢.r- no. 090000» a team
n~¢.Pv Ann.wv Aom.nv “No.v Amh.pv Am~.v
Fo. eao¢.n «we.. «0.- «the. «v.1 aoQ.- 0o. once»
Aoc.cv A0.-u Ace..wv anh.pv a—Q.v Ahn.nv
up. eeu~.¢~ oanoo.- two. No.. .00. mc.- sen~.- comuuumcaaaou
«ca.wv An¢.nv Aho.pv Aoo.~v num.pv Aoa.nv
no. eeu~.pp one. stop. nee. sewn. no. eeoo.- dawn:
«00.0.v Ao~.~v “kn.0v «Nm.pv A0¢.OV “om.0v
ma. ee-.m '00.- seep.o «0.0- o¢.o «Q. o-.- madmaxoh
nNM.Ou Ah—.pv Aho.wv acm.0v Amm.oV mop.0v
No. no.0 n6.o no.6 eaoo.o ow.o- hm.o. no.0 doouiusu
«a b<hmuu 3:05» uau mud<m choc «bu aha ecuuom

owuou Huge—wofisov PEG a a.“ canoauoar usovsoeoa
40.323 gunaouuod H3923

a 0.3mm.

49
factor for each individual firm's residuals was calculated
over the sample period. The correlation coefficients were

calculated as

tic
A c-2 ei,c91,c-1
pi: all (20/19) (48)

tfo
e e
t-l 1, t: 1, t

 

for each firm. Table 5 presents the average p for each
sector; The estimates of the p's ranged from .47 to .60. The
estimated p's were then used to transform the individual firm

data in the following manner:17

Yic'61Y1,c-—1=X1c'6ixi.t—1' 1‘9)

The subscript i refers to the individual firm.

Regressions were then run on the transformed data set,
which now contains only 19 observations for each firm. The
results of the regressions, after correcting for
autocorrelation, are reported in Table 6. After correcting
for autocorrelation, the personal tax rate was still
significantly' negative. in 'the same four sectors (metal,
communications, transportation, and lumber). Correcting for
autocorrelation also added one more significant estimate to

the corporate tax rate. The metal sector's coefficient

 

17 Again, this transformation is performed on all of the
data, and not just the X and Y vectors.

50
estimate is now significant, demonstrating an elasticity of
0.73. The autocorrelation correction also reduced the
magnitude of the elasticity in the lumber sector from 1.5 to
1.02.

The results for the capital gains tax rate are
interesting. The capital gains tax no longer demonstrates a
significant effect on the debt/capital ratio of the
communications sector. It does, however, have a significant
effect on the textile sector. F-tests on the vectors of tax
rates indicate that only four of the sectors' debt/capital
ratios are affected by the tax rates as a vector. This is a
reduction of four sectors from the initial results. This is
due to some of the explanatory power of the autocorrelation
initially being attributed to the tax rates. The
autocorrelation correction removes this.

The strong results for the metal sector, the largest
sector in terms of observations, may indicate that, by
examining a more aggregated view of the economy, one may get
more significant results. I aggregated the 12 sectors on
which the results have already been reported into six sectors:
Transportation and Communications, Paper and Wood Products,
Non-manufacturing, Metals and Fabrication, Chemical and
Refining, and Household Products. Regressions were run with

and without correcting for autocorrelation. Aggregation

51

 

Table 5
Average Autocorrelation Coefficients by Sector
Sector . _AELQLQQLD— MW
Food and Tobacco .56 .19
Textiles, Apparel, and

Leather .49 .83
Paper and Printing .60 .18
Petroleum Refining .55 .22
Chemicals, Rubber, and

Plastics .58 .23
Lumber, Furniture, Stone,

Clay, and Glass .48 .24
Metals, Machinery,

Instruments, and Misc.

Manufacturing .56 .22
Transportation Equipment .57 .12
Motor Vehicles .47 .23
Transportation, .55 .22

Communications, and Utilities
Wholesale and Retail Trade .57 .23

Services .50 .21

52

.aodoa .eacce u.sa_m e a modem “ocean oa_u caoc_. a
peach “some. oo_ca cuesacou ogu c. accuse acooaoa u _au uncmao .euwaoo co ouoa xeu .ocoaaoa a «hug non-a x0» ouacoacou n «bu mucous. uc0u_>_u

 

a unucouc_ co one; x.» .ac_ocex a a»; .ooaozucocaa c_ nowua_uaua-» u.o>o. goo ua uceu_e_cu_u nouoo_uc. c u.o>o. xmo an acou_»_co.o nouoo_uc_ cc
“~0.pv Rom.v “on.bv aon.v aen.v “00.0
00. ow.w e~0.- ~0.- aawu. N0. 0N. n¢.- mauu>eom
«ms-Nu Ane.wv abo.nv Aow.Pv ANN.FV as~.wv
N0. 50.? ee~0.- m0.- eecm. 0w.- s~o.- 0n. canoe
awn.» aho.~v Amo.—v a¢¢.v “00.0 “mo-v
no. 00.— F0. econ. cop.- m¢.- w0.p- vn.- modu_co> Laue:
Aph.v A00.~v A0~.¢v nm0.Pv Aoo.wv n-.~v
00. m0.— 000. stop. MN. «mm. e~0.P seno.- cones;
5.8 83 3nd :3 8m; 2:. 5 85.633
00. ecum.¢ cae0.- ~0.- sane. 0w. m¢.- ewh.- comu0ucoamcoc»
Amn.Nv «Fm.Pv AQF.¢V n0¢.0 Ao~.Fv “op.v
m0. 00.w ee~0. 00. sewn.- 0'. no. 00.- ocmcmwum
«nu. .anh.Fv “on.~v “no.0 Amn.pv no¢.v
m0. ho.- 000. ako. «sew. o-.- «0.- m—. ouoonoh a nook
ans. amm0.0 Ah¢.mv nmh.v nhw.v 500.Fv
F0. n~.w Q00. «00.- .600. 00. N0. 0«.- coach
Ann.~v “50.nv abn.nv AOM.v awo.v nhp.~v
no. and «£30.- eemwo. .53.- ..0. N0: tsp...- cannot-£300
amw.nv ~nm.nv n0¢.v AoF.~v “0F.~v «om.cv
n0. econ.0 compo. aeww. caMN. p90. sank. seNN.- douuz
a00.wv A04.~v Am0.0 ~00.Fv “00.90 “00.0
M0. enN.~ ~w0. nu. ~0.- awe. 00. 0m.- mudmuxoh
and. and.pv Anm.w0 “on.v Aco.v n0¢.v
F0. smo- 000. Como. N00. 0e0. ON. 0F.- doumlucu
«a hS—m-m “3...; ~00 mud; «.50 5.0 Eh LOuuom

Owuou Houunoo\unom m.auwm o a“ mammauo> unocsoaon

moduuaouuooouad How usauoouuoo Hound uuasaoa soduaouwoa
o mmxama

53

resulted in less significant coefficient estimates. In
addition, only two of the F-statistics on the vector of tax
rates were significant at the 90% level. These results
indicate that controlling for sectoral differences is
important.

In addition to the correction for autocorrelation and
aggregation, one- and two-period lag models were estimated

for every sector. These lag models were constructed as

Yit = B1X1c+pzxi,t-1+a1T+V1c (50)

for a one-period lag. Construction of the two-period lag
model was similar. The results of these regressions indicate
that current values of the tax rates have a larger impact on
corporate debt/capital ratios than lagged values. This result
is consistent with earlier work by Auerbach (1985).

Auerbach uses a panel data set on 200 corporations,
covering 1958-1977, to examine the speed of adjustment of
corporate leverage to changes in the corporate tax system. A
partial adjustment model is used to measure the impact of
changes in depreciation allowances on structures and
equipment, as well as variance and.growth.rates of earnings on
short- and long-term.debt. .Auerbach finds that firms are very
quick. to adjust. their' debt load. as changes in ‘the ‘tax
structure occur. Because Auerbach is more concerned with the
speed of adjustment than with the magnitude of change, his

results are not directly comparable to mine. Auerbach also

54

examines the effect of net loss carryforwards on corporate
debt policy. A firm which has a large carryforward may be
exempt from the corporate income tax. This may reduce the
incentive for a firm to use debt as a way to reduce its tax
liability, because the use of debt has no impact on a firm's
current taxable income, which is already zero. He finds that
the existence of net loss carryforwards does reduce a firm's
use of leverage.

In order to account for this finding in my model, the
data set was reduced by eliminating all firms which possessed
a net loss carryforward during the sample period. This
reduced the sample size significantly to 482 firms. .A new set
of regressions was run with these data. The results did not
differ substantially from the original regressions. The signs
and magnitudes of the coefficients in each sector were very
close to their values before adjusting for net loss
carryforwards. Therefore, these results are not presented

here.

3.3. Conclusions

The attempts at estimating the impact of changes in tax
rates on debt/capital ratios have shown that firms do not
exhibit a large deal of sensitivity to changes in statutory
tax rates. The most encouraging result is that the strongest

results in terms of significance and magnitude come from the

55
largest sector, the metal sector with 209 firms. The F-tests
indicate that the interaction of the personal and corporate
tax rates do affect the debt/capital ratios of the firms in
those sectors. The results also appear to point out the need
for further research in the area of empirical estimation of
the effects of taxation on corporate leverage decisions. The
results from this section may be used as base case parameter
estimates in the simulation model. Sensitivity analysis can
then be used to examine the effects of changes in firm

behavior on the tax integration proposals.

Chapter 4

DESCRIPTION OF THE SIMULATION MODEL

4.1. Introduction

The purpose of this section is to describe the CGE model
which is used to simulate the effects of corporate tax
integration in a dynamic framework. This model is a modified
version of the CGE model described in Ballard, Fullerton,
Shoven and Whalley ( 1985) . It incorporates features of my
model of firm financial behavior, described in Chapter 2, and
also the empirical results of Chapter 3. Because my model does
not focus on the inter-generational effects of corporate tax
integration, I choose to use an infinite-horizon model of
consumer behavior, rather than an overlapping generations
model. A. model incorporating infinite-horizon consumer
behavior is described in Ballard and Goulder (1985) and

Ballard (1990).

4.2. The Consumer's Problem

The consumer in the model strives to maximize lifetime

utility. The consumer is assumed to have an additively-

separable lifetime utility function of the form:

56

5'7

.. I 1
v‘ 23 W Wt" ‘5"

p

where p is the subjective rate of time preference and Ct is
consumption in time t.

An isoelastic utility function is chosen in order to
insure tractability. The consumer's instantaneous utility can
be represented as

- (Ct_Cct)1-O

‘ 1-6 ’ 5’1

 

(52)

Ut=ln(Cc-Cc‘) , 6:1,

where 6 is the inverse of the elasticity of substitution
between consumption in adjacent periods. Consumers have a
minimum amount of required consumption in period t of C*t.
The existence of a minimum required level of consumption may
reduce the excessive intertemporal sensitivity that sometimes
characterizes models of this type. See Starrett (1982).
Combining equation (51) and equation (52) yields the

consumer's lifetime utility function:

 

~ - ea 1 (Ct—Cct)1'5 53
U"; (mm-1 1-6 ' ( ’

58

It is important to note that p is assumed to be constant.
This assumption is necessary in order to make the consumer's
behavior dynamically consistent in the sense of Strotz (1955) .

The consumer maximizes his lifetime utility subject to a
budget constraint, which insures that the present value of
lifetime consumption is less than or equal to the present
value of lifetime wealth. The consumer's lifetime budget

constraint is given by

 

 

t

“1 H (hrs) " I] <1+r,>

8'1 8'1

" pth " thti-Yt
SW I
2 c 1 + Z... (54)

where pt is the price of consumption in period t, r is the

8
rate of return in period s, W1 is the value of initial
capital, wt is the wage in period t, Ht is the hours of labor
supplied in period t, and itt is the value of transfers
received in period t18. Note that there are no borrowing
constraints in this model.

I-It is specified exogenously for each period. It is

specified to grow at a constant rate E. This rate accounts

 

18 The wealth of physical capital in this model differs
from that used in earlier versions of the GEMTAP model. The
capital the consumer owns is financial capital. The consumer
is assumed to hold both corporate debt and equity. Because
the after-tax rate of return is equivalent on both
instruments, for a given set of taxes, the consumer
demonstrates no preference for either. This allows firms to
adjust their debt/capital ratios freely in response to changes
in the structure of the tax system.

59
for both population growth as well as productivity growth,
because labor is specified in efficiency units.

The model of consumer behavior will be well behaved as
long as E < r, the rate of return which prevails in the long
run. If this were not the case, the present value of wealth
would be infinite, and there would be no effective constraint
on lifetime consumption.

A Langrangean can be formed from equations (53) and (54)
with Ct as the choice variable. The Langrangean yields the

following first-order condition for consumption:

 

 

6L 1 . -a Ap-
———:-——————— Cf-C = .
30. (1+p)*1( ‘ J t (55)
H (1+r,)
8-1
This marginal condition is rearranged as follows:
AP (1+p )“1
C _Cs '0: t t .
( t C) (56)

C
H (1+r3)

8-1

One can then examine the relationship between consumption

in any two adjacent periods:

(CC- Cc‘)
(Cc-1'Ct-1‘)

Pt-l (1+Ic)1'i'
Pc (1+p)J

 

=[ (57)

 

Equation (57) can be solved recursively to yield an

equation of motion for discretionary consumption. This is

60

1
(Ct-Cc.) = (C1_C10)oc-E' (58’

where

C
P H (l+r,) (59)

= s-i

_l. ,
c P: (1+P)t'1

Discretionary consumption in period one can be found by
substituting the equation of motion into the lifetime budget

constraint. This substitution yields:

as C
2 [(WJIC‘+Yc-ptCt‘)1—I (1+r,) -1]+W1
C1-C1. = t=1 ' I 1 t 8:1 1 (so,
«‘12;th cTH (1+r,)'1.J

8-1

 

Equation (60) can then be used to solve for consumption in
period one, as well as consumption in future periods, given
the time paths for the price of consumption, the wage rate,

the endowments, and the rate of return.

4.3. Producer's Problem

This section describes the incorporation of the essential
features of the ‘model of firm investment and financial
behavior described in Chapter 2. This model has nine distinct

production sectors which are constructed as aggregates of the

61

original nineteen sectors of the GEMTAP model described in
BSFW (1985) . The aggregation is done using the capital stocks
of the sectors as weights. Table 7 lists the production
sectors and the GEMTAP components used in their
construction.19

The production sectors are classified as corporate,
noncorporate, or owner-occupied housing, for the purposes of
tax treatment. This classification makes it very easy to
follow the flow of capital between corporate and noncorporate
sectors. This model does not examine the effects of sectors
containing both corporate and noncorporate firms.20 The
production sectors, with the exception of the owner-occupied
housing sector, are classified as corporate or noncorporate
based.on the proportion of the total capital stock within each
sector that is held by corporate and noncorporate entities.
Sectors where less than 68 percent of the capital stock was

incorporated were placed in the noncorporate sector. Although

68 percent is a large cutoff, it was chosen to insure that

 

19 The ninth production sector in this model is
government enterprises. This sector functions exactly as it
does in the earlier versions of the GEMTAP model. For this
reason, it is excluded from the description of the production
sector.

20 The interaction of corporate and noncorporate firms
within the same sector has been investigated by Gravelle and
Kotlikoff(1989) . They find that taking into account the
existence of corporate and noncorporate capital in the same
sector is important. As noted earlier, their results may be
due to aggregation bias, or other factors besides the
existence of both corporate and noncorporate capital within
the same sector.

62

there would be several production sectors which were
classified as noncorporate. The implication of reducing the
required percentage of noncorporate capital is explored as
part of the sensitivity analysis. Data on the breakdown
between the corporate and noncorporate capital within each
sector were obtained from Fullerton and Henderson (1987) .
Table 8 presents data on the capital stocks and the division
between corporate and noncorporate ownership.

Production in each sector is specified by a constant

elasticity of substitution value-added function of the form:

0-1 0'1 0

VA=¢lnLT+(1-H)K7]fi. “1’

where ¢ and u are sector-specific production parameters and a
is the elasticity of substitution between labor and capital.
Each production sector produces a single output under
constant-returns-to-sca1e technology. Producers face gross—
of—tax input prices of PL" and PK". Each firm minimizes input
between the corporate and noncorporate capital within each
sector were obtained from Fullerton and Henderson (1987) .
Table 8 presents data on the capital stocks and the division
between corporate and noncorporate ownership.

Production in each sector is specified by a constant

elasticity of substitution value-added function of the form:

0-1 0-1 a

VA=¢[|JLT+(1—|1)K—°_)°_‘1, “2’

PRODUCTION SECTORS

EBQDQQIIQE_§E£IQB§
PRIMARY (1)
(2)
(3)
CONSTRUCTION (4)
NONDURABLES (5)
(6)
(7)
(3)
(9)
DURABLES (10)
(11)
(12)
(13)
TRANSPORTATION, (14)
COMMUNICATION, AND

UTILITIES
TRADE (15)
OWNER-OCCUPIED (16)
HOUSING (17)
SERVICES (18)
GOVERNMENT (19)

ENTERPRISES

63
TABLE 7

USED IN THE SIMULATION MODEL

GEMTAP CQMEQNEHIS

Agriculture, Forestry, and
Fisheries

Mining

Crude, Petroleum,and Gas

Construction

Food and Tobacco

Textiles, Apparel, and Leather
Paper and Printing

Petroleum Refining

Chemicals and Rubber

Lumber, Furniture, Stone, Clay,
and Glass

Metals and Machinery
Transportation Equipment

Motor Vehicles

Transportation, Communication,
and Utilities

Wholesale and Retail Trade

Real Estate
Finance and Insurance

Services

Government Enterprises

64

where o and n are sector-specific production parameters and a
is the elasticity of substitution between labor and capital.
Each production sector produces a single output under
constant-returns—to-scale technology. Producers face gross-
Of-tax input prices of PL’" and PK". Each firm minimizes input
costs per unit Of output subject to the value-added
constraint.

A Lagrangean can be set up with K and L as the Choice

variables.

0-1

_ o_-1 A
EE=P;K+pU‘L+1[¢()-L ° +(1-p)K a ) ad.” (63)

TABLE 8

Breakdown Of Capital Stock Classification by Sector

INDUSTRY CORPMTE NONCORPORATE HOJS 1 110 TOTAL PERCENT
CMPCRATE
Primary $88,274.80 $444,606.10 0.00 $532,880.90 16.57
Construction 13,410.74 6,458.87 0.00 19,869.61 67.49
londurables 158,272.50 2,843.17 0.00 161,115.60 98.24
Durables 199,882.40 3,485.08 0.00 203,367.50 98.29
Transportation, 315,647.60 24,586.53 0.00 340,234.10 92.77
Communication,
8 Utilities
Trade 132,785.80 32,384.05 0.00 165,169.90 80.39
Owner-Occupied 0.00 0.00 1,660,132.00 1,660,132.00 MIA
Services 66,412.73 45,275.36 111,688.10 59.46
assassssssass 83333333838=8 :==s:==:====: assassssxxsss usages-sag
TOTALS 974,686.57 559,639.16 1,660,132.00 3,194,457.73

* All figures in millions Of dollars

65
The Lagrangean yields the following first-order conditions for
labor and capital:

 

83 £'_1 .142 _L :1

E? P,“ = ~A¢(l—p.) [11L " +(1—p.)K ° ] 1’°K ° (54)
35¢ 11- fl ° ’_1
'5': PL‘ = -l¢uluL ° +<1-u)K ° 1 1"’13 ° “5’

Dividing 'the first-order' conditions 'yields ‘the following

marginal condition:

_‘_1
PK.= (1-”)K a .

p: 2.
L 11L"

1“)

 

This condition can be solved for capital or labor and then
substituted back into the constraint to yield the following

per unit factor demands:

 

 

DK=¢-1[I1{ (1‘11) PL }1-°+(1’1‘)] 1-a (67)
l4 P“
and
D =<b'1[ (1-p.) { Ll PX. )1‘°+u] 13" . (68)
L (l-u) PL‘

66

The tax treatment of the production sectors is very
important. The tax policies that are simulated involve
adjustments to both the corporate and individual marginal
income tax rates. This model allows a more complete
specification Of how Changes in a marginal tax rate will
affect the firm. In earlier versions Of the GEMTAP model, the
average tax rates on capital were assumed to equal the
marginal tax rates on capital.21 The tax rate on capital was
calculated by taking the total tax revenue attributed to the
sector's capital stock as a proportion Of the total capital
stock in the sector» The advantage of this method is that the
use of the average tax rate will yield the correct amount Of
revenue, i.e., the amount Of revenue raised.through the use of
the average rate will more closely approximate the amount of
revenue raised in reality. In my model, the marginal tax
rates are used tO calculate the average tax rate. When one
uses marginal effective tax rates that derive from the
statutory rates, the revenue raised will differ from the
amount Of revenue raised in the actual U.S. economy. The
disadvantage Of the standard GEMTAP approach is that it does
not correctly measure the effect Of a tax rate change at the
margin.

Many public finance economists have done calculations for

"marginal tax rates." Some Of the best work has been done by

 

21 An exception is the version GEMTAP used by Fullerton
and Henderson (1989).

67

Jorgenson and Yun (1986) and Fullerton (1984)22. These
authors use the cost-Of-capital approach, and I adopt that
approach for my model. A program for introducing a cost-Of-
capital formulation based on marginal tax rates was
constructed by Fullerton and Henderson (1987,1989). In their
model, the effective tax rate on capital (tn) is calculated as
the difference between the pre-tax rate Of return on capital
(9) and.post-tax rate of return on capital (3) as a percentage
Of the pre-tax rate Of return.

‘31: = _E_'_5 (59)
p

Their model uses a Hall-Jorgenson (1967) user cost Of
capital formulation. A firm pays q(1-k) for a new piece Of
capital, where q is the price Of capital, and k represents the
investment tax credit. My formulation differs slightly in
that firms face adjustment costs Of’Q per unit, which they are
able to expense. The after-tax adjustment cost per unit then
is qO(1-—u) , where u is the statutory tax rate. The firm
receives an annual gross rate of return on that piece of
capital Of c. This rate Of return is reduced annually by 6,
the constant exponential rate Of depreciation. The firm is
required tO pay annual property taxes at a rate of w on the
value of the piece Of capital at any point in time. The

return net Of depreciation and property tax is then taxed at

 

22 It is important tO note that these calculations
abstract from risk.

68
the statutory tax rate. The firm is able to take a
depreciation deduction of 6g; the present value of those
deductions is represented by z. The after-tax return on a
piece of capital is then discounted at the firm's after-tax
discount rate of r. In equilibrium, the net outlay for the
piece of capital must exactly match the present value of the
return from the capital. This equilibrium condition is

represented by equation (69):

q(1-k)+q¢(1—u) =f(1—u) (c-wq) e“"e‘“dt+uzq. (70)
0

The user cost Of capital pc is found by integrating equation

(69) and solving for c/q. The user cost of capital is

specified as:

pc== i:: (1—keuz)-+vv- O+(O+I)¢. (71)

 

Fullerton and Henderson use data on 38 different assets
in 18 production sectors, with data on both corporate and
noncorporate capital in all sectors. They use asset-specific
values for 6, k, z, and w. They also use sector-specific
discount rates, r. A common value Of i, the nominal interest
rate, is used in all sectors. In the current version Of the
GEMTAP model, there is only one capital good. However, in
order for my model to partially capture the impact Of the

differential tax treatment Of assets, all assets within a

5!

gr

the
sec
inc
bas-

cor;

69

sector are aggregated into a single capital good for each
sector. These composite assets contain the weighted average
tax characteristics of the individual assets within the
sector. It is implicitly assumed that the asset weights
within a sector remain constant throughout the course of
simulations. Table 9 contains a listing of the different
types of assets as well as their exponential depreciation
rates, investment tax credit rates, and property tax rates.

All assets in this economy are required to provide a net-
Of-all-tax rate of return (s) of 0.08.23 If i is the nominal
interest rate on a bond, then s would equal i(1-6) , where 0 is
the individual's marginal income tax rate. Because s is
specified exogenously, it is easy to calculate i as simply s
grossed up by one minus the individual's income tax rate or,
i = s/(l-O). Once the nominal interest rate is known, one can
then calculate each sector's requisite discount rate. Each
sector's discount rate is a‘weighted average of the portion Of
income taxed as interest on debt (deductible from the income
base) and the portion taxed as dividends (which is taxed as

corporate as well as personal income).

 

23 In the standard GEMTAP model the net of tax rate Of
return is equal to 0.04. However, that model abstracts from
depreciation. If the rate Of return from capital is not
increased, in the presence Of depreciation, there would be no
positive growth in the capital stock.

Assets and Their Tax Parameters

Asset
(1) Furniture and Fixtures
(2) Fabricated Metal Products
(3) Engines and Turbines
(4) Tractors
(5) Agriculture Machinery
(6) Construction Machinery
(7) Mining 8 Oil Machinery
(8) Metalworking Machinery
(9) Special Industry Machinery
(10) General Industrial Equipment
(11) Office and Computing Machinery
(12) Service Industry Machinery
(13) Electrical Machinery
(14) Trucks, Busses, 8 Trailers
(15) Autos
(16) Aircraft
(17) Ships and Boats
(18) Railroad Equipment
(19) Instruments
(20) Other Equipment
(21) Industrial Buildings
(22) Commercial Buildings
(23) Religious Buildings
(24) Educational Buildings
(25) Hospital Buildings
(26) Other Monfarm Buildings
(27) Railroads
(28) Telephone 8 Telegraph
(29) Electrical Light 8 Power
(30) Gas Facilities
(31) Other Public Utilities
(32) Farm Structures
(33) Mining, Shafts, 8 Wells
(34) Other Monbuilding Facilities
(35) Residential Facilities
(36) Inventories
(37) Monresidential Land
(38) Residential Land
Source: Fullerton and Henderson (1987)

70
TABLE 9

(1983)

Depreciation Rate
.110
.092
.079

.165
.118
.254
.333
.183
.075
.066
.150
.150

.025
.019
.019
.023
.045
.018
.033
.030
.030
.045
.024
.056
.029
.015
.000
.000
.000

Property Taxes

.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.00768
.01126
.01126
.01126
.01126
.01126
.01126
.01550
.01550
.01550
.01550
.01550
.01126
.01126
.01126
.01837
.00768
.00000
.00000

Tax Credit
.10
.10
.10
.10
.10
.10
.10
.10
.10
.10
.10
.10
.10
.10
.06
.10
.10
.10
.10
.10

8'88'888

71

In this model, I assume individual arbitrage. The
consumer in this model is assumed.to hold.debt and.equity from
all sectors and.to arbitrage away any differences in the rates
Of return” In the corporate sector, the discount rate on.debt
is simply i(1-u), where u is the corporate income tax rate.
The discount rate on dividends is 1, because dividend income
is not deductible from the corporate income base. In this
model, there are no retained earnings, so the entire return is
subject to personal taxation. The firm's total discount rate

is the weighted average of these two terms:

 

D D .
PKK+ (1 WU”. (72)

I = [(l-u) *
Here u is the marginal corporate income tax rate, D/PKK is the
firm's debt/capital ratio, and i is the nominal interest rate.
The same weighted average calculation is applied to the
owner-occupied housing sector and the noncorporate sectors,
except that they apply their marginal tax rates of th and tnc,
respectively.24
In order to calculate the discount rates, it is also
necessary to know the debt/capital ratios Of each sector; For
the corporate sectors and the noncorporate sectors, the base
case debt/capital ratios are taken from the empirical

estimation Of Chapter 3. Information on debt/capital ratios

 

24'The tax rates in the owner-occupied housing sector and
the noncorporate sectors are taken.to be a fixed proportion.of
the marginal tax rate on debt. The factors Of proportionality
are taken from Fullerton and Henderson (1986).

t1
cc

as

72

from all Of the sectors included in Chapter 3 was used. The
debt/capital ratios are aggregated down to the final
production sectors used in the simulation model, using the
GEMTAP capital stocks as weights. For the housing sector, the
primary sector, and the construction sector, data on
debt/capital ratios are taken from Goulder and Summers (1989) .
Equation (74) demonstrates the ability Of firms to adjust
their effective tax rates by adjusting their debt/capital
ratios.25

The owner-occupied housing sector, the primary sector,
and the construction sector are assumed to maintain constant
.debt/capital ratios throughout the course Of simulations.
This is necessary due tO lack Of empirical information Of
their financial behavior in response tO the tax system. All
corporate sectors and the noncorporate service sector are
assumed to adjust their debt/capital ratios in response to
Changes in the corporate and personal tax rates. The effect
of differing tax changes is simulated using the empirical
elasticities from Chapter 3. Sensitivity analysis is used to
examine the impact Of differing degrees Of responsiveness Of
debt/capital ratios in response to changes in personal and

corporate tax rates.

 

25 Note that equation (68) points out that a firm could
alter its effective tax rate by adjusting its mix Of assets
and taking advantage Of the differential tax treatment of
assets. The incorporation Of such behavior is, however,
beyond the scope of this paper.

73
Once the discount rate is calculated, one uses equations
(69) and (70) to solve for the effective tax rate on capital

in each sector.

4.4. The Government Sector

In addition to the consumers and the production sectors,
the model includes a government sector. The government
collects taxes on labor, capital, outputs, income, and
consumer purchases. The government uses its revenue to
purchase goods, to make transfer payments to consumers, and to
subsidize government enterprises. Government enterprises is
modeled as a separate production sector. In this model, the
government is constrained to maintain a balanced budget in
every period. This model does not allow for governmental
borrowing. Some other models have allowed for the use of
government debt, for example, Auerbach and Kotlikoff (1987).
The level Of governmental transfers is held constant, in real
terms, during simulations through the use Of a Laspeyres.price
index. Governmental exhaustive expenditures are held
constant, in real terms, by using the expenditure function
from a Cobb-Douglas utility function. In both the base case
and revised case, government expenditures and government
receipts will grow at the exogenously specified rate Of
growth, 6. The revised case simulations involve Changes in
both the individual as well as production sector tax rates.

In order to maintain government budget balance, tax revenue

74
must also remain constant. If a tax policy results in a
decrease in tax revenue from one source (corporate income
taxes for example), another source of revenue (consumers'
marginal tax rates for example), must be increased to maintain
a balanced budget. The two sources of revenue replacement
used in these simulations are the imposition of a lump-sum tax
on consumers or an increase in the consumer's marginal income

tax rate.

4.5. General Solution to the Model

At the beginning Of each period, eleven prices are called
out. They are a price Of labor (the numeraire), a tax price,
and nine prices of capital, one for each production sector.
Based on these prices, the consumer selects his level of
saving and consumption in that period. The level of
consumption and saving are based not only on current prices,
but also on his expectations Of future prices, which determine
his stream Of future income as well as future consumption
expenditures. In this model, the consumer is assumed to
Operate under myopic expectations. He believes that current
prices will hold for all future periods. Ballard (1987), for
example, has shown that.the structure:Of expectations does not
make much difference for simulations like those undertaken
here. However, Auerbach and Kotlikoff (1987) and Judd (1985)

suggest that the structure of expectations may be important if

e;
mc
sa
st
fOJ
amc
adj
equ.

Cons
tax
for
intei
rAtes
Perio
and 11

“Sing

75
announcement effects are involved, or if the policy Change is
only temporary in nature.

An important feature Of this model is the ability to
transform the input prices of capital and labor into
intermediate and final goods ‘prices, based on a fixed-
coefficient matrix. This is made possible through the
application Of the Samuelson Substitution Theorem (1951) .
Hence, consumers are able to convert factor prices into final
consumption prices.

The investment gOOd in this model differs from saving in
earlier versions of GEMTAP. In earlier versions of the GEMTAP
model, the entire amount Of saving was net saving, so that all
saving translated directly into additions to the capital
stock. In this model, the level Of saving accounts not only
for net additions to the capital stock, but also for the
amount Of depreciation in each sector, as well as the
adjustment costs Of purchasing and installing new capital
equipment.

At the same time that the consumer is making“ his
consumption decisions, firms, based on the input prices and
tax rates, are selecting their levels Of output and demands
for labor and capital, as well as their demands for
intermediate goods. Firms first examine the marginal tax
rates to see if there has been any change from the previous
period. If there has been a change, firms in the corporate
and noncorporate sectors adjust their debt/capital ratios

using the empirically estimated elasticities from Chapter 3.

ec
it
se
eq~
la}
the

4.6

0rde
Pric-
initj
The a
It a)

76
Based on the new debt/capital ratio and current prices, firms
are able to calculate their effective tax rates. The
effective tax rates are then used.tO calculate their levels Of
production. The levels Of output in each sector are converted
into demands for intermediate goods based on a fixed input-
output matrix.

The model is solved using a modified version of the
tatonnement algorithm of Kimbell and Harrison (1986). Prices
are called out, and excess demands for labor and capital are
then evaluated. If there is excess demand in any market,
prices are adjusted and the process is repeated. Equilibrium
is achieved when the total demand for labor in the economy
equals the total supply, and when.the total demand for capital
in each sector is equal tO the total supply Of capital in each
sector. Also, the government budget must be balanced. Oruze
equilibrium is achieved in a given period, capital stocks and
labor endowments are augmented.and.the process is repeated for

the next period.

4.6. Calibration

The base case simulations in this model are calibrated.in
order to insure a steady-state rate Of growth E, with unitary
prices for all inputs“ The growth rate is formed based on.the
initial level Of saving and capital contained.in the data set.
The assumption of unitary price in the base case is important.

It allows one to break down the data, reported in dollar

77
values, into prices and quantities, and also simplifies the

algebra of calibration.

4.6.1. W

In the base case, the capital stock in each sector is
required to grow at the growth rate Of E. Each sector faces
a sector-specific rate Of depreciation, as well as an
exogenously specified amount of adjustment costs per unit of
investment, 4. This requires that the gross rate Of
investment must.be enough.to provide net growth to the capital
stock at the steady-state rate. The gross rate of investment

then can be specified as

71540-01, (73)

where 6 is the sector-specific exponential rate Of
depreciation. Given the initial capital stock in the sector,
E, and 9, one can solve equation (72) for I, the gross level

Of investment.

 

(74)

Parameters Of the production function must also be
calibrated. It is assumed that firms act to minimize cost and
that they make zero profits. Using the CES value-added

function from equation (62), along with the assumption Of cost

78
minimization, one can use the first-order conditions Of
equations (63) and (64) to solve for equation (65). Equation

(65) can be rearranged to solve for u:

_1 _1
P ‘L "/p ‘K °
p.= ’- K . (75)

.1
1+PL’L ° /P,;K

 

QIH

In the base case, the net-Of-factor-tax prices are required to
equal unity. Using this assumption, P*K can be rewritten as
(1+tx) , and P‘"L can be rewritten as (1+tL) , where tK and tL are
the tax rates on capital and labor respectively. Equation

(74) can then be rewritten as

1 _1
p: (1+tL)L /(1+tx)K . (75)

.l
1+(1+tL)L °/ (1+tK)K

 

619

Information on tx, tL, K, and L are contained in the benchmark
data set" The elasticity of substitution is specified
exogenously. Given u and a, one can then use the zero profit
assumption tO solve for ¢.

Firms in this model demand not only labor and.capital but
also intermediate goods from.other industries. ‘Thus the zero-

profit condition and unitary prices imply:

o~1 o-1

(1+tL)L + (1+cK)K + A=¢[pLT+(1_p)K—r] 7.1, (77)

79
where A represents the vector of intermediate inputs.

Rearranging yields:

4’: (1+tL)L + (1+tk)K + A
£*_1 °_-1 _0_ ' (73)
[FL (I +(1-Il)K O ] 0-1

 

4.6.2. W

The consumption and savings decisions of the consumer
must also be calibrated to return the values Of consumption
and saving contained in the benchmark data set and also
provide for steady state growth, with unitary prices.

Recall that equation (60), reproduced below, is used to
insure that the consumer's initial level Of consumption

replicates the benchmark value:

as C

2 [(thtNYt-ptct‘) H (1+r3) '1}+W1
CI‘CI. = t=1 . f 1 c 831 1 - (79)
2 p50 {SH (1+rs)‘1j

C-1 8'].

 

where

C
P1 H (1+r8) (80)

881

Qt=— t_1 .
Pt (1+9)

 

80
The first parameter that is calibrated is the consumer's
subjective rate of time preference, p. The equation Of motion
(58) implies that, in the steady state

( 1+1F) %=1+E (81)
1+p '

 

where E is the steady-state growth rate, f is the steady-state
rate of return, and 6 is the inverse of the elasticity of
substitution between consumption in adjacent periods. 'Values
for E and f can be Obtained from the benchmark data set.

Rearranging equation (80) yields the base-case value for p:

zfifld
p n+5)“

(82)
The benchmark. value is also) given for first-period
required consumption (C*). C* is chosen at the value used by
Ballard and.Goulder (1985). Required consumption is set at 80
percent Of first period consumption, and 6 is equal tO 1.005.
This is done to reduce the sensitivity Of the model to large
intertemporal swings in consumption. Sensitivity analysis is
performed on the required consumption ratio in order tO
examine its effects on the welfare gains from changes in tax
policy. It is assumed that required consumption, as well as
transfers, grow at the steady-state growth rate. Using this

fact, one can.solve equation (78) for first-period.consumption

(C1):

81

fie
1+?)- (83)

 

Cl: (H‘+Y1-C'1‘) +W1(

In the standard GEMTAP model, this equation would return
the correct level of base-case consumption and saving. In my
model, however, the presence of depreciation and adjustment
costs will lower the value Of initial wealth and lead to a
lower growth rate. The lower growth rate, if one does not
adjust initial wealth, will imply a lower level Of initial
consumption. Using the level Of consumption from the base-
case data set, and the consumption parameters consistent with
the lower rate Of growth, one can solve equation (78) to yield

the correct level of initial wealth, W1:

0 1 t C '
w1 = (Cl-01‘)£{p,0 [5H (1+r8)‘1} - gunman-menu (1+r,)'1
2&1 '

8-1 t-l 8'1

(84)

Using the calibrated parameters and data from the base-
case data set, one can.then run a sequence Of equilibria which
will be used to compare welfare gains in simulations with

changes in tax policy.
4.7. Data

The majority Of data in this model comes from Scholz
(1987). Scholz builds a data set based on 1983 production,

consumption, income, and tax data from various sources. The

82
information on debt/capital ratios and debt/capital
elasticities is taken from the empirical estimation of Chapter
3, and Goulder and Summers (1989) . Information on the
percentage of corporate capital within each GEMTAP sector, as
well as information on depreciation rates, property tax rates,
and investment tax credits, comes from Fullerton and Henderson
(1984) . Table 10 lists base-case data for each production
sector. It is important to note that the use of effective tax
rates on capital reduces the differences between the taxation
of corporate and noncorporate capital. This is an important
factor in explaining the relatively small size Of some of the
welfare gains from the various tax policies simulated in
Chapter 5. The small difference in effective tax rates is due
to the existence Of the inclusion of property tax rates, as
well as the different composition Of assets among the

production sectors.

83

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 10
BASE-CASE DATA
Sector Sector Sector Sector Sector Sector Sector Sector
1 2 3 4 5 6 7 8
Tax .52868 .45990 .57479 .56483 .58933 .57545 .42050 .46489
Rate
Depreciation .111778 .091492 .057371 .066195 .105127 .069429 .026754 .09499
Rate
Investment 0.077 0.092 0.051 0.069 0.078 0.036 0.011 0.052
Tax
Credit
Debt/Capital 0.14 0.17 0.30 0.32 0.47 0.38 0.50 0.47
Ratio
Capitql 28957 7654 44184 12183 50622 38008 2111666 41827
Stock
Adjustment 0.112 0.039 0.012 0.003 0.089 0.002 0.019 0.051
Costs
Per Unit
Key to sectors:
Sector 1: Primary
Sector 2: Construction
Sector 3: Mondurables
Sector 4: Durables
Sector 5: Transportation, Communication, 8 Utilities

Sector 6: Trade

Sector 7: Owner-Occupied Housing

Sector 8: Services

* Capital stocks in millions of dollars.

 

Chapter 5
SIMULATIONS

5.1. Introduction

Simulations were run to investigate the effect of three
separate tax policies. The first plan was full corporate-tax
integration. Under this plan, the marginal corporate tax rate
is reduced to zero, and all income is taxed at the personal
level. Under full integration, all firms are assumed to
accept a zero debt/capital ratio. The bankruptcy theory Of
corporate finance is based on the premise that the tax
advantage Of corporate debt is Offset by the increased
probability of bankruptcy, as a firm's leverage ratio
increases. Under full integration, there is no longer any
advantage to using debt.

The second plan is a reduction in the corporate tax rate.
The statutory corporate tax rate is reduced from .46 to .35.
This is similar to the Tax Reform Act Of 1986. Under this
plan, the value of the interest deduction to firms is
decreased, and firms will seek lower debt/capital ratios.

The third plan is a dividend exclusion from the personal
income tax. Modeling of this plan involves reducing the
required discount rate Of firms in the corporate sector.
Because dividends are not taxed as personal income, the

required pre-tax rate Of return on dividends is now equal to

84

85

S

We The

s. The interest rate on debt still remains 1:

corporate discount rate then becomes 1 = —%4 (1-u) nun-71:) as.

A dividend deduction from the corporate tax base is not
explored. Because firms in this model are assumed to pay out
all earnings as dividends, a deduction from the corporate
income tax base is the same as full integration.26

Simulations were run with two versions of the model, the
nine-sector base model, and a four-sector model. The four-
sector model consists Of corporate, noncorporate, owner-
occupied housing, and government enterprises sectors. Firms
from the nine-sector model are placed in each Of the four
sectors based on their capital stock classification. All
simulations are run under the requirement of revenue
neutrality.

The effects Of changes in tax policy are evaluated by
examining a 40 year period, calculating equilibria spaced one

year apartc In every simulation, by the end Of the forty-year

period relative prices are changing very little. This is

 

26 Bagwell and Shoven (1991) have shown that there has
been a decrease in the importance Of dividends beginning in
the late 1980's. They point to the increase in the use of
share repurchases as a way Of moving income from the firm to
its stockholders. Their empirical evidence seems to call into
question the assumption Of a 100 percent dividend-payout
ratio. TO relax the assumption Of a 100 percent dividend-
payout ratio necessitates the incorporation of retained
earnings as a third source Of finance. It would also require
information on the changes in firms' dividend payout ratios
with respect to changes in the corporate and personal income
tax systems. Due to the lack Of empirical data on firms'
payout ratios, the assumption Of a 100 percent payout ratio is
maintained.

86
important as it indicates the economy has nearly approached
the new steady state. Although equilibria are actually
calculated for only fourty-years, the welfare evaluations are
performed by examining the infinite stream under the
assumption.that the economy has reached.a steady state for all
periods after the actual equilibrium simulations.

Evaluation of the gain or loss associated with the change
in tax policy is based on examination Of the consumer's
equivalent variation as a proportion Of base-case wealth. The
consumer's base-case level of utility is computed from the
base-case simulation, and compared to his utility level which
arises after the policy change. The consumer's endowment of
capital is augmented until the original level of utility is
achieved. The gains from the new policy are calculated as a
proportion Of base-case wealth and by taking the consumer's
equivalent variation as a proportion Of the discounted stream
Of base-case Gross National Product. In addition, the welfare
gains are measured as a proportion Of the capital-tax revenue

in the model.27

 

27 The tax revenue from this model is larger, by almost
22 percent, than the actual revenue generated from capital
taxation. Note, that capital taxes in this model include
state corporate taxes as well as property taxes. In my model
the total tax revenue from capital is $198 billion, compared
to actual capital tax revenue Of $162 billion used in the
original GEMTAP model.

87

5.2. Full Integration Results

Once the full integration plan is implemented, the demand
for capital in the corporate sectors increases dramatically
due to the lower cost of capital. Because capital in this
model is assumed to be sector specific, the prices Of capital
in the corporate sectors rise above their base-case prices Of
unity; In the non-durables sector, the price Of capital rises
to 1.34 in the first period. In other words, it increases 34
percent over its base-case price. The durables sector and
transportation, communication, and utilities sector show
initial increases in their prices Of capital Of at least 17
percent. The initial increases in the prices of capital in
the corporate sectors are quickly reduced as new investment
increases the supply Of capital in those sectors. After 10
years, the prices Of capital in the nondurables and durables
sectors fall below unity. After 40 years, the price of
capital in every sector is below unity. This is the result of
capital deepening in every sector. The price of capital in
the primary sector falls about 3 percent, to .969, in the
first.periodc 'The prices Of capital in the other noncorporate
sectors and the owner-occupied housing sector also decrease in
the first period, since the demand for capital is reduced.

It is important to note that full integration does not
equalize the cost Of capital across sectors. This is because
sectors still maintain sector-specific depreciation rates,

property tax rates, and adjustment costs. The majority of

88
initial investment takes place first in the corporate sectors.
The capital stocks in the owner-occupied housing sector and
the noncorporate sectors initially show very little increase,
due to the low level of investment in those sectors and the
presence of depreciation.

The largest increase in the capital stock, relative to
the base case, occurs in the durables sector, where the
capital stock increased one and one-half times as much as in
the base case. The smallest relative increase in the capital
stock Of an individual sector occurs in the owner-occupied
housing sector, where the capital stock increased about 70
percent as much as in the base case. The increase in the
housing sector is large in absolute magnitude, because Of the
large amount Of capital that is initially located in that
sector. The majority of the increase in the capital in that
sector, however, occurs late in the course Of the simulation.

The welfare gains from the full integration plan in the
nine-sector model were modest. With income-tax replacement
the (gains ‘were $402 billion, in 1983 dollars, which is
approximately .98 percent Of base-case wealth or .78 percent
of base-case GNP. The gains using lump-sum replacement were
slightly larger, 1.1 percent Of base-case wealth,
approximately $463 billion. The gains as a proportion of
capital tax revenue were 15.13 percent under income-tax
replacement, and 17.6 percent under the lump-sum replacement.
These dynamic welfare gains are smaller than those found in

earlier studies. Ballard, Fullerton, Shoven, and Whalley

89

[BFSW] (1985) find dynamic welfare gains of approximately
$1,064 billion, with lump-sum replacement, and $648.74
billion, with the income-tax replacement, in 1983 dollars.
One reason for the lower dynamic welfare gains is the
immobility Of capital in my model. Because the gains to
integration are unable to be realized right away, their
present value falls. In addition, my model has less
dispersion among capital tax rates.

Full integration simulations were also run with the four-
sector model. The welfare gains were lower, approximately
$369 billion with the income-tax replacement and $447 billion
with a lump-sum replacement. These gains are approximately
four to eight percent less than the gains in the nine-sector
model. An important reason for this appears to be that
adjustment costs play a relatively more important role in the
four-sector model than they do in the nine-sector model. This
issue is addressed subsequently. In addition, aggregation of
the production sectors also reduces the dispersion Of capital-
tax rates.

A four-sector model employing perfectly-mobile capital
was used to examine the effects Of the capital-mobility
restriction. The welfare gains using perfectly-mobile capital
were 1.04 percent Of base-case wealth under income-tax
replacement and 1.08 percent Of base-case wealth under lump-
sum replacement, ($426 billion and $443 billion). These gains
are larger than found when the capital mobility restriction is

employed. The gains are not extremely large. The existence Of

90

depreciation in this model lowers the gains from the
reallocation of capital. A large portion of the economy's
capital stock flows out Of the housing sector, where it is
depreciated at a relatively low rate and into the corporate
sector, where it is depreciated at a higher rate. The higher
depreciation rates in the corporate sector lead to a faster
depletion of the economy's capital stock, and therefore dampen
the welfare gains from full integration.

Fullerton (1983) incorporates industry-specific capital
and restricted capital mobility to examine the impact Of full
integration. In his model, he restricts the ability Of Old
capital to leave a production sector after a change in tax
policy. In his most restrictive case, full integration he
results in dynamic welfare gains Of $380 billion, in 1983
dollars. These gains are very close to mine. The welfare
gains in my model are further reduced, due to the presence of
adjustment costs and depreciation. Adjustment costs lower the
amount Of available capital for net investment into the
sectors yielding higher than average rates of return on
capital.

Another set Of full integration simulations was run using
the nine-sector and four-sector models. In this set of
simulations, firms were not able to set their debt/capital
ratios to zero instantly. Instead, firms were assumed to
adjust their debt/capital ratios based only on the estimated
elasticities. The welfare gains from this set of simulations

were larger, under both types Of revenue replacement. In the

91

nine-sector'model, the income-tax replacement simulations led
to welfare gains of $492 billion, about 1.2 percent of base-
case wealth. This was an increase Of almost 22 percent over
the initial full integration results. The welfare gains in
these simulations were approximately 18.7 percent Of capital
tax revenue. With the income-tax replacement, firms are able
to continuously adjust their debt/capital ratios, as the
income tax rates change. The use Of debt allows firms to
lower the cost Of capital further, as their discount.rate will
be lower than in the case where there is no debt, and hence,
no interest deduction. Because I use the estimated
elasticities instead of an explicit bankruptcy-cost function,
firms are able to receive the benefits of the interest
deduction without an explicit increase in the cost of using
debt. The increase in welfare in the nine-sector model, using
a lump-sum replacement, was substantially larger, $603
billion, about 1.5 percent Of base-case wealth. This was an
increase Of almost 30 percent over the original full
integration simulations. In this case, firms adjust their
debt/capital ratios only once, after the initial reduction in
the corporate income tax rate. This type of replacement does
not allow firms tO respond as fully as the income-tax
replacement. However, the lump-sum replacement is a less
distorting means Of revenue replacement.

The gains in the four-sector model, over the previous
four-sector full integration simulations, were not as large as

the'gains based on the nine-sector model. ‘Using an income-tax

 

re
fc
ad
el
th.
In
frc
the

deb-

92
replacement, the welfare results from full integration
increased almost 19 percent to $439 billion. The gains under
a lump-sum replacement showed an increase of less than five

percent.

5.3. Reduction of the Corporate Tax Rate

The welfare gains from the reduction in the corporate tax
rate were lower than the welfare gains under full integration.
This was expected, because, although the rate reduction does
lower the cost of capital in the corporate sector, it.does not
lower it as significantly as would outright elimination Of the
corporate income tax. In the nine-sector model, the welfare
gains were $340 billion under the lump-sum replacement. These
gains were 0.82 percent Of base-case wealth, or 0.6 percent Of
the discounted stream Of GNP. The gains were $242 billion
under'the income-tax replacement (about 0.58 percent.of'wealth
or 0.42 percent of GNP). The reduction in the corporate tax
rate with income-tax replacement provides an interesting case
for study. Under income-tax replacement, firms are able to
adjust their debt/capital ratios each period. Although the
elasticities are small, there is still a slight reduction in
the corporate debt/capital ratios in the corporate sectors.
In the nondurables sector, the debt/capital ratio decreases
from 32 percent to 29 percent, the largest change Of any of
the sectors. The durables sector shows an increase in its

debt/capital ratio due tO the presence Of a positive

93
debt/capital elasticity with respect to changes in the
personal tax rate.28 ‘The movement in the debt/capital ratios
is not enough, however, to make the welfare gains from an
income-tax replacement larger than the gains from a lump-sum
replacement.

In the four sector model, the debt/capital ratio in the
corporate sector falls from 36 percent to approximately 34
percent. The welfare gains in this model were $311 billion
under lump-sum replacement and $242 billion under income-tax
replacement. The price Of capital in the corporate sector
rises above unity, but it does not increase by as much as it

does under full integration.

5.4. Dividend Deduction

This plan yielded welfare gains that were larger than the
corporate tax reduction in both the nine—sector model and the
four-sector model. In the nine-sector model, welfare gains of
$360 billion and $307 billion were achieved under lump-sum and
income-tax replacement. The movement in the debt/capital
ratios was very small (usually less then a one-percentage-

point change).

 

28 Simulations were also run using a negative value (-
.07), for the elasticity Of the debt/capital ratio in the
durables sector, with respect to changes in the personal tax
rate. This was the sign predicted by the bankruptcy model.
This resulted in very little change in any Of the simulation
results.

56
HO
gr-

sec

gai
bil
bill
and
foun

bill

0f th

94

The movements in the prices of capital among the sectors
were less dramatic than under the full integration proposal.
The largest initial rise in the price Of capital occurred in
the trade sector, where the price rose approximately nine
percent above its base-case price Of unity. The trade sector
had the lowest debt/capital ratio Of any of the corporate
sectors. Dividend relief has a larger impact on this sector
because it finances a :relatively larger' portion of its
investment with equity. The price Of capital initially
decreased in the primary, construction, durables, and owner-
occupied housing sectors. The increase in capital stocks was
more concentrated in the transportation, communication and
utilities sector, where the capital stock increased almost 1.7
times as much as in the base case. The other corporate
sectors also experience increases in their capital stocks.
However, they were not as large relative tO the base-case
growth as in the transportation, communication, and utilities
sector.

The four-sector’1model ‘yields somewhat lower 'welfare
gains. Under lump-sum replacement, the gains were $344
billion. Using income-tax replacement, the gains were $311
billion. A dividend deduction plan was also modeled by BFSW,
and their welfare gains were significantly larger than those
found here. Their dynamic gains ranged between $592 and $709
billion, in 1983 dollars.

Table 11 presents a summary Of the welfare gains from all

of the base-case tax policy simulations.

 

"

Ca:
the

unc‘

95

5.5. Sensitivity Analysis

This section examines the sensitivity Of the results to
changes in various model parameters, specifically depreciation
rates, adjustment costs, debt/capital elasticities, and the
classification of corporate and noncorporate sectors. To test
the sensitivity of the model to changes in.the parameters, the

new'values of the parameters are used to calibrate a new'base-

Table 11

Summary Of Initial Simulation Results1

 

  
  
 
  
  
  
 

 

 

 

 

= a
Policy Nine-Sector Nine-Sector Four-Sector Four-Sector
Income-Tax Lump-Sum Income-Tax Lump-Sum
Replacement Replacement Replacement Replacement
Full $402 $463 $369 $447
Integration2 ‘
Full ‘ $492 $603 $439 $468
Int ration3
Dividend . $307 $360 $311 $344
Deduction
Corporate $242 $340 $242 $311
Tax Rate
Reduction ==
a

 

 

 

 

 

 

Welfare gains expressed in billions.
Under these simulations debt/capital ratios were set equal to zero.

Under these simulations debt/capital ratios were not set equal to zero.

case simulation. Revised-case simulations are then run, and
their results are evaluated relative to the base case formed

under the new model assumptions.

"I

IT
IT

wi

inc
not
mod
adj'
$75!
wea]
The
With
two)

incom

96
5.5.1. Adjustment_§osta

Adjustment costs prevent or make more costly the flow of
capital into a sector. The presence of adjustment costs
reduces the welfare gains Of the new tax policy. Sensitivity
analysis was performed first by doubling the adjustment costs
of each sector. A second set Of simulation was run with the
adjustment costs set equal to zero.

Sensitivity analysis was done comparing the results of
full integration simulations to the base-case results of
section 5.2. With doubled adjustment cost factors in each
sector, the welfare gains were lower. There was a reduction
in welfare gains Of almost 37 percent under lump-sum
replacement and 33 percent using the income-tax replacement.
These effects were slightly larger in the four-sector model,
with welfare reductions on the order of 43 percent.

The elimination Of adjustment costs caused a tremendous
increase in the level Of welfare in the four-sector model, but
not a very dramatic increase in welfare in the nine-sector
model. In the four-sector model, the elimination Of
adjustment costs resulted in welfare gains Of between $637 and
$755 billion-dollars (between 1.6 and 1.8 percent of base-case
wealth), over one and one-half times the base-case results.
The results in the nine-sector model were less dramatic.
With lump-sum replacement, there was slight increase, about
two percent, in welfare gains. The gain in welfare using an

income tax-replacement was almost five percent.

8C
69
pa.
WEI
pal

wei

5.5.

role

I‘EIat

97

The reason for this is due to the disaggregation of the
corporate sector. The corporate sectors which receive
immediate investment in the nine-sector model have less than
the average adjustment cost parameter Of the corporate sector
in the four-sector model. In the four-sector model, the
weighted average adjustment cost parameter of all corporate
sectors is used. This implies that adjustment costs as a
proportion Of new investment are higher in the four-sector
model than they are in the nine-sector model. Therefore,
elimination Of the adjustment cost parameters will lead to
larger welfare gains in the smaller model. The magnitude of
difference, however, still seems quite large, even given the
difference in adjustment cost parameters.

These results may be attributable to the fact that
adjustment costs in the four-sector' model are taken. as
aggregates Of the nine-sector model adjustment-cost
parameters, using' the capital stocks Of each sector as
weights. One would like to aggregate adjustment cost
parameters using the level Of investment in each sector as
weights. However, because levels of investment are not known

a priori, this method Of aggregation is not used.

5.5.2. Depreciation

Depreciation Of the capital stocks plays an important
role in this model. It is through depreciation that there is
a reduction in the capital stocks Of the sectors ‘with

relatively low net rates.Of return, IBecause Of the assumption

c-

e1

ti

in
de;
rai
cor
Pa:

mod

5.5.

the

98
of immobile sector-specific capital, there is no flow of pre-
existing capital between the sectors. The construction Of the
model limits the sensitivity analysis that can.be performed on
the depreciation parameters. Sensitivity analysis is
performed by examining the effects Of eliminating depreciation
from the model.

The model shows very little sensitivity to elimination Of
depreciation. Elimination Of the depreciation rates results
in slight decreases in welfare in both the nine-sector and
four-sector models. The reductions in welfare range from two-
to seven percent. The reason for this is that the reduction
in depreciation rates has a large effect on the consumer's
subjective rate Of timejpreferenceu This is due to the effect
of depreciation on the steady-state growth rate used in the
calibration Of the subjective rate of time preference. The
elimination Of depreciation increases the consumer's rate of
time preference from about .0001 tO almost .048. This
increase means that the consumer, in the model with no
depreciation, is discounting future utility at a very high
rate. Consequently, one does not get as large a change in the
consumer's behavior when there are changes in the tax
parameters Of the models, and the-gains that occur late in the

model are heavily discounted.

5.5.3 pebthapital Elasticities

The primary purpose Of this dissertation is to examine

the effects Of endogenous financial behavior on the welfare

In

La
t1
bt

no

in}

99
effects of various corporate tax policies. Sensitivity
analysis is done on the base-case debt/capital ratios, and on
the elasticity parameters.

The first set of simulations that was run was a full
integration simulation assuming that firms hold no debt. With
no debt in the model, the welfare gains from integration are
increased. The welfare gains, in the nine-sector model,
increased by almost 60 percent, under income-tax replacement.
Using income-tax replacement, the welfare gains were $644
billion. These gains are 1.6 percent Of base-case wealth, 1.1
percent of GNP, or 15.95 percent of capital-tax revenue. With
lump-sum replacement, the welfare gains from full integration
were $897 billion (2.2 percent Of wealth, 1.6 percent of GNP,
or 21.45 percent Of capital-tax revenue). The elimination of
debt in this model raises the effective tax rate on capital,
because firms are unable tO deduct any Of their distributions
from income. Therefore, all income coming from the corporate
sector is taxed twice. This should make the implementation Of
full integration even more important in reducing the
distortion between the return to corporate and noncorporate
capital. Partial integration (a reduction in the corporate
tax rate) with income-tax replacement, yields gains in welfare
that are larger than those in the original set of simulations,
but the gains are smaller than under full integration.

The gains in the four-sector model were also large. With
nO debt in the model, the increase in welfare gains under full

integration was almost 35 percent, when income-tax replacement

at

we
th-
pel
red
wel

in

and
play
gain

debt‘

100
was employed. Using lump-sum.replacement resulted in welfare
gains that were 42 percent larger than those in the original
simulations. The gains from the reduction in the corporate
tax rate averaged about 14 percent over the original
simulations, which are almost identical to the results from
the nine-sector model.

A set Of simulations was also run with the debt/capital
ratios in each Of the initial sectors set at twice their
original base-case values. In the nine-sector model, welfare
gains were reduced by almost 50 percent. Full integration
yielded welfare gains Of 0.45 percent Of base-case wealth,
about $181 billion, using lump-sum replacement. Income tax
replacement yielded even lower gains in welfare. The gains
from full integration were 0.29 percent Of base-case wealth,
about $116 billion.

The four-sector model showed similar reductions in the
welfare gains. With the original debt/capital ratios doubled,
the welfare gains from full integration decreased almost 62
percent, under both methods Of revenue replacement. The
reduction in the corporate tax rate resulted in smaller
welfare changes. The welfare gains were 42 percent less than
in the original model.

These results seem tO indicate that the presence Of debt,
and the proportion.of capital that is financed.with.debt, does
play an important role in determining the size of the welfare
gains from Changes in tax policies. The larger the

debt/capital ratios Of firms, the smaller is the distortion

101
between corporate and noncorporate capital, because a smaller
amount of income is subject to double taxation.

A final set of simulations was run to examine the impact
of changes in the:debt/capital elasticities on the size of the
welfare gains. The elasticity estimates from Chapter 3
indicate that firms may respond to changes in the personal and
corporate tax rates. The size of the responses is small, as
none of the estimates is greater than unity. Poterba (1991)
provides evidence that firms' financial decisions may be more
sensitive to changes in the tax system than my empirical

results indicate.29

He points out that the Tax Reform Act
Of 1986 was projected to increase federal corporate tax
revenue, but failed to do so. One of the primary reasons for
this was a large increase in the use Of corporate leverage.
This may indicate more responsive elasticities than were used
in the base-case simulations. In order to test the
sensitivity of welfare gains tO changes in the elasticities,
simulations were run involving the dividend-deduction plan,
and the lowering Of the corporate tax rate. Both plans were
run under income-tax replacement.

The elasticities were first increased.by a factor of ten.

This makes the firms very sensitive to changes in the tax

system. Within the first three periods Of both simulations,

 

29.As:more data become available regarding information on
debt/capital ratios, it would be Of interest to investigate if
firms have become more responsive to changes in the tax
system. A Chow test (Chow (1960)), could be run to see if
there has been.a change in firm behavior, compared to the data
used in my estimation.

01
di
Th
de
in

the

5.5

and

Cap)

orig

102
the debt/capital ratios of the firms were driven to zero. The
welfare gains under the lowering Of the corporate tax rate
were smaller than the gains in the central case with the
estimated elasticities. welfare gains, in the nine-sector
model, were $123 billion compared to $242 billion in the
standard model. Under the dividend deduction plan, the gains
were also smaller than in the standard model. The welfare
gains were $258 billion, compared to $307 billion in the
original model. These results indicate the importance of
correct specification Of firms' financial decisions within the
context Of general equilibrium models. This is even more
apparent after examining the welfare gains from the case where
the debt/capital elasticities are set to zero. The reduction
in the corporate tax rate results in gains of $335 billion.
These are considerably larger than those that emerge under the
other two elasticity specifications. The gains under the
dividend deduction plan were on the order of $360 billion.
These results appear to indicate that changes in the
debt/capital ratios throughout the simulations are important
in determining the actual size Of welfare gains, and that, if

these effects are ignored, the gains may be overstated.

5.5.4. Classification Of Corporate and Noncorporatg Sectgrs

The classification of production sectors into corporate
and noncorporate sectors was based on the proportion Of
capital owned by corporate and noncorporate entities. The

original rule was, if more than 68 percent Of the capital

103

stock was incorporated, the sector was classified as corporate
for tax purposes. This section examines the sensitivity of
the nine-sector model to changes in the classification of the
corporate and noncorporate sectors. A production sector, in
these simulations, was classified as being corporate if 50
percent or more Of the capital in that sector was
incorporated. This rule leaves only one noncorporate sector,
the primary sector. Full integration simulations and
simulations based on the Tax Reform Act Of 1986 were run on
the new model.

A reduction in noncorporate capital in the model resulted
in welfare gains. The welfare gains from both full
integration and a reduction in the corporate tax rate were
smaller, in every case, than found in the original
simulations. The full integration simulations resulted in
welfare gains Of $307 billion, using lump-sum replacement, a
reduction in welfare gains Of almost 24 percent. Full
integration with income-tax replacement resulted in welfare
gains that were lower, by almost 22 percent than the original
simulations. These changes are large. As more capital in the
model is incorporated, the capital missallocation distortion
between sectors is lessened. In addition, the overall
distortion from the corporate income tax is reduced. Thus,
the gains from removing the distortions are smaller. Also,
implementation Of both policies results in a greater loss Of
tax revenue than under the previous classification Of

corporate and noncorporate capital.

104

The reduction in the corporate tax rate from 46 percent
to 35 percent also resulted in decreased welfare gains,
relative to the original simulations. Lump-sum replacement
resulted in welfare gains of $280.5 billion, a decrease Of
almost $50 billion from the model with three noncorporate
sectors. The use of income-tax replacement resulted in even
larger reductions in the welfare gains. The new model
provided welfare gains Of $237 billion, a reduction Of almost
23 percent over the original model.

These results are important. They show that the
classification Of corporate and noncorporate capital in the
model is important in determining the benefits to changes in

corporate tax policy.

5.5.5. 5 m ' a 'O

In order to test the sensitivity Of my model to changes
in the amount of intertemporal substitution, the level Of
required consumption was adjusted and the full integration
simulations were repeated. In the original simulations, the
consumer was required to consume 80 percent Of base-case
consumption. The 80 percent consumption requirement was
chosen in order tO reduce the amount Of intertemporal
substitution and because it was the value used by Ballard and
Goulder (1985).

The first set Of simulations increases the amount Of
required consumption to 90 percent Of base-case consumption.

This-greatly'restricts the ability of the consumer to increase

105

his level of saving. With the higher required consumption
ratio there was a decrease in welfare from full integration.
Using lump-sum replacement, full integration resulted in a
decrease in welfare Of 0.13 percent Of base case wealth, about
$52 billion. This is a drastic decrease from the $463 billion
welfare gain found in the original simulations. The use of an
income-tax replacement resulted in even larger welfare losses.
The welfare loss from full integration was $245 billion, about
0.61 percent Of base-case wealth. Restricting the consumer's
saving decision plays an important role in the simulation
results.

Simulations were also run with the required consumption
ratio set at 60 percent Of base-case consumption, 40 percent
Of base-case consumption, 20 percent Of base-case consumption
and with no required consumption. As the required consumption
ratio was reduced, the welfare gains from full integration
increased substantially.

With the required consumption ratio equal to 0.6, the
welfare gains from full integration were 1.9 percent of base-
case wealth ($763 billion), under lump-sum, and 1.7 percent.of
base-case wealth ($692 billion), under income tax replacement.
These welfare gains are almost 1.7 times as large as the
original welfare gains.

With the consumption ratio set at 0.4, the welfare gains
were even larger. Full integration resulted in welfare gains
of between $1,280 and $1,353 billion. These are almost four-

fold increases Over the original simulations.

106

Full integration resulted in welfare gains Of $1,990
billion using lump-sum replacement, when the required
consumption ratio was 0.2. The gains using an income-tax
replacement were also quite large, 4.8 percent of base-case
wealth, $1,930 billion.

A final set of simulations was run with no required
consumption. These simulations allow a very high degree Of
intertemporal substitution. The welfare gains from full
integration were very large. The improvement in welfare ranged
between 6.4 and 6.6 percent Of base-case wealth. These gains
are between $2,573 billion and $2,654 billion.

This sensitivity analysis underscores the importance Of
the restriction Of the amount Of intertemporal substitution in
the model. If the consumer is free to save as much or as
little as he wants, the welfare gains increase tremendously.
This indicates that the intertemporal gains from full
integration are very large, and.account for a large portion of
the welfare gains in the simulations. It also indicates the
caution that must be taken in setting the amount Of required
consumption in the model. The consumer's response to
integration is crucial. This is even more apparent if one
examines the first-period savings elasticity of the consumer
under the different required consumption specifications.

There has been some controversy regarding the relative
importance of the static gains and the human-wealth effects
resulting from a change in tax policy. In the simulations

reported here, there appear tO be large dynamic effects. This

 

107
is underscored by the changes in the welfare gains under
different required consumption specifications. The static
gains also appear to be important as can be seen from the
sensitivity analysis on corporate and noncorporate
classification and comparing the results of the nine-sector
and four-sector models.

It is important to note that I am not able to explicitly
separate the dynamic from the static gains in the model.
However, it appears that in the central case both effects are
important.

With a required consumption ratio Of 0.8 the consumer
exhibits a first-period saving elasticity Of 0.77. However,
if there is no required consumption in the model, the
consumer's first-period saving elasticity increases to 20.9.
This seems to indicate that a required consumption ratio Of
zero results in extreme saving behavior. This is one reason

I chose tO set the required consumption ratio equal tO 0.8.

5.6. Comparison with Other Studies

There are a number Of other studies which have examined
changes in tax policy in a general equilibrium setting.
Galper, Lucke, and Toder (1988) examine the impact Of the Tax
Reform Act Of 1986, using a model with endogenous financial

behavior. They find static welfare gains Of approximately

108

$11.87 billion, in 1983 dollars.30 Their model has a number
of income Classes, and they find that the welfare.gains accrue
to individuals mostly in the upper income tax brackets. The
Tax Reform Act Of 1986 is similar to my simulation of the
reduction in the corporate tax rate.

Fullerton and Gordon (1983) examine full integration in
a model featuring endogenous financial behavior. Their model
is based on the bankruptcy theory. They find essentially no
welfare gains from full integration. The reason for this is
that their model focuses on the income risk Of firm.awnership.
As noted earlier, Bulow and Summers (1984) point out that
income risk is very small relative to capital riSk. Given
this fact, their results appear to be extreme in nature.

Jorgenson and Yun (1990) use an econometrically based
general equilibrium model tO measure the welfare gains of a
number Of comprehensive federal tax system reforms. Their
model has a single production sector composed Of corporate,
noncorporate and owner-occupied housing components (the same
model used in Jorgenson and Yun (1986)). They find welfare
gains that are extremely large. They predict dynamic welfare
gains from the Tax Reform Act of 1986, Of $1,561.8 billion
under lump-sum replacement, and $1,563.1 billion using income-
tax replacement. These are very large welfare gains relative

tO other studies. One reason for the large gains is that

 

30 If one were to assume a 4 percent discount rate and
the same return over an infinite period, this static gain
would correspond tO a dynamic gain Of approximately $296
billion.

109
their model allows for a great deal of intertemporal
substitution. My central-case is designed to reduce the
amount of intertemporal substitution. The sensitivity
analysis indicates that the assumption of no required

consumption would result in similarly large welfare gains.

5.7. Conclusion

The results Of my simulations point out the importance Of
a number Of things. First, the financial behavior Of firms
does appear tO play an important role in determining the
welfare effects Of federal tax policy. Second, adjustment
costs also appear to be important in determining how large the
welfare.gains will be, especially when.the tax reform leads to
a large desire for the flow Of capital between sectors. The
role of adjustment.costs, however, appears tO be'dampened when
the model is disaggregated. Along with adjustment costs, the
assumption Of sector-specific capital also appears to be very
important. Fullerton (1983) also emphasizes this point by
showing the reduction in welfare gains, when there is a
restriction on the flow of capital between sectors.

The empirical evidence Of Chapter 3 indicates that firms
do not show a large degree Of sensitivity to changes in
marginal tax rates. However, the estimation also indicates
that in some sectors tax rates do matter. Ignoring the

effects Of changes in tax rates on corporate leverage

110
decisions completely may result in an overstatement of welfare
gains from Changes in tax policy.

The gains from full integration, achieved in this study,
are generally smaller than those found in other studies, but
still appear to be significant. The sensitivity analysis with
respect.towthe required.consumption ratio indicates that it is
extremely important in determining the size Of the welfare
gains, and that if the required consumption ratio is reduced,
the welfare gains approach those found in other studies.

Due to the sensitivity Of the model to changes in the
various parameters the implementation of full integration
would appear somewhat unpredictable. Because Of the
sensitivity of the model, a partial integration solution would
appear a more satisfactory result. The fact that this model
abstracts from the international sector, and also does not
examine the distributional aspects Of changes in tax policy is
important. However, all forms of integration examined in

these simulations appear to be a step in the right direction.

REFERENCES

111

REFERENCES

Atkinson, A. B., 1971. "Capital Taxes, the Redistribution of
Wealth, and Individual Savings," Review of Economic
Studies, Vol. 38, pp. 209-228.

Atkinson, A. B., 1980. "Welfare Implications of the Taxation
of Savings," Economic JOurnal, Vol. 90, September, pp.
529-549.

Auerbach, Alan J., 1985. "Real Determinants of
Corporate Leverage," in Benjamin M. Friedman, Ed.,
es ' e ' s,
The University Of Chicago Press, Chicago, pp. 305-
325.

Auerbach, Alan J., and Kevin Hassett, 1989. "Investment, Tax
Policy, and the Tax Reform Act of 1986," presented at
Conference on the Economic Impact of the Tax Reform Act
Of 1986, Ann Arbor, MI, November 9-11, 1989.

Auerbach, Alan J., and Laurence J. Kotlikoff, 1987. Dyaaaig
Eiaaal Poligy, Cambridge University Press, Cambridge.

Auerbach, Alan J ., and Mervyn A. King, 1983. "Taxation,
Portfolio Choice, and Debt-Equity Ratios: A General
Equilibrium Model," The Quarterly Journal of Economics,
Vol. 98, pp. 587-609.

Bagwell, Laurie Simon, and John B. Shoven, 1991. "Cash
Distributions to Shareholders," Journal of .Economic
Perspectives, Vol. 3, NO. 3, Summer, pp. 129-140.

Ballard, Charles L., 1987. "Tax Policy and Consumer
Foresight: A General Equi l ibrium Simulation Study , "
Economic Inquiry, Vol. 25, pp. 267-284.

Ballard, Charles L., 1990. "On the Specification of
Simulation Models for Evaluating Income and Consumption

Taxes," in Manfred Rose, Ed., Heidelbegg Canfazanga an
Taxing goasamptioa, Springer-Verlag, Berlin, pp. 147-188.

Ballard, Charles L., Don Fullerton, John B. Shoven, and John

Whalley, 1985. A General Equilibrium Model £9; Tax 2011 gy
Evaluagioa, University Of Chicago Press, Chicago.

112

Ballard, Charles L., and Lawrence H. Goulder, 1985.
"Consumption Taxes, Foresight, and Welfare: A
Computational General Equilibrium Analysis," in John
Piggott and John Whalley, Eds. , w

' ' ' , Cambridge
University Press, Cambridge, pp. 253-282.

Ballentine, J.G., and Charles E. McLure Jr., 1980. "Taxation
and Corporate Financial Policy," Quarterly Journal of
Economics, March, pp. 351-372.

Bulow, Jeremy I., and Lawrence H. Summers, 1984. "The
Taxation of Risky Assets," JOurnal of.Political.Economy,
NO. 84, Vol. 2, pp. 20-39.

Chow, Gregory C., 1960. "Tests of Equality Between Sets of
Coefficients in Two Linear Regressions," Econometrics,
V01. 28, No. 3, pp. 591-605.

Feldstein, Martin, 1978. "The Welfare Cost of Capital Income
Taxation," Journal of Political Economy,‘Vol. 86, April,
pp. 829-851.

Feldstein, Martin, and Daniel Frisch, 1977. "Corporate Tax
Integration: The Estimated Effects on Capital
Accumulation and Tax Distribution of Two Integration
Proposals," National Tawaournal, Vol. 30, March, pp. 37-
52.

Feldstein, Martin, and Jerry Green, 1983. "Why do Firms Pay
Dividends?" The American Economic Review, March 1983, pp.
17-30.

Fullerton, Don, 1983. "Transition Losses Of Partially Mobile
Industry-Specific Capital," The Quarterly Journal of
Economics, February, pp. 107-125.

Fullerton, Don, 1984. "Which Effective Tax Rate?" National
Tax Journal, March, pp. 23-41.

Fullerton, Don, and Roger Gordon, 1983. "A Reexamination Of
Tax Distortions in General Equilibrium Models," in
Martin Feldstein Ed., Behavioral Simulagiga Maghgaa in

Tax Poligy Analysis, University Of Chicago Press,
Chicago, pp. 369-426.

Fullerton, Don, and Yolanda K. Henderson, 1984. "Incentive
Effects Of Taxes on Income from Capital: Alternative
Policies.in the 1980's," in.C.R~ Hulten.and I.V. Sawhill,

Eds., Tha Legacy of Reaganomics: Pgospeggiyes fa: Lang-
Ierm.Growth, The Urban Institute Press, Washington, D.C.

113

Fullerton, Don, and Yolanda K. Henderson, 1989. "A
Disaggregate Equilibrium Model of the Tax Distortions
Among Assets, Sectors, and Assets," International
Economic Review, Vol. 30, NO. 2, pp. 391-413.

Fullerton, Don, A. Thomas King, John B. Shoven, and John
Whalley, 1981. "Corporate Tax Integration in the United
States: A General Equilibrium .Approach," American
EConomic Review, September, pp. 677-691.

Galper, Harvey, Robert Lucke, and Eric Toder, 1988. "A
General Equilibrium Analysis Of Tax Reform," in Henry
Aaron, Harvey Galper, and Joseph Pechman, Eds., Uaaaay

one en 36' ' ab mSe .eHbr eII‘- e! no ,9! t
The BrOOkings Institution, Washington, D. C., pp. 59- -114.

Gerardi, Geraldine, Michael J. Graetz, and Harvey S. Rosen,
1990. "Corporate Integration Puzzles," National Tax
Journal, Vol. XLIII, NO. 3, pp. 307-314.

Gordon, Roger H., and.Jeffrey'K.IMacKie-Mason, 1990. "Effects
of the Tax Reform Act Of 1986 on Corporate Financial
Policy and Organizational Form," in Joel Slemrod Ed., Q9
IQKQ§_MQ££§I- MIT Press, Cambridge, MA, pp. 91-132.

Gordon, Roger H., and Burton G. Malkiel, 1981. "Corporate
Finance," in Henry J. Aaron and Joseph A. Pechman Eds.,
H w ax s ec co om c ehavio , The Brookings
Institution, Washington, D.C., pp. 131-198.

Goulder, Lawrence H., and Lawrence H. Summers, 1989. "Tax
Policy, Asset Prices, and Growth: A General Equilibrium
Analysis," Journal of Public Economics, Vol. 38, pp. 265-
296.

Gravelle, Jane G. , and Laurence J. Kotlikoff, 1989. "The
Incidence and Efficiency Costs Of Corporate Taxation when
Corporate and Noncorporate Firms Produce the Same Good,"
Journal of Political Economy, Vol. 97, pp. 749-780.

Hall, Robert, E., and Dale W. Jorgenson, 1967. "Tax Policy
and Investment Behavior," The American Economic Review,
Vol. 57, pp. 391-414.

Harberger, Arnold C., 1962. "The Incidence Of The Corporate
Income Tax," Journal of Political Economy, Vol. 70, pp.
215-240.

Harberger, Arnold.C., 1966. "Efficiency Effects.of Income from

Capital," in Marian Krzyzaniak Ed., WADE

gogpogatioa Income Tax, Wayne State University Press, pp.
107-117.

114

Hulten, Charles R., and Frank C. Wykoff, 1980. "The
Measurement of Economic Depreciation," in Charles R.
Hulten, Ed., ec' I fla n

, The Urban Institute Press,
Washington, D.C.

Jorgenson, Dale W., and Kun-Young Yun, 1986. "The Efficiency
of Capital Allocation," Scandinavian Journal of
Economics, NO. 88, Vol. 1, pp. 85-107.

Jorgenson, Dale W., and Kun-Young Yun, 1986. "Tax Policy and
Capital Allocation," Scandinavian JOurnal of Economics,
No. 88, Vol. 2, pp. 355-377.

Jorgenson, Dale W., and Kun-Young Yun, 1990. "Tax Reform and
U.S. Economic Growth," Journal of Political Economy, Vol.
98, NO. 5, Part 2, pp. 8151-8193.

Judd, K.L., 1985. "Short-run Analysis of Fiscal Policy in a
Simple Perfect Foresight Model," JOurnal of Political
Economy, Vol. 93, NO. 2, pp. 298-319.

Kimbell, L.J., and G.W. Harrison, 1986. "On the Solution Of
General Equilibrium Models," Economic Modeiing, July, pp.
197-212. -

King, Mervyn, 1978 . Oli a th ,
Chapman 8 Hall, London, England.

McLure, Charles E. Jr., 1979. Must Corngnata Ingona Ba Taxaa
Tyice2, The Brookings Institution, Washington, D.C..

Meyers, Stewart, C., 1977. "Determinants of Corporate
Borrowing," JOurnal of Financial Economics, Vol.5, pp.
147-175.

Miller, Merton H., 1977. "Debt and Taxes," The Journal of
Finance, Vol. XXXII, NO. 2 (May), pp. 261-275.

Modigliani, Franco, and Merton Miller, 1958. "The Cost of
Capital, Corporate Finance and the Theory of Investment, "
The American Economic Review, Vol. XLVIII, NO. 3 (June),
pp. 261-297.

Pechman, Joseph A., 1987. Federal Tax Polic , The Brookings
Institution, Washington D.C., pp.135-89.

Pereira, Alfredo M., 1988. "Corporate Tax Integration in The
United States: A.Dynamic General Equilibrium Analysis,"
University Of California at San Diego Working Paper. ‘

115

Poterba, James M., 1991. "Why Didn't The Tax Reform Act of
1986 Raise Corporate Taxes?" NBER Working Paper NO. 3940.
Cambridge, MA.

Poterba, James IM., and. Lawrence H. Summers, 1985. "The
Economic Effects of Dividend Taxation," in Edward I.

Altman and Marti G. Subramanyam, Eds., e v s
Wang-2. Richard D- Irwin. Homeward. IL. pp-
227-284.

Samuelson, Paul A., 1951. "Abstract of a Theorem Concerning
Substitutability in Open Leontief Models," in Tjalling

KOOpmans Ed., native Analysis in Enogncnion ana
Allgganign, John Wiley and Sons, New York.

Scholz, Karl, 1987. "Documentation for 'the 1983 General
Equilibrium Data Set," mimeo, Stanford University.

Shoven, John B., 1976. "The Incidence and Efficiency Effects
Of Taxes on Income from Capital," JOurnal of Political
Economy, Vol. 84, pp. 1261-1283.

Shoven, John B., and John Whalley, 1972. "A General
Equilibrium Calculation of the Effects of Differential
Taxation of Income from Capital in the U.S.," JOurnal of
Public Economics, Vol. 1, pp. 281-321.

Slemrod. Joel, 1983. "A General Equilibrium Model Of Taxation
with Endogenous Financial Behavior," in Martin Feldstein,
Ed., ehavio a Simu ation Met ds T 1
Analysis, University Of Chicago Press, Chicago, pp. 427-
458.

Starrett, David A., 1982. "Long Run Savings Elasticities in
the Life Cycle Model." Factor Markets Workshop Research
Paper No. 24, Stanford University.

Stiglitz , Joseph E. , 1973 . "Taxation, Corporate Financial
Policy, and the Cost of Capital , " Journal of Public
Economics, Vol 2, April, pp. 1-34.

Strotz, Robert H. , 1955. "Myopia and Inconsistency in Dynamic
Utility Maximization." Review of Economic Studies, 23,
pp. 165-180.

Summers, Lawrence H. , 1981. "Taxation and Corporate
Investment: A q-Theory Approach," in Brookings Papers
on Economic Activity, Vol. 1, pp. 68-136.

HICHIGRN STATE UNIV.

I (III III IIIIIIIIIII‘

2930079175

RRRIES
WI Nil
£353