.z, .

2.3
.D.‘

$.42

9.5.

.

in an
at

. . _

.

‘ ... xiv

4..
.u:

d...
39.3.. r. .
i}: .2,

aka.” .

$4“? :r

 

Tully;

'/

CCM

This is to certify that the

dissertation entitled

The Effects of Taxes
on Foreign Direct Investment in the United States

presented by

Taejoo Kim

has been accepted towards fulfillment
of the requirements for

Ph . D. degree in Economics

 

 

Mfwmx

Major professor

 

Date ’0 74/,” Zoof

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

 

 

 

uneasy "'
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DALE DUE DATE DUE DATE DUE

 

\ ,1 $151002???

 

()9
SEP 1 821392
«6/
to

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/DateDuep65-p. 15

THE EFFECTS OF TAXES ON

FOREIGN DIRECT INVESTMENT IN THE UNITED STATES

By

Taejoo Kim

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements

for the degree of

DOCTOR OF PHILOSOPHY

Department of Economics

2001

ABSTRACT

THE EFFECTS OF TAXES ON
FOREIGN DIRECT INVESTMENT IN THE UNITED STATES
By

Taejoo Kim

The main objective of this dissertation is to analyze the effects of US. taxation on
foreign direct investment in the US. Specifically, this dissertation tries to provide
empirical answers to some questions regarding the relationship between F D1 in the US.
and the US. effective corporate tax rate: (a) what is the qualitative response of F D1 in
response to a change in the effective tax rate?, (b) how does the foreign investor
country’s double-taxation-relief system affect the response of F D1 in the US. to the US.
effective tax rate?, (c) is the average effective tax rate or the marginal effective tax rate
the better explanatory variable for FDI behavior?, and ((1) what are the responses of
acquisition-type F DI and establishment-type FDI to both the average and the marginal
effective tax rate?

The estimation results indicate that overall foreign investment decreases in
response to an increase in the US. average effective tax rate, and that multinationals from
countries with worldwide system do not appear to adjust their investment, but
multinationals from countries with territorial system reduce their investment in the US.
in response to an increase in the US average effective tax rate. These results are

consistent with the Traditional View, but do not conform to the New View.

The industrial organization theory suggests that F D1 is closely related to market
imperfections, and FDI is a result of highly strategic decision-making of a multinational
corporation. Also, most F D1 is accounted for by acquisition of existing US. firms.
Therefore, FDI is not expected to respond to the marginal effective tax rate, which is
based on the perfect-competition assumption and measures the tax burden of a new
investment project. My estimation results appear to support this hypothesis.

Another estimation result of mine indicates that establishment-type F DI does not
respond to a change in either the average effective tax rate or the marginal effective tax
rate. I explain this result with the composite feature of establishment-type FDI: it is a new
investment and it is a mode of FDI.

On balance, this dissertation does not support the New View. I think the problems
of the New View arise from its assumptions: perfect capital mobility among countries
and among assets. Perfect capital mobility means that every adjustment process takes
place immediately. So, in the world of the New View, the economy is always in the
equilibrium state. However, in the case of direct investment, capital seems to move
slowly in response to a change in the relative tax rates between countries and between
assets. These capital movements, in turn, lead to changes in asset prices and in the pretax
returns of the assets. That is, the change in the pre-tax rate of return in response to a
change in tax rate is the result of the process of capital movements. But, the New View
thinks that every adjustment process occurs immediately, and the price and the pretax
return change immediately in response to a tax change. And then capital (direct
investment) moves again in response to these changes in asset price. The problems of the

New View seem to arise from this misunderstanding.

ACKNOWLEDGEMENTS

Many people helped to make this dissertation possible and I would like to thank
some of them. First, I would like to express my gratitude and sincere appreciation to my
advisor, Professor Charles L. Ballard, for his guidance, advice, and encouragement.
Without his invaluable advices and encouragement, the completion of this dissertation
would not have been possible. I would also like to thank other members of my
committee, Professor Leslie Papke and Jeffrey Wooldridge for their valuable comments.
In addition, I would like to thank Korean government (Ministry of Finance and
Economy) for giving me an opportunity to study abroad.

I thank my parents and parents-in-law for their continuing love and support. And
most of all I wish to thank my wife Sooyeon Lee and two sons, Konhyung and Sangwoo.
Last four years were in many ways harder for them than those were for me. I want to

thank them for their patience, understanding, and love.

iv

TABLE OF CONTENTS

LIST OF TABLES ........................................................................ vii
LIST OF FIGURES ....................................................................... ix
CHAPTER I
INTRODUCTION ........................................................................ 1
1. Foreign Direct Investment in the United States ................................. 1
2. Economic Effects of F DI ......................................................... 2
3. FDI and Taxes ....................................................................... 4
CHAPTER 1]
LITERATURE REVIEW AND DATA ISSUES ..................................... 9
1. Literature Review ................................................................... 9
2. Data Issues ........................................................................... 14
2.1. FDI Data ........................................................................ 14
2.2. Data on the Effective Tax rates .............................................. 18
CHAPTER III
THE NEW VIEW ON TAXES AND FOREIGN DIRECT INVESTMENT ...... 24
1. The New View ....................................................................... 24
2. Problems of the New View ......................................................... 26
3. Problems in the New View’s Empirical Analysis ................................ 30
3.1. Problems of the Average Effective Tax Rates by Tax Analysts ......... 31
3.2. Problems of the Way the New View uses Tax Analysts’ Data ........... 34
3.3. Replication Results of Swenson’s Estimation .............................. 39
4. Summary ............................................................................. 43
CHAPTER IV
OVERALL U.S. TAX EFFECT ON FDI IN THE US. ............................... 45
1. Tax Consequences of the US. Source Income of Foreign Investors ......... 45
2. Empirical Model Specification .................................................... 46
3. Data .................................................................................... 51
4. Estimation Results ........................................... -. ....................... 55

CHAPTER V
FDI IN THE US. AND THE INTERNATIONAL TAX SYSTEM

OF THE INVESTOR COUNTRY ....................................................... 61
1. Overview of the International Taxation System ................................. 61
2. The Response of FDI in the US by International Tax Regime ............... 62
3. Empirical Results of Previous Studies ............................................ 63
4. Model Specification and Data ...................................................... 65
5. Estimation Results .................................................................... 73
CHAPTER VI
TYPES OF FDI AND EFFECTIVE TAX RATES .................................... 80
1. FDI and the Measures of Effective Tax Rates .................................... 81
1.1. Characteristics of FDI ......................................................... 81
1.2. Composition of FDI ............................................................ 85
2. Establishment FDI and the Effective Tax Rates ................................. 90
2.1. Incentives to the Establishment F DI ......................................... 92
2.2. Empirical Analysis ............................................................. 95
CHAPTER VII
CONCLUSION ........................................................................... 100

APPENDIX: CALCULATION OF MARGINAL EFFECTIVE TAX RATE 107

1. Model .................................................................................. 107
2. Tax Variables ......................................................................... 110
3. Expected Inflation Rate ............................................................. 117
4. Other Variables ....................................................................... l 19
5. Calculation ............................................................................ 122
REFERENCES .............................................................................. 126

vi

LIST OF TABLES

Table 1. Summary of Empirical Studies ................................................

Table 2. Replication Result of Swenson (1994)‘s Table 3 ...........................

Table 3. Illustration of Swenson’s Calculation of the AETR
for Each Industry .................................................................

Table 4. Replication Result of Swenson‘s Industry-level Estimation ...............

Table 5. Replication Result of Swenson‘s Country-level Estimation ................

Table 6. FDI in the US. by Industry: 1979-1996 .......................................

Table 7. Average Effective Tax Rates by Industry ....................................

Table 8. Marginal Effective Tax Rates by Industry ...................................

Table 9. Test Results (Total FDI on Measures of the Effective Tax Rate) .........

Table 10.

Table 1 1.

Table 12.

Table 13.

Table 14.

Table 15.

Table 16.

Table 17.

Table 18.

FDI by Industry and the Average Effective Tax Rate .....................

F DI by Industry and the Marginal Effective Tax Rate .....................

Home Country’s International Tax System and After-tax Return .......
F DI in the US by Major Investor Countries: 1979-1996 ................
FDI from Territorial Countries and the AETR .............................
FDI from Territorial Countries and the METR ............................
FDI from Worldwide Countries and the AETR ............................
F DI from Worldwide Countries and the METR ...........................

Acquisition FDI and the AETR ..............................................

vii

13

36

37

41

44

52

53

54

55

58

60

63

67

74

75

78

79

88

Table 19. Acquisition F DI and the METR ............................................. 89

Table 20. Establishment FDI and the AETR .......................................... 98
Table 2]. Establishment FDI and the METR .......................................... 99
Table A—1. Corporate Income Tax Rates ............................................... 111
Table A-2. Investment Tax Credit Rates ............................................... 112
Table A-3. Depreciation Methods ....................................................... 116
Table A-4. Comparison of the Expected and Actual Inflation Rates ............... 1 18
Table A-5. Economic Depreciation Rates .............................................. 120
Table A-6. Asset Lifetimes ............................................................... 121
Table A-7. Asset Share .................................................................... 123
Table A-8. Capital Stock Share by Industry ............................................ 124

viii

LIST OF FIGURES

Figure I. MacDougall Model .............................................................
Figure 2. Foreign Direct Investment in the US: 1979-1996 ........................
Figure 3. The Avearge Effective Tax Rates ............................................
Figure 4. Comparison of the Marginal Effective Tax Rates .........................
Figure 5. Comparison of the AETR of Tax Analysts and that of Tax Foundation
Figure 6. FDI in the US. and the US. Balance on Current Account ...............
Figure 7. Effective Corporate Tax Rates in the US: 1979-1996 ....................

Figure 8. The US. Balance on Current Account with Seven Major Investor
Countries ........................................................................

ix

I7

21

22

33

50

68

69

CHAPTER I
INTRODUCTION

1. Foreign Direct Investment in the United States

Direct investment implies that a person has a lasting interest in and a degree of
influence over the management of a business enterprise in another country. For the
United States, ownership or control of 10 percentl or more of an enterprise’s voting
securities is considered evidence of such a lasting interest or degree of influence over
management. Thus, foreign direct investment (F D1) in the United States is defined as “the
ownership or control, directly or indirectly, by one foreign person of 10 percent or more
of the voting securities of an incorporated US. business enterprise or the equivalent
interest in an unincorporated US. business enterprise”2. Any investment that is not direct
investment by this definition is considered portfolio investment. In 1997, outlays of
foreign direct investors to acquire or establish businesses in the US. were $69.7 billion,
about 8.1% of domestic nonresidential fixed investment.

Foreign investors can undertake FDI in the US by acquiring shares of existing

US. firms, or by establishing new subsidiaries or branches. Quantitatively, acquisition is

 

' The lO-percent criterion is somewhat arbitrary, and may understate or overstate foreign control. However,
Graham and Krugrnan (1995) suggest that, in the aggregate, the danger of overstatement and
understatement is not that large. They point out that, on average. foreign parents controlled 77.5 percent of
their US. afiiliates in 1991.

2 Quijano (1990), p. 29.

the more important route of F DI. Figure 2 in Chapter 2 shows that the acquisition type of
FDI accounts for over 80 percent of total FDI. The choice between acquisition and new
establishment depends on many strategic considerations. The strategic considerations and

the reason why foreign investors prefer the acquisition route will be discussed in

Chapter 6.

2. Economic Effects of FDI

According to the neoclassical theory of the effects of F DI (MacDougall model3),
FDI leads to an increase in the output of the world as a whole. The gain from FDI is
shared by both the home country (investor country) and the host country (investee
country). In a world of perfect competition and no externalities, foreign direct investment
equalizes the marginal productivities of capital of both the investor and investee
countries. The efficiency in the use of capital improves, and the total output of the world
increases. The argument of the neoclassical model is easily explained by Figure 14.

In Figure 1, the vertical axes measure the marginal product of capital in the
investor country (MPKI) and the investee country (MPKZ) and the horizontal axis
measures the total capital endowment of the world. The lines ee ’ and dd’ represent the
marginal productivity schedule of the investor country and the investee country,

respectively. Before FDI takes place, the investor country produces Olebc of output with

 

3 MacDouga11(l960).
‘ Caves (1996), p. 194.

01c capital in combination with the given amount of labor, and the investee country (host
country) produces Ogdac with 02c capital. The price of capital is 01R] in the investor
country, which is lower than that in the investee country Osz. Now fc amount of FDI
flows from the investor country to the investee country, and the marginal productivity of
capital is equalized between the two countries at OR = 02R. As a result of FDI, the
output is Olegf in the investor country and Ogdgf in the investee country. Compared with
the total output before F DI, total world output is increased by the triangle abg. The output
of the investor country is decreased byfgbc, but as a return to FDI the investor country
receives fghc. The investor country enjoys greater national income than before FDI.

Similarly, the investee country enjoys a net increase of national income agh.

Figure 1. MacDougall Model

 

 

 

 

 

 

 

 

MPK1 d MPK2
e
(1 R2
R g h R
R1 / b
d ’ e
0/ C 02

Of course, the MacDougall model cannot capture all the economic effects of F DI,
because it is based on several strong assumptions: perfect competition, perfect capital

3

mobility, no extemalities, and no consideration of time (it is a static model). As a matter
of fact, FDI has an effect on the quantity and quality of employment, technology
diffusion, balance of payments, and other variables such as national security. In reality.
most debates on the effects of FDI concentrate on these external effects. Advocates of
FDI argue that FDI creates huge job opportunities in the US, or that FDI brings good
technology into the US and stimulates domestic R&D efforts, or that capital inflow
through F DI improves the balance of payments. In contrast, critics of FDI argue that
foreign firms tend to obtain more of their production inputs from abroad than US. firms,
and that the resulting reduced demand for the products of domestic suppliers costs the
US jobs and worsens the US. trade balance. They also argue that R&D by foreign-
controlled firms is skewed toward less challenging application-oriented technologies,
rather than more basic and scientific technologies. Due to the nature of the extemalities,
there is not much evidence on these arguments5 . Graham and Krugman (I 995) review

these issues and conclude that F D1 in the US. is not a bad thing on balance.

3. F DI and Taxes

There are a variety of reasons why multinational corporations (MNCs) invest
abroad: acquiring particular resources (labor, natural resources, technological capability,

or management skill) at a lower cost, supplying goods and services to markets in the host

 

5 For more detailed discussion about the debates and empirical evidence, see Graham and Krugman (1995,

chapter 3-6).

country or in adjacent markets, seeking benefits from economies of scale and scope, or
promoting long-term strategic objectives (such as. advancing international
competitiveness by acquiring the assets of foreign corporations)6. In addition to these
factors, the host-country tax system also potentially affects the attractiveness of
investment by foreigners.

There are two different views on how foreign direct investment responds to tax
variables: the traditional “choice between alternative investment locations” view and the
new “asset competition” view. While the traditional view argues that FDI in the US.
should respond negatively to an increase in the US. tax rate, the new view argues that
F D1 in the US. should respond positively to an increase in the US tax rate.

According to the “choice between alternative investment locations” view, foreign
investors compare the after-tax rate of return in the US (Rus[1-Tus]) and that in the
home country (RH[I-TH]), where R, is the pre-tax return in a country, and Ti denotes the
effective tax rate in a country. Foreign investors decide to undertake investment in the
location with the higher after-tax rate of return. In this view, an increase in the US. tax
rate reduces the after-tax rate of return, and leads to a decrease in FDI in the US.

(that is, TusT ——> FDI I). Hartman (1984) and Boskin and Gale (1987) support this view.
The “asset competition” view criticizes the “choice between alternative investment
locations” view in that it ignores the possibility that the pre-tax rate of return may be
affected by the change in tax rate. In the asset competition view, competition for tax-
favored assets causes the relative price of the tax-advantaged assets to rise, and this

means that the pre-tax rate of return of the tax-favored assets decreases. Therefore, an

 

‘5 Dunning (1992), pp. 56-61.

increase in the US tax rate decreases the price of US assets, and increases the pre-tax
rate of return on the US assets. Since foreign investors from countries with a territorial
tax system7 do not pay taxes on foreign income to their home country, they receive the
full U.S. after-tax return (Rusll-TUSD. Firms from countries with a worldwide tax system
pay taxes to their home country and get a foreign tax credit (RU5[1-TH]), so in general,
their total taxes are not affected by the US. tax changes. However, according to the asset
competition view, firms from countries with a territorial tax system are not greatly
affected by an increase in the US tax rate, but worldwide firms increase their investment
in response to an increase in the US. tax rate. Swenson argues that an increase in the US.
tax rate leads to an increase in the US. pretax rate of return. These changes do not affect
much the investment incentive of territorial firms. but they make worldwide firms
increase their investment in the US. Accordingly, total F D1 in the US, which is the sum
of FDI from the territorial countries and FDI from the worldwide countries, should
respond positively to the US. tax rate. The asset competition view is addressed in depth
in Chapter 3.

The main purpose of this paper is to analyze the effects of US taxation on F DI in

the US. More specifically, this paper tries to provide empirical answers to some

 

7 Multinationals are subject to taxation by their home countries, and by their host countries. To relieve the
double-taxation problem, each country adopts either a “worldwide” system or a “territorial” system.
Under the worldwide system, multinationals get credits for taxes paid to foreign countries, but under the
territorial system, multinationals do not have to pay taxes on foreign earnings. The US, the UK, and
Japan adopt the worldwide system. France, Germany, Canada, and the Netherlands adopt the territorial
system.

I This statement is based on the assumption that the rate of taxation of the worldwide country is greater

than that ofthe U.S.

questions regarding the relationship between FDI in the US and the US effective
corporate tax rate:

(a) Does FDI in the US respond positively or negatively to an increase in the US
effective corporate tax rate measure?

(b) How does the foreign investor country’s double-taxation-relief system affect the
response of FDI in the US. to the US effective corporate tax rate?

(c) Is the average effective tax rate or the marginal effective tax rate the better
explanatory variable for FDI behavior?

(d) What is the response of establishment-type F DI to both the average effective tax rate
and the marginal effective tax rate?

This dissertation is organized as follows. In Chapter 2, I review the relevant
literature and discuss the data issues. The argument of the New View and the empirical
analysis of Swenson (1994) are reconsidered in Chapter 3. In Chapters 4 and 5, I
investigate the effects of US. effective corporate tax rate on F D1 in the US, as well as
the effects of the foreign investor country’s international tax system on FDI in the US.
This paper is different from Hartman (1984), Boskin and Gale (1987), and Slemrod
(1990), in that I use acquisition and establishment data, while they use capital/low data.
Even though I use a different F DI data set, I get results that are similar to those of
Hartman, Boskin and Gale, and Slemrod. Also, this paper is different from Swenson
(1994) in that my calculation of the average effective tax rates is different from hers. I get
opposite results from those of Swenson (1994), even though I use the same FDI data.

I explain the data on FDI and the effective tax rates in Chapter 2. In Chapter 6, this paper
compares the average effective tax rate and the marginal effective tax rate in explaining
F DI behavior. Also in Chapter 6, I investigate the response of establishment-type FDI to

7

the average and the marginal effective tax rate. Finally, Chapter 7 provides a conclusion.

CHAPTER II
LITERATURE REVIEW AND DATA ISSUES

1. Literature Review

Early studies on the effects of taxation on FDI in the US concentrate on analyzing
the relationship between capital inflow and measures of the after-tax rate of return.
Hartman (1981, 1984), the first contribution, analyzes the effect of taxes on FDI, using
annual capital flow data9 for the years 1965 - 1979. He distinguishes FDI financed by
retained earnings from FDI financed by funds transfers from parent firms, and estimates
separate equations for these two types of F DI. He uses the after-tax rate of return realized
by foreign investors, the after-tax rate of return on overall US. capital, and the tax rate on
US. capital owned by foreigners relative to the tax rate on US. capital owned by US.
investors as explanatory variables. He excludes explicitly the home-country tax rate as an

explanatory variable, based on the arguments of Hartman (1985)”). He finds a positive

 

9 There are two different kinds of FDI data available: one is the acquisition and establishment data, and the
other is the direct investment capitalflows data (balance-of-payments data). The capital flow data consist
of the equity capital. the intercompany debt between US. affiliates and their parents, and the reinvested
earnings. If a foreign affiliate borrows in the US. in order to buy a machine, this transaction will not appear
in these data. For more detailed explanation for the capital flow data, see section 2.2.

'0 Hartman (1985) argues that the home country’s rate of tax on foreign source income and the presence or
absence of a foreign tax credit should be irrelevant to a mature foreign subsidiary’s investment and
dividend decision. He compares after—tax rates of return from two alternative investment decisions
(reinvestment or immediate repatriation) of a mature foreign subsidiary. The parent firm’s after-tax rate of
return from reinvestment is [(l-t)/(1-t*)][1+r*(l-t*)], and that from repatriation is [(1-t)/(l-t*)][1+r(l-t)],
where t is tax rate, r is rate of return, and * stands for a host-country variable. Therefore, the home-country

system of deferring taxes and providing a credit for host-country tax payment (worldwide system) induces

9

effect of the after-tax rate of return on both types of FDI, and the effect is much stronger
for FDI financed by retained earnings. He also finds a negative effect of the relative tax
term on F DI. These results suggest that FDI (especially FDI by retained earnings)
responds negatively to the US. average tax rate.

Boskin and Gale (1987) reestimate Hartman’s (1984) equations, using updated
series for the tax rate and the rate of return. Their qualitative results are consistent with
those of Hartman (1984), even though the estimated elasticities of F DI to the rate of
return are somewhat lower. They also try some different model specifications and
different sample periods. They conclude that although the results are somewhat sensitive
to sample periods and model specification, the qualitative conclusions of Hartman (1984)
are fairly robust.

Young (1988) uses revised data on foreign direct investment and rate of return for
the sample period 1953-1984 to estimate equations similar to those of Hartman. He
modifies the Hartman model by including the lagged dependent variable and US. GNP
as explanatory variables. His results indicate that foreign direct investment in the US.
through retained earnings appears to be more sensitive to tax rates than foreign direct
investment through transfer of new funds. This is consistent with the findings of
Hartman.

Slemrod (1990) points out that previous studies ignored the possibility that FDI

could be affected by home-country taxationl 1, and did not consider nontax determinants

 

multinational firms to invest abroad up to the point at which the after-foreign-tax return available abroad
equals the available domestic after-tax return. He explains that this conclusion arises from the unavoidable
nature of the tax for a firm with subsidiary earnings.

” Slemrod (1990) points out that Hartman‘s (1985) argument does not apply to an immature subsidiary’s

investment (FDI financed by funds transfers from the parent company), but only applies to a mature

IO

of F DI. Therefore, he investigates the effect of both US. and home-country taxation on
FDI in the US. Firstly, he changes the model specification by introducing a marginal
effective tax rate as a tax variable. rather than an average effective tax rate, and by
including other explanatory variables, such as the real exchange rate. The results are that
the US. effective tax rate has a negative effect on total FDI and FDI financed by transfer
of funds, but not on investment from retained earnings. Secondly, Slemrod disaggregates
FDI data by the major investing countries. It is expected that investors from territorial-
system countries would respond sensitively to the US. marginal effective tax rate, but
that for investors from worldwide-system countries, the US tax rate would be a less
crucial factor. However, Slemrod‘s estimation results generally do not support this
prediction. About the effect of the home-country taxation on FDI in the US, it is
expected that FDI from territorial countries should be positively related to the home-
country tax rate. But transfers of funds from worldwide countries should have a less
positive relation to the home-country tax rate, and retained earnings from those countries
should be unaffected. The regression results do not support these predictions. Slemrod
concludes that these results suggest that home-country taxation does not have much effect
on FDI in the US.

Auerbach and Hassett (1993) distinguish acquisition-type FDI from establishment-
type F DI. They argue that establishment FDI could be affected by the marginal effective
tax rate, and that acquisition FDI should be affected primarily by the laws regarding

mergers and acquisitions (M&A). They construct an investment model of FDI using

 

subsidiary’s investment (F Dl financed by retained earnings). So, he argues that it is worthwhile to
investigate empirically the effect of both the home country’s rate of taxation and its system of taxing

foreign-source income.

11

particular M&A tax-code provisions. The model is neoclassical in character, and is based
on the assumptions of the neoclassical model. Their simulation results. which are based
on their investment model of FDI, suggest that the Tax Refomr Act of 1986 (TRA86)
decreased the incentives for worldwide firms to invest in all assets other than equipment,
and that TRA86 decreased the overall investment incentives of territorial firms.
However, in reality, this prediction was not supported by the actual FDI trend. They
conclude that the FDI boom in the late 19803 might not have been significantly affected
by TRA86.

Swenson (1994) analyzes F D1 in 18 industries, for the years 1979 through 1991.
She uses average tax rates calculated by Tax Analysts. She finds that the response of
worldwide investors to the US average tax rate was positive and statistically significant.
For territorial investors, the effect of the average tax rate on FDI is much smaller and
statistically insignificant. These results are consistent with the argument of Scholes and
Wolfson (1992) that FDI by worldwide firms responds positively to the tax rate, and this
makes total FDI respond positively to the tax rate. I replicate her estimation, and the
results will be discussed in chapter 3.

Table 1 shows the differences in F DI data, the tax rate measures used, and the main

results, among the empirical studies above.

12

Table 1. Summary of Empirical Studies

 

 

 

 

 

 

 

 

 

 

Hartman Boskin & Young Slemrod Swenson
(1984) Gale (1987) (1988) (1990) (1994)
FDI data capital flow capitalflow capital flow capital flow acquisition and
establishment
(1965-1979) (1956-1984) (1953-1984) (1960-1987) (1979-1991)
Main A fter-tax After-tax A fter-tar ME TR AE T R
Explanarory Rate of return Rate of return Rate of return
Variable
Conclusion
(US. tax negative negative negative negative positive
Effect on
F DI)

 

"' AETR stands for the average effective corporate tax rate, and METR for the marginal effective corporate

tax rate

13

 

2. Data Issues

2.1. FDI Data

There are two different kinds of FDI data available: one is the acquisition and
establishment data, and the other is the direct investment capital flows data (balance-of-
payments data). Both data sets are available in the Survey of Current Business published
by the Bureau of Economic Analysis.

The acquisition and establishment data set consists of the actual outlays of foreign
investors to establish or acquire new US affiliates. Roughly speaking, establishment
involves creation of a new legal entity by a foreigner, and acquisition involves obtaining
a voting interest in an existing US. business by a foreigner. These data are available
since 1979. Swenson (1994) uses this data set.

The capital flow data consist of equity capital, intercompany debt between US
affiliates and its parents, and reinvested earnings. If a foreign affiliate borrows in the
US in order to buy a machine, this transaction will not appear in this data. Hartman
(I984), Boskin & Gale (1987), Young (1988), and Slemrod (1990) use this data set.

The acquisition and establishment data and the capital flow data provide different
measures of the annual growth in FDI in the US. The acquisition and establishment data
cover the actual outlays to establish or acquire new US. affiliates, regardless of how or
by whom the investment was financed. Thus the outlays may be made by either the
foreign parent or an existing U.S. affiliate, and the source of financing may be other than
the foreign parent group, such as local borrowing by existing U.S. affiliates. In contrast,
the balance-of-payments data cover only transactions between foreign parent groups and

US. affiliates. If, for example, a US. affiliate of a German chemical manufacturer

I4

acquired a US chemical company by borrowing funds in the US, the borrowed funds
would be included in the acquisition and establishment data but not in the capital flow
data, because the acquisition did not involve funds from the foreign parent”.

As mentioned above, the capital flow data cannot capture the foreign direct
investment financed by borrowing in the US. Young (1988) admits the limitation of the
capital flow data by saying “F D1 in this study is most accurately thought of as a financial
transaction. It does not necessarily mean purchase of real assets, such as plants and
equipment, and does not include funds borrowed in the US” Auerbach and Hassett
(1993) point out that the capital flow data are not directly related to physical investment,
which is of interest to the researcher, and on which are based the theoretical models to
form effective tax rates. Swenson (1994) also agrees that the acquisition and
establishment data provide a more accurate assessment of the tax effect on FDI. So, I
think the acquisition and establishment data set is more appropriate than the capital flow
data for explaining investment behavior. The acquisition F DI, the establishment FDI, and
the total FDI (sum of acquisition F DI and establishment FDI) can be broken down by
industries. And the total FDI can be disaggregated by some major foreign investor
countries. In chapter 4, I disaggregate total F DI (the sum of the acquisition FDI and the

establishment FDI) by four industriesl3 . And in chapter 5, I use the total FDI by seven

 

'2 Quijano (1990).

'3 I calculate the average effective tax rates on the basis of corporate income tax return data in Statistics of
Income (SOI) Bulletin published by the lntemal Revenue Service. So, my industry classification is
restricted by the industry classification of the SOI Bulletin (eight industries: Agriculture-Forestry-Fishing,
Mining, Construction, Manufacturing, Transportation and Public Utilities, Wholesale and Retail Trade,
Finance-lnsurance-Real Estate, and Services). And FDI data in Survey of Current Business have some data
suppressions to avoid disclosure of data of individual companies. Therefore, to avoid losing too much

information, I consolidate some industries into one industry. The four industries are (1) manufacturing, (2)

15

major investor countries. In chapter 6, I disaggregate the acquisition FDI and the
establishment FDI by four industries. According to these data sets. most FDI are

accounted for by the acquisition FDI (see Figure 2).

 

trade, (3) services and FIRE (finance, insurance, and real estate), and (4) other industries.

16

Figure 2. Foreign Direct Investment in the US. 1979 — 1996

 

90000

 

 

FDI in the US. in millions of US. dollars

 

 

 

LL «Sta?
+acq

i—N—est
”w l

 

 

 

 

17

2.2. Data on Effective Tax Rates

Since the tax burden of a corporation is dependent on many factors, such as
corporate financial policy, inflation expectations, uncertainty, and the tax code, it is
difficult to measure the exact burden of corporate income taxation. Especially, the tax
code has lots of complexities, such as various deductions, credits, and exemptions, which
make the statutory tax rate different from actual tax burden. The average effective tax
rate and the marginal effective tax rate are contrived to estimate the actual tax burden or
to measure the incentive effects of taxation.

The average effective tax rate is said to be a good measure of the tax burden of
existing firms, but it is not necessarily a good measure of the impact of taxes on the
incentives to make new investment. In contrast, the marginal effective tax rate measures
the estimate of the net tax burden on the expected income from a marginal investment. So
it is said to be a good measure of the incentive effect of taxes on new investment, but it
cannot be a good measure of the tax burden of existing businesses”.

Swenson ( l 994) uses the average effective corporate tax rates as the main
explanatory variable for FDI in the US. However, Slemrod (1990) argues that the
marginal effective corporate tax rate is a better measure of expected tax burden on a
prospective new investment, and uses the marginal effective corporate tax rate as the tax-
rate measure. All studies dealing with the effects of taxes on F DI use corporate level

effective tax rates rather than the total of corporate and personal level effective tax rates.

 

'4 Fullerton (1984), p. 23.

18

This is because foreign investors pay taxes at the corporate level, but not at the personal
level.

In order to see which tax rate measure is the more appropriate explanatory variable
for F DI behavior, I use both average effective corporate tax rates and marginal effective
corporate tax rates.

The average effective corporate tax rate is generally defined by actual corporate
taxes paid as a proportion of corporate income.15 The average effective tax rate is used to
measure the overall tax burden of existing firms. Tax Foundation’s calculation of US
average effective corporate tax rates is one of the most recent calculations of the AETR”.
Tax Foundation calculates the annual U.S. AETR for the years 1945 — 1998. I follow the
method of Tax Foundation in calculating the average effective tax rates. I calculate the
average effective corporate tax rate, based on Statistics of Income Bulletin (801 Bulletin)
data published by the lntemal Revenue Service (IRS). The average effective tax rate is
calculated by taking federal corporate income tax paid divided by net income. Figure 3
shows my calculation of the AETR for the US. as a whole. Table 7 in chapter 4 presents
my calculation of the average effective tax rates by industry. The calculation of the AETR
of Tax Analysts and that of Tax Foundation are compared in Figure 5 in Chapter 3.

The marginal effective tax rate measures the fraction of the real pretax rate of
return to a new investment that will be collected as taxes. It is used to measure the
incentive effect of taxes on new investment. One of the most recent calculations of the
marginal effective tax rates is the one by Gravelle (1994). She calculates the US overall

METR for the years 1953 - 1989. Figure 4 compares my calculation of the METR and

 

'5 Fullerton (1984), p. 24.
'6 Moody (1998).

19

that of Gravelle (1994). The major difference between my own calculation and
Gravelle’s calculation of the METR is the expected inflation data. While Gravelle (1994)
uses the expected inflation rates of the Drexel-Burnham-Lambert Decision Makers Poll, I
use the expected inflation rates of the Livingston Survey data maintained by the Federal
Reserve Bank of Philadelphia. The methodology and data used in calculating the METR

are presented in an appendix.

20

Figure 3. The Average Effective Corporate Tax Rates in the US.

40%
i 35%
30%
25%
20%
15%

10%

0%

Source

5%g

 

The Average Effective Corporate Tax Rates

 

 

 

 

 

 

 

 

 

 

 

1979 1980 1981 1982 1983 1984 1985 1988 1987

1988

1989

1990

1991

1992

1993

1994

1995

1998

: My own calculation, based on corporate tax return data in Statistics of Income Bulletin

published by the lntemal Revenue Service.

21

i
I

4'

Figure 4. Comparison of the Marginal Effective Corporate Tax Rates

70%

50%

40% -'

30%
20%
f 10%
i

0%

 

 

The marginal effective corporate tax rates

 

60% -

 

 

 

 

 

 

 

 

 

 

1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996

Sources: Gravelle (1994) and text.

22

l-O—KIM .. ‘
1+.-___._§:av_¢"_¢x

One of the most important differences between this paper and the paper of
Swenson (1994) is the calculation of the average effective tax rates. She uses the average
effective tax rates calculated by Tax Analysts. but I calculate the AETR using the
corporate tax return data from the Internal Revenue Service (IRS).

Tax Analysts use data for the 500 largest companies in calculating the AETR for
the years 1980 — 1984, and for the 1,000 largest companies for the years 1985 — 1989.
The data are derived from the Form IO-Ks that publicly held corporations are required to
file with the Securities and Exchange Commission (SEC). More detailed discussion of
the Tax Analysts’ calculation of the AETR will be found in chapter 3.

My calculation of the AETR depends on the data on corporation income tax returns
in SO] Bulletin. The data are broken down by eight broad industries. The data contain the
income tax returns filed by all US. corporations, so the number of firms included in the
IRS data is several million. In 1980, the number of returns filed was 3.7 million. and in

1996 it was 4.6 million.

23

CHAPTER III
THE NEW VIEW OF
TAXES AND FOREIGN DIRECT INVESTMENT

l. The New View

According to Scholes and Wolfson (1992). if two assets give rise to the same pre-
tax cash flows, but the cash flows to one asset are taxed more favorably than those to the
other asset, taxpayers will bid for the tax-favored asset. This competition for the tax-
favored assets causes the relative price of the tax-advantaged assets to rise. This
competition continues until the equality of the post-tax rate of return between the tax-
advantaged and tax-disadvantaged assets is restored. As a result, the price of the tax-
favored asset will increase, and the before-tax rate of return on the tax-favored asset will
decrease. Scholes and Wolfson refer to this reduction in the before-tax rate of return of
the lightly taxed asset as an ‘implicit tax’. They use an example of municipal bonds,
which are issued by state and local governments. The interest earned on municipal bonds
is exempted from federal income taxation. Investors bid up the price of these municipal
bonds, such that their before-tax rate of return is lower than the return on fully taxed
bonds

Scholes and Wolfson extend the ‘implicit tax’ hypothesis to the discussion of the
effects of taxes on FDI. They suggest that investors whose home country use a worldwide
tax system can respond positively to a change in the host country’s tax rate. They state
this as follows:

24

“One might be tempted to suppose that if tax rates are lower in a foreign country
than they are domestically, it follows that after-tax rates of return on marginal
investments should be higher in the foreign country. But this ignores the very real
possibility that pretax rates of return in the foreign country may be lower than that
available domestically. In countries where the tax rate on income is relatively low,
one would expect competition to force down pretax profitability. In other words, the
foreign investments will bear implicit taxes. --------- Implicit taxes can arise because
the foreign country encourages investment by offering generous tax benefits, and
competition for the right to garner these benefits results in lower before-tax rates of

return. --------- (for worldwide investors) it is only necessary to compare before-tax

rates of return for one-year horizon investment.”l7

Swenson (1994) makes Scholes and Wolfson’s argument in a more general form,
which predicts the response of investors from territorial countries as well as that of
investors from worldwide countries to a change in the US. tax rate. She also provides
empirical evidence supporting Scholes and Wolfson’s ‘implicit tax’ hypothesis in the
context of F DI and taxes. According to the argument of Swenson (1994), firms invest
until their after-tax rate of return on the investment is equal to the return on the passive
asset. In her own notation, Rus(1-Tus) = RF, where RF is the after-tax rate of return on an
alternative passive asset (or requisite return). In this view, an increase in the US. tax rate
(T as) increases the before-tax return on US. assets (Rug). Since foreign investors from
countries with a territorial tax system do not pay taxes to their home country, they receive
the full U.S. after-tax return (RT = RUSH-Tits] = RF). Firms from countries with a
worldwide tax system pay taxes to their home country and get a foreign tax credit, so

they ultimately earn RW = RUSH-Tug) + RUSTUS — RUSTH = RUSH-TH). According to

 

'7 Scholes and Wolfson (1992), pp. 253-254.

25

Swenson (1994), firms from countries with a territorial tax system are not greatly affected
by an increase in Tu; because their after-tax return on direct investment in the US. (R r =
RU5[1 -TU5]) is fixed at RF regardless of the US. corporate tax rate. But, firms from
worldwide tax countries increase their investment in response to an increase in Tug,
because an increase in the US. tax rate (Tug) leads to an increase in the US. pretax rate
of return (Rug). These changes do not affect the investment incentive of territorial firms,
but they do make worldwide firms increase their investment in the US. Accordingly, FDI

in the US should respond positively to the US. tax rate.

2. Problems of the New View

The conclusion of the new view is derived from two assumptions: perfect capital
mobility among countries, and perfect mobility between physical and financial assets.
Under the assumption of perfect international capital mobility, an immediate adjustment
in the pretax rate of return is possible. If the assumption of perfect capital mobility does
not hold, the conclusion that worldwide investors should respond positively to an
increase in the US. tax rate cannot be derived. However, what is generally observed and
accepted in the related literature is far remote from these assumptions. Under the
assumption of perfect capital mobility between physical and financial assets, the
conclusion that territorial investors should not respond to a change in the US. tax rate is
possible. And this assumption is also far from reality. Therefore, the “asset competition”
view can be said to have limitations in explaining the real behavior of direct investment,
because of its unrealistic assumptions. This section discusses the assumption of the new

26

view in more detail.

Capital mobility between countries

The asset competition view (the New View) can hold only under the situation of
perfect competition and perfect capital mobility between countries. Scholes and Wolfson
(1992, p. 84) state, “. . .we assume that markets are perfect. In this setting, no transaction
costs are incurred to undertake investments or to manage them. All investors are assumed
to possess identical information regarding the future cash flows from investment
alternatives. Moreover, investors act as though their behavior has no influence on the
prices at which assets can be bought and sold.” Swenson (1994, p. 249) states, “. . .we
begin with the premise that asset competition, and perfect capital mobility drive the after-
tax rate of return on US. assets into line with the world rate of return.”

However, the consensus in the academic literature is that capital is not perfectly
mobile among countries. Feldstein and Horioka (1980) and Mishkin (1984) report
empirical evidence suggesting that capital is internationally quite immobile. Feldstein and
Horioka analyze the relation between savings rates and investment rates of the 21 OECD
countries for the period 1960 — 1974. Their basic finding is that the correlation between
savings rates and investment rates is close to one (0.89). They conclude that this result is
not consistent with perfect capital mobility. Mishkin analyzes real interest-rate
differentials among seven developed counties for the period 1967 -— 1979. He finds that
there exist real interest-rate differentials across countries. Persistent differentials in real
interest rates are also inconsistent with perfect capital mobility.

More recently, F eldstein and Bacchetta (1990) find that, while there is still a

27

positive correlation between savings rates and investment rates, the size of the correlation
has decreased during the 19805. This appears to be have been due to (1) the relaxation of
explicit capital controls in some countries, (2) the ongoing integration of the world capital
market, and (3) the vast increase in the borrowing needs of the United States.
Nevertheless, the results of Feldstein and Bacchetta are still consistent with a substantial
degree of international capital immobility. They find that, even in the 19805, each
additional dollar of domestic saving is associated with more than 50 cents of additional
investment in the domestic economy. Frankel (1990) and Obstfeld (1993) provide
reviews of the capital-immobility literature. Obstfeld concludes that “Capital mobility
appears noticeably lower between industrial economies than it is within them, although
intereconomy capital mobility certainly has increased over time. . . .. It is doubtful that
capital will ever be as mobile between nations as it can be within them. The mere
existence of national governments sovereign within their borders means that no investor
can think about domestic and foreign assets in quite the same way.””3
Gordon and Bovenberg (1996) provide six possible explanations for the observed
capital immobility: Existence of capital controls, high transaction costs, exchange-rate
risk, fear of expropriation, asymmetric information across countries, and market power of

large countries. And they argue that the most plausible explanation is asymmetric

information between foreigners and domestic residents. They state this as follows:

“Investors, by living and working in a particular country, know much more about
the economic prospects of that country than they do about those in other countries.
When foreigners try to acquire a firm in the country, they can easily end up being
overcharged by domestic owners, who have access to better information not only

 

'8 Obstfeld (1993), pp. 66-67.

28

about that specific firm, but also about future government policies affecting the firm.
When foreigners buy domestic inputs or services or pay domestic income taxes, their
lack of information can again lead to overpayment. Foreigners’ lack of knowledge
can result also in a less efficient use of resources, due for example to their poorer
ability to forecast market demand in a new setting or to deal with idiosyncratic

aspects of the domestic contract law, the local distribution system and supply

network, and local customs governing labor relations.”'9

Substitutability between direct investment and portfolio investment

The perfect-capital-mobility assumption of the asset competition view is a result of
not distinguishing direct investment and portfolio investment. That is, the asset
competition view regards direct investment as a perfect substitute for portfolio
investment. However, the immediate or quick bid up for the tax-favored asset is possible
only in the case of portfolio investment. It is generally accepted that portfolio investment
is not a perfect substitute for FDI. Caves (1996) and Graham and Krugman (1995) list
several reasons why direct investment is different from portfolio investment. First of all,
the motivation of direct investment is different from that of portfolio investment. The
primary motivation of FDI is to extend control over foreign businesses, because this
extension of control can improve corporate strategic value in a variety of ways. The
primary motivation of portfolio investment is simply to earn higher returns. In other
words, portfolio investment pursues higher Short-term returns, while direct investment is
carried out pursuing long-terrn returns, including intangible benefits. Secondly, as
Auerbach and Hassett (1993) point out, a significant portion of FDI is financed locally. If

the motivation of FDI were simply the higher interest rate, FDI should not be financed

 

'9 Gordon and Bovenberg (1996), p. 1059.

29

locally. Thirdly, FDI among advanced countries typically moves in both directions across
national boundaries. If FDI were merely designed to find a higher rate of return, then F DI
should move unilaterally from a country with a lower rate of return to other countries
with higher rates of return. Finally, shifting capital between direct investment and
portfolio investment entails adjustment costs and time. Therefore, portfolio investment

cannot be a perfect substitute for direct investment.

Capital mobility between physical assets and financial assets

Swenson (1994) assumes that the after-tax rate of return on physical assets and
financial assets is always equal. In her own expression, the equality of Rugfl-Tus) = RF
always holds. But, this equation can only be satisfied under the equilibrium state after
every adjustment process finishes, or under perfect mobility between physical assets and
financial assets. In reality, the economy is not in an equilibrium state, even though it is
always moving toward an equilibrium state. And as physical investment is not a perfect
substitute for portfolio investment, capital is not so mobile between physical assets and

financial assets.

3. Problems in the New View’s Empirical Analysis

Even though the conclusion of the New View depends heavily on unrealistic
assumptions, Swenson (1994) provides some estimation results supporting the argument
of the New View. She gets a positive and significant FDI elasticity with respect to the

30

AETR. So, I try to replicate her estimation. By using a data set that is very similar to
hers”, I find a positive FDI elasticity with respect to the average effective corporate tax
rate. However, the elasticity is much smaller than hers, and it is statistically insignificant.
In addition, I find that there are some serious mistakes in her analysis that may make her
results unreliable. One problem is that the average effective tax rate calculated by Tax
Analysts is not an accurate estimate of the US. average effective corporate tax rate. That
is, Tax Analysts’ calculation of the average effective tax rate has some bias, and so it
does not seem that Tax Analysts’ average effective tax rate is a representative US.
average tax rate. The other problem is that Swenson uses Tax Analysts’ data very

arbitrarily in her analysis. These points will be explained in depth below.

3.1. Problems of the Average Effective Tax Rates by Tax Analysts

Tax Analysts’ main goal of their calculation of the AETR seems to be to calculate
large US. firms’ overall average effective tax rates. According to the methodology part
of the book Effective Corporate Tax Rates, Tax Analysts selected 500 or 1,000 of the
largest US firms, ranked in terms of gross sales and assets, as their sample firms, and
calculated overall US. average effective corporate tax rates for the years 1980 - 1989. In
turn, Tax Analysts grouped these firms into from 54 to 91 industrieszl, and calculated the

average effective corporate tax rates for each industry.

 

2° Swenson’s sample period is 1979-1991, and she says her data source is Eflective Corporate Tax Rates by
Tax Analysts (Swenson, 1994, pp. 254-255). But, I cannot get Tax Analysts’ average effective tax rate data
for the years of 1979, 1990, and 1991. According to Tax Analysts, they published six volumes of the book
Eflective Corporate Tax Rates and the books have the AETR data only for the years 1980-1989.
2' 54 industries for the years 1980 — 1984, 91 industries in 1985, 84 industries in 1986, 83 industries in

31

The first problem of using Tax Analysts’ AETR in FDI analysis lies in the fact that
Tax Analysts’ data are based on information in the financial statements of only the largest
US. firms. Firms have a tendency to make their financial statements look good for a
variety of reasons. And there is some evidence22 that large firms engage in more
deception. Therefore, it is difficult to say that the average effective tax rates calculated by
using Tax Analysts’ data are the representative average effective tax rates of the United
States. Accordingly. the result from the estimation with biased data cannot be a reliable
one. Figure 5 compares Tax Analysts" average effective corporate tax rates with Tax
Foundation’s average effective corporate tax rates”. It turns out that TA’S average
effective tax rates are substantially different from the average effective corporate tax rates
calculated using more information.

The second problem of Tax Analysts’ estimates for the average effective tax rates
is associated with the calculation of the average effective tax rates by industry. Tax
Analysts’ selected 500 or 1,000 largest US. firms and categorized them into 54 - 91
industries, according to the Standard Industrial Classification. This method may result in
more biased estimates for the industry-level average effective tax rates, because some
industries may include very few firms. If Tax Analysts’ purpose of calculating AETR
were a calculation of large firms’ AETR for the selected industry, a more appropriate
sampling method would be to select firms ranked in the top 20 or 30 percent in each

industry.

 

1987 and 1988, and 74 industries in 1989.
22 US. Department of Treasury (1999) and Manzon and Plesko (2001).
23 Tax Foundation (1998) calculated the US. average effective tax rates by using Statistics of Income data

(by IRS) for the years 1950-1994.

32

Figure 5. Comparison of the AETR of Tax Analysts and that of Tax Foundation

 

‘ Average Effective Corporate Tax Rates

 

l 35%

 

 

 

 

 

i
l 30%
25% -
20% l 7 g _ -
'—0— Tax Analysts
-If Tax Foundation
, 150A, ,_____g__#_____,________ _-‘ WW 7,. W L!_ .__,,_,A . - .__ . .. .,_i, _-M,_
1
100A, __.__ _______,_L__,_,_#___‘, ,,,, ._,__,__-_, , . ________ii _ - __
i
‘ 5%._____.i,, , , .,,,_,________,, _- ifi-__ , .. ”UL—H _ n
j 0% - - .
l 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
‘ YEAR

33

3.2. Problems of the Way the New View uses Tax Analysts’ Data

Since the industry classification of Swenson (1994) is different from that of Tax
Analysts, Swenson recalculates the average effective tax rates in the US. by using Tax
Analysts’ data. The only information about how Swenson estimates the average effective
tax rates by using Tax Analysts’ data is Table 3 in Swenson (1994), which compares the
two-year average of the effective tax rate before the Tax Reform Act 1986 (TRA 1986,
hereafter) and after TRA 1986. In order to find out how she uses Tax Analysts’ data and
check out whether my replication of her estimates for the AETR is correct, I try to
reproduce her Table 3. Table 2 compares the numbers in Swenson’s Table 3 and the
replication results of my own. I can reproduce her Table 3 except for the ‘petroleum
industry’, ‘finance industry’, and ‘insurance industry’. That is, the results of reproduction
are close for all industries except these three industries. She might have made errors in
calculating the average effective corporate tax rates of these three industries. Table 2
simply indicates that I understand how Swenson estimates the average effective tax rates
by using Tax Analysts’ data.

By reproducing Swenson’s Table 3, I find that she uses the AETR of a large-digit
industry to represent that of a small-digit industry. That is, she uses the AETR of a subset
of an industry to represent the AETR of the entire industry. In the case of the year 1988,
the number of industries used in Swenson’s analysis is 18, but that of Tax Analysts is 83.
To calculate the AETR for each of the 18 industries, Swenson should have regrouped Tax
Analysts’ 83 industries into 18 industries, and then should have calculated the AETR for
the small-digit industries using all of the information from the large-digit industries that
are included in that small-digit industry. To make this point clearly, and to Show how this

34

method distorts the AETR for each of the 18 industries, Table 3 is presented. Taking a
look at the ‘food industry’ of Swenson’s classification in Table 3, the ‘food & kindred
products’ industry, ‘miscellaneous food production’ industry, ‘dairy products’ industry,
‘meat products’ industry, and ‘beverages‘ industry of Tax Analysts’ classification are
included in the ‘food industry’. The average tax rate of the entire food industry is
30.32%, but that of food & kindred products is 33.46%. That is, Swenson uses the AETR
of the ‘food & kindred’ industry to represent that of the entire ‘food’ industry. The

situation of other industries is the same as that of the ‘food industry.’

35

Table 2. Replication Result of Swenson (1994)’s Table 3

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 3 in Swenson (1994) Replication Results
INDUSTRY (84+85).’2 (87+88)l2 (84+85)i2 (87+88)/2
Mining 29.2 17.3 29.2 17 3
Petroleum 45.9 35.9 16.3 26.0
Food 27.3 32.8 27.3 32.8
Chemicals 21.2 28.2 21.2 28.2
Metals 30.9 32.5 28.0 32.5
Machinery (Non-electrical) 24.6 26.0 24.0 26.0
Machinery (Electrical) 28.0 30.0 27.4 30.0
Printing & Publishing 34.3 38.2 34.3 38.2
Rubber & Plastics 18.7 22.3 18.7 22.3
Stone, Clay, and Glass 27.3 30.6 27.3 30.6
Instruments 22.7 24.9 22.7 24.9
Wholesale trade 26.0 . 33.2 26.1 33.2
Retail trade 33.5 37.1 33.5 37.1
Banking 9.9 18.1 9.9 18.0
Finance 16.0 19.4 17.7 16.2
Insurance 15.9 19.4 4.3 19.4
Real estate 21.2 24.1 21.2 24.1
Services 30.4 31.4 30.4 31.4

 

 

 

 

 

 

36

 

Table 3. Illustration of Swenson’s Calculation of the AETR for Each Industry

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Swenson's 18 industries Tax AIILIV'SIS. 83 industries (in 1988) AETR
Mining Mining(l3) 0.1078
Oil and gas extraction (8) 0 469
Total (21) O 1299
Petroleum Petroleum refining (26) 02702
Food Food & kindred products (7) 0.3346
Food production lvliscellaneous (19) 0.2833
Dairy products (4) 0,3806
Meat products (10) 0.2877
Beverages (7) 0.2801
Total (47) 0.3032
Chemicals Chemicals (20) 0.29
Chemicals; Miscellaneous (21) 0.2499
Drugs (17) 0.2634
Soap and other detergents (12) 0.2555
Total (70) 0 2715
Metals Metals: Fabricated (24) 0.3183
Metals: Primary (25) O 2226
Total (49) 0.2394
Non-electric Machinery and equipment: General industrial (12) 0.234
Machinery Machinery. Construction, mining, and materials handling (10) 0.2774
Machinery: Special industrial (3) 0.3656
Machinery and equipment. Metal working (3) 0.2464
Engines and turbines (7) 02569
Total (35) 0.2669
Electric Electrical computing and office machines (22) 0.1047
Machinery Electrical: Household appliances (2) 0.2716
Electrical components and accessories (15) 0.3828
Electrical and electronic machinery equipment. and supplies (8) 0.2579
Electrical machinery, equipment. and supplies: Miscellaneous. (9) 0.1349
Equipment-communication (19) 04235
Total (75) 0.2366
Printing & Printing“ Commercial (9) 0.3486
Publishing Publishing and printing-newspapers (13) 0.326
Publishing and printing-periodicals and books (8) 0.4023
Total (30) 0.3437
Rubber & Rubber (5) 0.2259
Plastics Plastics (6) 0.256
Total (11) 0,2445

 

37

 

Table 3 (Cont’d).

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Swenson's 18 industries Tax Analysts” 83 industries (in 1988) AETR
Stone, clay, Stone. clay. glass. and concrete products ( l3) 0 3416
and glass
Instruments Measuring and controlling instruments (26) 0201
Wholesale trade Durable goods (23) 0.3076

Nondurable goods: Drugs (4) 0.3754
Nondurable goods: Miscellaneous (20) 0.3201
Total (47) 03202
Retail trade Apparel and accessory stores ( l 1) 0.3171
Building materials and hardware stores (5) 0 2808
Department stores (9) 0.3138
Drug stores (6) 0.4243
Eating and drinking places (12) 0 3362
Grocery stores (19) 0.2671
Miscellaneous (14) 0.3734
Variety stores (13) 03148
Total (89) 0.3024
Banking Banking: National banks (50) 0.16081
Banking: State banks (40) 0.1889
Total (90) 0.1704
Finance Federal savings and loan (4) 0.4379
Credit agencies: Other than banks (1 1) 0.2791
Savings and loan holding companies (17) 0.2731
Financial services (6) 02227
Total (38) 02607
Insurance Insurance (36) 02471
Real Estate Real estate (7) 0.2316
Servrces Services: Advertising agencies (3) 04133
Services: Enginering and architectural (5) 0 3196
Services: Personal (4) 03198
Services: Computer and data processing (5) 0.1057
Services: Health services (10) 0.305
Services: Hotels and motels (2) 0.2589
Services: Miscellaneous (12) 0.3226
Services: Motion pictures (3) 0 31 12
Total (44) 0.2671

 

 

 

 

Foot notes) 1. Numbers in parentheses: number of sample firms.

2. Tax Analysts’ industry in bold font: Swenson has this industry represent the entire industry.

38

3.3. Replication Results of Swenson’s Estimation

Industry-level argrlysis

Table 4 compares my replication results of Swenson’s industry-level estimation and
Swenson’s original estimation results. The first two columns present a part of the original
estimation results of Swenson (1994, Table 4). These columns Show that the magnitudes
of the elasticities of the AETR are positive (around one: 0.75 - 1.13), and the elasticities
are statistically significant. Because column two is Swenson’s baseline specification, I
reestimate a second column with data for the period of 1980 —198924.

Column three and column four report the replication result. The coefficients in
column three are GLS estimates, and those in column four are OLS estimates with
heteroskedasticity-robust t-value. In both columns. the estimated elasticities of F DI with
respect to the AETR are positive, but they are very small and statistically insignificant.
Since F D1 in the US. might be affected by the US business situation, U.S. real GDP

might be thought to be a relevant control variable. And if US. real GDP is correlated

 

2’ The number of observations in Table 4 of Swenson (1994) is 229. Considering Swenson’s industry
classification (18 industries) and the period of analysis (13 years: 1979-1991), there must be 5 missing
observations in her data set. However, I find that there are more than 5 missing observations. In early
volumes of the Survey of Current Business (SCB for 1979, 1980, 1981, and 1982), there are only 13
industries that match with Swenson’s 18 industries. And there are also some missing observations arising
from data suppressions in the Survey of Current Business for the purpose of avoiding disclosure of data of
individual companies. In addition to missing observations in the FDI data, there are also some missing
observations in the AETR data. The AETRs calculated using Tax Analysts’ data indicate that the Banking
industry and the Insurance industry have negative tax rates for several years (early 19805). Taking

logarithm of the negative AETRs results in missing observations.

39

with one ofthe explanatory variables, omitting real U.S. GDP will cause biased
estimates. Therefore, as Hartman (1984). Boskin and Gale (1987). and Young (1988) do.
I try to add U.S. real GDP as a control variable. The fifth column shows the results of the
estimation with US. real GDP. U.S. real GDP appears to have a strong explanatory
power, and the F DI-elasticity with respect to the AETR becomes close to zero, and the
statistical significance of the estimates is reduced greatly.

As mentioned in the previous section, Swenson uses the AETR of a large-digit
industry to represent that of a small-digit industry. Common sense tells us to calculate the
AETRs for each industry by using all data of the sample firms in each industry.
Therefore, I correct the industry-level AETRS by dividing the US. taxes paid by firms in
each industry by the net income of the firms in each industry. To see if the use of the
corrected AETRs for each industry leads to the same result, I redo the estimation with the
corrected AETRS. Column six and column seven show the reestimation results with these
corrected AETRs. The F DI-elasticities with respect to the AETR are much smaller
(almost zero) than those in columns three and four. The estimation results using the
AETRS for each industry calculated more appropriately no longer support Swenson’s
conclusion. Column eight shows the results of the estimation with US. real GDP. Real
U.S. GDP appears to have a strong explanatory power, and the AETR variable turns out
to have no explanatory power.

In conclusion, Swenson’s estimation suffers from the mismatching problem that the
AETRs of the large-digit industry represent the small-digit industry. And her estimation
results also might have the ‘omitted variable bias.’ I conclude that I cannot replicate
Swenson’s industry-level estimation with data the set available, and that it is hard to think
of Swenson’s estimation result as reliable.

40

Table 4. Replication result of Swenson’s industry-level estimation

 

 

 

 

 

 

 

 

 

 

 

 

Swenson Replicated Corrected
(1994, Table 4)
GLS GLS GLS OLS OLS GLS OLS OLS
log(AETR) 1.13 0.75 0.23 0.24 0.14 -0.02 0.06 -0.03
(3.26) (2.30) (0.93) [0.93] [0.56] (-0.13) [0.23] {-0.13}
log(EX) -l.38 -l.42 -1.82 -1.79 -1.28 -l.96 -l.95 -l.38
(-2.38) (-3.01) (-5. l 3) [-4.22] {-2.83] (-5.50) [~5.15] {-3.19]
Time trend 0.08 0.08 0.16 0.15 -0.35 0.17 0.16 -0.38
(3.94) (4.69) (8.06) [4.75] {-2.96] (8.33) [5.08] {-3.17]
Industry no yes yes yes yes yes yes yes
dummies
log(GDP) - - - - 15.64 - - 16.38
[4.36] [4.73]
N 229 229 151 151 151 153 153 153
R-squared 0.16 0.45 - 0.55 0.59 - 0.55 0.60

 

* AETR and EX stand for the average effective tax rate and exchange rate, respectively.

" Number in ( ) is ordinary t-value and number in [ ] is heteroskedasticity-robust t-value.

41

 

Country-level analysis

Table 5 shows the replication result of Swenson’s country-level analysis. The first
column is a part of Swenson’s (1994) Table 6, which is her baseline specification. The
first column of Table 5 shows that, while the estimated F DI-elasticity with respect to the
AETR for investors from worldwide countries is positive (2.99) and statistically
significant, the elasticity for investors from territorial countries is close to zero (0.09) and
statistically insignificant. The second column shows my replicated estimation results,
which seem to support Swenson’s results that worldwide investors respond positively to
the US. effective corporate tax rate. but territorial investors do not respond to US.
effective corporate tax rate.

However, it is hard to say that Swenson’s estimation is reliable, even though
Swenson’s country-level estimation is qualitatively replicated. Since F D1 in the US.
might be affected by the US. business situation, U.S. real GDP might be thought to be a
relevant control variable. And if US. real GDP is correlated with one of the explanatory
variables, omitting U.S. real GDP will cause biased estimates. Therefore, I try to add U.S.
real GDP as a control variable. The third column shows the results of the estimation with
US. real GDP. Adding log(GDP) reduces the magnitudes of the FDI elasticities with
respect to the AETR for both worldwide and territorial investors. The estimated FDI
elasticity with respect to the AETR for worldwide investors is much smaller than before,
and is now statistically insignificant. And the estimated FDI elasticity for territorial
investors becomes negative, and is still statistically insignificant. On the other hand, U.S.

real GDP itself appears to have a strong explanatory power.

42

In addition, as mentioned earlier, Swenson uses Tax Analysts‘ data in calculating
the average effective tax rate. Tax Analysts’ data are based on information in the
financial statements of only the largest US firms. Since there is some evidence that large
firms engage in more deception, it is difficult to say that the average effective tax rates
calculated by using Tax Analysts’ data are the representative average effective tax rate of
the United States. Accordingly, the results from the estimation with biased data cannot be

considered to be reliable.
4. Summary

Theoretically, the prediction of the new view is based on two highly unrealistic
assumptions: perfect capital mobility among countries, and perfect mobility between
physical and financial assets. If these assumptions do not hold, it is not possible to derive
the predictions that worldwide investors should respond positively to an increase in the
US. tax rate, and that territorial investors should not respond to a change in the US. tax
rate. Instead, what is generally observed and accepted in the related literature is far
remote from these assumptions. .

And empirically, there are some serious problems in the estimation process of the
new view. The estimates for the average effective tax rates are not so reliable. An
important control variable is missing. And the estimation results of the new view cannot
be replicated.

In the following chapters, I analyze some issues regarding the relationship between
FDI in the US. and the US. effective corporate tax rate, with more appropriate estimates
for the average effective tax rate and more appropriate econometric models.

43

Table 5. Replication result of Swenson’s country-level analysis

 

 

 

 

 

 

Swenson
(1994, Table 6)
log(WWTAX) 2.99 3.83 1.56
(2.66) (2.57) (0.87)
log(TTAX) 0.09 1.16 -1 .11
(0.09) (1.05) (-0.73)
log(EX) -0.39 -l .75 -l .71
(-1.42) (-2.70) (-2.73)
Time trend yes yes yes
Country dummies yes yes yes
log(GDP) - - 19.41
(2.12)
N 69 57 57
R-squared 0.36 0.66 0.69

 

* WWTAX, TTAX, and EX stand for the average effective tax rate for worldwide worldwide

 

investors, that for territorial investors, and exchange rate, respectively.

** Number in ( ) is t-statistics.

44

CHAPTER IV

OVERALL U.S. TAX EFFECT ON F D1

1. Tax Consequences of the U.S. Source Income of Foreign Investors

According to the definition of foreign direct investment in Chapter 1, a foreign
person’s establishment of an unincorporated branch (same legal entity as the parent firm)
and investment in a U.S. corporation (separate legal entity) are both considered FDI in
the United States. The U.S. tax treatment of dividends and branch income is determined
by the U.S. source of income rule (lntemal Revenue Code, section 861 - section 865) and
the bilateral income-tax treaty between the U.S. and the foreign investor’s home country.

According to the U.S. source of income rule, the source of dividend income is
determined by the residence of the corporation paying the dividends. And the residence is
determined by the corporation’s country of incorporation. Therefore, a U.S. corporation
owned by foreign persons is subject to U.S. corporate income tax on its net income, and
in addition, the dividend income of the U.S. corporation is subject to flat 30%
withholding tax when dividends are repatriated. Usually, the withholding tax rate is
reduced to 0% - 15% by income tax treaties. For example, without treaty provisions, the
U.S. dividend income of a Canadian firm is subject to 30% withholding tax, but the U.S.-

Canada Income Tax Treaty reduces this rate to 5%25 .

 

2’ Protocol Amending the Convention Between the United States of America and Canada With Respect to

Taxes on Income and on Capital, Article 5.

45

Business profits of a U.S. branch are considered effectively connected income26.
Therefore, they are subject to U.S. corporate income tax, but not subject to withholding
tax. To equalize the tax treatment of foreign corporations operating through a U.S. branch
and those operating through a U.S. subsidiary, TRA 1986 introduced a new branch profits
tax, which imposes a 30% tax on the repatriated profits of a U.S. branch of a foreign
corporation”.

Therefore, the tax burden on the U.S. income of a foreign investor depends on the
rates of U.S. corporate income tax and withholding tax (or branch profits tax). However,

for simplicity, I ignore withholding taxes, as the previous studies do”.

2. Empirical Model Specification

In this section, I introduce an econometric model that is designed to explain overall
FDI in the United States. Thus, at this stage, all F D1 are considered together, regardless
of whether the FDI is in the form of an acquisition or an establishment, and regardless of
whether the investment originates in a worldwide country or a territorial country. In later

chapters, we will consider separate estimation for different categories of FDI.

 

26 A foreign person’s U.S. income that is “effectively connected” with the conduct of a trade or business in
the United States is subject to U.S. net—basis taxation. For more details, see U.S. Joint Committee on
Taxation (I999).

27 Slemrod (19901:), p. i 78.

28 For example, Hartman (1984) argues that he ignores withholding taxes because they are relatively

constant over time.

46

For the estimation of the overall tax effect on F DI in the U.S., I use the following

specification.

(1) log(FDIu): a0 + a) log(ETRp) + a2 log(BCA,) + a3 log(GDP,) + £61161: + up,
where

F D1,, = total foreign direct investment of industry i in year t,

ET R), = effective tax rate of industry i in year t,

BCA, = balance on current account in year t,

GDP, = U.S. real GDP in year t.

d, = industry dummy,

u), = error term.

The dependent variable is the log of foreign direct investment, and the main
independent variable is the log of the effective tax rates. In general, when a variable is a
positive dollar amount, the log is often taken. And when a variable is a proportion or a
percent, there is a tendency to use them in level form”. But I make the independent
variable (ETR: the effective tax rate) appear in logarithmic form, because the double-log
specification has a nice property (the coefficient of the independent variable in the
double-log model is an elasticity). I test whether the independent variable should take the
logarithmic form or the level form using the Davidson-MacKinnon test”. The result of
the Davidson-MacKinnon test suggests that the ETR can take either the logarithmic form

or the level form. The coefficient a 1 in the above model is the elasticity of foreign direct

 

2° Wooldridge (2000), p. 135.
3° Wooldridge (2000), pp. 283-284.

47

investment with respect to the effective tax rate. I use both the average effective tax rate
and the marginal effective tax rate as the independent variable, to see which effective tax
rate measure has a better explanatory power in explaining foreign direct investment

behavior. The rationales for the control variables are as follows:

(a) Real U.S. GDP: Foreign direct investment in the U.S. may be influenced by U.S.
economic conditions, and the economic conditions are measured by U.S. GDP. Therefore,
I enter real U.S. GDP as a control variable. Econometrically, if U.S. GDP has some
correlation with one of the explanatory variables, and if it has an effect on dependent
variable (F DI), omitting the GDP variable will cause biased estimates. Hartman (1984),
Boskin and Gale (1987), and Young (1988) also include a measure of aggregate economic
activity in the U.S. as a control variable in their models,31 although they use nominal
GNP. While they use nominal GNP, I use real GDP as the output variable. In the double-
log specification, entering nominal variables instead of real variables in both the right-
hand side and the left-hand side is the same as assuming that the price level is
uncorrelated with the nominal variable on the right-hand side. I test for the correlation by
including the log of the GDP-deflator in the double-log specification with the variable
defined in nominal terms. It turns out that the price level is correlated with nominal GDP.
Therefore, using the double-log specification with nominal F DI as the left-hand side
variable and nominal GDP as the right-hand side variable causes the “omitted variable

bias.”

 

3' Hartman (1984) and Boskin and Gale (1987) use log(FDI/GNP) as dependent variable in their models.

This is equivalent to including log(GNP) as an independent variable, with a fixed coefficient of 1.

48

(b) U.S. balance on current account: Figure 6 compares the trend ofFDI in the U.S. and
that of the U.S. balance on current account, during the 1979 — 1996 period. These trends
suggest that a surge in FDI in the U.S. approximately coincides with a surge in the U.S.
current-account deficit. This implies that foreign multinational companies invest a part of
the surplus from trade with the U.S. in purchasing U.S. firms and establishing U.S.
subsidiaries. In other words, FDI in the U.S. may have a negative relationship with the
U.S. current account balance. Therefore. I include the U.S. balance on current account as

a control variable.

(c) Industry dummies: There are many industry-Specific factors, such as the capital-labor
ratio, entry barriers, the market growth rate, and economies of scale, that determine the
characteristics of an industry. These factors differ among industries, and they may
influence the FDI. To control for these industry-specific factors that affect foreign direct

investment, I include industry dummies.

49

Figure 6. FDI in the U.S. and the U.S. balance on Current Account: 1979 — 1996.

100.000 -~—--- -»-— --———--- —— . . .r--. - . ._ ___ _.._

’ 50,000 -.

 

o , ___Mi,-__ _.,- _- _.__,_ ,-_.___ ,2. _.__ _ ._-_--_-______

i1979 1980 1981 1 1983 1984 1985 1986 1987 1988 1989 1990 9 1992 1993 1994 1995 1996

-50,000 .

militlons of U.S. dollars

400,000 '

 

-150,000 ~‘

 

 

 

 

-200.000 1 — m.,*_ ~- w- w--

Sources: Economic Report of the President and Survey of Current Business.

* ‘fdi’ and ‘bca’ stand for F D] in the U.S. and the U.S. balance on current account, respectively.

50

3. Data.

For the estimation of the overall tax effect on F DI. annual data on foreign direct
investment are needed. Earlier Studies that use the capital flow data analyze the overall
tax effect on FDI with annual FDI data. However, the acquisition and establishment data
by the BEA that I use are only available Since 1979, so I cannot get a sufficiently large
number of observations. To avoid the problem from small number of observations, 1
disaggregate the annual FDI data by four industries: manufacturing, trade, finance and
services, and other industries”. Table 6 shows FDI (the acquisition and establishment)
data for the four industries.

Table 7 and Table 8 Show my calculation of the average effective tax rates and the
marginal effective tax rates for the four industries33 . The methodology and the data
sources for my calculation of the average effective tax rates are provided in Chapter 2,
and those for my calculation of the marginal effective tax rate are provided in Chapter 2

and an appendix.

 

32 I calculate the average effective tax rates on the basis of corporate income tax return data in Statistics of
Income (80]) Bulletin published by the lntemal Revenue Service. So. my industry classification is
restricted by the industry classification of the SOI Bulletin (eight industries: Agriculture-Forestry-Fishing,
Mining, Construction, Manufacturing, Transportation and Public Utilities, Wholesale and Retail Trade,

F inance-lnsurance-Real Estate, and Services). In addition, the FDI data in the Survey of Current Business
have some data suppressions to avoid disclosure of data of individual companies. Therefore, to avoid losing
too much information, I consolidate some industries into one industry. The four industries are (1)
manufacturing, (2) trade, (3) services and FIRE (finance. insurance. and real estate), and (4) other
industries.

33 Table 8 shows that the marginal effective corporate tax rates do not vary much among industries after
1987. This is because the marginal effective tax rate for equipment and that for structures are almost same

after 1987.

51

Table 6. FDI in the U.S. by Industry: 1979-1996 (millions of U.S. dollars)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year Manufacturing Wholesale & FIRE & Services Others All Industries
Retail Trade

1979 4,170 890 NA NA 15,317
1980 3,629 1.221 NA NA 12 172
1981 8.074 859 NA NA 23,219
1982 2,379 1.146 4,974 2,318 10,817
1983 3,1 13 293 3.995 690 8,091
1984 3,106 1,994 5,099 4,998 15,197
1985 12,140 2,021 4,925 4.020 23,106
1986 16,772 6,889 13,184 2,333 39,177
1987 19,751 2,483 15,088 2,988 40,310
1988 36,136 10.476 17,742 8,337 72,692
1989 35,958 4,495 22,932 7,776 71,163
1990 23,898 2,926 32,251 6,857 65,932
1991 11,461 2,228 10,862 986 25,538
1992 6,014 954 5,801 2,564 15,333
1993 1 1,090 2.332 9,707 3,100 26,229
1994 21,218 3,698 14,481 6,229 45,626
1995 26,643 4.006 19,669 6,879 57,195
1996 27,835 7,734 34,775 9,587 79,929

 

Source: Survey of Current Business.

52

 

Table 7. Average Effective Tax Rates by Industry: 1979-1996

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Industry
Year
FIRE &

Manufacturing Trade Services Others
1979 0.2681 0.2677 0.2656 0.1195
1980 0.2602 0.2754 0.2496 0.2783
1981 0.2579 0.3086 0.2600 0.31 17
1982 0.3514 0.3313 0.2797 0.3865
1983 0.2573 0.3031 0.2219 0.4128
1984 0.2510 0.2955 0.2524 0.3676
1985 0.2231 0.4044 0.2034 0.4097
1986 0.2530 0.4343 0.1727 0.5606
1987 0.2340 0.3679 0.2127 0.3822
1988 0.2043 0.2927 0.2172 0.2815
1989 0.2159 0.3505 0.2091 0.3697
1990 0.2274 0.4192 0.2083 0.4044
1991 0.2427 0.4012 0.2190 0.3926
1992 0.2287 0.3182 0.2178 0.3773
1993 0.2198 0.3048 0.2137 0.3356
1994 0.2248 0.2633 0.2075 0.3031
1995 0.2081 0.2907 0.1583 0.3058
1996 0.2058 0.2826 0.1759 0.2945

 

Source: Text.

53

 

Table 8. Marginal Effective Tax Rates by Industry: 1979-1996

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Industry

Year

Manufacturing Trade FIRE & Others

Services

1979 0.3668 0.4114 0.3996 0.3898
1980 0.4109 0.4437 0.4351 0.4279
1981 0.3293 0.3709 0.3600 0.3508
1982 0.3579 0.3910 0.3823 0.3750
1983 0.3195 0.3610 0.3501 0.3409
1984 0.3247 0.3657 0.3549 0.3459
1985 0.3120 0.3569 0.3451 0.3352
1986 0.2825 0.3358 0.3217 0.3100
1987 0.3223 0.3217 0.3219 0.3220
1988 0.3263 0.3252 0.3255 0.3258
1989 0.3350 0.3325 0.3332 0.3337
1990 0.3256 0.3245 0.3248 0.3251
1991 0.3279 0.3267 0.3270 0.3273
1992 0.3237 0.3229 0.3231 0.3233
1993 0.3407 0.3413 0.3411 0.3410
1994 0.3364 0.3376 0.3372 0.3370
1995 0.3412 0.3417 0.3416 0.3415
1996 0.3332 0.3348 0.3344 0.3341

 

Source: Text.

54

 

4. Estimation Results

Because my data set has a cross-sectional component (4 industries) and a time-
series component (18 years), I test for heteroskedasticity and serial correlation. I use the
Breusch-Pagan test to test for heteroskedasticity“, and the AR(!) serial correlation test35
to test for autocorrelation.

In the case of the regression of FDI on the average effective tax rate and other
variables (equation (1) in section 2 in this chapter), I cannot reject either the
homoskedasticity assumption or the assumption of no serial correlation at the 10%
significance level. In the case of the regression of FDI on the marginal effective tax rate
and other variables, I cannot reject either the homoskedasticity assumption or the
assumption of no serial correlation at the 10% significance level. Table 9 summarizes the

test results.

Table 9. Test Results (total FDI on measures of the effective tax rate)

 

 

 

 

Dependent Tax-rate Heteroskedasticity Autocorrelation
Variable Measure p-value* AR(1)* * p-value* * *
AETR No 0.73 No 0.19 0.13
log(FDI) METR No 0.11 No 0.18 0.16

 

 

 

 

 

 

 

 

 

* p-value of LM statistic
** estimate for the coefficient of AR( 1) model

*"p—value of t statistic

 

3’ For the methods for testing for heteroskedasticity, see Wooldridge (2000), pp. 255-261.
35 For the methods for testing for serial correlation, see Wooldridge (2000), pp. 380-387.

55

Table 10 reports the results of the regressions of F DI on the average effective tax
rate and other variables. Adding log(GDP) to the regression (column 2) reduces the tax
elasticity to -0.74 and its significance level significantly to —l .26. And log(GDP) turns
out to have a significant effect on log(FDI). This suggests that log(GDP) is a relevant
control variable. Foreign investors appear to increase their direct investment in the U.S.
when the U.S. economic situation is good and the GDP elasticity is about 2.5. Adding
log(BCA)36 to the regression (column 3) increases the tax elasticity slightly to ——0.85 and
its significance level to —1.45. And log(BCA) itself has a negative and statistically
significant effect on log(FDI). This suggests that log(BCA) is also a relevant control
variable. Foreign investors appear to decrease their direct investment in the U.S. as the
U.S. current account balance improves. The specification in column four includes
log(EX)37 as a control variable. The log(EX) does not make any difference. When
log(EX) is introduced, the coefficients and t-values of the other variables remain almost
the same.

Column five and column six show the estimation results when the exchange rate,
instead of the balance on current account, is controlled for. The exchange rate seems to
have a negative effect on F DI (column 5). However, after including log(GDP) as an

explanatory variable, the negative effect of the exchange rate on FDI is significantly

 

’6 ‘BCA’ stands for balance on current account. The balance on current account is actually positive in some
years, and negative in others. To take the logarithm of the balance on current account, it is needed to make
the smallest value of each year’s balance on current account have a value of one. 1 add a constant of
$195,818 (millions of U.S. dollars) to each year’s balance on current account. Therefore, the estimated
coefficients need to be interpreted with care. Hartman (1984), Boskin and Gale (1987), Young (1988), and
Slemrod (1990) also use this transformation method.

37 ‘EX’ stands for the trade-weighted exchange rate.

56

diluted. The Sign of the coefficient is changed and its t-value is reduced quite
significantly. And comparing the R-squareds in column two and column six, it is difficult
to say that the log(EX) is a relevant control variable.

According to my baseline specification (column 3), the elasticity of total FDI with
respect to the average effective tax rate is —O.85. This estimate is not statistically
significant at the 5% or the 10% significance level, but it is marginally significant at the
15% level (p-value: 0.15). And the estimation result tells that the sign of the FDI
elasticity with respect to the AETR is not positive at the 10% significance level. The 10%
critical value for a one-tailed test with 60 degrees of freedom is -1.296. The t-value of the
FDI elasticity with respect to the AETR is —1.45. Therefore, the hypothesis that the F DI
elasticity with respect to the AETR has a positive sign can be rejected at the 10%
significance level.

These results are consistent with the traditional view that foreign investors tend to
undertake investment in the location with the higher after-tax rate of return. And these
results are exactly opposite to those of Swenson (1994). She finds a positive and

statistically significant effect of the average tax rate on foreign direct investment.

57

Table 10. FDI by Industry and the Average Effective Tax Rate*

 

 

 

 

 

 

 

 

 

Dependent Variable: log oftotal FDI (in 1992 dollars)
Independent
Variables
(l) (2) (3) (4) (5) (6)
-1.74 -0.74 -0.85 -0.87 -1.33 -0.76
log(AETR) (-2.94) (-l.26) (-1.45) (-1.46) (-2.13) (-1.26)
2.62 2.56 2.67 2.71
log(GDP) _ (3.89) (3.88) (3.37) _ (3.35)
-0.05 -0.05
log(BCA)” _ _ (-l.90) (-1.89) _ _
0.18 -1.09 0.14
log(EX) _ _ _ (0.28) (-1.80) (0.21)
industry
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.5516 0.6421 0.6627 0.6632 0.5745 0.6424
# of
Obs. 66 66 66 66 66 66

 

 

 

 

 

 

 

 

 

* The number in ( ) is the OLS t value.
AETR, BCA, and EX stand for the average effective tax rate, the balance on current account, and the
trade-weighted exchange rate, respectively.

"”" log(BCA)=log(195,8 l 8+balance on current account).

58

To see which effective tax rate is a better explanatory variable for FDI behavior, I
replace the AETR with the METR. Table 1 1 presents the results of the regressions of FDI
on the marginal effective tax rate and other variables. As with the results from the
regressions with the average effective tax rate, adding log(GDP) and log(BCA) to the
regression (column 2 and 3) reduces the absolute values of the FDI elasticities with
respect to the METR and their significance levels quite substantially. In fact, when both
log(GDP) and log(BCA) are added, the METR variable has wrong sign. Whenever
log(GDP) is included in the regression, the METR has a very small t-statistics. Therefore,
it is impossible to say that the marginal effective tax rate has an important role in
explaining the FDI behavior. These results suggest that the average effective tax rate is a
better measure of tax burden than the marginal effective tax rate in explaining the
variation of total foreign direct investment. The reason why the marginal effective tax
rate is not a good explanatory variable for foreign direct investment will be addressed in
Chapter 6.

The specification in column four includes log(EX) as a control variable. The
log(EX) does not make any difference. The coefficients and t-values of the other
variables remain almost the same as they were when log(EX) was not included. This is
the same as the result from the regressions of FDI on the AETR and other variables.
Column five and column six show the estimation results when the exchange rate, instead
of balance on current account, is controlled. The exchange rate seems to have a negative
effect on F DI (column 5). However, after including log(GDP) as an explanatory variable,
the negative effect of the exchange rate on FDI is greatly diluted. And comparing the R-
squareds in column two and column six, it is difficult to say that the log(EX) is a relevant
control variable. These results are the same as the regression results with the AETR.

59

Table 1 1. FDI by Industry and the Marginal Effective Tax Rate*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent Variable: log oftotal FDI (in 1992 dollars)
Independent
Variables
(l) (2) (3) (4) (5) (6)
-3.50 -0.63 0.27 0.36 -3.18 -0.69
log(METR) (-2.65) (-0.45) (0.18) (0.22) (-2.51) (-0.47)
2.81 3.05 3.14 2.73
log(GDP) _ (3 .90) (4.22) (3.16) _ (2.79)
-0.05 -0.05
log(BCA)" _ _ (-1.70) (- l .68) _ _
0.10 -1.43 -0.09
log(EX) _ _ _ (0.14) (-2.55) (013)
industry
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.5409 0.6340 0.6510 0.6511 0.5858 0.6341
# of
Obs. 66 66 66 66 66 66

 

 

 

 

* The number in ( ) is the OLS I value.

METR, BCA, and EX stand for the marginal effective tax rate,

the balance on current account, and the trade-weighted exchange rate, respectively.
" log(BCA)=log( 195,818+balance on current account).

60

CHAPTER V

FDI IN THE U. S.
AND THE INTERNATIONAL TAX SYSTEM
OF THE INVESTOR COUNTRY

1. Overview of the International Taxation System38

International transactions, by their nature, entail a double-taxation problem that
could result in a heavy tax burden for multinational enterprises. In general, there are two
different approaches to avoiding the double-taxation problem. Some countries, such as
the United States, the United Kingdom, Japan, and Italy, adopt the worldwide (or
residence-based) system. Under the worldwide system, the home-country government
taxes the worldwide income of its residents, and then allows a credit for foreign taxes
paid on foreign-source income. The worldwide system is based on the principle that a tax
system should not distort the investment location decision of a home-country firm
(capital export neutrality). Other countries, such as France and the Netherlands, adopt the
territorial (or source-based) system. Under the territorial system, the home-country
government taxes only home-country-source income, and foreign-source income is
exempt from the home-country taxation. The territorial system is based on the principle

that there should be no discrimination within a country between domestic and foreign

 

3" For more details, see Ballard (1999).

61

investors, or between foreign investors from different countries (capital import
neutrality). In the case of Canada and Germany, by statute, they adopt a worldwide
system. However, by income tax treaty with the United States, Canada and Germany
exempt U.S.-source income from its domestic taxation.39 Thus, Canada and Germany are
considered to have a territorial system. from the perspective of foreign direct investment

in the U.S.

2. The Response of FDI in the U.S. to the International Tax Regime

The effect of a host country’s tax rate on inward foreign direct investment depends
on the tax system of the foreign investor’s home country. For example, when the home
country of the foreign investor has a territorial system, the effective corporate tax rate on
FDI is equal to the tax rate imposed by the host country, because multinationals do not
pay taxes on foreign-source income to the home-country government. Therefore,
differences among host-country effective tax rates would be expected to have an effect on
the investment location decisions of firms from territorial tax countries. The effect of the
host country’s tax rate would be expected to have less influence on foreign investment
from countries that have a worldwide system with a foreign tax credit. In a simple case

without deferral“), if the host country’s tax rate is less than the home country’s tax rate,

 

3" Slemrod (1990b). p.179.

4° Under the worldwide system, a subsidiary’s profits are typically not taxed until they are repatriated to its
parent firm, whereas a branch’s profits are taxed as they accrue. This is called deferral. Deferral in the
worldwide system gives firms an incentive to accumulate profits in low-tax countries, rather than to

repatriate them to its home country. For more details, see Alworth (1988), Altshuler and Newlon (1993),

62

the effective tax rate on FDI is the home country’s. but if the host country’s tax rate is
higher than the home country’s tax rate (the firm is in excess-foreign-tax-credit position),
the effective tax rate on FDI is the host country’s. Therefore. for firms from a worldwide
tax regime, the effect of the host country’s tax system is filtered through the tax system of
the home country, and may be substantially mitigated. Table 12 summarizes the
relationship between the home-country tax system and the after-tax return on U.S.

investments.

Table 12. Home Country’s International Tax System and After-tax Return

 

Home country’s

 

international After-tax return on U.S. investments
tax system
Territorial RUS(1‘TUS )

 

RUS(1'TUS ) + RUS(TUS - TH) = RIJS(]'TII ) ifTus < TH
RUS(1'TUS ) lfTUS > TH

Worldwide

 

 

3. Empirical Results of Previous Studies

According to the traditional view, it is expected that investment by territorial firms
would respond negatively to an increase in the U.S. tax rate, and that FDI from
worldwide countries should be insensitive to U.S. tax rates, or less sensitive to U.S. tax

rates than F DI from territorial countries. On the contrary, the New View (associated with

 

Altshuler, Newlon, and Randolph (1995). and Hines and Hubbard (1990).

63

Scholes and Wolfson (1990)) argues that worldwide investors should respond positively
to the U.S. tax rate, while territorial investors should not be sensitive to the U.S. tax rate.

Slemrod (1990) analyzes the effect of the home-country tax system on FDI in the
U.S. He uses the capital-flow data for four territorial countries and three worldwide
countries. He finds a significantly negative effect of U.S. taxes on F DI from only two
countries (France, Germany) out of four territorial countries, and from only two countries
(Japan, and the UK) out of three worldwide countries. In addition, he does not find any
support for the prediction that FDI from territorial countries should be more sensitive to
U.S. tax rates than FDI from worldwide countries. Slemrod interprets these results as
meaning that, because multinationals take full advantage of deferral and sophisticated
accounting and financial strategies, the home country’s system of alleviating double
taxation is not an important determinant of FDI.

Swenson (1994) uses the acquisition-and—establishment data for manufacturing-
sector FDI from four territorial countries and two worldwide countries. She finds a
positive and significant effect of U.S. taxes on F DI from worldwide countries. However,
in the case of territorial countries. the tax effect is much smaller than it is for the
worldwide investors, and it is rarely significant. These results conform to the prediction
of Scholes and Wolfson (1992) that worldwide investors respond positively to the host-
country tax rate, while territorial investors are not sensitive to the host-country tax rate.

While Slemrod does not find any significant effect of the home-country tax system
with the capital flow data, Swenson finds significant results supporting the argument of
Scholes and Wolfson (1992) with acquisition and establishment data. This chapter

investigates the effect of the home-country tax system on FDI in the U.S. with acquisition

64

and establishment data and with the average effective tax rate, which is different from

that used by Swenson.

4. Model Specification and Data

For the estimation of the effect of foreign investor country’s double-taxation-relief

system on FDI in the U.S., I use the following specification.

(2) log(FDIc, )= ap+ a;log(ETR,) + aglog(GDPJ + a 3 log(BCAa) + Zacdc + uc, ,
where

F DIC, = foreign direct investment of country c in year t,

E TR, = effective tax rate in year t,

BCAC, = U.S. balance on current account with country c in year t,

GDP, = U.S. real GDP in year t,

d, = country dummy,

uc, = error term.

The dependent variable is the log of FDI of country c in year t, and the main
independent variable is the log of the effective tax rate in year t. The rationales for the
control variables are the same as those in the previous chapter, except for the country
dummies. Country-specific factors, such as differences in language, differences in
geographical location, and cultural affinity, might have an influence on the

communication and information cost and the transaction cost of FDI. To control for these

65

country-specific factors, I include the country dummies.

The data used in this chapter are slightly different from those used in the previous
chapter. While the dependent variable in Chapter 4 was total outlays of FDI
disaggregated by industry, it is total outlays of FDI from seven major investor countries
in this chapter. The outlays of F DI by the seven major investor countries (XFDIC) account
for about 70-80% of total FDI (XFDli). Table 13 shows the acquisition and establishment
F D1 in the U.S. for seven investor countries.

I use the annual average and marginal effective tax rates for all industries as the tax
variable. Figure 7 presents my calculation of the annual average and marginal effective
corporate tax rates of the U.S. I provide the methodology and the data sources in Chapter
2 and an appendix.

For the estimation of equation (2), I use the U.S. balance on current account with
each foreign investor country, instead of total U.S. balance on current account. The
historical data on the U.S. balance on current account with major trade-partner countries
are downloadable from the home page of the Bureau of Economic Analysis. Figure 8
shows the movements of the U.S. balance on current account with the seven major

investor countries, along with F DI from the countries.

66

Table 13. FDI in the U.S. by Major Investor Countries: 1979-1996 (millions of dollars)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Territorial Countries Worldwide Countries
Year
Canada France Germany Netherlands UK. Japan Italy
1979 1,414 262 2,436 4,955 2,511 257 NA
1980 1,956 600 1,424 1,650 3,066 596 97
1981 6,084 903 1,149 572 6,178 616 NA
1982 1,196 455 601 330 3,128 587 213
1983 1,072 295 584 492 2.366 392 NA
1984 2,587 330 685 562 3,714 1,806 45
1985 2,914 754 2,270 771 6,732 1,152 NA
1986 6,503 2,491 1,351 4,700 8,572 5,416 166
1987 1,276 2,044 4,664 391 15,142 7,006 268
1988 11,360 4,199 2,090 2,214 22,559 16,188 313
1989 4,403 3.469 2,435 3,629 23,047 17,410 436
1990 3,430 10,217 2,363 2,247 13,096 19,933 1,786
1991 3,454 4,976 1,922 1,661 2,169 5,357 435
1992 1,351 406 1,964 1,331 2,255 2,921 2281
1993 3,797 1,249 2,841 2,074 8,238 2,065 375
1994 4,128 1,404 3,328 1,537 17,261 2,715 412
1995 8,02 1,129 13,1 17 1,061 9,094 3,602 NA
1996 9,70 6,021 12,858] 6,476 14,757 8,813 NA

 

Source: Survey of Current Business, various issues.

67

Figure 7. Effective Corporate Tax Rates in the U.S.: 1979 - 1996

 

45%

40%

35%

30%

25%

20%

15%

10%

5%

0%

 

 

.._. _. ..___4..___ _ .. .-i

 

 

 

 

 

 

 

 

 

x_

 

 

x A .
V v f

1979 1980 1981 I982 1983 1984 1985 1986 1987 1988 1989 1990

year

1991 1992 1993 1994 1995 1996

 

 

+ METR
+ AETR

 

 

 

Source

: text and appendix

68

Figure 8. The U.S. Balance on Current Account with the Seven Major Investor Countries

Millions of U.S. dollars

14,000

and F DI from the Countries (Sources: Survey ofCurrent Business and www.

12000 -~

10.000
8,000
6.000

4,000 -

2.000 -

O

-2.000 -
4,000 -'
—6,000 ‘
-8,000 '

12,000

10,000

8,000

6,000 «

4.000 <

        

 

 

       

 

bea.doc. gov)
CANADA
-—BCA
[ft-FDI
‘1—1979—Tn198‘0 7W;8.1.~T1982 ”198’ “T984798; 1988* 198 A “994 l 1995 " 1998
YEAR
FRANCE
J
1 i
l—I—BCA l
l—a—FDI

Millions of U.S. dollars

2.000 <

-2,000

4,000

   

 

1979 1980 1981 1982 198 1984 1985 1988 1987 198

1

 

 

 

69

Millions of U.S. dollars

Millions of U.S. dollars

15.000

10,000

5,000 «

-5,000

-10.000 .

-15.000

-20,000

-25.000

20,000

 

A

 

1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 991 1992 1993 1994 1995 1996

—:-BCA
-—i=oi

 

 

YEAR

Nethelands

 

18,000 4

16.000

14,000 <

12.000 .

10.000 4

8.000 .

6.000
4.000

2.000

A

 

1‘

l

 

 

 

1979 1980 1981 1982 1983 1984 1985 1987

YEAR

1988 1988 1989 1990

 

7O

 

25,000

20.000

10.000

5.000

Millions of U.S. dollars

-10.000 <

' -15.000

~20.000

15,000 4

-5.000 -

 

L

L

A

 

1979 1980 1981 1982 1983

1984

1985 1986

1987 1988 1-.g 3 1994 1995 1996

 

 

 

 

30,000

20,000 «

10,000 «

40.000

-30.000

Millions of U.S. dollars

-50,000

-70.000

YEAR

 

-20,000 «

40,000 1

-60,000 -

 

1979

1980

1981

   

1982

   

1983 1984

1985 1986 1987 1988 1989

1990 1991 1992 1993 1994 1995 1998

' + total

 

 

 

 

71

3,000

2,000

1,000

-1.000

-3.000

Miilons of U.S. dollars

4,000

-5.000 4

-6.000

 

-2.000 4

 

1986

1987

1988

1989

1990

1991 1992 1993 1994

 

 

72

5. Estimation Results

Effective Tax rates and FDI of Territorial Investors

To see how the foreign investor country’s double-taxation-relief system affects
FDI in the U.S., I run regressions of FDI from the territorial countries and regressions of
FDI from the worldwide countries on the effective tax rate measures, separately.

Table 14 presents the estimation results of the regressions of FDI from territorial
countries on the average effective tax rate. I test for heteroskadasticity and serial
correlation, and it turns out that there is little evidence of heteroskedasticity or
autocorrelation. Column two, column three, and column four show that adding log(GDP),
log(BCA), and log(EX) do not have a great effect on the FDI elasticity with respect to the
AETR for investors from territorial countries. These results suggest that the average
effective tax rate is the most important factor for territorial investors’ investment-decision
making. According to the estimation results from my baseline specification (column 3),
the average-effective-tax-rate elasticity of FDI from territorial countries is negative (-3.5)
and statistically significant (t-value: -2.69). This result is consistent with what is expected
by the traditional view: territorial firms should respond negatively to an increase in the
U.S. effective corporate tax rate.

Table 15 reports the estimation results of the regressions of FDI from territorial
countries on the marginal effective tax rate. Adding control variables changes the sign of
the METR elasticity and reduces its statistical significance level quite substantially.
According to the estimation results from my baseline specification (column 3), the
marginal effective tax rate has little power in explaining the FDI behavior.

73

Table 14. F DI from Territorial Countries and the Average Effective Tax Rate*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent Variable: log oftotal FDI (in 1992 dollars)
Independent
Variables .
(I) (3) (3) (4) (5) (6)
-3.90 -3.47 —3.54 -3.18 -3.40 -3.13
log(AETR) (-4.62) (-2.66) (-2.69) (-2.37) (-3.64) (-2.34)
0.42 0.38 0.23 0.29
log(GDP) _ (0.43) (0.39) (0.24) _ (0.29)
-0.06 -0.07
log(BCA)“ _ _ (-0.80) (-0.92) _ _
-0.66 -0.63 -0.61
log(EX) _ _ _ (-1.24) (-1.20) (-l .15)
country
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.3669 0.3686 0.3747 0.3894 0.3804 0.3812
# of
Obs. 72 72 72 72 72 72

 

* The number in ( ) is the OLS I value.

AETR, BCA, and EX stand for the average effective tax rate, the balance on current account, and the

exchange rate, respectively.

** log(BCA)=log(26,320+ba1ance on current account).

74

 

Table 15. F DI from Territorial Countries and the Marginal Effective Tax Rate*

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Dependent Variable: log of total FDI (in 1992 dollars)
Independent
Variables
(1) (2) (3) (4) (5) (6)
-1.93 1.22 1.56 0.34 —2.34 -0.01
log(METR) (-1.47) (0.79) (0.97) (0.19) (-l.88) (-0.004)
2.82 2.94 2.05 1.93
log(GDP) _ (3.34) (3.42) (1.95) _ (1.86)
-0.07 -0.07
log(BCA)” _ _ (-0.83) (-0.82) _ _
-0.88 -1.56 -0.89
log(EX) _ _ _ (-1.43) (-3.01) (-1.45)
country
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.1909 0.3077 0.3149 0.3362 0.2937 0.3292
# of
Obs. 72 72 72 72 72 72

 

* The number in ( ) is the OLS i value.

METR, BCA, and EX stand for the marginal effective tax rate.

the balance on current account, and the exchange rate, respectively.

" log(BCA)=log(26,320+balance on current account).

75

 

Effective Tax rates and FDI of Worldwide Investors

Table 16 presents the results of the regressions of FDI from worldwide investors on
the average effective tax rate. Adding log(GDP) to the regression (column 2) reduces the
absolute value of the AETR elasticity and its significance level substantially. And
log(GDP) turns out to have a significant effect on log(FDI). Adding log(BCA) to the
regression makes the AETR elasticity smaller, and reduces its significance level further.
Log(BCA) also has a significant effect on log(FDI). These results suggest that the
average effective tax rate is not an important factor for worldwide investors’ investment-
decision making. According to the estimation results from my baseline specification
(column 3), the average-effective-tax-rate elasticity of FDI from territorial countries is
positive (0.45) but its t-value is quite small (0.25). This result is consistent with the
expectation of the traditional view that FDI from worldwide countries should be
insensitive to U.S. tax rates, or less sensitive to U.S. tax rates than F DI from territorial
countries.

Table 17 presents the results of the regressions of F DI from worldwide investors on
the marginal effective tax rate. The estimation results from the baseline specification
(column 3) show that marginal effective tax rate has little power in explaining F DI
behavior. This result is similar to the result from the regression of territorial FDI on the

marginal effective tax rate.

Conclusion

Comparing the elasticity of FDI in column (3) in Table 14 and that in column (3) in

Table 16, we see that FDI of territorial countries responds negatively to an increase in the

76

U.S. average effective tax rate. while the response of FDI from worldwide countries to
the U.S. average effective tax rate is insignificant. This finding is consistent with the
expectations of the traditional view that the investment by the territorial firms would
respond negatively to an increase in U.S. tax rate, and that F DI from worldwide countries
should be insensitive to U.S. tax rates, or less sensitive to U.S. tax rates than FDI from
territorial countries.

Column three of Table 10 shows that overall FDI is negatively affected by the
AETR (elasticity: -0.85), but the effect is only marginally significant. This estimate may
be interpreted as a mingled effect of the results in Table 14 and Table 16. That is, the
overall elasticity of FDI with respect to the AETR is the result of the combination of
territorial investors’ sensitive response and worldwide investors’ insensitive response.

Comparing the elasticity of FDI in column (3) in Table 15 and that in column (3) in
Table 17, the marginal-effective—tax-rate elasticities of F D1 of both worldwide and
territorial investors are statistically insignificant. So, we cannot say that the marginal
effective tax rate has any relationship with F D1 of either territorial firms or worldwide
firms. This finding suggests that the marginal effective tax rate is not a good explanatory
variable for FDI behavior. This result is generally similar to the result from the overall

analysis in Chapter 4, and the reason is addressed in Chapter 6.

77

Table 16. FDI from Worldwide Countries and the Average Effective Tax Rate*

 

Dependent Variable: log oftotal FDI (in 1992 dollars)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Independent
Variables
(1) (2) (3) (4) (5) (6)
-3.45 0.77 0.45 0.44 -2.07 0.74
log(AETR) (-2.68) (0.42) (0.25) (0.25) (-1.56) (0.41)
4.23 3.87 3.06 3.29
log(GDP) _ (2.95) (2.75) (2.04) __ (2.15)
-0.15 -0.14
log(BCA)“ _ _ (-1.89) (-1.76) _ _
-0.83 -1.44 -0.94
log(EX) _ _ _ (-1.43) (-2.51) (-1.58)
country
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.6951 0.7463 0.7663 0.7774 0.7342 0.7656
# of
Obs. 48 48 48 48 48 48

 

* The number in ( ) is the OLS I value.

AETR, BCA, and EX stand for the average effective tax rate, the balance on current account, and the

exchange rate, respectively.

** log(BCA)=log(7l,747+balance on current account).

78

 

Table 17. F DI from worldwide countries and the Marginal Effective Tax Rate*

 

Dependent Variable: log oftotal FDI (in 1992 dollars)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Independent
Variables
(1) (2) (3) (4) (5) (6)
-5.55 -2.46 -l.54 -l.93 -4.33 -2.77
log(METR) (-3 .39) (-1.24) (-0.76) (-0.96) (-2.72) (-1.42)
2.87 3.05 2.05 1.76
log(GDP) _ (2.48) (2.68) (1.58) _ (1.36)
-0.13 -0.1 1
log(BCA)” _ _ (-1.66) (-0.46) _ _
-0.91 -1.41 -l.02
log(EX) _ _ _ (-1.55) (-2.74) (-1.74)
industry
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.7187 0.7541 0.7691 0.7820 0.7656 0.7706
# of
Obs. 48 48 48 48 48 48

 

* The number in ( ) is the OLS t value.

METR, BCA, and EX stand for the marginal effective tax rate,

the balance on current account, and the exchange rate, respectively.

" log(BCA)=log(71,747+balance on current account).

79

 

CHAPTER VI

TYPES OF F DI AND EFFECTIVE TAX RATES

This chapter seeks to find out the answer to the question raised in Chapter 4 and
Chapter 5: “Why is F DI affected more significantly and strongly by the average effective
tax rate than by the marginal effective tax rate?” Swenson (1994) estimates not only the
effect of the average effective tax rate on F DI but also the effect of the marginal effective
tax rate on F DI. She finds a positive significant effect of the AETR on FDI but a negative
significant effect of the METR on FDI. She explains the negative METR-elasticity of
F DI with the limitations of the METR itself. That is, she says that she does not accept the
negative response of FDI to the METR as a reliable one, because she does not think the
METR is a good measure of tax burden. In my own estimation, I find a positive (0.27)
and quite insignificant (t-value: 0.18) coefficient. This suggests that the marginal
effective tax rate has nothing to do with FDI. This chapter tries to present a systematic
explanation for the relation between FDI and the measures of effective tax rates. In doing
so, this chapter takes advantage of the industrial organization theories of FDI.

This chapter also explores empirical evidence regarding the conjecture of Auerbach
and Hassett (1993). They conjecture that establishment FDI could be affected by the
marginal effective tax rate, but that acquisition FDI should not be affected by the
marginal effective tax rate. The second goal of this chapter is to answer the question,

“Does establishment FDI respond to the marginal effective tax rate?”

80

1. F DI and the Measures of Effective Tax Rates

As is shown in Chapter 4 and Chapter 5, our baseline estimates indicate that F D1 in
the U.S. does not respond significantly to the marginal effective tax rate. This
phenomenon may be explained by considering the characteristics of F DI and the
composition of F D1. While FDI is closely related to market imperfections, the concept of
the marginal effective tax rate is based on the assumption of perfect competition. While
most F D1 is accounted for by acquisition-type F DI (purchase of existing U.S. firms), the
marginal effective tax rate measures the tax burden of a new investment project. This
section discusses why the marginal effective tax rate has little power in explaining FDI

behavior, in terms of the characteristics of FDI and the composition of FDI.

1.1. Characteristics of F DI

Foreign direct investment is closely related to the concept of control and ownership
and the special characteristics of multinational corporations (MNCs, hereafter). That is,
in general, F D1 is undertaken by MN Cs for the purpose of obtaining control and
ownership of a business in a foreign country. The fact that MNCs are undertaking FDI is
very important in explaining the response of F D1 to the effective tax rates. According to
modern theories of FDI, the monopolistic characteristics of MNCs enable or drive MN C5
to make FDI. Therefore, the marginal effective tax rate, which is based on the assumption
of a perfect market, has limitations in explaining FDI, which is related to market

imperfections. To make clear this point, it is needed to review the industrial organization

81

theories of F D1. 1 summarize the industrial organization theories of F DI, based on

Graham and Krugman (1995), Casey (1998), and Caves (1996).

The monopolistic advantage theory: Hymer (1976)“ and Kindleberger ( 1969)

The monopolistic advantage theory explains how MNCs. which are in a
disadvantageous position compared to local firms, are able to do foreign investment.
A firm that undertakes investment in a foreign country faces some additional costs that a
local firm, which is limited to the local market, does not incur. These costs are related to
geographical distance between home and host country, exchange-rate risk, and
differences in languages, cultures, technical standards, and customer preferences. For a
MNC to overcome these disadvantages and invest in a foreign country, there must be
some advantages over its local rivals. These advantages take the form of economies of
scale or of superior technology. The conclusion of this theory is that FDI takes place

because the MNCs have enough monopolistic advantages over their local competitors.

The oligopolistic reaction theory: Knickerbocker (1973)

The oligopolistic reaction theory tries to explain the observation that large U.S.
multinationals tend to follow one another into foreign markets. A firm operating in an
oligopolistic industry is compelled to follow its rival into a foreign market, even though

the firm’s assessment of the expected profit of the project in the foreign market may not

¥

4' Originally, Stephen Hymer’s Ph.D. dissertation was written in 1959. It was published posthumously in

I 976.

82

be so good. This oligopolistic follow-the-leader behavior is motivated not by the desire to
earn more profit, but by the desire to avoid getting shut out of a new market by an
aggressive rival (leading firm). If the leader establishes a subsidiary in a foreign country,
the rivals recognize that this investment might give the leader first-mover advantage and
knock out their export business in the foreign country. These considerations dispose the

rivals to imitate the leader and establish their own subsidiaries in the foreign country.

The internalization theory: Buckley and Casson (1976)

Internalization theory criticizes the ”monopolistic advantage theory,’ on the ground
that it does not give an adequate explanation for why the MNCs choose F DI, rather than
alternative modes of servicing the foreign market (such as export and licensing).
According to internalization theory, for a MNC to service a foreign market via direct
investment, rather than exporting or licensing, there must be some internalization
advantage. That is, there must be economics for a MNC to exploit a market opportunity
through internal operations rather than through exporting or licensing. These economies
might be associated with costs of contract enforcement, or maintenance of quality or
other standards. If these costs are absent, firms use licensing or exporting as a means of
serving the foreign market. For example, a MNC can establish a subsidiary to maintain
dependable suppliers of intermediate goods or to protect leakage of technology

(maintaining technological monopoly).

83

The eclectic theory: Dunningfl 992)

According to the eclectic theory, for FDI to take place there should be three
advantages: ownership-specific advantages, intemalization-specific advantages, and
location-specific advantages. Firstly, an MNC should possess some advantages that its
local rivals do not have, such as product differentiation, advanced technology, and special
marketing and managerial skills. Secondly, there should be some internalization
incentives. That is, FDI should be more profitable than licensing or exporting. Thirdly,
the foreign country should have some specific characteristics (e.g., low labor costs,
abundance of raw materials, infrastructure, etc.) that can be linked with the other two
advantages. For a MNC, which has owner-specific and intemalization-specific

advantages, to invest in a certain country, the country must have some characteristics to

attract the MNC.

By reviewing the industrial organization theories of FDI, we can understand that
FDI is closely related to market imperfections and that F D1 is a highly strategic behavior.
Only firms with monopolistic advantages can undertake FDI. The firms in an
oligopolistic industry may have incentives to undertake FDI. Maintaining technological
monopoly is a source of market power in the product market, and thus it gives rise to an
incentive to undertake FDI.

In addition, FDI entails externalities. Examples include the introduction of new

technology (e.g., just-in-time inventory method from Japanese automakers) and the

84

training of workers who may transfer their skills elsewhere”. So, we can conclude that
FDI is closely related to market imperfections, and that FDI is a result of the strategic
decision-making of a MNC. However, the concept of the marginal effective tax rate is
based on the assumption of perfect competition and no uncertainty, and takes account of
a very limited number of tax parameters. This discrepancy between the two can be an

explanation of why F DI does not respond to the marginal effective tax rate.
1.2. Composition of FDI

As shown in Figure 2 in Chapter 2, most F DI (over 80%) is accounted for by
acquisition-type F D1. The preference for the acquisition mode can be accounted for by
the advantages of acquisition-type FDI over establishment-type FDI. The choice between
the acquisition route and the establishment route depends on the factors of time, risk, and
the market for takeover. According to Caves (1996), the acquisition route is very quick
and less risky than the new-establishment route. The MNC that buys the foreign firm also
buys access to the valuable information and the existing distribution network of the
foreign firm. It also acquires the local management team, who are aware of the local
market environment. These factors reduce the risk of foreign investment. Of course, the
acquisition is constrained by the availability of target firms. In contrast, the establishment
route requires a long time to achieve the desired size and profitability, and it is much

riskier than the acquisition route.

 

’2 Graham and Krugman (1995), p. 59.

85

If foreign investors, who are attempting to purchase an existing U.S. company or
acquire some control over a U.S. company, were to consider the tax burden of the
acquisition project, they would consider the current tax burden of the target company, not
the tax burden from a new establishment project. However, the typical calculation of the
marginal effective tax rate measures the tax burden of a new investment project. In
addition, the marginal effective tax rate. which considers only a limited number of tax
parameters, such as the investment tax credit, depreciation allowances, and the statutory
tax rate, cannot capture all the factors that are relevant for acquisition-type FDI. In sum,
while most F D1 is accounted for by acquiring existing firms, the METR measures the tax
burden of a new investment project, considering only a few tax parameters. This logic
can be another explanation to the question of why the marginal effective tax rate is not a
good explanatory variable for F DI. In order to see if this hypothesis is correct, I run a
regression of acquisition-type FDI on the AETR and on the METR separately. 1 use the
following specification that is the same as equation (1) in Chapter 4, except for the

dependent variable. The dependent variable is acquisition-type FDI. rather than total F D1.

(3) log(Acquisition,,)= a0 + a) log(ETR,,) + a2 log(BCA,,) + a3 log(GDP)) + 201d. + up,
where

Acquisition), = acquisition type FDI of industry i in year t,

ETR), = effective tax rate of industry i in year t,

BCA, = balance on current account in year t,

GDP, = U.S. real GDP in year t,

d,- = industry dummy,

up = error term.

86

Table 18 and Table 19 present the results ofthe regressions of acquisition FDI on
the average effective tax rates and the marginal effective tax rate, respectively. Column
three of Table 18 (my baseline specification) shows that after controlling for GDP and the
U.S. balance on current account. the estimated elasticity ofthe acquisition FDI with
respect to the AETR is —0.78. Thus, the sign and magnitude of this coefficient are
consistent with my hypothesis that acquisition FDI is influenced by the average effective
tax rate. However, the coefficient is not statistically significant (t-value: -1.12 and p-
value: 0.27). With this estimation result, we cannot reject the null hypothesis that the
elasticity of acquisition FDI with respect to the AETR is zero.

Column three of Table 19 shows the estimated elasticity of the acquisition F DI
with respect to the METR is 0.23. Thus, this coefficient is not only small in magnitude,
but it is also of the wrong sign. In addition, the coefficient is not even close to statistical
significance. Its t-value and p-value are 0.13 and 0.89, respectively. Therefore, we can
certainly not reject the null hypothesis that the METR elasticity is zero.

My hypothesis is that acquisition FDI cannot be accounted for by the marginal
effective tax rate, but can be accounted for by the average effective tax rate strongly.
These results give at least some support to the hypothesis. Comparing column three of
Table 18 and that of Table 19, the magnitude of the estimated elasticity of acquisition
FDI with respect to the AETR is larger than that of the estimated elasticity of acquisition
F DI with respect to the METR. The AETR elasticity is of the predicted sign, although’it
falls somewhat short of statistical significance. The METR elasticity, however, is very
small and quite insignificant. Therefore, I can say that the estimation results modestly

support my hypothesis. At least, the empirical evidence is not against my hypothesis.

87

Table 18. Acquisition FDI and the Average Effective Tax Rate“

 

 

 

 

 

 

 

 

 

Dependent Variable: log of acquisition FDI (in 1992 dollars)
Independent
Variables
(l) (2) (3) (4) (5) (6)
-1.36 -0.66 -0.78 -0.79 -l .09 -0.67
log(AETR) (-1.91) (—0.94) (-1.12) (-1.11) (-1.48) (-0.93)
2.65 2.67 2.73 2.70
log(GDP) _ (2.99) (3.06) (2.72) _ (2.65)
-0.05 -0.05
log(BCA)” _ _ (-1.63) (-l.62) _ _
0.09 -0.94 0.08
log(EX) _ _ _ (0.12) (-1.24) (0.10)
industry
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.4860 0.5638 0.5862 0.5864 0.5013 0.5639
# of
Obs. 56 56 56 56 56 56

 

 

 

 

 

 

 

 

 

* The number in ( ) is the OLS t value.
AETR, BCA, and EX stand for the average effective tax rate, the balance on current account, and the
trade-weighted exchange rate, respectively.

*"' log(BCA)=log(l95,81 8+balance on current account).

88

Table I9. Acquisition FDI and the Marginal Effective Tax Rate*

 

 

 

 

 

 

 

independent Dependent Variable: log of acquisition FDI (in 1992 dollars)
Variables
(1) (3) (3) (4) (5) (6)
-2.84 -0.64 0.23 0.24 -2.98 -0.78
log(METR) (-l.80) (-0.39) (0.13) (0.13) (-1.94) (-0.43)
2.76 3.06 3.07 2.62
log(GDP) _ (2.92) (3.19) (2.47) _ (2.16)
-0.05 -0.05
log(BCA)“ _ _ (-l.45) (-1.43) _ _
0.01 -1.34 -0.17
log(EX) _ _ _ (0.02) (- l .87) (-0.19)
industry
dummies Yes Yes Yes Yes Yes Yes

 

 

 

 

 

R-squared 0.4820 0.5515 0.5758 0.5758 0.5157 0.5578
# of
Obs. 56 56 S6 56 56 56

 

 

 

 

 

 

* The number in ( ) is the OLS I value.
METR, BCA, and EX stand for the marginal effective tax rate,
the balance on current account, and the trade-weighted exchange rate, respectively.

” log(BCA)=log(l95,8]8+balance on current account).

89

2. Establishment F DI and the Effective Tax Rates

According to the conjecture of Auerbach and Hassett (1993), establishment FDI
could be affected by the marginal effective tax rate, but acquisition FDI should not be
affected by the marginal effective tax rate. They state, “. . .the theoretical discussion and
empirical analysis of the impact of taxation on FDI has treated the problem as one of
acquiring new capital, even though this is only one of the possible modes. The mode
of investment chosen affects tax liability differently because the choice to acquire a U.S.
company will depend on the U.S. merger law governing, for example, step-up in basis
and transfer of tax benefits, whereas investment in new capital will depend on the
statutory tax rate, the investment tax credit, and depreciation schedules.”43 I agree with
the second part of their conjecture, and I have explained this in terms of the
characteristics of FDI (industrial organization theory of F DI). In addition, I have provided
some empirical evidence that the acquisition FDI does not respond to the marginal
effective tax rates, but modestly responds to the average effective tax rate. However, I do
not agree with the first part of their conjecture. In principle, Auerbach and Hassett are
making a valid point regarding establishment FDI. In principle, a marginal calculation is
the correct one to make for a new establishment. However, in practice, the marginal
effective tax rate is calculated under a set of very strong assumptions, so that the METR
is not a very effective explanatory variable. My hypothesis is that the establishment-type
FDI does not respond strongly to either the marginal effective tax rate (as measured) or to

the average effective tax rate.

 

’3 Auerbach and Hassett (1993), p. 123.

90

First of all, why does the establishment FDI not respond to the marginal effective
tax rate? Even though the establishment FDI is a new investment. the establishment F D1
is also a mode of F DI. That is, establishment FDI is also related to imperfect competition,
and it is a highly strategic behavior. Therefore, by the same reason I explained in the
previous section, the marginal effective tax rate is unlikely to be a good explanatory
variable for establishment FDI.

In addition, the effects of establishment F DI on an industry or market are different
from those of acquisition F DI. In general, the acquisition route of FDI does not bring new
capacity to the market in question. Of course, in some cases, the MN Cs may bring an
expansion of production capacity by vitalizing the acquired company or enhancing the
productivity. However, the establishment route of F DI adds production capacity to the
market. It also entails some external economies - the creation of job opportunities and the
training of workers. Therefore, each level of government in the U.S. offers investment
incentives to foreign investors. Especially, state and local governments provide financial,
tax, and other incentives to stimulate the local economy. These investment incentives
might make the marginal effective tax rate faced by foreign investors different from the
marginal effective tax rate calculated from some limited number of tax parameters.
Consequently, the METR that can be calculated using available data may not capture
some important aspects of the actual marginal incentives facing firms.

Secondly, why does establishment F DI not respond to the average effective tax
rate? As explained in Chapter 2, the average effective tax rate, which is defined as the
ratio of taxes paid to net income, is a good measure of the tax burden of existing firms,

but it cannot be a good measure of the impact of taxes on the incentives to make a new

91

investment. Accordingly, the establishment FDI, which is new investment in nature, does
not respond to the average effective tax rate of existing U.S. firms.

In sum, since establishment FDI is also related to market imperfections and
strategic behavior, and it is attracted by special investment incentives, the METR cannot
be expected to do a good job of explaining establishment F DI. And since establishment
FDI is new investment, it is not to respond to the AETR. Therefore, neither the METR
nor the AETR is expected to explain establishment (greenfield investment) —type FDI. To
give support to my hypothesis, I discuss the foreign investment incentives programs
provided by state governments in section 2.1. And to test whether my hypothesis is
supported by empirical evidence, I try to estimate the elasticity of establishment F DI with

respect to the AETR and that with respect to the METR in section 2.2.
2.1. Incentives to the Establishment FDI

In general, the attitudes toward FDI of each level of government in the U.S. depend
on the types of F DI. In general, most countries have a tendency not to welcome the
transfer of ownership or control of domestic firms from domestic citizens to foreigners.
Accordingly, acquisition of domestic firms by foreigners is restricted, or at least it is not
promoted. For an extreme example, foreign control in nuclear energy, broadcasting,
telecommunication, gas pipelines, and mining certain minerals and fuels is restricted in
the United States”. However, a new investment, which adds production capacity and

entails external economies of job creation and technology diffusion, is often promoted by

 

’4 Graham and Krugman (1995), pp. 122-123.

92

local governments. A number of state and local govemments in the United States provide
a variety of investment incentives for foreign investors. The state and local governments
provide these incentives to stimulate the local economies. Therefore, the primary targets
of these incentives are the establishment-type F DI. rather than the acquisition-type FDI.
The state and local governments do not consider the acquisition of an existing firm by a
foreign investor as an investment that is eligible for the incentive programs, because it is
nothing but a change of ownership. Graham and Krugman (1995) use an example of
Honda Motor Company in explaining how state governments are eager to attract foreign
direct investment. “. .. it was widely reported that at least three subnational governments
— the state governments of Ohio and Pennsylvania and the provincial government of
Ontario — competed to attract the large Honda Motor Company facility that eventually
was located in Marysville, Ohio.”45 These investment incentives consist of various forms
of tax relief, financial assistance, and provision of infrastructure and land free of charge.
The following measures to attract foreign direct investment are excerpted from Poniachek

(1986) and Casey (1998).

(a) Financial assistance: Most states offer some type of financial aid to new investment
within their borders. Many states have established loan funds and mortgage guarantee
programs to help firms locating or expanding within the state. Four types of financing are
commonly used - industrial revenue bonds and mortgages, loan guarantee programs,

direct loan programs, and development credit corporations.

 

’5 Graham and Krugman (1995), p. 141.

93

(ti
neg

Cli

(b) Tax incentives: Various kinds of State and local tax incentives are available on a
negotiated or case-by-case basis. Among them are tax reductions and exemptions, tax
credits, and preferential assessment. These are usually offered as exemptions or deferrals
from property, inventory. sales, or income tax. Many states have enacted legislation to
exempt new companies from various state taxes for specified periods. Some states
provide a corporate income tax exemption for qualified companies. Preferred
property-tax assessment is commonly applied to machinery and equipment. Business
purchases may be exempted from sales tax. Reduced income tax is sometimes provided
for out-of-state sales. and preferential income tax write-offs may be given for selected

types of investment.

(c) Infrastructure incentives: Infrastructure incentives include state government support
of transportation, electric power, communications, and other forms of social overhead
capital and public services, such as rail and highway connections, free land for industrial

development, and low-cost sites in industrial parks.

((1) Training programs and labor incentives: These incentives include payment for the
testing and screening of work-force personnel, and subsidies for job advertisements.
Sometimes local educational institutions support curriculum efforts related to the needs

of the incoming foreign companies.

(e) Natural resource incentives: These incentives involve government assistance in
providing foreign businesses with industrial resources such as land and energy. In areas
where state environmental controls are more rigid than federal standards (e. g., pollution

94

control), exemptions or special exclusions to state law are sometimes granted to the
foreign producers.

All of these investment incentives provided by local governments make the
expected tax burden of a new investment project faced by a foreign investor different
from the marginal effective tax rate calculated with only a limited number of tax
parameters. Therefore, establishment FDI is not expected to respond to the marginal
effective tax rate, as the METR is usually measured. In conclusion, my hypothesis
regarding the relationship between establishment FDI and the effective tax rates is that
neither the AETR nor the METR will do a good job of explaining establishment-type

FDI. In the next section. I present empirical support for this hypothesis.

2.2. Empirical Analysis

To test my hypothesis, I run a regression of establishment-type F DI on the AETR
and on the METR separately. 1 use the following specification that is the same as
equation (3) in section 1 of this chapter, except for the dependent variable. The dependent

variable is establishment-type FDI, rather than acquisition-type FDI.

(4)10g(Establishment,,)= a0 + a) log(ETRp) + a2 log(BCA,) + a3 log(GDP)) + Edd,- + u,,,
where

Establishment,, = establishment-type F D1 of industry 1 in year t,

E TR,-, = effective tax rate of industry 1 in year t,

BCA, = balance on current account in year t,

GDP, = U.S. real GDP in year t,

95

d,- = industry dummy. and

u), 2 error term.

Table 20 and Table 21 present the results of the regressions of establishment F DI
on the average effective tax rates and the marginal effective tax rate, respectively.
Column three of Table 20 (my baseline specification) shows that, after controlling for
GDP and the U.S. balance on current account, the estimated elasticity of establishment
FDI with respect to the AETR is —0.60, but it is not statistically significant (t-value: -1.01
and p-value: 0.32). With this estimation result, we cannot reject the null hypothesis that
the elasticity of establishment FDI with respect to the AETR is zero even at the 30%
significance level.

Column three of Table 21 shows the estimated elasticity of the establishment F DI
with respect to the METR is —0.49, and that its t-value and p-value are -0.33 and 0.74,
respectively. Therefore, we cannot reject the null hypothesis that the elasticity is zero.

Comparing column three of Table 20 and that of Table 21, the magnitude of the
estimated elasticity of establishment F DI with respect to the AETR is slightly larger than
that of the estimated elasticity of establishment FDI with respect to the METR. And the
p-value of the METR elasticity is about two times higher than that of the AETR
elasticity. The estimated elasticity of establishment FDI with respect to the AETR looks
better than the estimated elasticity of establishment F DI with respect to the METR.
Therefore, these estimation results do not strongly support my hypothesis that
establishment FDI cannot be accounted for very well by either the METR or the AETR.
However, I can say that the estimation results modestly support my hypothesis. While
establishment FDI can only be explained very weakly by the AETR, establishment FDI

96

cannot be said to be explained by the METR. At least, the empirical evidence suggests
that establishment FDI does not respond to the METR. which is opposite to what is
expected by Auerbach and Hassett (1993).

Columns three and four of Table 20 and Table 21 show that the balance of current
account does not play a significant role in explaining establishment F DI. This is in
contrast to the results in Table 18 and Table 19, in which the balance of current account is
shown to have a marginally significant effect on acquisition F DI. This is an intuitive
result. The balance-of-current-account variable is associated with the idea that, when the
United States incurs a large current-account deficit, foreign investors are provided with a
greater opportunity to acquire U.S. assets. This explanation of the role of the current
account balance is more powerful in the case of acquisition investment than in the case of

establishment investment.

97

Table 20. Establishment FDI and the Average Effective Tax Rate*

 

 

 

 

 

 

 

 

 

 

Dependent Variable: log of establishment FDI (in 1992 dollars)
Independent
Variables
(1) (2) (3) (4) (5) (6)
-l.28 -0.58 -0.60 -0.54 -0.89 -0.52
log(AETR) (-2.09) (-0.99) (-1.01) (-0.90) (-l .45) (-0.87)
2.65 2.65 2.38 2.37
log(GDP) _ (3.58) (3.56) (2.79) _ (2.81)
-0.01 -0.01
log(BCA)” _ _ 60.34) (0.38) _ _
-0.46 -1.36 —0.46
log(EX) _ _ _ (068) (-2.15) (069)
industry
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.7871 0.8305 0.8311 0.8327 0.8051 0.8322
# of
Obs. 56 56 56 56 56 56

 

 

 

 

 

 

 

 

"‘ The number in ( ) is the OLS I value.
AETR, BCA, and EX stand for the average effective tax rate, the balance on current account, and the
trade-weighted exchange rate, respectively.

** log(BCA)=log(l95,818+balance on current account).

98

Table 21. Establishment F DI and the Marginal Effective Tax Rate*

 

 

 

 

 

 

 

 

 

independent Dependent Variable: log of establishment F D1 (in 1992 dollars)
Variables
(ll (3) (3) (4) (5) (6)
-2.75 -0.57 -0.49 -1.17 -2.93 -1.17
log(METR) (-2.03) (-0.41) (-0.33) (-0.72) (-2.31) (-0.79)
2.74 2.77 2.10 2.10
log(GDP) __ (3.48) (3.40) (2.00) _ (2.10)
-0.004 -0.0002
log(BCA)* * _ __ (-0.15) (-0.006) _ _
-0.76 -1.70 -0.76
log(EX) _ _ _ (-1.03) (-2.88) (-1.06)
industry
dummies Yes Yes Yes Yes Yes Yes
R-squared 0.7861 0.8278 0.8279 0.8317 0.8165 0.8317
# of
Obs. 56 56 56 56 56 56

 

 

 

 

 

 

 

 

 

"‘ The number in ( ) is the OLS I value.
METR, BCA, and EX stand for the marginal effective tax rate,

the balance on current account, and the trade-weighted exchange rate, respectively.
** log(BCA)=log(l95,8l8+balance on current account).

99

CHAPTER VII
CONCLUSION

(1) The estimation results of this dissertation (Chapter 4) suggest that overall foreign
direct investment in the U.S. decreases 0.85 percent with a one-percent increase in the
U.S. average effective tax rate. This estimate is not statistically significant at the 5% or
the 10% significance level, but it is marginally significant at the 15% level (p-value: 0.15,
t-value: -1.45). And the Sign of the FDI elasticity with respect to the AETR is not positive
at the 10% significance level. The 10% critical value for a one-tailed test with 60 degrees
of freedom is -1.296. Therefore. the hypothesis that the FDI elasticity with respect to the
AETR has a positive sign can be rejected at the 10% significance level.

These results are weakly consistent with the traditional view, in which foreign
investors tend to undertake investment in the location with the higher after-tax rate of
return. And these results are opposite to the new view, in which FDI responds positively
to the host-country tax rate. In addition. I find that FDI does not respond to the marginal
effective tax rate. The marginal-effective-tax-elasticity of FDI is close to zero (0.27) and

statistically insignificant (t-value: 0.18).

(2) The estimation results in Chapter 5 suggest that multinationals from countries with a
worldwide tax system appear not to adjust their investment in response to changes in tax
rates in the U.S., but multinationals from countries with territorial system reduce their

investment in response to an increase in the U.S. average effective tax rate. The average-

effective-tax-rate elasticity of FDI from territorial countries is negative (-3.54) and

100

statistically significant (t-value: -2.69). The average-effective-tax-rate elasticity from
worldwide countries is positive (1.56) but statistically insignificant (t-value: 0.97). These
results are consistent with the prediction of the traditional view that the investment by
territorial firms would respond negatively to an increase in the U.S. tax rate. and that FDI
from worldwide countries should not respond sensitively to the U.S. tax rates. These
results are opposite to the argument of the new view, in which the worldwide investors
respond positively to the host-country tax rate, and the territorial investors are not

sensitive to host-country tax rate.

(3) This paper also gives an answer to the question of why neoclassical investment theory
(the cost-of-capital theory) cannot be easily applied to foreign direct investment behavior.
I find the answers in the characteristics of F DI and the composition of FDI in the U.S.

Firstly, F D1 is closely related to market imperfections, and F D1 is a result of
strategic decision-making on the part of an MNC. However, the concept of the marginal
effective tax rate is based on the assumption of perfect competition. Secondly, the
calculation of the METR considers only a limited number of tax parameters. Thirdly,
while most FDI is accounted for by acquisition of existing firms, the METR measures the
tax burden of a new investment project.

My estimation results weakly support the last point. The elasticity of the
acquisition FDI with respect to the AETR is negative (078) and statistically
insignificant (t-value: -1 . 12). But, the elasticity of the acquisition F DI with respect to the
METR is close to zero (0.23) and statistically insignificant (t-value: 0.13). Although both
the estimated AETR-elasticity and the estimated METR-elasticity are insignificant, the
former appears to have better explanatory power than the latter.

101

flno-sfifi
. ..

 

(4) I test the conjecture of Auerbach and Hassett (1993): establishment FDI could be
affected by the marginal effective tax rate. but acquisition FDI should not be affected by
the marginal effective tax rate. I agree with the second part of their conjecture, and I
explain this in terms of the characteristics of FDI and the composition of FDI in the U.S.
However, I do not agree with the first part of their conjecture. My hypothesis is that
establishment-type FDI does not respond to either the marginal effective tax rate or the
average effective tax rate.

Firstly, since establishment F D1 is also related to market imperfections and it is a
highly strategic behavior, the marginal effective tax rate, which is based on perfect
competition and considers only a few tax parameters, is unlikely to be a good explanatory
variable for establishment FDI. Secondly. the average effective tax rate, which measures
the tax burden of existing firms, is unlikely to be a good measure of the impact of taxes
on the incentives to make new investment. That is, since establishment FDI is new
investment in nature, it does not respond to the average effective tax rate of existing U.S.
firms.

And I find the empirical estimation results supporting my hypothesis. The elasticity
of establishment FDI with respect to the AETR is negative (060) and statistically
insignificant (t-value: -1.01). And, the elasticity of establishment FDI with respect to the
METR is a little smaller (-0.49) and also statistically insignificant (t-value: -0.33).

At least, we can say that establishment FDI does not respond to the METR, which is

opposite to what is expected by Auerbach and Hassett (1993).

(5) The difference between this paper and the earlier papers belonging to the traditional

view is that this paper uses the acquisition and establishment data rather than capital-flow

102

data. As Young (1988), Auerbach and Hassett (1993), and Swenson (1994) point out. it is
more appropriate to use acquisition and establishment data in explaining investment
behavior. Even though this paper uses different data, the results of this paper are
consistent with the arguments of the traditional view.

The estimation results of this paper are exactly opposite to those of Swenson
(1994). She finds a positive average-tax-rate effect on F DI, and she finds a positive
average-tax effect on FDI by foreign firms from worldwide tax regimes, and an
insignificant average-tax-rate effect on FDI by foreign firms from territorial tax regimes.
The different point of this paper from Swenson (1994) lies in the calculation of the
average effective tax rate. She uses average effective tax rates calculated by Tax Analysts,
but I calculate the AETR based on the data in the Statistics of Income Bulletin by the

lntemal Revenue Service.

(6) In order to see what makes the results of this paper different from those of Swenson, I
collect Swenson’s data set and replicate her estimation. As a result of replication, I find
that the F DI elasticity with respect to the AETR turns out to be smaller (0.24) than her
original number (0.75), and statistically insignificant. However, I find that there are some
serious mistakes in her analysis that may make her results unreliable. One problem is that
the average effective tax rate calculated by Tax Analysts may not be a good measure for
the actual tax burden of U.S. firms. Tax Analysts calculate the average effective tax rate
by using information in financial statements, and only the largest U.S. firms are included
in their sample. Because there is some evidence that big firms engage in more deception,
it does not seem that the Tax Analysts’ average effective tax rate is a representative U.S.
average tax rate. The other problem is that Swenson uses Tax Analysts’ data very

103

 

arbitrarily in her analysis, by having the AETR of a subset of an industry represent the
AETR of the entire industry. I redo the estimation with the corrected AETRs. The
estimation results, using the AETRs for each industry calculated more appropriately, no
longer support Swenson’s conclusion. I conclude that Swenson’s industry-level analysis
suffers from the mismatching problem that the AETRS of large-digit industries are used
to represent small-digit industries, and that it is hard to think Swenson’s estimation result

is reliable.

(7) The problems of the New View arise from its assumptions. The New View assumes
that capital is perfectly mobile among countries and among assets. And it assumes that
every adjustment process take place immediately. So, in the world of the New View, the
economy is always in the equilibrium state. However, it is generally accepted that capital
is not so mobile among countries and among assets.

In the case of direct investment, capital seems to move slowly in response to a
change in relative tax rate between countries and between assets. These capital
movements, in turn, may lead to changes in asset prices and in the pretax returns of the
assets. That is, the change in the pre—tax rate of return in response to a change in tax rate
is the result of the process of capital movements. But, the New View thinks that every
adjustment process occurs immediately, and the price and the pretax return change
immediately in response to a tax change. And then capital (direct investment) moves
again in response to these changes in asset price. Every problem of the New View seems

to arise from this misunderstanding.

104

 

(8) Lastly, I want to point out that foreign direct investment is an investment itself. but it
is significantly different from domestic investment in general meaning. In economic
analysis, investment is defined as an increase in the capital stock of an economy or an
increase in a fixed asset (PPE: property, plant, and equipment) of a business. However,
foreign direct investment includes both the increase in capital stock by foreign investors
and the transfer of ownership or control of domestic firms from U.S. citizens to
foreigners. The most important portion of FDI is accounted for by the transfer of
ownership or control. Therefore, in economic analysis, foreign direct investment should
be treated differently from domestic investment. My estimation results, which suggest
that the response of FDI to the marginal effective tax rate is different from the commonly
expected response of investment to the marginal effective tax rate, can be explained by

this characteristic of FDI.

105

 

 

APPENDIX

106

APPENDIX

CALCULATION OF MARGINAL EFFECTIVE TAX RATES

1. Model

The marginal effective tax rate measures what fraction of the real pretax rate of
return to a new investment will be collected as taxes. The marginal effective tax rate
accounts for tax benefits and penalties on income from a new investment that do not take
the form of changes in the statutory tax rate. The calculation of the marginal effective tax
rate assumes a perfectly competitive firm in a world with no uncertainty“. In a world
without uncertainty a firm will invest a unit of capital up to the point where the present
value of future receipts is equal to the cost of acquiring the asset. The model in this
appendix draws on Gravelle (l 982)’s reformulation of the neoclassical model of Hall and
J orgenson (1967).

The rental price of capital services is defined by the equality between the cost of
acquisition of an asset at time 0, and the present value of future rental prices. In the

absence of taxes, this equality can be written as

(A1) q = Jce'é’e”'e""””dt = c]e“”""dt
0 0 r + 5

 

’6 Fullerton and Henderson (1987).

107

where q = acquisition cost.
6 = the rental price of a new unit of capital,
r = the real return or discount rate,
7r= the expected inflation rate. and

c5 = the rate of depreciation of asset.

Rearranging terms, equation (A1) can be expressed as follows:

1

(A2) C =<r+6>

This means that the rental price is equal to the sum of the rate of return and the
depreciation rate, for each dollar of capital (q is normalized to $1).

In the presence of taxes, q can be written as follows.

or:

(A3) q = Ic(l — u)e‘5’e”'e'("’”)'dt + uq jD(t)e""’”"dt + kq
0

0

c(l — u)J'e"(""s)’dt + uqz + kq
O

C(I—u)
= +uz+k ,
r*+5 C] q

 

where u = statutory tax rate,

108

mfiflg‘nm-a-m~-n u. a. A 9‘!
an... A o I ' I _

 

D(t) = depreciation deductions allowed at time I per dollar of investment,
k = the investment credit rate.
r* = the real after-tax discount rate, and

z = present discounted value of depreciation allowances.

Rearranging terms, equation (A3) can be expressed as follows:

(A4)

C_
q l—u

When the equation (A2) and (A4) are combined, it is possible to express the real
pretax return (r) as a function of the required after tax real return (r*), the economic

depreciation rate (5), and tax variables (u, z, k):

*
(A5) rzr—l——+—5—(l—uz—k)—5

— ll
The marginal effective tax rate is defined as follows:

r—r*

(A6) u* = —,

where u* = marginal effective tax rate.

109

2. Tax Variables

Statutory tax rate (u)

Corporate income is subject to federal, state, and local income taxes. For the
federal income tax, most corporations do not reach the top marginal tax rate, but the great
bulk of corporate capital is held by firms that do. So I can model the top marginal
corporate income tax rate as federal corporate tax rate.47 Jorgenson and Yun (1991)
estimated state and local tax rates. Since the state and local tax rates show rising trends, I
assumed 9% for 1987-1990 and 9.5% for 1991-1996. Because of the deductibility of state
tax from federal tax, the combined tax rate can be estimated as follows: u = up(1-u5) + us,
where u; is the federal corporate tax rate and us is the state corporate tax rate. Table A-1

shows my calculation of the statutory tax rates.

Investment tax credit rates

Before 1986, the tax code included an investment tax credit (ITC), which permitted
a firm to subtract some portion of the purchase price of an asset from its tax liability at
the time the asset was acquired. The TRA 1986 eliminated the investment tax credit.

Table A-2 shows the investment tax credit rates by asset types.

 

‘7 See King and Fullerton (i984), p 202.

110

“WWI“ 7‘7,

 

Table A-1. Corporate Income Tax Rates

 

 

 

year Federal State and Local Federal, State. Local
1978 0.48 0.0645 0.514
1979 0.46 0.0657 0.496
1980 0.46 0.0724 0.499
1981 0.46 0.0773 0.502
1982 0.46 0.0885 0.508
1983 0.46 0.0863 0.507
.1984 0.46 0.0834 0.505
1985 0.46 0.0874 0.507
1986 0.46 0.0896 0.508
1987 0.34 0.0900 0.399
1988 0.34 0.0900 0.399
1989 0.34 0.0900 0.399
1990 0.34 0.0900 0.399
1991 0.34 0.0950 0.403
1992 0.34 0.0950 0.403
1993 0.35 0.0950 0.412
1994 0.35 0.0950 0.412
1995 0.35 0.0950 0.412
1996 0.35 0.0950 0.412

 

 

 

 

Source : Jorgenson and Yun (1991) and text.

111

 

Table A-2. Investment Tax Credit Rates

 

 

 

 

 

 

Asset type 1978- 1 980 1981-1986 1987-

EQUIPMENT

Autos 3.3 6 0
Office/computing equipment 10 10 0
Trucks, buses, and trailers 6.6 10 0
Aircraft 7 10 0
Construction machinery 6.6 10 0
Mining/oilfield equipment 10 10 0
Service industry equipment 10 10 0
Tractors 9 10 0
Instruments 1 0 10 0
Others 10 10 0
General industrial equipment 9.1 9.6 0
Metalworking machinery 8.6 9.4 0
Electric transmission equipment 10 10 0
Communications equipment 10 10 0
Other electrical equipment 10 10 0
Furniture and fixture 10 10 0
Special industrial equipment 10 10 0
Agricultural equipment 10 10 0
Fabricated metal 10 10 0
Engines and turbines 10 10 0
Ships and boats 10 10 0
Railroad equipment 10 10 O
STRUCTURES

Public utility 10 10 0

 

Source: Gravelle (1994) and text.

112

 

Depreciation Allowances

Historically, the United States has adopted various methods of depreciation:
straight-line method, sum-of-the-years’-digits method (SYD), and declining-balance
methods. Gravelle (1994) summarizes the depreciation methods adopted by the U.S.
Table A-3 shows the depreciation methods historically adopted by the U.S. tax code.

The straight-line method allows the taxpayer to deduct l/ T of the purchase price of
the asset each year for Tyears. The asset is fully depreciated after T years. The formula
of the present discounted value of depreciation allowances under the straight-line

4 .
scheme 8 can be derived as follows:

The declining-balance method allows the taxpayer to deduct a every year from .the
remaining value of the asset, where a denotes the depreciation rate for tax purposes. For
the double-declining balance method, the taxpayer can deduct a (=2/T) of the purchase
price in the first year, and he deducts 2/ T of the remaining depreciation basis in later

years. The formula of the declining-balance method49 is as follows:

 

w
z = Jae "“”’”dt =
0 a+ p

 

’8 King and Fullerton (1984), p. 20.
’9 King and Fullerton (1984), p. 20.

113

 

f

For the sum-of-the-years’-digits (SYD) method. the taxpayer starts by calculating
the sum of 1 through T. Let the sum be SUM. The taxpayer can deduct T/S UM of the
purchase price in the first year, (T -1 )/S UM in the second year, and l/SUM in the last year.
If T=3, for example, the purchase price is allocated as 3/6, 2/6, and 1/6 across the three
years. Under the SYD scheme, the asset is fully depreciated, as under the straight-line

scheme. The formula can be derived as follows:

 

 

'I‘ T
2 = ]——.——e(€ _ 1) “”dt = 3—]e"”dt — 2, jte‘p’dt
0 T 0 T. 0
Jtdt
O
_ ~eT -eT_
=30 e )+ 2.<Te‘p’+e 1>
T p?” p
2 1 —)l'
=——1—— 1— ‘
pTl th e )1

If we know the value of p and T, we can calculate the present discounted value of
depreciation allowances under each method. The discount rate (p) of the corporations is
the sum of the real after-tax discount rate (r*) and the expected inflation rate (it). The real
after-tax discount rate is equal to a weighted average of the rate of return on debt and
equity”:

r*=f(i(l-uF)-1t)+(l-f)E,

 

5° See Gravelle (1994), p. 291.

114

where f is the fraction financed by debt. i is nominal interest rate, up is the corporate
statutory tax rate, and E is the return required by stockholders prior the personal level

taxes.

115

Table A-3. Depreciation Methods

 

 

 

 

 

Asset type Year Depreciation Methods
Equipment 1978-80 SYD
(including
Public Utility 1981 Double declining balance
Structures)
1982-86 150% declining balance
1987- 3, 5, 7, 10-year class;
double declining balance
15, 20-year class;
150% declining balance
Structures 1978-80 Nonresidential structures;
150% declining balance
Residential;
SYD
1981-86 175% declining balance
1987- Straight line

 

 

Source: Gravelle (1994).

116

qfsrrff _ «-.m _'

3. Expected Inflation Rate

In calculating the marginal effective tax rate, Hendershott & Hu (1981) calculate
the expected inflation rate using the adaptive expectations model (a geometric weighted
average of inflation of the previous seven quarters). Gravelle (1994) uses the expected
inflation rates of the Drexel-Bumham-Lambert Decision Makers Poll for the period
1979-1989, and those calculated, based on Hendershott and Hu’s adaptive expectations
model, for the period 1953-1978.

The data of the Drexel-Bumham-Lambert Decision Makers P011 is no longer
available since 1989. And the adaptive expectation model has lost its theoretical
reliability. So, I use the Livingston Survey data, which is historical data on the expected
inflation rates maintained by the Federal Reserve Bank (FRB) of Philadelphia. The F RB
of Philadelphia collects forecasts of inflation and other economic variables from
economists in academic profession and business. The Livingston Survey data are
frequently used in empirical studies in macroeconomics5 1. Table A-4 compares the
expected inflation rates from the Livingston Survey and actual inflation rates. For the
years 1979-1981 and 1988-1991, the actual inflation rates were higher than the expected
inflation rates. And for the years 1982-1987 and 1992-1996, the actual inflation rates

were lower than the expected inflation rates.

 

5' For example, Clarida (2000).

117

Table A-4. Comparison of the expected and actual inflation rates

 

Livingston Survey

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year (CPI) Actual CPI Actual GDP Deflator
1979 70% 11.3497 8.5
1980 9.629 13.4986 9.2
1981 10.269 10.3155 9.4
1982 7.198 6.1606 6.3
1983 5.131 3.2124 4.3
1984 5.319 4.3173 3.8
1985 4.781 3.5611 3.4
1986 3.954 1.8587 2.6
1987 3.689 3.6496 3.1
1988 4.095 4.1373 3.7
1989 4.769 4.8183 4.2
1990 3.964 5.4032 4.3
1991 3.835 4.2081 4
1992 3.364 3.0103 2.8
1993 3.387 2.9936 2.6
1994 3.183 2.5606 2.4
1995 3.469 2.8340 2.3
1996 2.879 2.9528 1.9

 

118

Sources: Economic Report of the President and www.phil.frb.org.

 

4. Other Variables

Hulten and Wykoff (1981) estimate the economic depreciation rates for 34 asset
categories. For the economic depreciation rates for each asset, I assume that Hulten and
Wykoff (1981)’s estimated values still hold. Gravelle (1994) recalculates the economic
depreciation rates for 28 asset categories based on Hulten and Wykoff’s estimated values.
Gravelle (1994)’s economic depreciation is presented in Table A-5. Lifetime of assets is
reported in Table A-6. For nominal interest rates, I use Baa bond rates from Economic
Report of the President 1999.52 For the required return to stockholders prior to personal

taxes, I set it equal to the real after-tax rate of return, plus a risk premium of 4 percent.53

 

5’ See Gravelle (1994), p. 293.
’3 See Gravelle (1994), pp. 291-293.

119

Table A-5. Economic Depreciation Rates

 

 

 

 

Asset Type Economic Depreciation
Rate
EQUIPMENT
1. Autos 0.3333
2. Office/computing Equipment 0.2729
3. Trucks, Buses, and Trailers 0.2537
4. Aircraft 0.1818
5. Construction machinary 0.1722
6. Mining/oilfield equipment 0.1650
7. Service industry equipment 0.1650
8. Tractors 0.1633
9. Instruments 0.1473
10. Others 0.1473
1 1. General industrial equipment 0.1225
12. Metalworking machinery 0.1225
13. Electric transmission Equipment 0.1 179
14. Communications equipment 0.1179
15. Other electrical equipment 0.1 179
16. Furniture and fixtures 0.1 100
17. Special industrial equipment 0.1031
18. Agricultural equipment 0.0971
19. Fabricated metal 0.0971
20. Engines and turbines 0.0786
21. Ships and boats 0.0750
22. Railroad equipment 0.0660
STRUCTURE
1. Others 0.0663
2. Industrial 0.0454
3. Public utility 0.033
4. Commercial 0.0316
5. Farm 0.0230
6. Residential 0.0237

 

Source: Gravelle (1994).

120

 

Table A-6. Asset Lifetimes

 

 

 

Asset type 78-80 81 82-83 84-86 87-92 93-

EQUIPMENT
Autos 3 2.5 2.5 2.5 5 5
Office/computing equipment 7 4.5 4.5 4.5 5 5
Trucks, buses, and trailers 5 4.5 4.5 4.5 5 5
Aircraft 9.2 4.5 4.5 4.5 6.8 6.8
Construction machinery 5 4.5 4.5 4.5 5 5
Mining/oilfield equipment 9.2 4.5 4.5 4.5 7 7
Service industry equipment 9.9 4.5 4.5 4.5 7 7
Tractors 7.1 4.5 4.5 4.5 6.4 6.4
Instruments 10.3 5.1 5.1 5.1

5( 12%) 5(12%)

7(81%) 7(81%)

20(7%) 20(7%)
Others 8.8 4.5 4.5 4.5 7 7
General industrial equipment 9.9 4.9 4.9 4.9

3(4%) 3(4%)

5( 14%) 5(14%)

7(73%) 7(73%)

20(9%) 20(9%)
Metalworking machinery 7.8 4.2 4.2 4.2 6.7 6.7
Electric transmission equipment 13.8 7.3 7.3 7.3

5(5%) 5(5%)

7(53%) 7(53%)

20(42%) 20(42%)
Communications equipment I 1.5 4.5 4.5 4.5 5 5
Other electrical equipment 9 4.5 4.5 4.5 7 7
Furniture and fixture 8 4.5 4.5 4.5 7 7
Special industrial equipment 9.2 4.5 4.5 4.5 6.5 6.5
Agricultural equipment 8 4.5 4.5 4.5 7 7
Fabricated metal 14.2 6.6 6.6 6.6

5(18%) 5(18%)

7(40%) 7(40%)

20(42%) 20(42%)
Engines and turbines 18.1 10.8 10.8 10.8

15(25%) 15(25%)

2(75%) 2(75%)
Ships and boats 16 4.5 4.5 4.5 10 10
Railroad equipment 15 9.5 9.5 9.5 7 7
STRUCTURES
Others 31 15 18 19 31.5 39
Industrial 36 15 18 I9 3 1.5 39
Public utility 22 12.9 12.9 12.9

15(25%) 15(25%)

2(75%) 2(75%)
Commercial 37 15 18 19 31.5 39
Farm 20 15 18 19 20 20
Residential 31 15 18 19 27.5 27.5

 

 

 

 

 

 

 

Source: Gravelle (1994).

121

 

5. Calculation

I calculate marginal effective tax rates for 28 assets, using the information
described above. Then I calculated the METR for two broad assets (equipment and
structures). I use Gravelle (1994)’s asset share calculation (Table A-7). To calculate the
METR by industry, appropriate weights are needed. I use Gravelle (1994)’s capital stock
share (Table A-8) as the weights. My calculation of the METR for 4 industries is

presented in Table 8 in Chapter 4.

122

Table A-7. Asset Share

 

 

Asset type Asset Share
EQUIPMENT
Autos 4.6
Office/computing equipment 5.0
Trucks, buses, and trailers 8.3
Aircraft 1.9
Construction machinery 3.4
Mining/oilfield equipment 1.7
Service industry equipment 3.2
Tractors 2.9
Instruments 6.5
Others 2.7
General industrial equipment 6.8
Metalworking machinery 6.7
Electric transmission equipment 5.4
Communications equipment 10.5
Other electrical equipment 1.8
Furniture and fixture 5.8
Special industrial equipment 6.6
Agricultural equipment 4.9
Fabricated metal 3.3
Engines and turbines 2.0
Ships and boats 2.3
Railroad equipment 4.4
STRUCTURES
Mining 7.6
Others 1.3
Industrial 16.7
Public utility 27.8
Commecial 40.6
Farm 6.1

 

 

 

 

Source: Gravelle (1994).

123

Table A-8. Capital Stock Share by Industry

 

 

 

 

 

Industry Equipment Structures

Agriculture 0.479 0.521
Mining 0.389 0.611
Construction 0.83 8 0.162
Transportation 0.766 0.234
Public Utility 0.174 0.826
Trade 0.299 0.701
Finance, Insurance, and Real Estate 0.354 0.646
Services 0.354 0.646
Manufacturing 0.508 0.402
Food 0.444 0.556
Chemicals 0.557 0.443
Metal 0.540 0.460
Machinery 0.53 1 0.469
Other Manufacturing 0.495 0.505

 

 

Source : Gravelle (1982).

124

.5341".

or} 7".3.

Ls...

LIST OF REFERENCES

 

F.7- r
l

REFERENCES

Altshuler, Rosanne, and T. Scott Newlon (1993), “The Effects of U.S. Tax Policy on the
Income Repatriation Patterns of U.S. Multinational Corporations.” In Studies in
International Taxation, edited by Alberto Giovanini. Glenn Hubbard, and Joel Slemrod:

77-114. Chicago: University of Chicago Press.

Altshuler, Rosanne, T. Scott Newlon, and William C. Randolph (1995), “Do Repatriation
Taxes Matter? Evidence from the Tax Returns of U.S. Multinationals.” In The Eflects of

Taxation on Multinational Corporations. edited by Martin Feldstein. James R. Hines, Jr.,
and R. Glenn Hubbard: 253-276. Chicago: University of Chicago Press.

Alworth, Julian S. (1988), The Finance, Investment and Taxation Decisions of
Multinationals. New York: Basil Blackwell.

Auerbach, Alan J., and Kevin Hassett (1993), “Taxation and Foreign Direct Investment in
the United States: A Reconsideration of the Evidence.” In Studies in International
Taxation, edited by Alberto Giovanini, Glenn Hubbard, and Joel Slemrod: 119-147.
Chicago: University of Chicago Press.

Auerbach Alan J., and David Reishus (1988), “The Impact of Taxation of Mergers and
Acquisitions.” In Mergers and Acquisitions, edited by Alan J. Auerbach: 69-85. Chicago:
University of Chicago Press.

Auerbach, Alan J., and Joel Slemrod (1997), “The Economic Effects of the Tax Reform
Act of 1986.” Journal of Economic Literature 35: 589-623.

Ballard, Charles L. (1999), “International Aspects of Fundamental Tax Reform.” Mimeo,
Michigan State University (November 23).

Boskin, Michael, and William G.Gale (I 987), “New Results on the Effect of Tax Policy
on the International Location of Investment.” In The Effect of Taxation on the Capital
Accululation, edited by Martin Feldstein: 201-19. Chicago: University of Chicago Press.

126

 

Buckley, Peter J ., and Mark C. Casson (1976), The Future of Multinational Enterprise,
London: Macmillan.

Casey, William L. (1998), Beyond the Numbers: Foreign Direct Investment in the United
States. London: Jai Press.

Caves, Richard E. (1996), Multinational Enterprise and Economic Analysis. Cambridge:
Cambridge University Press.

Clarida, Richard (2000). “Monetary policy rules and macroeconomic stability: Evidence
and some theory.” The Quarterly Journal of Economics, February: 147-180.

Dunning, John H. (1992), Multinational Enterprises and the Global Economy. Addison-
Wesley Publishing Company.

F ahim-Nader. Mahnaz (1999), “Foreign Direct Investment in the United States.” Survey

of Current Business, June: 29-3 7.

Feldstein, Martin, and Philippe Bacchetta (1990), “National Saving and International
Investment.” In National Saving and Economic Performance, edited by B. Douglas
Bemheim and John B. Shoven: 201-226. Chicago: University of Chicago Press.

Feldstein, Martin S., and Horioka Charles (1980), “Domestic Savings and International
Capital Flows.” Economic Journal, June: 314-329.

Frankel, Jeffrey A. (1990), “Quantifying lntemational Capital Mobility in the 19805.” In
National Saving and Economic Performance, edited by B. Douglas Bemheim and John
B. Shoven: 227-270. Chicago: University of Chicago Press.

Fullerton, Don (1984), “Which Effective Tax Rate.” National Tax Journal 37: 23-41.
Fullerton, D., and Y. K. Henderson (1987), “The Impact of Fundamental Tax Reform on

the Allocation of Resources.” In The Eflects of Taxation on Capital Accumulation, edited
by M. Feldstein: 401-443. Chicago: University of Chicago Press.

127

Gordon, Roger H., and A. Lans Bovenberg (1996), “Why is capital so immobile
internationally? Possible explanations and implications for capital income taxation.”

American Economic Review, December: 1057-1075.

Graham, Edward M., and Paul R. Krugman (1995), Foreign Direct Investment in the

United States. Washington, DC: Institute for lntemational Economics.

Gravelle, Jane G. (1982), “Effects of the 1981 Depreciation Revision on the Taxation of

Income from Business Capital.” National Tax Journal 35: 1-20.

Gravelle, Jane G. (1994). The Economic Effects of Taxing Capital Income. Cambridge,
Massachusetts: The MIT Press.

Hall, Robert E., and Dale W. Jorgenson (1967), “Tax Policy and Investment Behavior.”

American Economic Review 5 7: 391-414.

Hartman, David G. (1981), Domestic Tax Policy and Foreign Investment: Some
Evidence, NBER Working Paper no. 784, Cambridge, Mass: National Bureau of
Economic Research.

Hartman, David G. (1984), “Tax Policy and Foreign Direct Investment in the United
States.” National Tax Journal 37: 475-87.

Hartman, David G. (1985), “Tax Policy and Foreign Direct Investment.” Journal of
Public Economics 26: 107-21 .

Hendershott, Patric H., and Sheng-Cheng Hu (1981), “Investment in Producer’s
Equipment.” In How Taxes A fleet Economic Behavior, edited by Aaron, Henry J., and
Joseph A. Pechman: 85-129. Washington DC: The Brookings Institution.

Hines, James R. (1999), “Lessons from Behavioral Responses to lntemational Taxation.”
National Tax Journal, June: 305-322.

Hines, James R., Jr., and R. Glenn Hubbard (1990), “Coming Home to America:
Dividend Repatriations by U.S. Multinationals.” In Taxation in the Global Economy,

edited by Assaf Razin and Joel Slemrod: 161-207. Chicago: University of Chicago Press.
128

Hulten, Charles R., and Frank C. Wykoff (1981), “The Measurement of Economic
Depreciation.” In Depreciation, Inflation, and The Taxation of Incomefrom Capital,
edited by Charles R. Hulten: 81-170, Washington, DC: The Urban Institute Press.

Hymer, Stephen H. (1976), The International Operations of National F irms, MA: MIT
Press.

Jorgenson, Dale W., and Kun-Young Yun (1991), Tax Reform and the Cost ofCapital.

London: Clarendon Press.

Kindleberger, Charles P. (1969), American Business Abroad: Six Lectures on Foreign
Direct Investment. New Haven, CT: Yale University Press.

King, Mervin A., and Don Fullerton (1984), The Taxation of Income from Capital.
Chicago: University of Chicago Press.

Knickerbocker, Frederick T. (1973), Oligopolistic Reaction and Multinational
Enterprise. Boston: Harvard University Graduate School of Business Administration.

MacDougall, G.D.A. (I960), “The Benefits and Costs of Private Investment from
Abroad: A Theoretical Approach.” Economic Record, March: 13-35.

Manzon, Jr., Gil B., and George A. Plesko (2001), “The Relationship Between Financial
and Tax Reporting Measures of Income.” Mimeo, Massachusetts Institute of Technology

(January).

Mishkin, Frederic S. (1984), “Are Real Interest Rates Equal Across Countries? An
Empirical Investigation of lntemational Parity Conditions.” Journal of Finance,
December: 1345-1357.

Moody, Scott (1998), Special Report: The Federal Corporate Income Tax Since World
War 11. Washington DC: Tax Foundation.

Obstfeld, Maurice (1993), “International Capital Mobility in the 1990s.” National Bureau
of Economic Research Working Paper No. 4534 (November).

129

Poniachek, Harvey A.(l986), Direct Foreign Investment in the United States. Lexington,
Mass: D.C. Heath and Company.

Quijano, Alicia M. (1990), “A guide to BEA statistics on foreign direct investment in the

United States.” Survey of Current Business. January.

Scholes, Myron S., and Mark A. Wolfson (1992), Taxes and Business Strategt
Englewood Cliffs, NJ: Prentice Hall.

Slemrod, Joel (1990), “Tax Effects on Foreign Direct Investment in the United States:
Evidence from a Cross-country Comparison.” In Taxation in the Global Economy, edited

by Assaf Razin and Joel Slemrod: 79-117. Chicago: University of Chicago Press.

Swenson, Deborah (1994), “The Impact of U.S. Tax Reform Act on Foreign Direct
Investment in the United States.” Journal of Public Economics 54: 243-66.

Tax Analysts (1986), Effective Corporate Tax Rates 1980-1 984. Arlington, VA.

Tax Analysts (1986), Effective Corporate Tax Rates 1985. Arlington, VA.

Tax Analysts (1987), Effective Corporate Tax Rates 1986. Arlington, VA.

Tax Analysts (1988), Effective Corporate Tax Rates 198 7. Arlington, VA.

Tax Analysts (1990), Effective Corporate Tax Rates 1988. Arlington, VA.

Tax Analysts (1991), Effective Corporate Tax Rates 1989. Arlington, VA.

Tayor, Herb (1992), “The Livingston Surveys: A History of Hopes and Fears.” Business
Review, January/February: 15-27.

U.S. Joint Committee on Taxation (1999), Description and Analysis of Present-Law

Rules Relating to International Taxation.

U.S. Council of Economic Advisers (1999), Economic Report of the President.
Washington, DC: Government Printing Office.

130

U.S. Department of Commerce, Bureau of Economic Analysis (various years), Survey of
Current Business. Washington DC.

U.S. Department of Treasury (1999), “The Problems of Corporate Tax Shelters:
Discussion, Analysis and Legislative Proposals.” Washington, DC: U.S. GPO.

Wooldridge, Jeffrey M. (2000), Introductory Econometrics: A Modern Approach.
Cincinnati, Ohio: South-Western College Publishing.

Young, Kan H. (1988), “The Effects of Taxes and Rates of Return on Foreign Direct
Investment in the United States.” National Tax Journal 41: 109-21.

131

illiliiiiiliilliiiittlili