PLACE N RETURN BOX to roman this checkout from your mood.
TO AVOID FINES Mum on or before data duo.

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

MSU I: An Affirmative Action/Equal Oppommny Institution
W

 

 

m1

INTERNATIONAL BILATERAL TRADE AND INVESTMENT NEGOTIATIONS:
THEORY, FORMAL MODEL, AND EMPIRICAL EVIDENCES

By

Erick Duchesne

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

DOCTOR OF PHILOSOPHY

Department of Political Science

1997

ABSTRACT

INTERNATIONAL BILATERAL TRADE AND INVESTMENT NEGOTIATIONS:
THEORY, FORMAL MODEL, AND EMPIRICAL EVIDENCES

By

Erick Duchesne

This dissertation represents a theoretical, formal, and empirical analysis of
bilateral international trade and investment negotiations. It focuses on interrelations
between negotiation structure and bargaining process. The structure of negotiation is
defined in terms of trade interdependence, structure of trade, institutional constraint,
political necessity, and societal support. The process of bargaining is illustrated by a
Stahl-Rubinstein formal model, which emphasizes international breakdown costs,
domestic costs for delays, and asymmetry of information. The model is tested with an
ordered multinomial logit analysis of the United States reliance on Section (Super) 301
retaliation and reciprocity policy instrument. The results indicate that it is a combination
of all dimensions of the theory and parameters of the formal model that best explains the

outcome of an international bilateral trade and investment negotiation.

©Copyright by
Erick Duchesne
1997

A la mémoire de mon pere.

iv

ACKNOWLEDGEMENTS

Many friends, mentors, and valued colleagues have provided me with their
priceless help in all stages of this dissertation. What first started as a term paper in Prof.
Scott Gates’ Research Seminar at Michigan State University laid the foundations for this
dissertation. Later, I was fortunate enough to be invited to the International Relations
Game Theory Summer Program at the Hoover Institution. That is where the enlightening
lectures of Prof. Bruce Bueno de Mesquita, Jim Morrow, and Cliff Morgan provided the
sparks to pursue a formal study of international trade negotiations. Also at the Hoover
Institution, I had the opportunity to meet Prof. Bill Clark of the Georgia Institute of
Technology. Bill and I have been corresponding regularly on the nature of international
negotiations. These discussions led to a joint paper on an application of the Stahl-
Rubinstein bargaining model to the Canada-United States Free Trade Agreement. I am
deeply indebted to Bill for his help with the second chapter of this dissertation.

When this dissertation was in its embryonic stage, I profited from the generous
comments of Profs. John Aldrich, Bruce Bueno de Mesquita, Brian Gaines, Bob
Muncaster, Dina Zinnes, and all the participants at the Junior Masters Class at the
University of Illinois at Urbana-Champaign. Several useful comments also came out of
a series of presentations I made at the weekly International Relations seminar at Michigan
State University. Special thanks to Scott Gates and Gretchen Hower for organizing the
seminar.

Numerous colleagues at Michigan State University lent me a hand at various

stages of the research. Brandon, Bryan, Roger, Sara, Sugato, be assured of my lasting
love and friendship. I would also like to thank Prof. Anthony Flemming-Blake and my
good friend Percy "Goatman" Hamming, without whom this dissertation would have been
as easy to read as a Chinese manual for miniature models, and Prof. Donald Ipperciel for
helping me putting all the pieces of the puzzle together at the formatting stage.

No one deserves more gratitude than Profs. Scott Gates, Gretchen Hower, and
Chuck Ostrom for their constant support over the last four years. They all went well
beyond the call of duty to supervise my progress. At times, they were my best
supporters. At other times, they were my toughest critics. Without them, this
dissertation would have never been possible. I owe them a lasting gratitude.

Finally, I would like to express all my love for my mom and my sisters Michelle,
Bérangére, Roselyne, Sylvianne, Jacynthe, Helene, and France for their blind faith in

me.

vi

TABLE OF CONTENTS

LIST OF TABLES ....................................... x
LIST OF FIGURES ...................................... xii
CHAPTER 1
INTRODUCTION ........................................ 1
Dissertation’s Goals .................................. 1
Plan of Study ..................................... 5
CHAPTER 2
STRUCTURE OF NEGOTIATION AND BARGAINING PROCESS ........ 17
Theoretical Considerations .............................. 17
The essence of negotiation ......................... 17
The contextual elements of analysis of
international negotiation ........................... 22
On the notion of power ...................... 23
Foreign policy analysis ....................... 27
Domestic/intemational linkages .................. 28
Putnam’s two-level of analysis metaphor ............ 30
The elements of analysis of international
bargaining process .............................. 36
The Nature of international trade bargaining .......... 37
CHAPTER 3
THE STAHL-RUBINSTEIN BARGAINING MODEL AND PROPOSITIONS . . . 42
The Stahl-Rubinstein Model ............................. 42
Propositions, Model, and Operationalization ................... 47
Contextual dimension ............................ 48
International ............................. 48
Domestic ............................... 56
Process dimension .............................. 64
Relationship between context and process ................ 77
Nature of Outcome in International Economic Negotiations .......... 86
CHAPTER 4
STRUCTURE OF EMPIRICAL TEST AND HYPOTHESES ............. 98
Empirical Test Strategy ............................... 98

vii

Method of Analysis ................................. 102

Dependent Variable ................................. 104
Failure: Challenging European Community export
subsidies for barley ............................ 105
Nominal success: A GATT illegal ban on imports of
cigarettes by the Thai government .................... 105
Partial success: Korea’s tariffs and quotas on
table wine imports ............................. 106
Success: Taiwan’s import duties on household
appliances .................................. 107
Independent Variables ............................... 108
International dimension .......................... 108
Interdependence .......................... 108
Trade structure and threat credibility .............. 111
Domestic dimension ............................ 112
Institutional constraint ...................... 112
Political necessity ......................... 115
Societal support .......................... 117
CHAPTER 5
EMPIRICAL TEST: SECTION (SUPER) 301 ..................... 120
Introduction ..................................... 120
Power and Dependence .............................. 121
Structural power .............................. 121
Economic power .............................. 126
Dependence ................................. 128
Empirical Results of Principal Model ...................... 133
Descriptive statistics ............................ 135
Multivariate Analysis ........................... 141
Method ............................... 141
Ordered multinomial logit results .................... 145
Elements of analysis of theoretical model
and parameters of formal model ................ 172
From the general to the specific: US bargaining success
versus Canada, Japan, and the European Union ....... 178
An Alternate Model of Compliance ....................... 199
CHAPTER 6
CONCLUSION ....................................... 217
APPENDIX A
THEORETICAL PROPOSITIONS ............................ 227
APPENDIX B
DESCRIPTION OF VARIABLES ............................ 230

APPENDIX C

LIST OF HYPOTHESES ................................. 235
APPENDIX D
DATA SET .......................................... 237
APPENDIX E
CALCULATION OF CASES CORRECTLY PREDICTED ............. 243
APPENDIX F
LOGIT ESTIMATES .................................... 249
LIST OF REFERENCES .................................. 253

ix

Table 1:

Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

Table 7:

Table 8:

Table 9:

Table 10:

Table 11:

Table 12:

Table 13:

Table 14:

Table 15:

LIST OF TABLES

Stahl-Rubinstein Three Rounds Sequential Game .......
American Institutional Division .................

Level of Success for American Negotiators
When Using Section (Super) 301 ................

Level of Success For American Negotiators by Targets
Economic Power Hypothesis: Target’s GNP .........

Dependence Hypothesis: Targets’ Exports to the
United States ............................

Dependence Hypothesis: Share of GNP Accounted
For By Exports to the US. ...................

Summary of Level of Analysis, Elements of Analysis,
Parameters, and Variables ....................

Descriptive Statistics .......................
Bivariate Analysis: Table of Means ..............
Interdependence Model: Empirical Estimates .........
Interdependence Model: Marginal Impacts of Regressors

Yearly Economic Data and Yearly Level of Success
by American Negotiators .....................

Interdependence Model Performance for Cases
Involving Canada .........................

Interdependence Model Performance for Cases
Involving Japan ..........................

114

123

124

128

130

131

134

136

146

149

165

179

181

Table 16:

Table 17:

Table 18:

Table 19:

Table 20:

Table 21:

Table 22:

Table 23:

Table 24:

Table 25:

Table 26:

Table 27:

Table 28:

Interdependence Model Performance for Cases
Involving the European Union .................

Assessment of US Bargaining Success Vis—a-Vis
Different Targets .........................

Compliance Model: Empirical Estimates ...........
Compliance Model: Marginal Effects of the Regressors

American Institutional Division .................
Data Set ...............................
Interdependence Model: Cases Correctly Predicted .....
Null Hypothesis: Empirical Estimates .............
Compliance Model: Cases Correctly Predicted ........
Logit Analysis: Success/Failure .................
Logit Analysis: Failure ......................
Logit Analysis: Success .....................

Ordered Multinomial Logit Analysis With
Three Categories .........................

xi

204

232

237

244

246

247

249

250

Figure 1:
Figure 2:
Figure 3:
Figure 4:
Figure 5:
Figure 6:
Figure 7:
Figure 8:

Figure 9:

Figure 10:
Figure 11:
Figure 12:
Figure 13:
Figure 14:
Figure 15:
Figure 16:
Figure 17:

Figure 18:

LIST OF FIGURES

General Bargaining Model ..................... 13
Contextual Dimension ....................... 14
Process Dimension ......................... 15
Detailed Bargaining Model .................... 16

Stahl-Rubinstein Bargaining Model Under Perfect Information 90

International Dimension ...................... 91
Institutional Dimension ....................... 92
Political Dimension ......................... 93
Societal Dimension ......................... 94
Effect of Discounting ........................ 95

Effect of Discounting With Different Discount Parameters . 96

Bargaining Model With Asymmetric Information ....... 97
Marginal Impact of Trade Interdependence .......... 210
Marginal Impact of Trade Interdependence (By GNP) . . . 211
Marginal Impact of Presidential Approval .......... 212
Marginal Impact of the Misery Index ............. 213

Marginal Impact of Trade Balance (Interdependence Model) 214

Marginal Impact of Targets’ Exports ............. 215

xii

Figure 19: Marginal Impact of Trade Balance (Compliance Model) . . 216

xiii

CHAPTER 1

INTRODUCTION

1- Dissertation’s Goals

Following Nixon’s measures in 1971 and the demise of the Bretton Woods
system, the United States gradually shifted from a spirited commitment to multilateral
trading institutions to a more aggressive bilateral stance. Also in 1971, The United
States experienced its first overall trade deficit of the century. Its declining universal
economic clout, a more hands-on policy by the Congress, and dissatisfaction regarding
multilateral negotiations are some of the factors that explain why the US turned to a
managed-trade strategy, or what was coined by some as "aggressive unilateralism"
(Bhagwati and Patrick, 1990). The main targets of the United States’ government were
its major trading partners: Canada, the European Union, and Japan‘. While the Reagan
administration successfully negotiated an agreement with Canada (CUSFTA) that
encompasses most of the financial and commercial interactions between the two

countriesz, its economic relationship with Japan has remained chaotic with accusations

 

‘ Almost 50% (37/75) of the cases treated in this dissertation are related to trade
relations between the United States and these three targets. This warrants a special
attention to these international trade partners.

2 It should be noted, however, that the agreement, as well as the North American
Free Trade Agreement (NAFTA), do not put an end to all trade conflicts between the
two neighbors. For instance, at the moment when this dissertation is written Ottawa and

l

2

of unfair trade regularly stigmatizing the Japanese, nonetheless sectoral agreements
between the parties do occur at regular intervals3. In contrast, the US. relationship with
the European Union seems to be more antagonistic; "fortress Europe " is a catch phrase
that has made its way into the corridors of the White House as well as into the hallways
of major political science and economics departments across the United States“. Yet,
economic agreements have resulted from specific negotiations between the EU and the
United States, and, despite CUSFTA and NAFTA, trade disputes between Canada and
the United States occur periodically.

What can explain such discrepancies in the successes achieved by the United
States with its major trading partners? Several hypotheses can be tested. Could these
results be a mere projection of the military and/or economic size of the protagonists?
Could it rather be a matter of trade interdependence? What about the impact of domestic
regimes or partisan trade politics?

This dissertation proposes that one has to delve into negotiation theories to
account for such an inconsistency in the results obtained by the American negotiators.
By turning to the study of the contextual and procedural elements of negotiation, not only

can we shed light on these bilateral economic encounters, but we can also gain insight

 

Washington are still embroiled in vigorous discussions regarding salmon fishing on the
West Coast. ,

3 One of the most notorious case concerns negotiation over supercomputers and
satellites (Bayard and Elliot, 1994, chapt. 5; Office of Technology Assessment, 1991;
Tyson, 1992). These, of course, are not unique cases since this dissertation treats twelve
different cases of trade conflicts with Tokyo.

4 In fact, 29% (22/75) of all cases treated in this study include trade conflicts
between the United States and the BC or an EC member.

into the process of negotiation itself.

The general aim of this dissertation is to develop a framework for a theory of
international negotiations. The next step is to develop a formal model that accounts for
the main elements of the theory. Finally, the theory and model must be tested
statistically on a large number of cases. To reach these goals, this dissertation examines
some political and economic factors, domestic as well as international (negotiation
context), that influence how the negotiators voluntarily distribute the proceeds from
international trade and investment cooperation (bargaining process). The crux of the
argument is that international and domestic contexts of the trade relationship between two
nation-states alter the process of negotiation, which, in turn, affects the respective
bargaining leverage of the state actors in an international trade and investment negotiation
(see Figure 1)5. Eventually, the outcome of negotiation itself is dependent on these
respective bargaining leverages. Even though all these dimensions are significant, by
themselves, they have only limited explanatory power; it is only through a thorough
examination of the interaction between these dimensions that we can start uncovering the
nature of the international negotiation process. My analysis shall not be static, but
dynamic. Henceforth, we need to tackle all the questions enumerated above at once. By
concentrating on only one or two aspects of trade politics, we may mistakenly
misperceive the tree from the forest, thereby presenting a flawed picture of international
trade negotiations.

This dissertation provides a first step in this direction. Nonetheless, it is still a

very important step. This dissertation constitutes the first study that attempts to move

 

5 All figures are listed at the end of their respective chapter.

4

from a theory of international trade negotiations to a formal model depicting the situation
and, finally, an empirical test on the theory and model. It is still, however, only a first
step, because not all aspects of the framework represented in Figure 1 are treated. In
a more realistic environment where negotiators haggle over multiple issues over a long
period of time, it makes sense to represent the situation as a repeated game. Under such
conditions, the bargaining outcome of a previous encounter may influence the context as
well as the process of an actual negotiation. For instance, the impact of the CUSFTA
was felt keenly by the representatives of Canada, Mexico and the United States when
trying to put forth the NAFTA. Not only were many of the negotiators the same, which
help lay the foundation and trust necessary to reach a second agreement, but also much
of the political and economic context had already been altered by the CUSFT A. One
could even wonder if the NAFTA would have been possible without the CUSFTA? This
example tends to demonstrate that the iterated (or repeated“) nature of bargaining is an
important aspect that all must keep in mind when studying international negotiations.
This, however, would add many complexities to a theory and a model which already
include many factors. As a first attempt at providing a theory, model and empirical test
of international trade negotiations, this dissertation is attentive to the phenomenon of
iterated (or repeated) nature of bargaining, even though it will not be explicitly modeled.

For similar reasons, the impact of bargaining process on context of negotiation
is not included explicitly in the model. Nevertheless, for future studies, it will become

imperative to take into account the fact that the way that bargaining goes has some salient

 

‘5 The distinction being that in an iterated game the same players meet again in the
next round. In a repeated game, only one of the player has to be present in the next
round.

5

influence on the context of negotiation. Examples abound. A government may use
international bargaining in order to send a strong message to its domestic constituents.
This strategy was used by Gorbachev when negotiating the INF treaty (Eichenberg, 1993)
and by Sadat at Camp David (Stein, 1993). Consequently the relationship between
domestic interests and the possibilities for international accord is not unidirectional.
Chief of governments can use international bargaining as a tool leading to a
restructuration of domestic coalitions. In some instances, this influence attempt can be
directed towards the domestic society of a target country. This was, according to Odell
(1993), one of Reagan’s most salient objectives during the negotiations with Brazil
regarding its computer policies.

Even if the impact of bargaining outcome on bargaining process and negotiation
structure, as well as the impact of bargaining process on negotiation context are two
salient segments of a fully developed theory of international negotiations, we must
perforce recognize that it would be a momentous and time-consuming task to test their
significance on a statistical basis. Consequently, this dissertation pertains to the influence
of negotiation context on bargaining process. These two phenomena affect the bargaining
leverage of negotiators. Finally, as illustrated by Figure 1, the study establishes a direct
connection between leverage and bargaining outcome, such that negotiators with more

bargaining clout obtain a larger "share of the pie ".

11- Plan of Study
In chapter 2, before getting into the specific nature of international trade

negotiations, this study concentrates on the essence of negotiation structure and

6

bargaining process. This distinction between the two concepts is clearly indicated. The
chapter presents a brief review of past and recent studies that put forth a series of
recommendations for the study of bargaining processes. A common finding of most of
these studies is that the discipline is in dire need of an integrated, "middle range" theory
that bridges the gap between technical formal studies developed mainly in the fields of
economics that concentrate on the process of negotiation and verbal, descriptive analysis
of one or few case studies that turn their attention to the outcomes of negotiation’.
These studies also point to the cooperative, non-zero—sum, informative, and time-
dependent nature of negotiation.

Following those recommendations, I introduce a theoretical foundation of such a
"middle range" theory from the perspective of international trade negotiations. First, I
present the general armature of my theory by introducing, in broad terms, a description
of negotiation structures and bargaining processes and their impacts on the outcome of
international trade negotiations. The first leg of the theory is made up of the contextual
elements influencing the output of international negotiations (Institutional constraint,
political need, societal support, trade interdependence, and trade structure). This is
followed by an assessment of the procedural elements of bargaining among international
negotiators (Risk attitude, information, delays). The interrelation between "context" and

"process" is then be explored. Finally, I explore the impact of these two dimensions on

 

7 That is not to say that there is no need for more formal analysis that sometimes
provide a useful simplification of the real world, or for detailed studies of single cases
that may unveil non-obvious, and sometimes meaningful, elements of analysis that
students of negotiation may have omitted previously. As an example, even though they
provide a highly stylized treatment of a negotiation outcome, it is hard to deny the
consequential contribution of Bueno de Mesquita and Lalman (1992) on the matter.

7

the bargaining leverage of negotiators, and in turn, how these affect the output of
international trade and investment negotiations.

The study then reverts to a more detailed analysis of the negotiation and
bargaining, as well as their effect on outcomes. First, the specific contextual elements
affecting the output of international trade and investment negotiations (Figure 2) are
investigated. There, I make a distinction between domestic and international contextual
dimensions. The domestic dimension is subdivided into institutional, political, and
societal elements of analysis, while the international dimension is subdivided into the
interdependence and bilateral trade structure elements of analysis. For this section, I
provide a justification for the choice of these dimensions and relate them to current
literature on bargaining theory and international relations. This stage concludes with a
claim for the use of the "two levels of analysis" metaphor as a useful representation of
the contextual elements of analysis influencing the process and outcome of international
trade negotiations.

The next section of chapter 2 focuses on the bargaining process (Figure 3). In
other words, my attention turns to alternative demands, concessions and give—and-take
processes, which affect a negotiation outcome. Again, I provide a brief discussion of
recent studies that have had a direct influence on the theoretical foundation of this study.
In this discussion, I conclude that one of the most efficient ways to represent the essence
of the bargaining process comes from a sequential representation, generally known as a
Stahl-Rubinstein game, which has been developed in the field of economics since the
mid-seventies.

This model is the focus of in chapter 3. The model assumes a certain reification

8

by treating the two sets of negotiators as two individuals having in front of them several
contractual agreements. Both have a strong incentive to reach an agreement, but their
interest is not entirely analogous. Given this situation, "What will be the agreed
contract, assuming that both parties behave rationally?" (Rubinstein, 1982: 97). It should
be clearly noted that this model does not answer the positive question (what is the
agreement reached in practice) or the normative question (what is the just agreement),
but what would be the agreement if both sides behave rationally. This constitutes a
notable precision. Given that negotiators in the real world do not always behave as
utility maxirnizers, we can not expect a perfect fit between the formal model and the
empirical test. At best, we must strive to obtain the best estimated fit between the two,
that is, the best approximation of the "real world". Furthermore, the sequential model
accounts for the negotiators’ international costs related to a breakdown in negotiations,
domestic costs for delaying", and level of information regarding the negotiating strength
of their opponent(s). Consequently, the Stahl-Rubinstein sequential model makes up a
nice bridge between the theory of bargaining and negotiation developed in this
dissertation and its empirical verification.

Chapter 3 also comprises the main propositions of this research. The first six
propositions relate to the structure of negotiation and its relationship with negotiators’
bargaining strength: Pl) international trade interdependence; P2A) international trade

complementarity; P2B) international threat credibility; P3) institutional domestic

 

8 It can be alternatively modeled with a discount parameter. Indeed, the higher the
domestic costs for delaying that negotiators incur the more impatient they become.
Hence, the more impatient they become the smaller is their discount parameter, i.e, the
more they devalue the future.

9

constraint; P4) domestic political necessity; and P5) domestic societal support. The next
three concentrate on the process of bargaining: P6) international breakdown costs; P7)
domestic costs for delays in reaching an agreement; and P8) information about other
side’s "strength". Finally, the last six purport to the impact of negotiation structure on
bargaining process: P9) international trade interdependence and international breakdown
costs; P10A) international trade complementarity and international breakdown costs;
PIOB) international trade complementarity and possibility of reaching an agreement;
P11) domestic institutional constraint and information about other negotiators’ strength;
P12) domestic political necessity and domestic costs for delays in reaching an agreement;
and P13) domestic societal support and domestic costs for delays in reaching an
agreement. The chapter concludes with guidelines for the appraisal of an international
trade and investment bargaining outcome. Falsifiable conditions, as well as indications
of how this study could be replicated or applied to cases not discussed here, are
presented. A special emphasis will be put on how, specifically, the outcome variable is
measured. A more detailed representation of the model is presented in Figure 4.

In chapter 4, I indicate how the theory and model can be applied to examples of
international trade negotiations. By fixing its attention on some specific incidents of
international trade negotiations, not all aspects of the theory are empirically tested. This
derives from a conscious choice. I attempted to develop a theory that can be applied to
a wide array of negotiations instances; as general or as specific as possible. At the limit
the model introduced in this dissertation could be implemented, with some basic
modifications, with any type of international negotiations, including security issues.

Here, the focus is rather specific. Consequently, some adaptations of the theoretical

10

framework were in order. Accordingly, this chapter starts by presenting a "strategy of
verification", that is, a presentation of a framework that will be used in the next chapter
for the purpose of empirical testing. It uses a data set partially constructed from a
research done at the International Institute for Economics on Washington’s use of Section
(Super) 3019. Out of this study by Bayard and Elliot (1993), I selected seventy—five
cases of trade negotiations involving the threat of use of reciprocity and retaliation by the
United States. Second, I shall indicate the methods of analysis that I intend to use.
Since the dependent variable, the level of bargaining success by American negotiators,
has four ordinal categories, I chose to test the propositions with an ordered multinomial
logit analysis.

Chapter 4 concludes with a description of the variables and discussion of the
hypotheses. These variables bridge the gap between the theoretical propositions and the
statistical analysis of the American use of Section (Super) 301. In consequence, all
hypotheses are organized according to each individual variable and correspond to a
contextual dimension of the theory (Figure 2). First, the trade interdependence
dimension is represented by the ratio of a target’s share of US overall export over a
similar measure for the American’s share of a target overall exports (DEP). Another
measure of this dimension accounts for the ratio of United States’ share of gross national
product (GNP) accounted for by exports to a target over a target’s share of GNP
accounted for by exports to the United States (DEPGNP). Second, the bilateral trade

structure (or American threat credibility) dimension is represented by a dummy variable

 

9 Note, however, that many variables used in this dissertation were not selected by
Bayard and Elliot.

11

representing the type of issue involved in the dispute (ISSUE). Third, the institutional
constraint dimension, in regards to the United States, considers the level of division
among its domestic political institutions (DIVIDED). For a target, this dissertation
borrows and reorganizes some data from the POLITY 111 project to appraise its level of
domestic institutional constraint (IN SCON ST). Fourth, the value of the contested goods
(DVCGOOD) and a misery index (Unemployment + Inflation = MISERY) are used to
calculate the United States’ economic necessity. The utilization of a threat of retaliation
or a direct retaliation by Washington (RETAL) corresponds to the political necessity
dimension. Fifth, and finally, the societal dimension is enacted by the trade balance
between the United States and a target (TBAL) and American presidential approval
(APPROV). These ten variables and hypotheses are statistically tested in the next
chapter.

Chapter 5 pertains to the results of the statistical analysis. It starts with some
basic test of alternative theoretical explanations of outcomes of international trade
negotiations (Structural power, economic power, and dependence). It is followed by the
results of some simple descriptive statistics and an ordered multinomial logit analysis for
the previous theoretical propositions and hypotheses. Six of the ten variables included
in the equation show a good significance. The marginal, individual, impact of each of
these pertinent variables is also evaluated. A special sections analyze the level of success
of American trade officials when resorting to Section (Super) 301 against different
targets. Finally, the chapter concludes with an alternative model, based on compliance
instead of interdependence, which seems better suited to the specific nature of the data

set used for the empirical analysis of outcomes of international trade and investment

12

negotiations. The implication of the results of the two models is then discussed.

The concluding chapter recaps the main findings of the dissertation. It goes back
to the theoretical foundations of the dissertation and assess its usefulness for our
comprehension of international negotiations. It also points to its shortcomings and

proposes some avenues for future research endeavor.

l3

_mno_>_ mEEQOm .mcmcmo ” P

93E

225w £5 E .3333 #0: 35:26.61 A ................

 

 

oEooSO
m:_:_mm._mm

 

 

22am 9: c. neoaxm 2:20:23. Al

 

 

 

 

 

 

 

 

 

 

flEEMmlhm
ho
......... mmooen.
-- - «mafia...—
A mEEmmEm
....................... <
............. coamzomwz
do

38:30

 

 

 

14

93025 moot.
6.9.35

L

co_mcmE_Q 8:99:00 “w 939”.

oocmbchmEoE
_

65:62.95

 

toaazm gmmmowz Efibmcoo
.ngoom _8=__on_ 38:23:.
_ m _
ozmmEoD

_

 

_
#meoo

15

:o_mcoE_D 80005 ”a 050$
26._ :91 Smtmaé oEmEE>m< Seton.

_ii _ _

22% Lou
Bmoo ozonoD coszLBE

 

 

33 :9:

eBOUxmoLm Lou
wfimoo $.8szsz

.

 

 

 

 

wwwooi

l6

 

 

8582
828.30 II

9:08me

 

59:0

_m_uo_2 mEEQOm 8:900 ”v 059”.

 

 

526d
mcEEmcmm
o>=omqmmm

\

 

 

 

omega;

/

 

>660 L0...
330 02me00

 

 

 

 

SEEP—E

 

 

 

 

cBODxmmLm
Lou wflwoo
_mco_..me..c_

 

 

 

380.5

A\\\

If

 

 

 

toaqzw Kaboom

 

 

 

\Ewwmomz _on_on_

 

 

 

Efibmcoo 6:25:55

 

 

 

 

o:moEoo

 

 

mug—03.5w $06;-
.9225

 

 

 

mocmbchmEQE

 

 

ficozquE.

nxmucoo

 

 

CHAPTER 2

Structure of Negotiation and Bargaining Process

1- Theoretical considerations
A) The Essence of Negotiation

It is commonplace to use the terms "negotiation" and "bargaining"
interchangeably. In an effort at clarification, I adopt John Cross’ (1969: 7) definitions:
"[T]he term "bargaining" will refer to the process of demand formation and revision
which provides the basic mechanism whereby the parties converge toward an agreement,
while "negotiation" will refer to the whole situation within which bargaining occurs."
It should be clear to the reader by now that this study is not only interested in bargaining
per se, but also concerned with the whole negotiation procedure. This distinction is more
than trivial. If I had been only preoccupied with the bargaining aspect of international
trade negotiations, this study would start with the Stahl-Rubinstein model, without paying
any attention to the structural and contextual elements of analysis justifying the magnitude
of the parameters of the formal model.

According to I. William Zartman, the great majority of negotiations studies,
"from the implicit wisdom of De Calliéres [1716] and De Felice [1778] to the explicit
analysis of Nash [1950, 1953] and Rapoport [1960], " imply the existence of common and

conflicting goals among parties (1979: 9). Indeed, if the parties had only conflictual

17

18

goals, there would be no need for negotiations, and there would be no process to
evaluate”. Anytime we observe the beginning of a negotiation process, "there is pn'ma
facie evidence of at least one common goal (the agreement itself)" (Zartman, ibid. ). On
the other hand, if the situation were one of common goals only, it would be only a
matter of finding the optimal agreement. The players would willingly offer all the
information necessary to the other parties, in order to reach an agreement as soon as
possible. In such a case, the bargaining process could be represented as a coordination
game and better be demonstrated by cooperative game theory. It would be only a matter
of "discovery" of the common interest“.

Beyond the category of things that no one cares about, there are complementary
interests of the bargaining parties. These are goods that are valued more by one player
than by the other. These values are the core of the negotiating process, because they can
be exchanged as tradeoffs against each other during the bargaining process (Zartman,
1976: 9). In game theoretic terms, we sometimes refer to these values as side payments.

The meeting point of divergent, convergent, and complementary interests is seen

in this study as the intersection of policy options. A value can be assigned to each of

 

1° Even in the case where the United States, using Section 301, threatens
international actors of unilateral retaliation, common goals always exist, such that the
threat can be seen as the starting point of the bargaining process. On the subject, see
Bayard and Elliot (1994).

“ This should not be interpreted as saying that the " search" and "discovery" of
common interest does not exist or is non-interesting. It is an important aspect of
negotiation. A bargaining process can be productive. For instance, classical liberal
economists, from David Richardo to Milton Friedman, claim that "freer" international
trade leads to an increase of the "pic of international goods" available to all nation-states.
I must mention, however, that while this model concentrates on the distributive aspects
of bargaining, it starts with the premise that the process of negotiation itself creates
opportunities for a productive enhancement of the amount of goods to be distributed.

19

these intersecting policy options. Again, we must turn to the wisdom of Zartman to

capture the essence of the allocation and reallocation process in a bargaining encounter:

Although it is in both parties’ interest to reach agreement on an
acceptable reallocation of values, it is also the interest of each to
end up with as much of the pie as it can or to give up as little and
gain as much as possible, depending on whether the reference is
to a single contested value or to several exchangeable or
complementary values (1976: 9).

This depiction of the bargaining process denotes the non-zero sum nature of
negotiation. In other words, if an agreement is reached, both parties win. Negotiation,
therefore, can be partially interpreted as a cooperative enterprise; that is a situation in
which all parties reap benefits from reaching an agreement (Cross, 1969: 4). Agreement
is Pareto-superior to nonagreement. In formal analyses, the non-agreement outcome is
conveniently set at zero. In empirical studies, this does not have to be the case. The
non-agreement outcome must take into account the possibility of retaliation. Sometimes
the threat of retaliation is implicit and we refer to it as "tacit bargaininglz" (Downs and
Rock, 1990); at other times the threat is very explicit and specific, such as when the
United States threatens to use Section 30113 (Bayard and Elliot, 1994). Furthermore,

to yield a non-zero sum, "either things must be valued differently by the different parties

 

‘2 This notion of tacit bargaining should not be confused with Schelling’s (1960: 54-
80) more specific interpretation. For Schelling tacit bargaining refers to a process in
which the negotiators select in isolation their possible agreement points. When no
coincidence in their choices occurs, no cooperation is achieved at all.

‘3 This element is introduced in the model later through the players international
costs regarding a breakdown in the negotiations.

20

or there must be side payments that are newly available because of the agreement"
(Zartman, 1976: 10). Hence, in order to reach an agreement, each party must be willing
to give up some of its less valued items in exchange for things it prizes more. It may
be also willing to give up some of its valued items in order to get the remainder of the
items.

Even in the face of the mixed-motive nature of the process, if negotiation were
only a matter of finding an acceptable reallocation of a given set of values between the
owners, we would only need to construct two simple utility scales indicating how much
worth the players associate with each outcome, and this would lead to a knife-edged
effect of negotiation (Zartman, 1976: 13). In such a scenario, the result would be the
Nash solution, located at a point where the product of what the parties value is the
greatest. However, given it is in the players’ interest to obtain more by giving less, it
is also in their interest to control the information regarding their utility list or function
(Homans, 1961: 61). It is also in their own interest to inflate the value they associate
with each item being negotiated. This explains why in a negotiation encounter, the
parties have tendency to go from some extreme points, where they tend to put a high
value on all or most items being negotiated, and gradually converge towards a "middle
ground " that better reflects their true utilities. Regardless of their effort to hide their true
preferences, structural elements may favor one player over the other when time comes
to hold on private information”. Thus, the negotiation encounter must be seen as a

process occurring under imperfect information, where the respective parties control the

 

‘4 For instance, I argue below that the institutional constraints on the ratification
process of an international economic agreement plays a role regarding the capability of
players to hide their true preferences.

21

veracity and the amount of information convened to other parties. Given its
"communicative" nature, the exchange of partial information is at the core of the
negotiation process.

In a certain sense, information is power. In the verbal and tacit exchanges that
comprise negotiation, a player’s goal is to manipulate information in order to affect the
other player’s evaluation of the values involved, in such a way "to bring about
convergence or agreement at a point more favorable to one side than the other"
(Zartman, 1976: 15). Information is power, such that it can be defined as the
"volitionally controlled ability of one party to produce such movement or re-evaluation
on the part of the other party, often more generally as the ability of one party to cause
another behavior in an intended direction" (Zartman, 1976: Ibid).

The ability to use information as a source of power may often depend on the skills
of the negotiators. In this study, I do not use such an approach. The power of
information is derived out of the internal structure of the government and its relationship
with the civil society. These factors constitute the resources that the negotiator can have
in regard to power. That is to say that information as power is relational; depending on
the respective structure of government and state/civil society relationship, some
negotiators may hold more informational power than others.

Another very important characterization which I apply to the bargaining process
is that it is fundamentally time dependent. There are costs associated, in terms of dollars
(or utilities) as well as in terms of postponement of the consumption of the good being
negotiated, with the bargaining process. It is these costs that motivate the whole process;

it does not only matter that an agreement is reached, it also matters when an agreement

22

is reached. Because it is widely acknowledged that there are some potential benefits
associated with an agreement for all parties (otherwise there wouldn’t be any
negotiation), all players prefer to reap the benefits earlier than later. In other words,
future benefits are discounted from the present. Furthermore, "there is a fixed cost of
bargaining which recurs in each time period" (Cross, 1969: 13). Bargaining itself is
costly. In the model introduced below, a primordial aspect influencing the distribution
of the benefits accruing from an agreement is that the negotiators need not face the same
costs for a delay in the negotiations. A negotiator who is more patient, that is, whose
costs for delay are lower, is more likely, ceteris paribus, to reap more profits out of the
agreement.

It appears unrealistic to build a deterministic model, with knife-edged outcomes,
that takes into account all the intricacies and complexity of the bargaining process. What
I propose in this dissertation is an ideal-type, a certain representation of the " real world,"
as well as a committed effort to build a model that represents this template of the
empirical world. In the pages that follow, I strive to identify meaningful and testable--
even if not always easily measurable-~dimensions of negotiations structure and bargaining

process.

B) The Contextual Elements of Agilysis of International Negotiations15
The purpose of this section is to yield an explanation of the research problem that

frames my theoretical analysis and to place this project into the general literature on the

 

‘5 For excellent discussions of the structure of the bargaining situation, see Roth
(1985) and Binmore and Dasgupta (1987).

23

level of analysis of international interactions. Waltz’s (1959) three-images metaphor is
among the first representations of international interactions to reveal the complexity of
international phenomena. Nevertheless, despite a recognition of the multi-faceted nature
of international relations, much of the discipline is dominated by a structural notion of
power revolving around such theories as balance of power (Claude, 1962; Haas, 1953;
Singer, Bremer and Stuckey, 1972), hegemonic leadership (Gilpin, 1981; Thompson,
1988), and power transition (Organski, 1968; Organski and Kugler, 1980).

Consequently, I now turn my attention to the saliency of the concept of power.

i) 0n the Nation of Power

Despite the absence of a clear consensus regarding its meaning, the notion of
power is the core concept for realists as well as neorealists. In a large part, given that
a political policy seeks either to keep power, to increase power, or to demonstrate power
(Morgenthau, 1985), it is generally assumed that an actor with the larger amount of
power has the advantage over an actor with a smaller amount. In other words, by
classical "power politics" theory, the stronger actor with the greater capabilities, will by
definition prevail in any encounter (military or otherwise) with a weaker actor.
Negotiation theorists who adopt such relational conceptualization (Iklé, 1964; Lall, 1966)
do not see power acting in any different way in international negotiations than in any
other aspect of international interactions (Habeeb, 1988:10). For them, the stronger
state, by tautological definition, will win in a bargaining situation (ibid.:3).

Unfortunately, even if it seems that the notion of power would lead us to a good

understanding of bargaining outcomes, the world is strewn with counter-intuitive cases

24

where the party that is believed to be weaker, measured through military and other
classical elements, had the upper hand in an international encounter or at least did not
fare worse than its opponent (Habeeb, 1988; Paul, 1994; Wriggins, 1987; Zartman,
1987)”. Similar results are derived from episodes of international trade negotiations.
For instance, John Odell (1993) finds that the United States was more successful when
negotiating with the European Community regarding its decision to elevate trade barriers
on US. feedgrains after Portugal and Spain joined the EC in 1986 than it was when
dealing with Brazil when the Latin American country introduced a national program
designed to promote its national computer industry. Without any doubts, any classical
aggregate measure of power17 would assign greater power to the European Community
than it would to Brazil.

Despite their widespread use, there is little consensus on the essence of power
relationships. The tautological nature of a definition of power as the ability of one actor
to get another actor to act differently without the actions of the former”, is due to the
"failure to distinguish clearly enough between a bargaining outcome and the bargaining
process which leads to it" (Cross, 1969: 17). A definition of power can indeed only take
us so far as the starting point if it only concentrates on the description of the outcome

of a bargaining episode. A more helpful definition of power should concentrate on the

 

‘6 The best-known examples being of course the Vietnam War and the Soviet-Afghan
Conflict.

‘7 See for instance the assessments of aggregate power by Cline (1975), Jones
(1954), Knorr (1970: chaps 2-3), and Organski (1968: chaps 6-8).

‘8 This is paramount to Dahl’s classic definition of power as "A’s ability to get B
to do something that B would not otherwise do" (1957: 201).

 

25

determinants of the outcome, not the outcome itself. This explains why this study
focuses on the structural and contextual elements of analysis that emphasize the ability
of one player to impose heavy costs for the burden of delay in the negotiation on the
other player, coupled with its own insensitivity to costs imposed on oneself. The
difficulty, then, is to uncover and select the proper elements which are appropriate for
a fruitful definition of the notion of power. That’s the task that is left for remainder of
this dissertation”.

It is indeed meaningful to incorporate a notion of power in any conceptual
framework interested in the study of international interactions in general, or international
bargaining outcomes in particular”. The idea of an aggregate structural power can be
helpful when one is interested in defining a power structure of the international system
into great, medium, and weak powers. It can be useful in providing an overall picture
of the actors’ positions in a self-help international system (Waltz, 1979), but it is of less
use when applied to international negotiations. Using aggregate structural power for such

matters assumes highly fungible power resources (Baldwin, 1979: 192). Yet power may

 

‘9 One element that is left out in this study is the idiosyncratic nature of the strategy
used by negotiators. Its role is important, as demonstrated by Lax and Sebenius (1986)
and Zartman and Berman (1982), but it is very difficult to transpose its influence
systematically to a large number of cases. Nevertheless, we should keep in mind that
the tougher the negotiators, the greater their opportunity to get an agreement close to
their position, but the less their ability of getting an agreement at all. Here, I take such
a factor into account only implicitly by assuming that the "toughness" of a negotiator is
derived from institutional constraints, political necessity, domestic support, as well as
international context, not from personal attributes.

2° Such conceptualization of power has been used successfully to forecast negotiation
outcomes. See, inter alia, Bueno de Mesquita, Newman, and Rabushka (1985). It
should be noted, however, that in their forecasting approach, these authors attribute a
large influence to the domestic bargaining setting on the outcome of an international
confrontation. In this sense, it is not so different than the approach of this dissertation.

26
not be highly fungible (Ibid: 165; see also Goldmann, 1979; Hart, 1976; Hoffmann, 1975;

Sprout and Sprout, 1965). Keohane’s (1989:62) "modified structural research program"
implies that structure of influence that works in one area may be irrelevant in another
(See also Keohane and Nye, 1977:46-54). National resources, in order to have
significant influence in bargaining, must create options that are useful in the context of
a specific issue-area that nations face (Lockhart, 1979:92; Zartman, 1991: 68). In
consequence, even though aggregate structural power should not be considered as a
"meaningless" concept, as stated by Baldwin (1979: 193), it should be seen as no more
than the foundation of an actor’s strategy in a specific bargaining situation (Habeeb,
1988:18; Lockhart, 1979: 90).

Such strategy may involve an attempt by strategists to link one particular issue
with other issue areas so as to achieve maximum advantage from their whole array of
international interactions (Tollison and Willett, 1979). Conversely, according to the
interdependence paradigm proponents, it is rather the power balance in a specific issue-
area, not its aggregate nature, that determines the outcome of interaction in that area.
It seems clear to them that different issue areas have different political structures that
may be more or less insulated from the overall distribution of military and economic
capabilities (Keohane and Nye, 1977:50). This research adopts some of the neoliberal
claims stated above by assuming that the structure of international trade between
negotiating nations is among the principal elements of analysis influencing the bargaining
outcome of international trade and investment negotiations. This representation of issue-
specific power is included further in the formal analysis of international trade

negotiations.

27

ii) Foreign Policy Analysis

Another shortcoming of the neorealist enterprise is its reification of the state. It
is a foregone conclusion that a scientific look at " inside the state" is necessary to acquire
a better knowledge of international interactions. The long tradition of Foreign Policy
Analysis (FPA) clearly points in this direction. In his seminal "pre-theories and theories
of foreign policy ", James Rosenau (1966) was among the forerunners to recognize the
need for a multilevel and multicausal explanation of international relations in general, and
foreign policy in particular (see also Snyder, Bruck, and Sapin, 1954 and 1963; Sprout
and Sprout, 1956, 1957, and 1965). Valerie Hudson (1995) provides a comprehensive,
but still incomplete, overview of an impressive theoretical and empirical progress in FPA
in last thirty-five years. The range of research goes from psychological explanations (de
Rivera, 1968; George, 1969; Singer and Hudson, 1992) to the presidential use of force
(James and O’Neill, 1991; Ostrom and Job, 1986), from bureaucratic interpretations
(Allison, 1971; Halperin, 1974; Hilsman, 1987) to cultural analyses of foreign policy
(Almond and Verba, 1963, Pye and Verba, 1965; Sampson, 1987). These constitute
only a very small subset of the research in FPA, but they share a common goal with this
study and other literature21 in the field not mentioned here; an assumption "that the
source of much behavior and most change in international politics is human beings,

acting individually or in collectivities" (Hudson, 1995:210).

 

2‘ Although not all share the behavioral assumption of other non-systemic (or
partially non-systemic) explanations of international interactions, such as expected utility
(Bueno de Mesquita, 1981), crisis decision-making (Lebow, 1981; Snyder and Diesing,
1977), regime type (Doyle, 1986; Maoz and Abdolali, 1989; Rummel, 1983; Russett,
1990), public Opinion (Morgan and Bickers, 1992; Shapiro and Page, 1988), and state
strength (Katzenstein, 1978; Krasner, 1978), have been prominent in the international
relations literature.

28

Yet, despite impressive results, an analysis of international politics that focuses
only on domestic factors may suffer from the same shortcomings as structural
explanations. For instance, any consequential empirical foreign policy research needs
to turn its attention to "low politics" transactions like trade flows (Schrodt, 1995). But
just as importantly, as James Rosenau suggested nearly thirty years ago with his elaborate
taxonomy of "linkage politics" (1969, see also his 1973), the interrelation between
domestic politics and the international environment needs to be assessed theoretically and

empirically by the political science academic community”.

iii) Domestic/International Linkages

Generally, interpretations of the linkage between domestic and international
politics have been cast in terms of the domestic causes and international effects, as in
Waltz’s (1959) "second image" (See also Hagan, 1986; Morgan and Campbell, 1991;

Morgan and Bickers, 1992; Ward and Widmaier, 1982)”. Reversed explanations have

 

22 Robert Putnam affirms that Rosenau’s effort " generated little cumulative research,
except for a flurry of work correlating domestic and international ’conflict behavior’"
(19882430). I have expressed, above, my skepticism regarding the linkage between
issue-areas, but I believe that linkages between domestic and international politics can
lead to fruitful developments. In fact, Putnam’s two-level of analysis metaphor is a
prime example of such flourishing results in the fields of international relations and
comparative politics.

23 Hagan argues that domestic conflict affects the degree of independence,
commitment, and intensity in a nation’s foreign policy behavior, but not always its
involvement in foreign conflict (1986: 291). Morgan and Campbell (1991) demonstrate
that for major powers (including nondemocracies), higher levels of decisional constraint
lead to a lower probability that conflicts will escalate to war. They do not find such
correspondence for minor powers. Morgan and Bickers (1992) add to Morgan’s previous
work that state leaders will treat an erosion of domestic support more seriously
(especially for actions short of war) when it comes from within segments of society that
are critical of the leader’s ruling coalition than when it comes from other domestic

29

focused on international causes and domestic effects. For instance, Wilkenfield and
Zinnes (1973) utilized factor analysis to propose a Markovian interpretation indicating
that the level of foreign conflict behavior affects the changes or transitions over time
between the levels of domestic conflict behavior. Studies on the impact of the
international environment on domestic politics gained some theoretical standing later with
Gourevitch’s "second image reversed " (1978). More recently, case studies representing
diverse methodological approaches, have displayed a theoretical sophistication on the
intemational-to-domestic causal connection by concentrating on the impact of the
international economy on domestic politics and domestic political economy (Alt, 1987;
Evans, 1979; Gourevitch, 1986; Katzenstein, 1985). However, as noted by Putnam
(1988: 433), these works elude any type of reciprocal causation.

Attempts have been made to look at the simultaneous and reciprocal cause/effect
relations between domestic and international environments. Works on regional
integration have focused mainly on the impact of parties and interest groups on the
process of European integration, but they have also recognized, with the notion of
" spillover " , the feedbacks between domestic an international developments (Deutsch et
a1. , 1957; Haas, 1958). Later studies in the domain shift the attention to the evolution

of the new supranational institutions and were the forerunners of transnationalism studies,

 

groups. As for Ward and Widmaier (1982), they do not find any evidence to support the
notion that as a consequence of high levels of domestic turmoil and conflict, a nation-
state will export foreign conflict behavior in the form of involvement in serious dispute.
In sum, even with the bulk of studies in the domain, the state of discipline shows only
mitigated support for extemalization of moderate internal conflict to the political use of
force abroad ("diversionary theory of war"), and only for a short—term period ("rally-
around-the-flag" effect). For review articles see Levy (1989), Stohl (1980), and Zinnes
(1980).

30

especially through the analysis of international regimes (Krasner, 1983). Unfortunately,
this again, diverted the attention from domestic factors. On a more theoretical level, the
"agent-structure" connection have been explored (Carlsnaes, 1992; Dessler, 1989;
Wendt, 1987). This line of work suggests that the structure of the international system
should be seen as a mean for action rather than only an environment in which action
takes place. However, the debate is mainly at the ontological and epistemological level,

and a rigorous empirical framework has yet to be provided.

iv) Putnam ’s T wo-Level of Analysis Metaphor

A more promising area of research stems from Putnam’s (1988) two level of
analysis metaphor. Putnam’s work constitutes an interesting attempt to bridge the gap
between domestic and international levels of analysis“. He focuses his attention on an
important but neglected aspect of international relations theory: international negotiation.
At level one, the world of structural realism, there are interactions between international
actors. At level two, the world of domestic politics, negotiators are accountable to a
wider internal audience. He defines the logic of the two-level game in the following

way:

 

2‘ Even though their work is nOt specifically about international trade negotiations,
Bueno de Mesquita, Newman, and Rabushka (1985), and Bueno de Mesquita (1990)
denote that the domestic bargaining setting acts as a salient determinant of the negotiators
set of demands in an international encounter. They add also, similarly to Putnam, that
the international process determines, often through bargaining, the outcome of an
international confrontation.

31

At the national level, domestic groups pursue their interests by

pressuring the government to adopt favorable policies, and

politicians seek power by constructing coalitions among those

groups. At the international level, national governments seek to

maximize their own ability to satisfy domestic pressures, while

minimizing the adverse consequences of foreign developments

(432).

He adds that the unusual complexity of this game stems from the idea that a rational
move at one board may be impolitic at the other board. Consequently, "there are
powerful incentives for consistency between the two games" (432).

Putnam’s game implies that the possibility of agreement is limited to overlaps of
what is acceptable to the winning coalitions (level 2 game) in each of the parties in the
negotiation (level 1 game). This area is called a "win-set". Thus by definition, any
agreement must fall within the Level 2 win—sets of each party, and the larger the win-sets
are, the more likely is an agreement at Level 1. The win-sets are important for two main
reasons: First of all, the decision-makers at the Level 1 game have to take into account
that they will need a ratification of their agreement. The more formal the ratification
process, the more constrained are the decision-makers in their international negotiations.
The second reason is "that the relative size of the respective Level II win-sets will affect
the distribution of the joint gains from the international bargain" (440)”. This last point
is fundamental for anyone who wants to understand the logic of the two-level game: the
larger the win-set of a negotiator, the more he can be pushed around by the other Level

I negotiators. Conversely, a small domestic win-set adds to the negotiation leverage of

a negotiator because he can always say that "this proposition looks fine to me, but I will

 

25 This general principle was first noted by Schelling (1960: 19-28).

 

32

never be able to have it ratified at home! ". Hence, " [a] government that is internally
divided is more likely to be able to strike a deal internationally than one that is firmly
committed to a single party" (445)“.

Moreover, this raises the specter of " involuntary defection" when the failure to
ratify an agreement is not due to an "egoist" action in the absence of an enforceable
contract, but rather an incapacity to get the domestic constituents to ratify the agreement
(438). It is rational for a negotiator to use a "tying hands" strategy, an attempt to
constrict his own "win-set, " in order to induce other negotiators to compromise at a point
closer to his preferences (Moravcsik, 1993: 28)”.

The exploitation of asymmetrical information is therefore a key issue.

Asymmetrical knowledge about the possibility of ratification makes it possible for a

 

2“ Putnam’s seminal article and the edited volume that ensued (Evans, Jacobson, and
Putnam, 1993) have stimulated a flurry of research, empirical, formal and theoretical,
on the causal relationship between the domestic and international environments. Iida
(1992b) and Cowhey (1993) have expanded the notion of "involuntary defection". On
application to American international trade issues, see Duchesne and Clark (1995), Kraus
(1993) and Schoppa (1993), or on the European Union, see Alt and Eichengreen (1989),
Huelshoff (1993) and Schneider (1993). For studies on international security issues, see
Knopf (1993), Morrow (1991), as well as part II of Evans, Jacobson and Putnam (1993).
For theoretical (mostly game theoretic) developments the reader should consult the
following: Iida (1991, 1992a), Kilgour (1991), McGinnis and Williams (1992, 1993), Mo
(1990, 1993), Pahre (1992, 1993), and Starr and McGinnis (1992). For a similar
approach in the area of comparative politics, consult Tsebelis (1990).

27 Peter Evans (1993: 399) indicates that the strategy of "tying hands" is infrequently
attempted and usually not effective. He adds that the strategy suggested by Schelling is
logically plausible "but lacks efficacy in practice. " Evans makes a good point.
However, it may be true that the strategy is rarely used purposely, but the fact remains
that the perception of a constricted "win-set" for the other side’s negotiators may have
an effect on the bargaining process. For instance, when negotiating with their American
peers, Canadian negotiators were worrying about the ratification role of US. Congress
leading to the CUSFTA agreement (Hart, 1994: passim). This, in turn, may have added
to the American’s bargaining clout.

 

33

negotiator to use deliberate misinformation as a negotiation tactic”. This asymmetry
of information regarding the chance of ratification of an agreement is often derived from
the domestic institutional structure of government. For instance, the role of Congress
as the "bad cop " during the CUSFTA negotiations with Canada, may have put the
American negotiators in the driver’s seat. There was little doubts in any American
negotiator’s mind that an international agreement would be ratified by the Canadian
Parliament, while in a Canadian negotiator’s mind there was always some doubts about
the true preferences of the American Congress (Duchesne and Clark, 1995). Therefore,
this seems to indicate that the domestic institutional structure of ratification of an
agreement is a driving force behind the negotiation strategy and leverage of respective
statesmen in an international negotiation.

Transnational linkages are another important aspect of the two-level of analysis
metaphor. As Andrew Moravcsik (1993: 15) puts it, "[t]he statesman can also target
policies directly at domestic groups in foreign countries, seeking allies ’behind the back’
of his international counter part. " Yet, a statesman does not always have to launch a
"covert" operation to facilitate his task. Often, an international alliance is natural. John
Odell (1993), for instance, notes that an important reason why the United States had
difficulty imposing sanctions on Brazil after it decided to put forth a national program
supporting its computer industry at the expense of foreign firm in 1985, is that large
American multinational corporations in the computer industry had already impressive

investments in the Latin American country. That natural alliance created a "negative

 

2“ For a skeptical view regarding the ability to deliberately use deception in an
international setting, see Bueno de Mesquita and Lalman (1990).

34

reverberation" that put the American negotiators at a certain disadvantage vis-a-vis the
Brazilians.

A large part of this analysis constitutes an effort to model international
negotiations by assessing the interaction between domestic and international elements of
analysis. Thus, the main question is not to wonder if a theory of international trade
negotiation should start from an international or domestic point of view, but to strive to
provide a framework that encompasses and pays attention to the interrelation between the
two levels of analysis. Therefore, the elements of analysis included in Putnam’s
metaphor are a driving force behind the model of international trade and investment
negotiations that I introduce below. Besides the influence of the size of the win-set, the
possibility of involuntary defection, and transnational linkages, Putnam’s model denotes
the importance of domestic preferences, coalitions and institutions, the negotiators
strategies, as well as the relevance of uncertainty about the other negotiators
preferences”.

These structural elements also correspond to the essence of negotiation introduced
earlier. First, the two-level of analysis metaphor, through its emphasis on domestic
preferences accounts for the three types of goal-conflicting, common, and
complementary-—pursued by the statesmen. Second, the convergence of respective policy
options can be studied according to the respective win-sets of the negotiators. This, in
turn, is an aspect of this research that I will consider below when I examine the nature

of the outcome of an international trade and investment negotiation. Third, through

 

29 These factors will be discussed further, as they are an integral part of the research
design of this dissertation.

35

transnational linkages and reverberation, the two-level of analysis metaphor informs us
on the specificity and credibility of the threats that actors may choose to use during an
international meeting. Fourth, and finally, the important concept of information as
relational power is derived from the institutional domestic structure of government and
associated with the possibility of voluntary or involuntary defection. It is therefore my
contention that the two-level of analysis metaphor is an adequate representation of the
structural nature of international trade and investment negotiations.

In the model that follows, the logic of two-level games is also used to depict the
relationship between the context30 and process of international trade negotiations.
Putnam’s metaphor provides the theoretical backdrop used not only to provide a better
understanding of the contextual dimensions of international trade negotiations, but also
to demonstrate how the context affects the process of international bargaining.

The model that ensues is not, however, a straightforward application of the two-
level game metaphor. If it was, it would be concerned not only with the effect of
domestic policies on the outcomes of international bargaining, but also with the
consequence of international moves on the ability to achieve domestic goals. The latter
strategy, called "synergistic issue linkage" by Putnam (1988: 447-48), more precisely,
constitutes an attempt by a statesman to " gain approval for an important domestic
measure by linking it to an attractive international agreement" (Moravcsik, 1993: 25).
This strategy is not explicit in the model used here. The "dependent" variable being the

outcome of international trade negotiation; special attention is not paid to the effect of

 

3° Note however that in this case the context of the international negotiation is not
derived uniquely from domestic factors. The nature of the trade relationship between the
trading nations is another important contextual element of analysis.

36

such an international negotiation outcome on the future domestic coalition formation. It
is nevertheless acknowledged that statesmen when negotiating an international trade
agreement keep an eye on the effect of such an international outcome on the domestic
coalition formation, and this, in turn, affects directly the bargaining process. Thus, this
aspect is modeled explicitly in this study. For instance, when negotiating the CUSFTA,
Brian Mulroney, for electoral considerations, linked the issue of an international trade
agreement with his forthcoming attempt at reelection (Duchesne and Clark, 1995). The
general model that I propose takes into account how such electoral considerations31
affected how the Canadian government approached the negotiation, but it does not
consider the effect of the outcome of the negotiation on the Canadian electoral results.

The nature of international bargaining as a process can then be modeled as a
Stéhl-Rubinstein sequential game. So, let’s now turn to the process of international

bargaining.

C) The Elements of Analysis of the International Bargaining Process.

The purpose of the previous section was to unveil the structural elements of
international negotiations as well as to demonstrate that issues of international trade
bargaining are related to mainstream research in international relations. It is unfortunate,
however, that international bargaining seems to be studied in some "vacuum," such that

the relationships to important issues of international relations and comparative politics is

 

3‘ It is not, however, included in specific empirical test regarding Section (Super)
301. It is ascertained here that such electoral pressures are only felt in instances of
"big", encompassing, international negotiations, such as a trade agreement or the decision
to join an international organization.

37

rarely assessed. Putnam’s two—level games -and their "offsprings"- represent a laudable
effort attempting to link the two research streams. One of this research’s goals is to
follow the same path. Also, a good amount of research on international negotiations has
been done in the fields of economics or psychology, with a special focus on bargaining
process. Consequently, this investigation tries to bridge the gap between those
disciplines and political science research.

This section is intended to present briefly the methods of analysis that have been
used in the study of international bargaining. This review is far from comprehensive as
it only attempt to identify some of the variables that I shall include in the research design

of this study”.

i) The Nature of International Trade Bargaining

The bargaining situations of interest to this research are those "in which the
ability of one participant to gain his ends is dependent to an important degree on the
choices or decisions that the other will make" (Schelling, 1960: 5)”. Interdependence
among negotiators points to the fact that one actor cannot gain access to the coveted
goods without the cooperation of other actors. Consequently, the true nature of

bargaining power in international trade negotiations is not that one state can credibly

 

32 For a comprehensive review see Habeeb (1988: chaps. 1-3), Kemper and Kemper
(1994: 1-28), Kremenyuk (1991). The most complete bibliography on international
negotiation is Lakos (1989).

33 Alternatively, following Nash we can use the term "bargaining" to refer to a
situation in which (i) individuals ("players ") have the possibility of concluding a mutually
beneficial agreement, (ii) there is a conflict of interests about which agreement to
conclude, and (iii) no agreement may be imposed on any individual without his approval
(Osborne and Rubinstein, 1990).

38

deny another state the entire gain from trade, "but merely the difference between the
second state’s entire gain from trade, and the gains it would receive if the first state
exercised its bargaining power over the terms of trade" (Wagner, 1988: 478). Also, in
this research, I adopt a variant of the classical liberalist assumption that free trade
increases the amount of international goods available to the individual actors (Hirschman,
1945). Therefore, what is negotiated is the division of the surplus created by freer trade.
Yet, returning to the section on the nature of international negotiation, we have seen that
the negotiators are pursuing different sets of goals. If it was only a matter of finding the
optimal intersection of policy options--that is, increasing the pie of international goods
through cooperation--, a strict classical liberal world view would indicate the relevant
nature of international trade. If international trade was rather a zero-sum interaction that
is, the negotiators having only conflicting goals, then a neomercantilist interpretation
would be more appropriate. It is, however, the mix of conflicting, common, and
complementary goals pursued by the negotiators that makes these strict interpretations
of international trade inappropriate.

On one hand, one could be prone to express the opinion that a classical liberal
world view better represents the nature of the CUSFTA negotiations. A better look at
the CUSFTA informs us about the important conflicting goals pursued by the Canadian
and American negotiators (Duchesne and Clark, 1995). On the other hand, we could
assume that when the American government threatens to use Section 301, it is a clear
expression of aggressive unilateralism, and thus a neomercantilist representation of the
situation is more appropriate. Yet, even in these cases, the fact that negotiations take

place after the threat is a clear indication that there are common interests in getting an

39

agreement rather than going forth with the threat. Thus, a neomercantilist interpretation
may not be suited even for those extreme cases.

There exists a large consensus regarding this definition of bargaining, but diverse
methods, models and theories have been proposed in order to find out which factors
determine the outcome bargaining, that is "who gets what." I have indicated earlier that
the concept of power was the leading indicator of determining the respective share of the
"pie" of international goods (Iklé, 1964; Lall, 1966). These diplomatic historical reports
of international negotiation fail to account for those instances when a seemingly "weaker"
power has received a larger share of the bargaining outcome (Habeeb, 1988; Odell,
1993; Paul, 1994). Therefore, a model of bargaining must include some factors that
contemplate different types of asymmetries in order to explain some of the counter-
intuitive cases.

The sequentiality of bargaining is addressed rather effectively by the
concession/convergence theorists. They posit that the parties start at some point of
stalemate and, in an action-reaction process they converge at a different rate towards an
agreement (Bartos, 1974, 1977; Contini, 1958; Cross, 1969, 1978). Even though the
convergence theorists indicate that the weaker party concedes more and at a faster rate
(Hicks, 1932; Pen, 1952; Zeuthen, 1930), they do not have much to say about what
makes the parties concede at a typical rate (Habeeb, 1988: 12). Nevertheless, the
concession/convergence theory alerts us to the pertinence of time in any type of

negotiation. It points to the proposition that the party whose costs of holding out are

40

greater generally concedes at a faster rate than his or her opponent“. It is my belief
that it is through the study of the interaction between structure and process that one can
determine the factors that affect the concession rate. The relationship between some
structural factors, such as political necessity and societal support, and the concession rate
or discount factor will be discussed in the modeling section of this dissertation.

The social-psychological literature on negotiation is oriented towards the behavior
and personality traits of the negotiators (de Calliéres, 1919 (1716); Deutsch, 1973)
characterized as "hard-line and soft-line" (Snyder and Diesing, 1977: 308),
" interpersonal " and "motivational orientation" (Rubin and Brown, 1975). The general
assumption of this approach is that "under conditions of unequal relative power among
bargainers, the party with high power tends to behave exploitatively, while the less
powerful party tends to behave submissively, unless certain special conditions prevail
[specifically, coalition formation by the weak] " (Rubin and Brown, 1975: 199). Again
the problem resides with a loose interpretation of the notion of power and the ad hoc use
of alternative factors [such as coalition formation] to explain bargaining outcomes. These
factors should not become part of a model only when they appear useful, but must be an
integral part of a better specified model. This means, again, that one must look at the
interaction between structure and process to explain the behavioral traits of the
negotiators.

Spector’s (1978) analysis of "need orientation" points in the right direction.

Spector studies the "personal" needs of a negotiator and their influence on the outcome

 

3" This is accounted for by either the discount rate or domestic costs in a Stam-
Rubirrstein sequential bargaining model.

 

41

of negotiations. My intent is not to use such a measure in the model below, given its
idiosyncratic nature, but to rather focus on the needs associated with the entire
negotiating parties. The needs of the negotiating parties, or governments, are associated
with their relationship with the civil society. These needs may depend on the
concentration of support from the business community and unions, or it can depend on
electoral considerations. Furthermore, as indicated previously, transnational linkages and
reverberation, may also affect the needs and the specificity of the threats of the
negotiating parties. This, in turn, will affect the domestic costs for any delays. For
example, a set of negotiators representing a country and, more that often, a single party,
that needs an agreement to enhance its electoral position on the domestic checkerboard,
will tend to face higher costs associated with a delay in the negotiations. In other words,
the negotiators will tend to accept a "suboptimal" agreement in order to avoid a delay
that might jeopardize their electoral chances. The set of negotiators that does not face
such an electoral constraint will tend, ceteris paribus, to be more patient, and will use
a more aggressive strategy to get to a deal that looks more optimal to them.

The next chapter unveils a formal model, called Stahl-Rubinstein, that accounts

for the interrelation between structure of negotiation and process of bargaining.

 

CHAPTER 3

The Stahl-Rubinstein Bargaining Model and Propositions

I- The Stahl-Rubinstein Model35

The procedural elements of analysis (costs, sequentiality, asymmetry) are
important aspects of a Stahl-Rubinstein bargaining approach. Furthermore, this approach
constitutes an interesting attempt at portraying the time dimension associated with
bargaining. It is thus my intention to use a Stahl-Rubinstein sequential game to represent
the bargaining aspects of international trade and investment negotiations.

Before getting into the specific description of the Stéhl-Rubinstein approach, let’s

start with its main assumptions”:

A1: Bargaining involves two players, Sven (S) and Ollie (0).

A2: Bargaining concerns one issue”. All considered contracts are Pareto-optimal
provided agreement is reached immediately, i.e., only agreements along a
contract curve are considered.

 

35 See Figure 5 for the basic structure of the model.
36 These assumptions are adapted from Stahl (1994: 3-4).

37 This is a controversial assumption when used to describe an international trade
negotiation. I shall discuss this issue below when I describe how I intend to measure the
outcome of an international trade agreement.

42

43

Each party uses a maximizing behavior in the sense that each party strives to
maximize its ordinal utility”.

Each party has correct expectations about the other party’s behavior, that is, each
party knows that the other party behaves according to A3. E. g. , Sven knows that
Ollie maximizes.

Each party has correct expectations about the other party’s expectations. E.g.,
Sven knows that Ollie knows that Sven maximizes and Ollie knows that Sven
knows that Ollie knows that Sven maximizes, etc., ad infinitum (common
knowledge).

The bargaining process is sequential; the parties alternate bidding so that each
party knows exactly what the other party has bid earlier. Thus the game becomes
one of perfect information”.

Each party knows the payoffs obtainable from each agreement, not only by itself,
but also by the other party, thus providing for a game of complete information.

Bargaining takes place over a number of periods, all of finite length. If we
assume that Ollie is able to accept an offer (Pj) by Sven in period j, we assume
that Ollie makes a comparison between the payoffs of Pi in period j and in period
j+1. Then, in regards to a specific order of bidding, for Ollie the comparison
regarding his payoff will be between payoffs in periods 1-2, 3-4, then 5-6, while
the comparisons for Sven will concern the periods 2-3, 34, then 5-6.

Bargaining takes place over a number of periods, but not necessarily of the same
length. If we assign all payoffs to the very end of the periods (as in A3,), then
for a given order of bidding, the periods for which the length has consequences
for Ollie are the even numbered periods 2, 4, 6, while the periods of importance
for Sven are the odd numbered ones. Thus, as long as we look for the
determination of the solution for a specific order of bidding, we can assume
different lengths for odd and even numbered periods (that is a different discount
rate)”.

 

38 In my discussion of the measurement of the outcome, I’ll indicate that cardinal

utility can be assumed to be a monotone transformation of money.

39 This assumption is relaxed below by the assumption that one (or two) party does

not know about the "strength" of the other party, that is, it doesn’t know about the true
preferences of the other party.

4° Therefore, it is an advantage to have a low periodic discount and hence a short

period between one’s own acceptance time and the corresponding time for the other
party. Likewise, it is an advantage if one can prolong the time between when the

44

A9: The payoffs for both Sven and Ollie of an agreement on a certain alternative
decrease over time. With a party’s payoff of an agreement on alternative x in
period j as v(xj), we assume that v(xj) > v(Xj,1) for every x and for every j.

After the presentation of its main assumptions, we can now turn to a verbal
interpretation of the Stain-Rubinstein approach, and a discussion of its main parameters.
The Stahl-Rubinstein model consists of a series of offers made alternatively by the
players in a negotiation“. The players can take action only at times in the (infinite) set
T= {0,1,2 ..}. In each period t e T one of the players, say i, proposes an agreement (a
member of X) and the other player (1) either accepts the offer or rejects it. More
specifically, an offer can be accepted, which ends the game, a counter-offer can be made
(at time t+1), or the players can use an outside option (b) if a stalemate in the
negotiation occurs“. In general cases the outside option is the status quo ante, but as
indicated previously, when the bargaining is accompanied with an implicit threat of
retaliation, the outside option may be perceived by the negotiators as an outcome worse
than the possibility of signing a suboptimal outcome.

Under perfect information, the logic of the Stirhl-Rubinstein equilibrium is that

the player who makes the first offer uses backward induction to submit an offer that

 

opponent can accept a bid and when oneself next can accept (Carlsson, 1986; Osborne
and Rubinstein, 1990: 54). In other words, it pays for one player to perceive that the
"pie" of coveted goods shrinks at a slower rate for himself than the perception of the
other player. For instance, Hart (1994) noted the tendency for the American negotiators
to delay their answers to a Canadian offer, while the latter showed great impatience,
during the CUSFTA negotiations.

4‘ For a more detailed analysis of the Stahl-Rubinstein approach, consult the original
texts: Rubinstein (1982), Stilhl (1972).

‘2 The outside option is not represented in the basic structure of Figure 5.

45

would be accepted by the other player if the negotiations were to drag on. The discount
factors represent the time pressures facing the respective players“. Therefore, because
of the need to reach a quick agreement, it pays to see the payoffs received by the player
who makes the offer equal to the payoffs that he would receive if the negotiation were
to last until a satisfactory agreement was reached by both players. In other words,
because the pie is shrinking as time goes on (an agreement in the future is worth less
than an agreement in the present), the player makes an offer in the present that contains

more concessions to the other player from a "bigger pie", which is commensurate to a

 

‘3 Stahl (1994: 11) indicates that the essence of the Stahl-Rubinstein approach does
not have to include a discount parameter in order convey the time pressure on the
bargainers. Let’s take a simple case to illustrate the point. Let’s assume that farmer
Sven produces cabbages at a cost of 50 cents a pound, and produces carrots at a cost of
75 cents a pound. On the other hand, farmer Ollie produces cabbages at a cost of 75
cents a pound and carrots at a cost of 50 cents a pound. Therefore, it would be
advantageous for both farmers if farmer Ollie was ready to buy his cabbages from farmer
Sven at any price between 51 cents and 74 cents a pound. On the other hand, it would
also be advantageous for both farmers if farmer Sven was ready to buy his carrots from
farmer Ollie at any price between 51 cents and 74 cents a pound. According to the
comparative advantage concept developed by classical liberal economists, both players
would be made better off by agreeing on an acceptable term of trade. However, every
year, when Fall harvest comes, they may argue on an acceptable term of trade (e. g.
Farmer Ollie might be willing to sell his carrots at a price of no less than 70 cents a
pound and would not pay more than 55 cents a pound for farmer Sven’s cabbages, while
farmer Sven would not sell his cabbages for any less than 60 cents a pound and would
not pay more than 65 cents a pound for former Ollie’s carrots). The problem is that for
every year that they cannot reach an agreement, both farmers are losing on a potential
gain from an agreeable term of trade. This shows that even without a discount parameter
there is a certain time pressure on the farmers to reach an agreement as quickly as
possible.

The discount parameter or cost for delay comes into play in the following
situation. Let’s assume that Olga, Ollie’s wife, understands the terms of trade involved
and scolds farmer Ollie every Fall for not reaching an agreement with farmer Sven. On
the other hand, Svetlana, farmer Sven’s wife, has a profound distrust for farmer Ollie,
and insists that her husband reaps the maximum benefits from the terms of trade.
Consequently, farmer Ollie will have a high discount rate (or will face high costs for
delaying), while farmer Sven is ready to hold out as long as possible to reach a better
agreement. Hence, farmer Sven has a bargaining advantage over farmer Ollie.

46

future offer with less concessions, but from a "smaller pie ".

Under perfect information an agreement would occur without delays. However,
incomplete information situations can explain costly delays. There have been several
attempts to extend the Stahl-Rubinstein sequential bargaining model to cover cases of
limited information (Sobel and Takahashi, 1983; Fudenberg and Tirole, 1983; Cramton,
1984; and Rubinstein, 1985).44 Limitations in information tend to have the most
profound impact on a model’s outcome when they are asymmetric. Hence, I suspect that
the highest degree of value added would come from examining an aspect of the
bargaining process where one side possessed private information. It can be argued that
the differences in domestic political institutions generate this type of asymmetry.
Specifically, the difference between executive and legislative relations, and the possible
impacts of these on the process of treaty ratification, are likely to be consequential. The
model proposed here can be used to examine the effects of these differences on the
ratification process. It can also serve as a referent to explore how these institutional
differences can be expected to influence inter-state bargaining and the nature of the
agreement that is eventually reached.

A Stahl-Rubinstein representation can also consider the possibility that the players
have different costs involved with the possibility of a breakdown in the negotiations
(Osborne and Rubinstein, 1990: 71-73). I shall further discuss the relationship between
breakdown costs and structural elements of analysis.

There are two other parameters of the Stahl-Rubinstein approach that I do not

 

4‘ See Sutton (1986) for an interesting review that attempts to integrate these “non-
cooperative” or “sequential” bargaining models with the earlier axiomatic approach to
bargaining theory (Nash 1950, 1953; Harsanyi 1967).

47

include in the empirical part of this research: the order of offer (that is who goes first)
and finite v. infinite games. Stirhl (1994) discusses the effect of who goes first when he
compares his model to Rubinstein’s approach. It turns out that "who goes first" has an
influence on the outcome in the Rubinstein’s model (1982), but it does not in Stéhl’s
(1972) model. He then proposes a synthesis of both models that can alleviate the
problem. As for the infinite v. finite case, it turns out that there may be some
quantitative effect on the outcome, but no qualitative effect. That is, the magnitude of
the difference in what the players get may differ slightly, dependent on the modeling
approach, but it will not affect which player gets more than the other depending on the
magnitude of the other parameters.

The Stahl-Rubinstein approach seems to be an appropriate representation of the
bargaining aspects of negotiation introduce at the beginning of this paper. The discount
factor (or cost for delay) is a good explanation of the convergence rate of the negotiators
towards an agreement. It can also be modeled in such a way to take into consideration
the effects of limited information, as well as statesmen’s different costs associated with
delays and breakdowns. But more importantly, it focuses on the element of time through
its sequential nature. The relation between the procedural nature of the Stéhl-Rubinstein
bargaining approach and the structural aspects of negotiation revealed by the two-level
of analysis is explored below. For now, I turn to a more specific discussion of the

dimensions of the model.

48

II- Propositions, Model, and Operationalization

The previous sections have provided the foundations for the model that I introduce
below. In these sections, 1 have tried to capture some of the limitations of current
research regarding international interactions in general and international bargaining more
specifically. The remainder of this dissertation explains and justifies each element of the
model. I shall proceed by stating each particular proposition, then I shall provide a little
discussion on the effect of each parameter, and finally, I shall present how each
parameter will be "tested. "

Part A) discusses the structural dimensions of international trade negotiations,
starting with the international aspects and following with domestic parameters. Part B)
turns to the process of international trade bargaining. Part C) explores the relationship
between the structure of international negotiation and the process of trade bargaining.
Finally, part D) presents a discussion of the nature of the outcome of international trade

negotiation (dependent variable).

A) ertextual Dimension

The contextual dimension relates to the nature of the structure of international
trade negotiations previously discussed. I look first at the international structural aspects
of negotiation, before turning my intention to its domestic components. This section is
an extension of the theoretical discussion of the two-level of analysis metaphor proposed

by Putnam (1988).

49

i) International (See Figure 6)
The two international structural elements of negotiation that I include in my model
are the trade dependence and the bilateral structure of trade between two nation-states.

Let’s first address the nature of interdependence.

O Interdependence45

P1: In a bilateral trade and investment negotiation, a nation-state that is less dependent
on the other state for its commercial exchanges has more bargaining power than
the second nation-state.

Many political scientists focus their attention on the contextual dimension of
international negotiations“. A wide array of studies mention that asymmetrical
dependence is an important factor that must be considered when assessing the outcome
of international negotiations. This comes as no surprise given that since the end of
World War 11 international relations have been marked by significant liberalization and
expansion of world trade. With an increasing dependence on international commerce,
national economies have become more sensitive to each other. The aftermath of
increasing commercial interdependence are fascinating in themselves, but more engaging
is the question of which countries are more dependent on the international economic

system, and how does that affect their relative bargaining power in international

 

‘5 A list of all propositions is presented in Appendix A.

4° Gary Goertz (1994) provides, arguably, the most comprehensive discussion of
contextual theories of international relations.

50

negotiations.

These questions are at the basis of this study. It starts with the hypothesis that
the less dependent a state is towards anorher state (and the entire international economic
system) regarding its economic exchanges, the more leverage it has when entering an
international negotiation concerning trade and investment with the latter. It is therefore
hypothesized that the larger a state’s share of trade and investment resulting from
international exchanges with another state, the more vulnerable it becomes to the
consequences of a breakdown of the commercial relationship between the two countries.
If the second state is not as reliant on the first for its external exchanges, the more likely
it is to find alternatives to trading with the first state. Therefore, each party’s bargaining
strategy is affected by its perceived vulnerability to retaliation.

This asymmetry of economic dependence lays the foundation for the actors’
strategies in any specific bargaining situation. Indeed, the less dependent state is more
likely to impose heavy costs for the burden of delay on the other player, coupled with
its own insensitivity to costs imposed on oneself. The credibility of a threat made by one
of the player to walk out of the bargaining table and impose sanctions partially depends

on the relative nature of trade dependence of the two states“7 .

- Operationalization:
A first measure of interdependence can be set in terms of geographic

concentration of exports and investment:

 

‘7 It is only partially the effect of trade dependence because it depends also on the
domestic distribution of benefits and costs that stems from agreement or nonagreement.
This element of analysis is explored by proposition P5 (Societal Support).

51

i) B’s share of A’s international trade and foreign direct investment output (%).
(A’s dependence)
ii) A’s share of B’s international trade and foreign direct investment output (%).
(B’s dependence)

iii) The ratio of i) over ii).

It can then be said that the trading partners are symmetrically dependent if the
value of iii) is roughly equal to 1. Nation A is said to be more dependent upon nation
B (instead of the reverse) if the value is greater than 1. Conversely, nation B is said to
be more dependent upon nation B (instead of the reverse) if the value is situated between
0 and 1.

However, this does not present the entire interdependence picture. A nation-state
may be highly dependent on another state for its external exchanges, but international
trade may only be a small subset of its entire economic strength. Therefore, it is also
important to incorporate a nation’s GNP in our test of trade interdependence.
Empirically, I suggest a look at the following:

i) A’s share of GNP that is accounted for by its exports and foreign direct

investments in country B (%). (A’s dependence)

ii) B’s share of GNP that is accounted for by its exports and foreign direct

investments in country A (%). (B’s dependence)

iii) The ratio of i) over ii).

Again, it can be said that the trading partners are symmetrically dependent if the

52

value of iii) is roughly equal to l. Nation A is said to be more dependent upon nation
B (instead of the reverse) if the value is greater than 1. Conversely, nation B is said to
be more dependent upon nation A (instead of the reverse) if the value is situated between

Oand 1.

O Complementarity of Trade

P2A: When the nature of the trade relationship between two countries is
complementary, an international trade and investment agreement is more likely
to occur.

Another international dimension that is considered in this work is the bilateral
trade structure between the negotiating parties. It has more to do with the possibility that
an agreement will be reached, but when coupled with the "political necessity" and
"societal support" dimensions, it can also affect the bargaining leverage of the negotiating
parties. It is assumed that when there is a high complementarity of trade between two
nations the negotiators are more likely to reach an agreement and meet their objectives.
Complementarity is defined in terms of what the trading partners produce. High
complementarity is assumed when the second country produces goods the domestic
population of the first country needs, and vice versa. For instance, country A might be
rich in primary products, while country B produces more manufactured products. When
the parties are negotiating over goods that both countries produce, the negotiators are
more tenacious and an agreement is less likely (thus the difficulty of attaining respective

objectives). The situation is thus more difficult for the negotiators who make a demand

53

for a concession while they have a great political need for an agreement (See the political
necessity dimension below)“, as well as when they ask for their counterparts to restrain
the flow of goods that are competing with domestic production. Furthermore, when a
high complementarity of trade exists, negotiators of nation A can find consumer groups
inside nation B that are likely to support their position, thus weakening nation B’s
negotiators bargaining position (International linkages). Even when two countries
produce similar goods, public support in the other country can be found if the goods
produced in the first country are cheaper to manufacture. However, this support can
easily be upset by domestic groups, such as unions, that would prefer to protect their
home industry at the expense of higher prices for the goods. This, again, reveals the
importance of looking at international trade negotiations in terms of a two-level of

analysis metaphor.

- Operationalization

The export concentration dimension can be unveiled by exploring the geographic
concentration in sources of supply (Holsti, 1978: 516). A first-cut at the problem is to
look at the geographic concentration of supply in terms of the three main sectors of the

economy :

 

‘3 For instance, a higher complementarity of trade between Canada and the United
States can partially explain why the latter has been more successful in its negotiations
with the former than it has been with Japan and the European Community. It is true that
Canada and the United States produce many similar goods, but a large part of the
negotiations between the two countries revolves around the United States’ willingness to
invest in the development of Canadian resources in exchange for a greater access for
Canadian—based companies to the US. market in order to assure them a greater economy
of scale for their'sales.

54

i) B’s share of A’s source of supply in the primary sector, and vice versa (%)
ii) B’s share of A’s source of supply in the secondary sector, and vice versa (%)

iii) B’s share of A’s source of supply in the tertiary sector, and vice versa (%)

This measure gives us a good indication of the complementarity of trade between
the two countries. Especially, it indicates the possibility of a more encompassing trade
agreement between two countries. For instance, it can be assumed that an agreement is
more likely when nation A is an important source of supply of primary goods for nation
B, while nation B is an important source of secondary products for nation A.

In more restrictive cases, such as a specific trade dispute, it would be worthwhile
looking at the specific concentration of trade of the disputed products. It is especially

the case when attesting the credibility of a threat as indicated below.

0 Trade Complementarity and Threat Credibility

P2B: A threat of trade retaliation is more credible when the targeted product is also
produced in the targeting country and/or can be imported from a third country.

Complementarity of trade can also be used as an indication of the credibility of
retaliation threats. It can be hypothesized that a threat is more credible when it is
targeted at a product that is also produced by the threatening country or that can be found

in any third country”. In such case, the threatening country does not have a great need

 

‘9 This is especially relevant for primary goods.

55

for the targeted product. Therefore, it is important to look at the concentration of
production of the targeted product. Sometimes, alternative goods can be used (or bought
from another country) to replace the targeted product. This also needs to be considered.
Of course, this measure is not always appropriate. It can better be used for instances
when the United States threatens to use Section 301, than it would for a free trade

agreement negotiation. For appropriate cases, I intend to look at the following measures:

- Operationalization
i) Concentration of production of the product in contention by the targeted
country (% of world production)

ii) Possibility of alternative sources of supply for the importing country

To make the verification of this hypothesis more general, we must also take into
account cases for which a threatening country explores the possibility of limiting the
export of a specific product to a targeted country. This could apply to the case where
a country or a group of countries (cartel) control a large proportion of the production of
a specific product, such that, by limiting its supply, it (they) can control world prices.
The obvious example that comes to mind is the price control of petroleum products
exercised by the Organization of Petroleum Exporting Countries in 1973. In this
instance, the Operationalization is the following:

i) Concentration of production of the product in contention by the exporting

country (% of world production)

ii) Possibility of alternatives sources of supply for the importing country

56

ii) Domestic

The domestic dimension is divided into three elements of analysis: institutional
constraint, political necessity, and societal support. The first element is a more
permanent fixture of domestic politics that can be applied to any cases of international
trade negotiations of a specific nation—state (or group of nation-states). The last two are
issue-specific oriented. The reasons for dividing the domestic dimension into three
categories are twofold. First, these elements have different (and often opposite) effects
on the structure of negotiation. Second, they affect the process of bargaining in a
different manner. The relationship between domestic contextual dimensions of
negotiation and the procedural dimensions of bargaining are explored later. For now,
I turn to the description of the domestic elements of analysis, starting with institutional

constraints.

0 Institutional Constraints

P3: The higher the institutional domestic constraint faced by negotiators, the higher
their bargaining leverage in an bilateral international trade and investment
negotiation.

This dimension pertains to the possibility of " involuntary defection. " Involuntary
defection reflects the behavior of an agent who is unable to deliver on a promise because
of failed domestic ratification. In some sense, this dimension is related to the "veto"
power that some domestic institutions may have. It can be seen as a source of bargaining

power since the negotiators, when facing strong domestic institutional opposition, can

57

always argue that agreement suggested by the other side’s negotiators is reasonable but
has little prospect of being ratified at home. The focus here is on institutional relevancy
and restriction on bargaining authorities. The more complex and uncertain the domestic
ratification process is, the more the negotiators can use this "trump card," and more
leverage they have in the international negotiation.

It is often assumed that such institutional constraint is, de facto, higher in
democracies that in non-democracies. This dichotomy democracy/autocracy can however
be misleading. It shows mitigated results in terms of willingness of the regime to go to
war, and it has not yet been tested adequately in regards to matters related to trade.
Therefore, I intend to shy away from such categorization and focus instead on
institutional political opposition.

It appears obvious that institutional political opposition comes necessarily from
other political parties (which in some sense takes us back to the democracy/autocracy
debate). It is indeed important to look at the role that may be played by other parties in
terms of the ratification of a trade accords", but intra party dissension must not be
overlooked. There is presumably an important difference when the ruling party or group
is a cohesive organization, rather than when the ruling party or group is organized
around structural factions.

Official political opposition may not always present a clear picture of institutional

opposition. Regime fragmentation should be explored. First, one needs to look at

 

5° There is an important difference in terms of political opposition if the regime has
no opposition parties or if the opposition parties control the legislature as in the case of
a minority government in a parliamentary democracy or in when the legislature is
controlled by one party while the executive is controlled by another in a presidential
system. Minor and major party opposition should also be distinguished.

58

regime fragmentation in terms of the degree of cohesiveness of the ruling party (group
configuration). The regime can be dominated by a cohesive single party, by a single
party with internal factions, or by a coalition of autonomous parties or groups. Second,
the presence or not of a predominant actor may have an impact on the level of
institutional opposition. Third, and more importantly, policy polarization over trade
matters must be considered. In fact, it may be true that the level of institutional domestic

opposition is high, but the actors may or may not be divided over foreign trade matters.

- Operationalization“:
To recap, here are the factors that I suggest should be evaluated:
Fragmentation: i) Group Configuration
— Regime is a single party or cohesive group
- Regime is a single party or group with internal faction
- Regime is a coalition of autonomous parties or groups
ii) Actor Predominance
- Politically predominant actor is present
- Politically predominant actor is absent
iii) Policy Polarization
- Actors are not divided over foreign trade matters

- Actors have broad differences on foreign trade issues

 

5‘ When one of the negotiating factions is an international institution, such as the
EU, the following Operationalization, with slight changes, can still be used.

59

Political Opposition: i) Intra Party Dissension
- Ruling party or group is a cohesive organization
- Ruling party or group is organized around structural
factions
ii) Other Political Parties
- No opposition parties
- Minor party opposition: 1—32 percent of the seats
- Strong party opposition: has 33-49 percent of seats

- Opposition parties control the legislature

0 Political Necessity

P4: In a bilateral international trade and investment negotiation, the greater the
necessity for an agreement for one party, the less its bargaining leverage.

This category of the domestic dimension refers to the political needs of the
government. Two main instances of political necessity are identified in this research:
electoral and economic. First, the international negotiation may be an important part of
the electoral platform of the incumbent government. In such a case the electoral
objective may come to surpass the trade objective. Therefore, the negotiators may be
pressed by the ruling party to come to a quick end to the negotiation in order to boost
the electoral chances of the party and this may come at the expense of the fulfillment of

trade objectives. This seemed to be the case of the Mulroney regime when negotiating

60
the CUSFTA with the United States. Furthermore, not only did the need for an

agreement come during an electoral period, but also the CUSFT A negotiations were a
very salient part of the Conservatives electoral platform.

Second, the political needs of the government can also be interpreted as the
necessity for an agreement in order to salvage what is perceived by the chief of
government as a bad economic situation. In other words, an agreement can be seen as
a way to ratchet up the domestic economic situation. The focus here therefore is the
domestic leaders’ perceived effect of the international trade agreement on domestic
economic revitalization. This perceived need for an agreement is a source of weakness
in an international negotiation because the negotiators see an agreement as a panacea for
a difficult domestic economic situation. Consequently, they consider that while a delay
exists in the signing of an agreement the domestic economic situation worsens. The
negotiators must then weigh the effects of a delay in the reaching of an agreement with
the results of a quick agreement that do not meet an optimal amount of their demands.
Of course, negotiators that do not view an international commercial agreement as a
panacea for a difficult economic situation have the luxury of delaying the negotiations
until they get closer to an agreement that contains the majority of their demands. Hence,

their bargaining leverage is increased.

- Operationalization
This category of the domestic dimension is rather difficult to measure, especially
in terms of economic needs. Because what is mostly important is the perceived need of

the chief of governments and his close associates, objective measures of economic

 

61

performance may not represent the most adequate representation of political necessity.
Nevertheless, it can be considered as a point of departure that must be associated with
a close reading of the official declaration of government members. Given this caveat,

here is how I suggest to "measure" this category of the domestic dimension:

Electoral Necessity: i) Existence of an forthcoming election

ii) Saliency of the international trade agreement for electoral

purpose

Economic necessity: i) Growth rate
ii) Unemployment Rate (Yearly change)
iii) Inflation rate
iv) Saliency of the international trade agreement in regards to

economic growth

0 Societal Support

P5: In a bilateral international trade and investment negotiation, the greater the
societal support for negotiators, the greater their bargaining leverage.

This dimension is the most difficult to estimate. It refers to a govemment’s
relationship with its civil society. This category of the domestic dimension is more

Specifically related to the case of negotiation at hand than the previous category. While

 

62

the previous category focused on the "needs" of the negotiators, this category
concentrates on the distributive effects of the potential agreements on the different
societal groups. The distinction is more than trivial. Moravcsik (1993: 26) points out
that "the more diffuse the costs and benefits of the proposed agreement, the more
possibilities for the statesmen to target swing groups and gain support at relatively low
cost. " This statement indicates the saliency of the scope of the issue negotiated. In the
case of a negotiation involving many issues, such as CUSFTA or NAFTA, the statesmen
are more at ease to target groups that would support their aims and are hence more likely
to succeed in swinging target groups towards their own views about the potential
agreements. Conversely, in more obtuse instances, such as the use of Section (Super)
301 by the United States, domestic support and opposition are more concentrated and
groups that would bear the highest costs from the potential agreement are likely to battle
fiercely against the proposed agreement. Therefore, the political necessity dimension
bears more influence on the outcome of a negotiation than the societal support dimension
when the negotiation encompasses a large number of issues, and vice versa. This is the
main justification explaining why I chose to separate these two categories of the domestic
dimension of international trade negotiation.

The success of the negotiators can be contingent on their resourcefulness, "that
is, on recognizing options usable in the conflict episode at hand and gaining acceptance
for the use of these options within the decision unit" (Lockart, 1979:133).
Resourcefulness affects the ability of negotiators to withhold an agreement or put an end
to the negotiations, or at least threaten to do so. A high level of resourcefulness

increases the bargaining power of the negotiators as they can count on the "willingness"

 

63

of the domestic society to suffer from the consequences of a failed negotiation. It is
easier for negotiators to take a tough stance during a negotiation when they know that the
population, the elite, and the bureaucracy they represent, share their view and would
rally to their support in the event that they should have to walk out of the negotiating
table. When domestic actors are highly committed to a set of strategies used by their
negotiators, they highly value the outcome pursued by the negotiators, and the latter can
use a more tenacious and dedicated strategy in order to reach a more optimal outcome.
On the contrary, "[o]n the basis of Olsonian collective-action analysis, it seems
reasonable to expect that concentrated groups that are disadvantaged by an agreement
[. . .] will become intransigent and influential opponents of agreement" (Moravcsik, 1993:
26).

Negotiators can also enjoy an increased bargaining support when a coalition inside
the other side’s domestic society supports their objectives (reverberation). For instance,
if their desire is to open another country’s market to cheaper commodities produced in
their country, they may find support from consumers groups and businesses operating in
the imports industry of this specific target country. Opposite to Evans’ (1993) view, I
don’t find it necessary to unfold an explicit attempt by statesmen to sway the opinion of
a targeted group in another country, in order to assess the importance of reverberation.
From my standpoint, the fact that such a strategic alliance exists is enough to influence

the nature of the negotiation process.

- Operationalization

In order to appraise the effect of this dimension I propose to look at the
relationship between the government and the business community, and between the
government and trade unions. When available, I plan also to look at surveys in order
to appraise the support of public opinion. The transnational linkages between statesmen

and foreign groups are also assessed.

Government/Domestic Interaction i) Support of Business Community
ii) Support of Trade Unions
iii) Support of Consumers Groups

iv) Role of Public Opinion

Type of Issues Negotiated i) Scope of the Negotiation

ii) Domestic Distributive Effects

Reverberation and Targeting i) Existence of "Natural" Alliances with Foreign
Groups
ii) Explicit Attempts to Influence Foreign Groups

by Statesmen (and vice versa)

B) Process Dimension (See Figure 3 in chapter 1)
I have so far covered and proposed some justifications for my choice of the

contextual dimensions of international trade negotiations. The following section

65

concentrates on the justification of the procedural elements of bargaining. The three
main components that I discuss are international breakdown costs, domestic cost of delay

(or discount parameter), and information.

0 Breakdown Costs

P6: In a bilateral international trade and investment negotiation, the smaller
international breakdown costs for negotiators, the larger their bargaining leverage.

Economists interested in negotiations focus their attention on the process of
bargaining. They have more particularly studied the effect of the negotiators’ costs
associated with a stalemate in the negotiations. In this study I consider these costs in
terms of the possibility that the negotiations will reach a stalemate in the near future, that
is, the next round of negotiation. Negotiators that fear an impasse in the negotiations,
are more likely to accept a suboptimal agreement, especially if they consider that a "non-
agreement" is worse than a suboptimal agreement (that is an agreement that is not
entirely satisfactory to them). It is generally true that a settlement of a dispute or a
successful trade negotiation is better for both sides to an impasse in the negotiations,
otherwise there would be no need for a negotiation.

However, two points should be made clear. First, the consequence of a standoff
does not have to be the same for each group of negotiators. Negotiators may have
different alternatives, or a different "security point. " An actor may be able to achieve

his preferred outcomes in any particular issue area if he can develop alternative

66

relationships in which that outcome is more easily available. Therefore an actor may be
able to achieve much of his objectives by not negotiating, or by stalling the negotiations.
This action is related to the point at which the negotiator would prefer stalemate over
negotiation (security point). Therefore, the side that has more alternatives as its security
point is reached, has an advantage over the opposite side. In other words, the
availability of alternatives may thus increase an actor’s bargaining power by decreasing
its reliance on the other actor, and, in turn, by decreasing its costs associated with a dead
end in the negotiations. Second, an actor may enter into a negotiation in order to avoid
a tacit threat of an aggravation in the relations between two countries. The possibility
of a trade war if no agreement is reached can be seen as a tacit threat. In such case, the
negotiators may come to believe that the alternative is not between an agreement and the
status quo, but rather between an agreement and an outside option that appears worse
than the status quo. In some cases, tacit bargaining takes place whenever a state attempts
to influence the policy choices (or bargaining options) of another state through actions
exogenous to the negotiations, rather than by relying on the formal negotiating process
(Downs and Rock, 19903). For instance, the though American unilateral economic
stance with other international partners, prior to and during its negotiation with Canada,
constituted an implicit threat directed towards the Canadian negotiators. The threat was
real given that it would impose some costs on both parties, even though Canadians would
have had to bear most of the costs of a deterioration in the trade relationship between the

two countries.

67

- Operationalization

An important development in the Stahl-Rubinstein sequential bargaining game is
the inclusion of the possibility of a breakdown in the negotiations or an outside option
(Crawford, 1987; Fudenberg, Levine, and Tirole, 1987). This modification implies that
at the end of each offer there is a chance move that ends the game with a probability of
q 5 (0,1)”. Assuming that the players do not care about when an agreement is reached,
the pressure on the players to reach an agreement is not their impatience but the risk that
the negotiations will break down. I will leave for future work the discussion of the effect
on the outcome of the inclusion of a risk of breakdown in the model”. What is of
interest at this point is the different assessment of the players regarding the probability
of an impasse. If we assume that in a game of complete information the most efficient
outcome and unique equilibrium is (x',y") (where x' and y’ are the respective shares for
P1 and P2), Osborne and Rubinstein (1994: 129-30) demonstrate that, when including the
probability of an impasse, the utility derived from this efficient agreement becomes (l-q)
#1 (x') and (l-q) M (y') respectively. Hence, this shows that higher costs associated with
a stalemate (that is higher values of q) accounts for smaller share of the "pie".

Given this parameter, it is in the advantage of a negotiator to act as if he was just

about to walk away from the table of negotiation. This strategy, when credible and

 

52 There could be different reasons why at the end of a period the negotiation could
terminate. The main reason would be that the player whose turn it is to make an offer
may consider walking away from the negotiation table.

53 A good demonstration can be found in Osborne and Rubinstein (1990: 54-63, 71-
76).

68

successful, put the onus on the other side, and is likely to lead to more concessions“.
The utilization by Washington of Section (Super) 301 is a clear demonstration of this
phenomenon. When US trade representative perceive that some actions, constituting an
impediment to US trade and investments abroad, have been taken by international actors,
they may decide that a case is "worthy" of being pursued. However, before using,
without consulting the target, some reciprocity or retaliation, negotiations begin. It is
in interest of both the US and a target to resolve the issue. Nevertheless, both sides
(especially the United States) must act "tough", threatening to walk away from the
negotiation table at any moment. This bargaining tactic is at the core of any type of
negotiation encounter: from union representatives bargaining for better' wages with
employers to a six year old kid bargaining with is dad to have the permission to see "The
Lion King" for the hundredth time; from a prospective buyer trying to get some extra
options from a car salesman to a group of terrorist negotiating with a government the
conditions of their surrender. I will indicate below, in the specific case of international

trade and investment negotiations, what affects the credibility of such threats.

0 Domestic Costs of Delay (or discounting)

P7: In a bilateral international trade and investment negotiation, the higher the
domestic cost of delay for negotiators (or lower their discount rate), the less
bargaining leverage they have.

 

5“ This means, once again, that it is in the negotiators interest to reach an agreement
rather than reaching a stalemate. If the converse was true, not bargaining encounter
would occur.

69

Negotiator who perceive that they suffer high domestic costs for delaying the
international negotiations are in a position of weakness. The higher those domestic costs
(at each round) the more impatient they become and they are more likely to accept a
suboptimal agreement. Alternatively, the negotiators’ patience can be modeled in terms
of a discount rare. The discount rate represents how future rewards derived from an
agreement are valued relative to the valuation of the present reward. The higher the
discount rate, more important is the future. In other words, the higher the discount rate
for an actor, more patient he is during the negotiation.

Figure 10 illustrates the situation. Consequently, negotiators that have a higher
discount rate than opposing negotiators, due to their patience, have more bargaining
leverage that the latter. If both set of negotiators has the same discount rate, the Stéhl-
Rubinstein solution is identical to the Nash solution, that is, it is the outcome that
maximizes the product of the differences between the utility each set of bargainers
assigns to the agreement reached and the utility they assign to nonagreement”. An
advantage of the Stirhl-Rubinstein approach over the Nash solution is that the latter can

not take into account a situation when bargainers have different discount rates.

- Operationalization
The domestic costs of delay are at the heart of the Stahl-Rubinstein model
presented below: " The key to bargaining power here, in Rubinstein’s model, and in other

variations given comes from the ability to put the onus of waiting entirely on the other

 

‘5 A discussion and comparison of the StAhl-Rubinstein’s and Nash’s solutions can
be found in Binmore, Rubinstein, and Wolinski (1985).

70
party " (Kreps, 19902565). By comparing Figure 10, where both players have the same

discount rate, to Figure 11, where Player 1 has a smaller discount rate than player 2, it
can be seen that in the second case Player 1 receives a smaller share of the "pie" than
in the first case. I shall give a "real world" example of asymmetric costs of delaying in
the section entitled "Political Necessity and Domestic Costs for Delaying. " At this point,
I present an hypothetical situation to underscore how a Stirhl-Rubinstein sequential
bargaining game can be applied to international trade negotiations.

A Stahl-Rubinstein game can be applied to international trade negotiations between
two nation-states in the following way: One of the nation-states, which we can label
"Player 1" is considered "weak", while the second, labeled "Player 2" is considered
"strong". The strength of the players is established according to their discount rates“.
A player with a small discount rate has large domestic costs associated with delays in the
negotiations. This means that this player highly discounts future agreements, and prefers
to settle rapidly. A player with a large discount rate, values a future agreement as
almost the same way he values an immediate agreement. Hence, Player 1 is "weak"
because he has a small discount rate (or large domestic costs associated with negotiations
delays), while Player 2 is " strong " because she has a large discount rate (or low domestic
costs associated with negotiations delays). Accordingly, Player 1 is at a relative
disadvantage in any international bargaining situation. The size of the discount factor in
international trade negotiations is derived from political or economic domestic factors,

such as forthcoming elections or difficult economic times. The following demonstration

~

56 Let’s imagine, for instance, that a player values an agreement at t=0 as "1" and
his discount rate is .5. A future agreement (t=1), modeled as the next round, is worth
51 to him, that is, a value of .5.

71

assumes perfect and complete information: players know their position in the game, the

rules of the game, their preferences and their opponent’s preferences.

* Players

Player 1 (who is considered weak) and Player 2 (who is considered strong)

* Information

Perfect and complete.

* Actions and Events
- Game begins by an offer by Player 2.

- Game ends in agreement when a country accepts the other country’s offer.

* Discounting
An agreement at time t is worth 6, and 62 as much to Player 1 and Player 2
respectively at time t+1 (Note: 6, is small and 62 is large, i.e., the future is highly

discounted for Player 1, it needs a "quick" agreement).

* Game sequence
The heart of the Stahl-Rubinstein game can be depicted in a three rounds
sequence. The Bargaining game when Player 2 moves first and knows that Player 1 is

weak is summarized in the following table.

72
TABLE 1

Stahl-Rubinstein Three Rounds Sequential Game

 

 

 

 

Offer by Time Player 2’s Share Player 1’s Share
Player 2 t x 1-x

Player 1 t-1 62x 1-62x

Player 2 t-2 1-61(1-62x) 61(1-62x)

 

 

 

 

 

 

By backwards induction, we assume that the Player 2 offers (x, I-x) at time t in some
subgame and Player 1 accepts. Then Player 1 could have already made an offer (62x,
I-ézx) that the Player 2 had accepted at H. Likewise, the Player 2 could have already
offered and Player 1 accepted [(1-6,(1-62x), 5,(I-62X)] at t-2. Setting the bargaining
shares of the Player 2 equal at time t and t-2, we can solve for the Stéhl-Rubinstein

equilibrium:

solving for x when x = 1-61(1—62x),

we obtain r257 = (1-60/(1-5162)

Consequently, Player 2 offers, in a subgame perfect equilibrium (r2, I-rz) which is

 

57 Under perfect information, (r2. l-rz) represents the equilibrium offer. That is, the
minimum offer (reservation price) acceptable to the Player 1 when we introduce costly
delays. In Figure 10, X0 corresponds to the share of the "pie" for Player 2, when an
equilibrium offer of (r2, l-rz) is made by player 2. Because in Figure 10 the two
discounts rates are equal at the equilibrium Player 2’s becomes r2: 1-¢S+62 (because
61:62). In Figure 11, the share are as indicated in Table 1.

73

accepted by Player 1. This offer corresponds to the respective shares (payoffs) for the
Player 2 and Player 1. It can be seen, by comparing Figure 10 and 11 that Player 1 gets

a larger share of the "pie" when his discount rate is the same size as Player 2.

0 Information

P8: In a bilateral international trade and investment negotiation, the more information
negotiators have about domestic costs of the other party (or its "strength"), the
more bargaining leverage they have.

This aspect of the process of bargaining refers to the ability of a negotiator to
perceive the true preferences of the other negotiator. In this study a negotiator can be
perceived as being "weak" or "strong. " A strong negotiator, that is one that will only
accept an agreement that satisfies most of his demands, has more bargaining leverage.
Conversely, a weak negotiator, that is a negotiator that is willing to accept a suboptimal
outcome, is in a disadvantageous bargaining position. A negotiator that can hide his true
nature has more ability to reach his objectives, even if he is "weak. " The "strength" of
a negotiator is determined by the contextual dimensions introduced above. Therefore,
a negotiator who does not suffer from an important trade reliance vis-a-vis the other side,
who has a wide institutional opposition, who is not in a dire need of an agreement for
an electoral or economic purpose, and who can count on a large support from the
domestic polity, is considered to be "strong". The nature of the information in the
bargaining process is said to be complete when both sides know the true nature of the

other side; asymmetric when only one side knows the true nature of the other side; and

74

incomplete when neither side knows the true nature of the other side. Thus, when a
bargaining model assumes complete information on the part of two bargainers, it implies
that they should agree to the predicted outcome immediately. However, complete
information is not a realistic assumption and it can only be used as a first look at a
bargaining situation. More realistically, a bargaining situation should be modeled in
terms of incomplete information, but in some circumstances it can be modeled as an
asymmetric situation when one player has much more information about the true nature

of the other player than vice versa.

- Operationalization

Figure 12 shows the first two periods of a sequential bargaining game with one-
sided incomplete information. In this game, Player 2 is informed about all aspects of the
game, while the other is unsure of the preferences of his opponent. We have seen
previously that when both players are completely informed about their opponents’
preferences, it is realistic to infer that an agreement will be reached immediately. This
is not the case when information is incomplete. Indeed, the main reason to study a
bargaining game with incompletely informed players is to explain delays in reaching an
agreement.

In a game of alternating offers with incomplete information each move is seen as
a message used to communicate information to the other player. This aspect of
negotiation is essential according to Osborne and Rubinstein (1990: 91): "Each player
may try to deduce from his opponent’s moves the private information that the opponent

possesses; at the same time, he may try to make his opponent believe that he is in a

75

better bargaining position than he really is. "

The basic model of alternating offer under incomplete information is closely
related to that of complete information (see description of Stéhl-Rubinstein game above).
Once again, the players propose agreements at time T = {0,1,. . .} as indicated by Figure
12. If an agreement is accepted at time t, then the outcome is (x,t). The players have
time preferences with a constant cost of delay. Specifically, Player i’s preferences over
X x T are represented by the utility function xi - cit where ci is Player i’s bargaining
cost.

To represent the effect of one—sided incomplete information, we assume that
Player 1 is uncertain of Player 2’s bargaining cost. c2, the bargaining cost of Player 2
can take two values cL and c”, where 0 < cL < cl < c“. With probability 1r", Player
2’s bargaining cost is CH, and with probability 1 - 1rH it is CL, where 0 < 1r” < 1.
Player 2 knows his own bargaining cost, as well as that of Player 1.

The assumption that cL < cl < cH means that Player 1 is in a weak bargaining
position when matched with an opponent with a low bargaining cost (CL) and in a strong
position when facing an adversary with a high bargaining cost (cu)58. Thus when Player
2 has the benefits of asymmetrical knowledge regarding Player 1’s preferences, that is,
when "Player 1 is unsure of Player 2’s type, Player 2 has every incentive to convince
Player 1 that his bargaining cost is CL" (Osborne and Rubinstein, 1990: 93).

Figure 12 represents this situation by introducing two players in the role of Player

 

5" Bear in mind the effect of fixed domestic costs in a game played under perfect
information: When it is common knowledge that Player 2 has low bargaining costs,
Player 1 obtains a small positive payoff only because she has the advantage of making
the first offer. If it is common knowledge that Player 2 has high bargaining costs, then
Player 1 obtains all the pie (Osborne and Rubinstein, 1990: 49).

76

259. One of them (2L), whom I call "strong", has bargaining costs of cL. The other
(2"), whom I call "weak", has a bargaining costs of C“. At the beginning of the game
the types of Player 2 are selected by nature with probability 1n, for 2H and probability 1 -
«H for 2L. The fact that Player 1 is not informed of the selection of Player 2’s type is
represented by an information set a t = 0. The dotted line connects the first two nodes
at which Player 1 has to make a choice. The branches labeled Xo represent a typical
offer of Player 1 in period 1 and the branches labeled Xl represent an offer by Player 2
in period 1. As in the game with complete information, an offer can be accepted,
yielding to an outcome (X”, N), or it can be rejected. A rejected offer leads to a
subsequent counteroffer.
The particularity of this game played under asymmetric information is that Player
1 must act according to his beliefs about Player 2’s type and previous moves made by
Player 2, given the history of then game. As the game goes on, Player 1 can update his
beliefs, but once he has made his mind regarding Player 2’s type, his beliefs remain
constant regardless of Player 2’s actions.
Using "trembling hands" perfection to resolve this type of game, Bill Clark and
I (1995) have provided a formal demonstration indicating that when Player 1 has a small
chance to accept a tough offer from a weak Player 2, it pays for Player 2 to bluff, even
if Player 1’s prior beliefs suggest that he is dealing with a weak Player 1. Consequently,
trembling hand perfection suggests a pooling equilibrium whether or not 1n. is large or

low: even when Player 2 has large bargaining costs (2“), he will behave as if he had low

 

59 For formal proofs see Osborne and Rubinstein (1990: Chapter 5) and an
application to trade negotiations seen Duchesne and Clark (1995).

77

bargaining costs (2,). Therefore, asymmetry of information endows Player 2 with
increased bargaining leverage, whether he is strong or weak if there is small chance that

Player 1 believes that he is strong.

C) Relationship Between Context and Process (See Figure 4 in chapter 1)

Numerous studies addressed the structural nature of negotiation and the process
of bargaining. This research only starts to scratch the surface in an attempt to unveil the
problems involved in the study of international trade negotiation. Its value added
contribution, however, comes from an effort to systematically bridge the gap between the
structure of negotiation and the process of bargaining. This relationship is explored in

the next pages.

0 Interdependence and Breakdown Costs

P9: In a bilateral international trade and investment negotiation, negotiators who
represent a country that is highly dependent on the other negotiating country for
its supply of international goods have a higher international cost for a breakdown
in the negotiations.

Negotiators who represent a nation-state in an international economic negotiation
are likely to have higher international costs associated with a negotiation impasse if their
nation—state is highly reliant on its trading partner for its exports and investments. Their
problem is accentuated if their country is outwardly oriented, that is a large share of the

country’s GNP comes from trade. If other negotiators represent a country that does not

78

depend as much on another country for its external exchanges, it can be said that these
negotiators have more bargaining leverage due to the fact that they will have less
international costs associated with a standoff in the negotiations. These negotiators are
in a position of strength because their country is not hit as hard as the previous country
if the negotiations break down or if a trade war is initiated. These negotiators can
therefore bargain from strength and make implicit threats of trade retaliations if their
demands are not met by the negotiators representing the more dependent state.

Let’s take a simple example for clarification. Two farmers living on a small
isolated island, Sven and Ollie, produce only two products, peanuts and pineapples.
Sven can produce peanuts at a very efficient price, but it costs more for him to cultivate
pineapples than it would cost him to buy pineapples from Ollie. Ollie can produce
pineapples at a very cheap rate, but can get peanuts only from Sven, who has the
monopoly on the production of this commodity. In terms of comparative advantage, it
is preferable for both farmers to negotiate a trade agreement for the exchange of the two
commodities. However, Sven has more bargaining power because in the eventuality that
no agreement is reached he can still produce his own pineapples or find another market
for his pineapples (Farmer Olaf also cultivates pineapples). Ollie does not have the same
luxury. He cannot find alternative markets for his peanuts. Therefore, Sven has more
"alternatives" or a higher security point than Ollie, and Ollie is facing higher
international costs than Sven regarding a breakdown in the negotiations. This is an
extreme example, but I hope that it conveys adequately the general logic sustaining the

concept of international costs for breakdown and interdependence of trade.

79

O Bilateral Trade Structure and International Costs for Breakdown

P10A: In a bilateral international trade and investment negotiation, a high
complementarity of trade leads to high international breakdown costs.

P10B: In a bilateral international trade and investment negotiation with a high
complementarity of trade, an agreement is likely to occur.

The nature of the goods negotiated also affect the process of an international trade
and investment negotiation. A negotiator who asks the leaders of another country to limit
some of their exports to his country or who requires a market opening for his country’s
products and investment is in a position of weakness. It is even more the case when the
targeted products are produced by both countries (non complementarity), as in the case
of the constant U.S. pressure on Japan to limit its export of some manufactured products
into the United States. Such issues of "unfair trade" are often harder to deal with.
When the nature of the goods traded are complementary (e.g. natural resources in
exchange for manufactured products) both sides may be more dependent on the other side
in terms of their trade and are more willing to reach an agreement. Therefore, the more
complementary the traded goods are, the higher are the international costs identified with
a breakdown in the negotiations. As in the example used above, if the farmers were both
producing peanuts and pineapples at different costs they would still benefit from a trade
agreement, but they would rather go their own way and break up the negotiations instead

of accepting an unfair deal.

80

0 Institutional Constraints and Information

P11: In a bilateral international trade and investment negotiation, the more domestic
institutional constraint negotiators are faced with, the less information about their
type they convey to other negotiators.

Negotiator benefits from good bargaining leverage when the other side is unsure
about the true nature of their domestic actors who must ratify an agreement.
Negotiators’ type depends on the level of domestic institutional constraint they are faced
with. Institutional domestic constraint is not only different depending on the level of
democratization, such that a study of the dichotomy democracy/autocracy may not be
very revealing. For instance, any international actors negotiating with American
representatives are never entirely sure about the preferences of the latter since
international trade treaties must be ratified by Congress. This information matter puts
American negotiators in a position of strength. It is indeed a well-known fact that policy
disagreements between branches of government are much more common in presidential
systems than parliamentary systems (Stepan and Skatch, 1993) and thus the possibility
for “involuntary defection, ” (Putnam 1988; Iida 1992a, 1992b) to occur at the ratification
stage is potentially much greater in the former than the latter.60 While not absolute, the
power of a cabinet to navigate a proposal through a majority held parliament is
considerable. Thus, in the CUSFTA case for instance, it is not unreasonable to make

the assumption that there is more uncertainty regarding a potential disagreement between

 

6° Recall that Putnam defines involuntary defection as “the behaviors of an agent
who is unable to deliver on a promise because of failed ratification”(1988:438).

81

US. negotiators and the Congress than Canadian negotiators and Parliament.
Therefore, in this case, in order to examine the effects of the potential for
"involuntary defection" on the content of the agreement, one would examine a finite
horizon sequential bargaining model in which the US congress exchanges offers with the
Canadian negotiators. In this game, the American negotiators are well-informed about
Canada’s preferences. Conversely, Canadian negotiators do not have precise information
about Congress’ preferences. They know that Congress prefers a " good agreement" (one
that allows the US. to reap a large share of the benefits of agreement) to no agreement,
but they do not know whether Congress prefers a "bad agreement" (one that does not

allow the US. to reap a large share of the benefits from agreement) to no agreement."1

9"

Thus, the Canadian negotiators are unsure of Congress type".

However, official political Opposition or domestic division of power is not the
only institutional source of uncertainty. Often, in a parliamentary system (or even in
more autocratic regimes), the opposition may come from inside the party holding power.
For instance, in Japan there exists an important intra-party division regarding issues of
foreign trade, while in Canada there is a strong party discipline in Parliament and there
is little doubt about where the position of the legislative position on foreign trade lies in
comparison to the position of the executive. The possibility of involuntary defection is

even greater when dealing with a group of sovereign entities such as the European

Union. The bottom line is, as suggested by Putnam’s (1988) two-level of analysis

 

6‘ Alternatively, it can be said that the Canadians are uncertain about Congress’
valuation. A "weak" Congress attaches more value to an agreement than a "strong"
Congress. Thus making a stalemate in the negotiations less appealing. On the concept
of valuation see, among others, Fudenberg and Tirole (1983).

82

metaphor, that the more diffuse is the domestic ratification process, the more bargaining
power that country’s negotiators are likely to have in their international dealings. It is

even more important when the nature of the information is asymmetrical.

0 Political Necessity and Domestic Costs of Delay

P12: In a bilateral international trade and investment negotiation, negotiators who have
great political need for an agreement face important domestic costs for delays in
the negotiations.

A government that counts on a trade agreement to foster its electoral position or
to ratchet up its perceived staggering economy is in a weak bargaining position. This
government faces a certain domestic cost for any delays in the negotiations. The more
salient is the negotiation on the electoral platform, the higher the domestic costs for the
government. In such circumstances, the negotiators representing the government may
be willing to accept an agreement that only partially fulfill its objectives in order to be
able to use the agreement as an electoral argument. Alternatively, it can be modeled in
terms of the negotiators’ patience. When facing high political pressure for an agreement,
the negotiators value the present much higher than the future, that is they highly discount
the future.

Let’s take for instance the circumstances under which the Canada—United States
Free Trade Agreement was negotiated. In Canada, the support for the CUSFTA " seemed
soft, but published polls indicated that it was more popular than [Brian Mulroney’s

party]. Emphasis on the issue thus looked like a promising lever for ratcheting (sic) that

83

party’s back up" (Johnston et al., 1992: 19). Of course, because many Canadians
believed that the United States would give up little while Canada would barter away its
sovereignty, supporters of the agreement were vulnerable to accusations of having sold
out the national interest. This was an argument familiar from earlier trade initiatives.
But the Conservatives might still reasonably have calculated that electorally the CUSFTA
would save, not sink, them. However low it might seem, support for free trade was at
least stronger than the support for the Conservatives. Furthermore, the issue was closely
linked to Mulroney’s neoconservative agenda. His party’s agenda was premised on the
idea that Canada must privatize, deregulate, enter trade liberalization agreements,
institute international as well as government-industry advisory committees based on the
US. model, and spend millions on trade missions. As a result, the largely private
decisions of the market was to automatically, and somewhat miraculously, produce
economic growth, employment, and prosperity for Canada (Wilkinson, 1993: 36). The
idea was not so much different than "Reaganomics" and "trickle down economics", but
Canada did not have the independent power of the US. Congress to strike a balance
between international liberal trade and nationalist protection measures to protect and
assure the level of competitiveness of its national industry. The polls showed that in a
"free trade versus protection" issue, the Canadian population came out in support of free
trade in a proportion of approximately 40 percent of the electorate (Johnston et al., 1992:
145-46). This was substantially larger than the Conservative party obtained in published
polls from late 1985 to the spring of 1988. Consequently, the free trade discussions were
an integral part of the Conservatives’ electoral platform, and it was of foremost

importance that the party could argue that important progress had been made in the

84

negotiations. In contrast, in the United States, the CUSFTA was seen as an important
issue, but the electoral survival of the leaders of the executive branch did not revolve
around the progress of the free trade negotiations. Consequently, the Canadian
government was more eager than Washington to reach an agreement, and this, in turn,

further reduced its bargaining leverage.

O Societal Support and Domestic Costs of Delay

P13: In a bilateral international trade and investment negotiation, the greater
negotiators’ societal support is, the less domestic costs for delays in the
negotiations they are faced with.

The societal support component of the model is the flip side of the previous
dimension. It relates to the willingness of the domestic society to incur short term costs
for a delay in the negotiations in exchange for long term benefits. In other words,
negotiators can count on domestic support for a delay in the negotiations until they can
meet a great majority of their objectives. Consequently, a high level of societal support
diminishes the level of domestic costs that the government is facing.

Again, let’s take an example from the CUSFTA to demonstrate the effect of the
societal dimension. It is common knowledge that American companies "owned over 40
percent of Canada’s manufacturing output, and engulfed Canadian culture via its
television programs, movies, videos, and magazines" (Wilkinson, 1993: 34). There is
thus no element of surprise in the fact that the transnational corporations (TNC)

community in Canada, both foreign and domestically owned, saw, and still see, its long-

85

run well-being associated with the ups and downs of the American economy. They have
more interest in supporting the economic vision of Washington rather than Ottawa’s.
According to Wilkinson, " it is thus worth noting that firms operating in Canada, even
when Canadian-owned and/or -controlled, have (unlike their counterparts in the United
States, The EC, and Japan, where loyalties are strong), had little, if any, commitment
to building the Canadian economy " (1993: 36). In fact, the point is made by a myriad
of Canadian analysts that the CEO’s of these firms, as any rational actors, had as a prime
concern not the betterment of Canada, but their own continuity, growth, and profits
(Grant, 1965; Kierans and Stewart, 1988; Shortell, 1991). Therefore, given that these
companies see their fates associated with the state of the American economy, it would
rationally be optimal for them to support an agreement that would be in line with
American interests, rather than Canadian objectives.

The most important partners of the Canadian government was the business
community, more specifically the numerous MNCs operating in Canada. As observed
above, the loyalty to building the Canadian economy of the CEO’s of these firms, even
if Canadian-owned, is questionable. In contrast, in the United States an elaborate
structure of business-government cooperation has been developed under GATT
negotiations. South of the border, the loyalty of "Corporate America", is not as
questionable. What is typically thought to be good for General Motors is good for
America! As Sylvia Ostry, principal Canadian negotiator of the NAFTA, has
commented, "to the outside observer [the business-government cooperation in the United

States] presents an impressive image of ’U.S.’ Inc." (1990: 25).

86

111- Nature of Outcomes in International Economic Negotiations

I have so far discussed all the independent elements of analysis that I believe
influence the partition of the surplus derived from an international trade and investment
negotiation. A very critical question remains. How do we measure the outcome of a
trade negotiation? That is, how can we find refutable and replicable conditions that
indicate which party reaped the most of the benefits from the agreement? This is no
small task. If we were only interested in the success or not of negotiations, the
assignment would be much easier. We could affirm that negotiations were successful or
unsuccessful, depending on whether a single agreed upon value has been adopted as the
result of the bargaining: "[A]n agreement is acceptable as prima facie evidence of
"success" since it can be assumed that no party would agree to a value that he viewed
as being worse than the value of nonagreement" (Zartman, 1976: 7-8).

The problem with explaining the distributive effects of negotiation is an age old
one. Unfortunately, very little progress has been made in this direction. The
predicament resides in the fact that, too often, researchers have sought their answers in
terms of single case studies "and [have been] thrown back to situational and historical
descriptions of essentially unique events " (Zartman, 1976: 18). When the attention has
turned to more abstract analysis of universal situations either in symbolic or mathematical
terms, the interest has swerved towards theoretical underpinnings of negotiations, rather
than towards a single negotiation. This leads us back to the fundamental questions that
we were asking at the onset of chapter one. What are the best variables for analyzing
the process of bargaining? How can theoretical variables be translated into practical

terms? The modus operandi of this analysis proposes that an answer must be sought in

87

the interplay of the negotiation structure and bargaining process. As such, a series of
dimensions of negotiation and bargaining variables have been identified in the previous
sections. It is not my contention that this research design is the universal answer that
students of negotiations have sought for eons. My goal is to provide a logical structure
for the interrelation between the structure of negotiation and process of bargaining. An
adequate measure of the distributive effects of an outcome62 (and de facto a theory of
negotiation) will only arise when explanations will be cast in terms of fully
operationalizable variables that can be applied to real cases. Seen in this light, this
model represents an important step in a research programme, but the project can only
come to fruition when its research design is extended to a large number of cases with
fully operationalizable and refutable variables.

With these caveats in mind, I can introduce how I intend to "measure" the
distributive effects of international trade and investment bargaining. The most logical
answer is to assume that the party that has more power will receive more of the benefits
derived from the agreement. However, we have seen above that the tautological nature
of such approach does not reveal the true nature of negotiation. The problem remains
one of defining power before-—and hence, independently--of outcome. That’s what most
of the previous pages have been about.

I propose to measure the outcome of an international trade negotiation as the

 

"2 The distributive effects of negotiation that I intend to measure are those that
accrue to the nation-states involved in the bargaining process. It seeks to answer the
question "Which country reaps most of the benefits that are produced by the agreement?"
I do not intend to look at the distributive effects of the agreement in relation to domestic
groups. This aspect of the research is only considered in terms of determining which
type of domestic support will the governmental negotiators get during negotiations.

88

intersection of policy options proposed by the negotiators. By policy option I mean a set
of prescriptions suggested by one of the parties regarding the future economic
relationship between the parties. By focusing on the monetary aspects of the intersection
of different suggested policy options, we can assign a dollar value. Such an approach
is appealing for its simplicity, but " it is dependable only for the most purely economic
situations" (Cross, 1969: 11). Nevertheless, even though it may not be extended to any
instances of negotiation“, it appears that such a measure provides an adequate template
for international trade and investment negotiation outcomes.

In most cases, such as when the United States threatens to use Section 301
(Bayard and Elliot, 1994), dollar values can easily be assigned to the intersection of
policy options sought by the respective negotiators. Even when the negotiation is a much
more complicated process, such as the CUSFTA, this approach has a clear advantage:
multiple issues can be brought back into a single issue dimension through their transfer
into a single utility dimension (dollar values). One obvious drawback of this approach
is that when applied to complex cases involving multiple issues, it becomes nearly
impossible to come close to the real dollar figures associated with the intersection of
policy outcomes. However, we must keep in mind that we do not need a knife-edged
measure of the outcome of an international trade negotiation, but we can find an almost
as satisfying answer by discovering a measure of the outcome that indicates a relative
partition of the benefits derived from the outcome. Furthermore, some of the more

complex cases have led to a great number of assessments, and it is thus relatively

 

63 It has also the drawback of not considering the personal motivations of the
negotiators.

89

uncomplicated to assess dollar values to the intersection of different policy options.
Therefore, for each case, the still-imperfect solution would be to identify the respective
policy options that were pr0posed by the parties and then evaluate the dollar value of the
intersection of the policies chosen as an outcome.

***

This research represents a modest attempt at moving beyond the concept of
structural power in order to explain the outcome of international commercial negotiations.
It borrows from studies in political science focusing on the context of those negotiations,
and from studies in economics that concentrate more on the process of negotiation. It
represents an attempt to bridge the gap between the two approaches. Its interpretation
is limited if it is not tested "statistically. " With this in mind, I now turn my attention to

the empirical aspects of my dissertation.

9O

coszBE Seton. one: .082 @252“..me Eoficfisméfim Hm 939“.

so 32 :Newé

/\>

91

co_mcoE_o .mcoszoE no 939“.

 

120 .6 xx. cozmz EEO
EEEoEoEEoO m< mtoaxm 55> moot. Co
.82 bszoEoEEoo o>=ooqmom 9me 9:03on

P _ r _

 

 

93025 moot.
_99w__m deco—6:36.05:

F x .

 

3:02.095“:—

92

co_m:oE_n_ _mco:3:wc_ K 930E

 

 

 

onma _mo:__0n_ co_m_>_n_ coEmNthn. oocmEEoooi 8:933:00
EEO Econ. SE. 5:8 8:94 0.396
_ _ _ m _
co:_wooo_O _mo:__0n_ co:mEoEmSu_

 

_

 

L
822262

93

Political Necessity

 

 

l i

Electoral Economic
Performance

Figure 8: Political Dimension

94

co_wc0E_Q _m:0_0ow Mm 059“.

 

 

 

 

 

 

 

9.85
9:39:05 5:30 00:90 0:25 \EcsEEoo
9:098. 5:98.05"; 0:002:00 oaoom 0__n:n_ 00:50:00 000:. wonmzm
80:00.0:
08:00.0: 00:02 .6 0:00Eoo
_0co:mEoE_ 08:. ::0EEo>oG

 

 

 

0>_omom _0:0_0om

#2

 

95

 

 

 

Figure 10: Effect of Discounting

 

96

 

 

 

Figure 11: Effect of Discounting With Different Discount Parameters

97

rue

8:02:25 0_:0EE>0< 55> Enos. mcEfigmm ”N: 059d

 

CHAPTER 4

Structure of Empirical Test and Hypotheses

1- Empirical Test Strategy

The previous chapters introduced a theory and a model of international trade and
investment negotiations. The theory, especially, presents an ideal-type of the structure
of international negotiations. As such, an empirical test that would follow closely the
propositions enumerated in Chapter 3 requires an important time consuming effort,
especially in terms of data gathering activity, that is better suited for a research group.
Nevertheless, this theory of international trade negotiations could easily be applied to an
in—depth analysis of a few cases. When dealing with a great number of cases, the task
is more difficult“.

With these caveats in mind, I decided that the next best option was to look for a
data set that includes some of the variables--or, at least, some proxies--needed to test the
model and propositions of Chapter 3. Two members of the Institute for International
Economics, Thomas O. Bayard and Kimberly Ann Elliot, published a book (1994) that
evaluates the degree of success for the American negotiators when they resorted to the

Section (Super) 301 policy instrument to get foreign trade partners to liberalize their

 

6‘ Some related works applied a cost—benefit analysis to trade protection and
enhancement (Baldwin, 1985; Destler and Odell, 1987; Lavergne, 1993; Noland, 1997),
while others looked at institutional features of domestic policies (Grossman and Helpman,
1994; Magee, Brock, and Young, 1989; Rodrick, 1994). This study incorporate the two
approaches.

98

99

commerce. Their research will serve as a baseline for the empirical testing of this
dissertation. Still, we need to be cautious with the interpretation of the results. The use
of Section (Super) 301 is a very special case of international trade negotiation; an
instance where the Americans already dealt the cards by accusing a target of unfair trade
practices. As such, it is very different from circumstances when both parties enter
willingly into a negotiation, in order to reach an agreement beneficial to all in terms of
liberalization of trade. Nevertheless, this use of retaliation and reciprocity by Americans
represents a suitable test of the theory and model discussed in the two previous chapters:
First, one of the players, the United States, is the same for all cases. It is especially
interesting, given that the United States is considered to be the most powerful economic
and military power on the planet (hegemon). As such, we would expect, if we believe
in the notion of aggregate power (or fungibility of power), that the United States would
come on top in all situations of negotiations. Such is not the case. Second, Bayard and
Elliot demonstrate that the United States obtain various degrees of success with "small",
as well as with "large " economic powers. This indicates that a unique reliance on trade
interdependence for an explanation of the outcomes of international trade negotiations,
can be misleading. Consequently, we also need to look at domestic institutions to add
some value to our analysis. Third, Bayard and Elliot’s results also show that the
American negotiators obtain varying degrees of success with the same target. This,
again, show the inefficacy of a model that does not go beyond institutional constraints
variables. Thus, there exists a need to look at a case by case analysis of domestic
support from the population and political necessity for the negotiators. This last level

of analysis is the most difficult to attain when time comes to test the some hypotheses

100

representing the propositions of this dissertation. For this reason, it is probably the
weakest part of the test that I propose below. Nevertheless, if we obtain good results by
using some proxies, it augurs well for a fully developed analysis coinciding more closely
with the variables, indicated in the previous chapter, regarding the domestic support and
political necessity dimension of the theoretical model.

Before pursuing with the description of the variables included in the empirical
test, a few words on the Section (Super) 301 policy instrument are necessary 65. After
the end of World War II, US trade policy rested firmly on multilateral institutions such
as the General Agreement on Tariffs and Trade (GATT) and the International Monetary
Fund (IMF). This commitment to multilateral instruments of trade weakened at the
beginning of the 1970’s. From that point on, the United States engaged itself in a
tougher bilateral stance. The main reasons for a decline of multilateralism are "the
decline in US hegemony, the increase in international economic interdependence, and the
fact that many Americans now view the field on which the trade game is played as
unfairly tilted against US competitors" (Bayard and Elliot, 1994: 9).

In the wake of the collapse of the Bretton Woods international economic system
and the oil shock of 1973, the American Congress became more assertive. This new
optic regarding international trade was well articulated in 1973 by, then Secretary of
Treasury, George Schultz, in congressional hearings on what became the Trade Act of

1974:

 

65 For a detailed assessment of the evolution of Section (Super) 301, see Bayard and
Elliot (1994), especially Chapter 2.

101

Today, economic power is not concentrated in the United States

alone as it was thirty years ago. Great centers of wealth have

grown up in Europe and Japan... However, along with this

diffusion of power has gone a reluctance to remove restrictions

that are contrary to the principles of an open world economy... In

this changed world of economic equals we need to deal with those

restrictions, and we need new rules to assure equality of

responsibility. (US House of Representatives, Committee on Ways

and Means, 1973: 159)66
The Trade Act of 1974, under Sections 301-302, "expanded discretionary authority to
retaliate against unjustifiable and unreasonable foreign barriers" (Bayard and Elliot, 1994:
24). More specifically, it authorized action against foreign exports subsidies. The Trade
Act of 1974 was expanded by the Trade Agreement Act of 1979; in which the President
was given the authority to enforce trade agreements. It also allowed for detailed
procedures for investigation (including deadlines for action), and the use of available
settlement procedures (Idem. ). later, retaliation in the service sector was authorized by
the Trade and Tariff Act of 1984 (Idem). Finally, "Super 301" was created by the
Omnibus Trade and Competitiveness Act of 1988, which shifted the authority to retaliate
from the president to the United States Trade Representative (USTR), specific to some
presidential directives (Idem). It also made retaliation against "unjustifiable" practices
mandatory.

Hence, in order to test my theoretical and formal models, I intend to use the
information included in Bayard and Elliot (1994) work on the United States use of

Section (Super) 301. They identify 91 cases between 1974 and June 1994, from which

75 led to negotiations with known results between the United States and the targeted

 

6" Cited in Ibid, p. 13.

102

country“. Consequently, I am testing the level of success attained by American
negotiators in terms of achievement of their objectives. For each case, Bayard and
Elliot have collected the following information: period of case, type of product, actual
value of US exports to the target (specific contested product), existence of a GATT panel
to resolve the case, existence of a GATT ruling, negotiating objectives success for
American negotiators, and degree of trade liberalization resulting from Section (Super)
301. Some of this information is useful for the empirical testing that follows in the next

chapter.

11- Method of Analysis.

Given that the dependent variable contains four (ordered) categories (see section
III), I use a multinomial ordered logit analysis."8 In this test of American negotiating
success (see next section), a multinomial logit or probit analysis would be inadequate,
because it would fail to account for the ordinal nature of the dependent variable.
However, ordinary regression analysis would err in the opposite direction. An ordinary
regression would treat the difference between a level of success of 4 and 3 as the same

as a difference between a level of success of 2 and 3. The difference between these

 

67 The cases excluded are indicated in Bayard and Elliot (1994: 59, n.3). I retained
their cases # 301-83, 301-88 and 301-89, because they provide the assessment on the
information I need. Their concern is that they do not have enough information regarding
trade liberalization following the use of Section 301. However, they provide the
information on the value of the contested goods and the level of success for the American
negotiators, which is what I need for my analysis. I also treated their case # 301-85 as
two cases regarding intellectual property negotiations with India because they provide
different assessments of the American success; one on copyright and film quota, and one
on pharmaceutical.

‘58 The statistical package I used was STATA 4.0 for Windows.

103

levels of success are not necessarily the same; they depend on the estimated cut points
(it’s). These it’s are unknown parameters to be estimated with B in the following

equation:

y‘=flx+€ (1)

as usual y‘ is unobserved. What we observe i869

y=0 ify’ 30
y-1 if0<y’ s u,

y-2 ifu1<y’su2 (2)

t

2):] ifpj-15y

It is therefore clear that there is not necessarily an equal difference between cut points
(it’s), thus the superiority of an ordered logit (or probit) analysis over a multivariate

regression analysis for the testing of the propositions of this dissertation.

 

“9 In the following equations, the mean and variance of e are normalized to zero and
one. Greene (1993: 673) indicates that the model can also be estimated with a
logistically distributed disturbance. He adds that this is a "trivial modification of the
formulation appears to make virtually no difference" (Idem). When we assume normal
distribution, we do an ordered multinomial probit analysis, while when we assume
logistically distributed error terms, we do an ordered multinomial logit analysis. A
visual inspection of the data convinced me of a logistic distribution of the data. I also
ran the model with ordered logit and probit, and had slightly better results with a logit
analysis. I shall give a more detailed description of ordered multinomial logit analysis
in the next chapter.

104

III- Dependent variable.

The previous chapters discussed how we can appraise the success of negotiators
in a bargaining encounter. For the remainder of this dissertation the focus is on the
bargaining success of a group of negotiators representing Washington. Now, the
question becomes: which aspects of the structure of negotiations affects the bargaining
success of the US. trade representatives?70 This, in itself, is no small task.
Fortunately, we can count on Bayard and Elliot (1994) monumental study of Section
(Super) 301 to provide a solid foundation on which we can build. Hence, in order to
measure the United States’ level of success in their negotiations with targeted countries

after the implementation of Section (Super) 301, I use Bayard and Elliot (1994: 63)

categories71 :

Failures (coded as 0): The case was not implemented to US satisfaction,
or it was circumvented in some other way.

Nominal success (coded as 1): A case in which an agreement was reached but not
implemented to US satisfaction.

Partial success (coded as 2): A case in which only some of the US objectives
were implemented.

Success (coded as 3): A case in which the US objectives were totally or

largely implemented.

 

70 Obviously, ultimately students of international negotiations need to develop a
framework an empirical analysis that can appraise the level of success for all sides of the
bargaining table. At this time, with the data at hand, it is only possible to assess the
bargaining success of Washington.

7‘ Bayard and Elliot use a probit regression analysis by dividing the dependent
variable into two categories (success and failure). It may be because they did not get
significant results by using their four initial categories. It may also be because they did
not have the program to run an ordered multinomial logit. I shall demonstrate in the
next chapter that we obtain better results by using ordered logit than by using probit or
logit analysis.

105

To help the reader understand the meaning of each degree of success, I give here
a brief example of each category. These examples are taken from assessments by Bayard

and Elliot (1994).

A) Failure: Challenging European Communig] (EC) Export Subsidies for Barley (Bayard
and Elliot, 1994: 378-79)

In November 1975, Great Western Malting Co. filed a petition claiming that EC
subsidization of malt exports to Japan and other countries was an unfair practice. The
company complained that it lost almost all of its Japanese market due to EC export
subsidies. The USTR did not file an official complaint under the GATT dispute
settlement procedures, but brought the problem to the attention of the European
Community Commission in 1976. Then, claiming that the Tokyo Round Subsidies Code
had addressed the issue in 1977, action under Section 301 was no longer necessary.
Thus, the complaint was dropped, effective 19 June 1980. However, the EC practice
continued. In 1992, the United States Department of Agriculture (USDA) announced,
in retaliation, a new package of Export Enhancement Program (EEP) for barley malt.
Bayard and Elliot assessment (1994: 379):

The Subsidies Code did not resolve the problem of agricultural

export subsidies, and average US exports declined by $1 million

in 1977-79 compared with 1974-76, while EC exports to Japan
returned to previous levels after a temporary decline in 1977.

B) Nominal Success: A GATT IllegaJ Ban On Imports of Cigarettes By the Th_a_i
Government (Bayard and Elliot, 1994: 449-50)

In April 1989 the US Cigarette Export Association (CEA) filed a petition
complaining that Thailand maintained an effective ban on imports of cigarettes and

prohibited foreign investment in cigarette manufacture. Thailand initially rejected an

106

American effort to create a panel to discuss the issue, but acquiesced when the issue was
brought to a GATT panel in April. In September the GATT panel ruled that Thailand’s
practice breached Article XI of GATT and were not justified under Article XX
(exceptions to protect health and safety), but denied an American claim that Thailand’s
excise tax violated the national treatment requirements of Article III. Given this mixed
result before a GATT panel, the USTR proposed a retaliatory "hit list" to enforce
compliance by the Thais. Following the threat, Bangkok expressed a willingness to
negotiate and bilateral negotiations started. On November 23 1990, USTR announced
that the case was dropped after the Thai government announced that it would no longer
ban the importation of foreign cigarettes. It would appear that this would constitute a
case of success for USTR, but the further developments led Bayard and Elliot (1994:
449) to propose an assessment that only claim nominal success for the Americans:

Thailand eliminated the practices that had effectively prevented

cigarette imports, but GATT-legal tariffs and taxes kept US

exports low. By summer 1991 US cigarette brands were still only

available in duty-free shops in Thailand because the combination

of a (nondiscriminatory) 55 percent excise tax on top of a 30
percent import duty priced them out of the market.

C) Pagiafl Success: Korg’s Tariffs 3nd Quotas On Table Wine Inmorts (Bayard and
Elliot, 1994: 442-43)

In April 1988 the Wine Institute and Association of American Vintners filed a
petition complaining that Korean tariffs and quotas on table wine imports (and other
measures) impeded access to the Korean wine market. The next year the United States
and Korea reached an agreement to provide nondiscriminatory treatment for wine

imports. Bayard and Elliot indicate that a good settlement was agreed upon, but there

107

are still some problems with implementation (1994: 443):

Though the liberalization Korea promised was significant, there

have reportedly been problems with implementation [...]. The

limited trade data available do not show much of an increase in US

exports, certainly not to the levels the petitioners anticipated [...].

Nevertheless, the case is judged as a partial success.

D) Success: Taiwan’s Import Duties On Household Ampmnces (Bayard and Elliott, 1994:
383-84)

Lai Fu Trading Co., an American enterprise based in Taiwan, filed a complaint
in March 1976, claiming that recent increases in import duties on certain household
appliances (consumer electronics especially), discriminate against US. commerce in the
Taiwanese market. Taipei retorted that the measure was only temporary and was only
in effect for a short period. Taiwanese representatives added that their objective was to
deal with balance of payments pressures resulting from the global recession. At first,
USTR accepted the Taiwanese argument, but when it became apparent that the tariffs
were still in place long after the balance of payment rational had disappeared, the United
States opened bilateral consultations with ROC seeking removal of those surcharges. The
charge was dropped in December 1977, when Taiwan agreed to lower its tariffs at least
to their level prior to 1975.

Bayard and Elliot (1994: 384) assessment:

US negotiators not only achieved a reversal of the increased

tariffs, the tariffs on some items were lowered still more. Trade

data also show US exports of consumer electronics increasing
nearly 40 percent [...] after the tariffs were lowered [...].

Let’s now turn to the description of the independent variables.

108
IV - Independent Variables72

A) IntemLiongl Dimension

i- Interdependence

The general proposition here is that in "a bilateral trade and investment
negotiation, a nation-state that is less dependent on another state for its commercial
exchanges has more bargaining power than the second nation-state". For this specific
analysis, we should reread the proposition as "the lower the relative trade dependence
the United States has upon the target, the most successful it will be in achieving its
objectives. "

One way to evaluate a trade interdependence between two nations is to compare
the respective share of total exports of one country that comes from exports to the other
country. That is, for each country involved in a dispute, we need to calculate the
percentage of its external output that is accounted for by its exports to other countries
being part of the negotiations. Finally, we take the ratio of these calculated values. In
the specific case of this dissertation, the following hypothesis73 regarding direction of

trade and bargaining position is proposed:

H1 When a larger share of the United States’ exports are going to a target
country (or group of countries) than the share of a target country’s (or
group of countries’) export going to the United States, American
negotiators are in a weakened bargaining position when they use Section
(Super) 301.

 

72 See Appendix B for a glance at a description of all the variables included in the
empirical (statistical) test.

73 For a list of all hypotheses see Appendix C.

109

and to account for this variable we need to calculate the following":

USTOUT: Annual United States’ Trade Output (in billions)
TATOUT: Annual Target’s Trade Output (in billions)
USEXP: Annual United States’ Exports to the Target (in billions)

TEXP: Annual Target’s Exports to the United States (in billions)

TSHOUT: T arget’s Share of United States’ Trade Output (USEXP/USTOUT * 100)
(%)

USHOUT: United States’ Share of Target’s Output (TEXP/TATOUT *100) (%)

DEP: Ratio of TSHOUT over USHOUT

It can then be said that the trading partners are symmetrically dependent if the
ratio is roughly equal to 1. The United States is said to be more dependent upon the
target (instead of the reverse) if the ratio is greater than 1. Conversely, the United States
is said to be less dependent upon the target (instead of the reverse) if the ratio (r) is: 0
> r > 1. Consequently, the United States is assumed to have less bargaining leverage
in the first scenario than in the second scenario. In terms of the statistical analysis, I
expect the variable to post a negative sign.

However, a nation-state may be highly dependent on another state for its external
exchanges, but international trade may only be a small subset of its entire economic

strength. Therefore, it is also important to control for a nation’s export as a percentage

 

7‘ The source for USTOUT, TATOUT, USEXP, and TEXP is FMI, Directions of
Trade... , country tables. I used the DOTS World Total. A special table from the same
document indicate the European Union (or EEC) exports. When the negotiations lasted
more than a year, I calculated the yearly average. The values are indicated in billions
of current US dollars. The trade output of Taiwan is not indicated in the FMI document.
For this I used The Republic of China, 1992, Taiwan Statistical Data Book, Taipei,
Economic Planning Council, Executive Yuan, for the 1976/77, 83/84, and 86 data. For
the 1992 data, I used The Republic of China, 1994, Monthly Bulletin of Statistics of the
Republic of China, Vol. 20, No 1 (Jan), Taipei, Directorate-General of Budget,
Accounting & Statistics, Executive Yuan.

1 10
of its GNP (DEPGNP). Hence, I propose the following hypothesis in regards to trade

dependence :

H2: When a larger share of the United States’ gross national product (GNP)
is generated by its exports to a target country (or group of countries) than
the share of a target country’s (or group of countries’) gross national
product (GNP) generated by its exports to the United States, the American
negotiators are in a weakened bargaining position when they use Section
(Super) 301.

To account for this variable these additional indicators are needed:

USGNP: Annual United States’ GNP (in billion)

TGNP: Annual Target’s GNP (in billion)”

USXGNP: United States’ Share of GNP accounted for by Exports to Target
(USEXP/USGNP * 100) (%)

TXGNP: Target’s Share of GNP accounted for by Exports to the United States
(TEXP/TGNP * 100) (%)

DEPGNP: Ratio of USXGNP over TXGNP

Once again, it can then be said that the trading partners are symmetrically
dependent if the ratio is roughly equal to 1. The United States is said to be more
dependent upon the target (instead of the reverse) if the ratio is greater than 1.

Conversely, the United States is said to be less dependent upon the target (instead of the

 

75 The source for the GNP of the US and targets is taken from the World Bank’s
World Table. This document does not have the GNP in US current dollars. However,
it has the GNP in current US dollars per capita and the population. Thus I have
multiplied these two values. The values indicated are in current billions of US dollars.
No aggregate data for the European Union are indicated. I had to add the GNP of each
member of the Union in order to get an aggregate value. When the negotiations lasted
more than a year, I calculated the yearly average. GNPs for the USSR and Taiwan were
not available in the IBRD document. Therefore, I had to rely on Republic of China,
Taiwan Statistical... and Monthly Bulletin..., op. cit. , for the data on Taiwan. For the
1977/78 GNP of the Soviet Union I relied on United States Central Intelligence Agency,
1979/80, National Basic Intelligence Fact book, Washington, DC. , Central Intelligence
Agency. For its 1979 GNP, I used Jennifer Carr, 1982, The USSR: A Statistical and
Marketing Review, Warwick, U. of Warwick Statistics Service, table 2.2, p. 22.

111

reverse) if the ratio (r) is: 0 > r > 1. Consequently, the United States is assumed to
have less bargaining leverage in the first scenario than in the second scenario. In terms

of the statistical analysis, I expect that the variable will post a negative sign.

ii- Trade Structure and Threat Credibility.

The general model indicates that an agreement is more likely if the trade structure
among the two negotiating parties is complementary. This variable is less salient in the
specific model. Indeed, the specific model is concerned with the possibility of trade
reciprocity and retaliation by the United States. The real issue then is to find under
which circumstances is the United States able to commit a credible threat of retaliation.
Ideally, in terms of trade structure, a good measure would be to look at the concentration
of production of the targeted product by the threatened country and the possibility of
alternative sources for the United States. That is, if the targeted country is a large
producer of the product targeted by the US, it becomes difficult for the latter to make
a credible threat of imposing a tariff on the targeted product. Such measure, however,
may be difficult to find in the short term and it does not apply to all cases.
Consequently, this endeavor is postponed until post-doctoral research.

Another, and easier way to measure the threat credibility of retaliation of the
United States is to concentrate on the cases in which the United States can exert a direct
and unilateral retaliation. As such, these cases refer to instances where the United
States can use commensurate tariffs and/or quotas directly on the targeted product. In
other cases, such as intellectual property or technical, environmental and health

standards, direct commensurate measures of retaliation are harder to evaluate and

112
therefore the threat is less credible. Consequently, I use Bayard and Elliot’s (1994)

dummy variable (ISSUE) that is scored as a 1 if the trade barrier in dispute is a tariff

or an import or export quota and 0 otherwise. The specific hypothesis is the following:

H3: The American negotiators have more bargaining leverage in their use of
Section (Super) 301 when the issue debated is about tariffs and/or quotas
than when the issue debated is about intellectual property or technical,
environmental and health standards.

and the variable created is:

ISSUE: Issue concerns a tariff or an import or export quota (coded as 1, 0
otherwise).

The sign of this variable should be positive in the statistical analysis.

B) Domestic Dimension

i- Institutional Constraint

For this variable, I think that the best test would be to replicate Hagan’s (1993)
measure of political opposition. However, this amounts to a time-consuming enterprise
and I do not intend to pursue it at the dissertation stage of this research agenda. In the
meantime I propose to proceed as described below.

This variable is concerned with the level of politization of the issue being
negotiated. It refers to what Putnam (1988) coined " involuntary defection. " This refers
to the fact that an agreement may be signed, but not ratified. Hence, there is always the
possibility that domestic political opposition will "veto" the agreement. This domestic

political opposition is considered as a source of bargaining power in the international

113

bargaining process because negotiators may say something like "this agreement looks fine
to us, but it doesn’t stand a chance of being ratified at home. " Consequently, their
bargaining leverage is increased and we can advance the following proposition: "The
higher the institutional domestic constraint faced by a negotiator, the higher her
bargaining leverage in a bilateral international trade and investment negotiation. "

I propose to use two different indicators, one for the United States and one for
the target. For the United States I will rely on a measure used by Lohmann and
O’Halloran (1994) called DIVIDED. This indicator measures the level of division
among different branches of government. The most divided government is when the
presidency is controlled by one party and both Houses come from another party. The
government is also divided, but with a lesser degree of opposition from the legislative,
when the latter is divided. It is undivided when the Senate, the House, and the
presidency is controlled by the same party.

The following table recaps the nature of the indicator“:

 

7" When the negotiations last more than one year, I use the date when an agreement
was signed. As the breakpoint of a change in the Presidency I use January 20 following
each electoral year, which is when the new President takes power. The similar
breakpoint for Congress is January 3 of each odd year.

TABLE 2

American Institutional Division

 

President Senate Majority House Majority DIVIDED
DEM DEM DEM -1

DEM REP DEM 0

DEM DEM REP 0

DEM REP REP 1

REP DEM DEM 1

REP DEM REF 0

REP REP DEM 0

REP REP REP -1

The most divided government being represented by a 1 and the least by a -1, we
should expect a positive value for this indicator. Therefore, I propose the following

hypothesis:

H4: The more divided the United States’ government is, the higher rate of
success for the American negotiators in their use of Section (Super) 301.

The other side of the coin is the level of domestic institutional opposition faced
by the target country’s negotiators. For this purpose, my intention is to use the data
provided by the POLITY III project. I use two variables found in Maoz and Russett
(1993). The first one is monocratism (MONO), which is a five-point ordinal scale
representing institutional independence of the chief executive. The second, centralization
(CENT), is a three-point ordinal scale which distinguishes between unitary and federal

political system. These two measures are summed up over their categories to create an

115

overall scale of institutional constraint ranging from 2 (a system lacking any form of
constraint) and 8 (a highly constrained system). To represent the institutional constraint
dimension, I created a dummy variable (INSCONST), which takes the value of 0 when
the sum of CENT and MONO varies from 2 to 4, to represent a low constraint political
system. For added values ranging from 5 to 8, it is coded as an 1 to represent high
constraint political system. The cases involving the EC are coded as a 1, because they
are highly decentralized and the institutional decision-making process is highly diffused.
The variable should have a negative value; the more constrained is the target, the less

success should the American negotiators have, and the hypothesis is read as:

H5: The higher the level of institutional constraint for the targeted country, the
less bargaining leverage for the American negotiators in their use of
Section (Super) 301.

ii— Political Necessity

This variable measures "how badly " an agreement is needed by the negotiating
parties. Given that the United States is always the "demander" in the cases that I study,
I propose to focus of the political needs on the American negotiators. One possible
indication of political necessity is the use or an implicit threat of retaliation by the United
States (RETAL). Retaliation is a tool made available to the United States Trade
Representative (USTR) in cases involving the use of Section (Super) 301, among others.
The logic would indicate that it is only when American negotiators are in a dire need for
an agreement that they use such a policy instrument. Thus, revealing their "weak type".

Indeed, the use of a retaliation indicates that all else failed. A threat or use of retaliation

116

is coded as a "1" and as a "0" otherwise. Given this variable brings to the fore a certain
weakness demonstrated by the American negotiators, we must anticipate that its sign will

be negative in the statistical analysis.

H6 The use of a threat of retaliation or a direct retaliation in cases involving
the use of Section (Super) 301 is an indication of bargaining weakness on
their part.

The "economic political necessity " variable is appropriate in this case. I join
Bayard and Elliot (1994), by assuming that the greater the dollar value of the contested
trade relationship between the United States and the target, the more difficult it would
be for the American negotiators to reach an agreement, and in turn, their objectives.
Thus "small cases" are more easily resolved to the satisfaction of the US. negotiators
than "big cases. " In order to measure the " severity " of a case, Bayard and Elliot (1994:
355-69) used the actual value of US. exports (in terms of the good or service targeted)
to the target after the case was concluded. I intend to utilize a similar measure. I
created a dummy variable (DVCGOOD) for "big cases" which take the value of 1 when
the amount of contested trade is 100 million dollars or larger, and 0 for "small cases"
when the value of contested trade is less than 100 million dollars. This variable should

have a negative sign and the hypothesis is the ensuing:

H7 The higher the value of the contested goods or services leading the use of
Section (Super) 301, the less bargaining success for the American
negotiators.

1 17
Another variable, a misery index (Keech, 1995:52-53) can be used to evaluate the

economic political necessity for the American negotiators (MISERY). The misery index
is calculated by adding the seasonally adjusted unemployment rate of the United States
to its inflation rate". The assumption is that American negotiators are in a position of
weakness when the misery index is high. In this case, the US. negotiators are in a dire
need of an agreement and they may be induced to make more concessions because they
see an agreement as a panacea for bad economic performance. Therefore the hypothesis

should be read as:

H8 A higher misery rate (Inflation + Unemployment) in the United States
leads to less bargaining success for the American negotiators when they
use Section (Super) 301.

The sign of MISERY should be negative in the statistical analysis.

iii- Societal Support
This variable can only be measured adequately by an in-depth analysis of each
case. However, I will again turn to Bayard and Elliot (1994) in order to find an

adequate "proxy" that can be used to measure the effect of this dimension on the level

 

77 As a baseline to code the seasonally adjusted unemployment rate and inflation rate
I used the date an agreement was signed between the United States and the target. When
it occurred between January and April, I used the rates of the previous year. When it
occurred between May and August, I calculated the mean values of the rates for the
previous and the current year. For the last four months, I used the rates of the current
year. The source for the unemployment rate and inflation rate is U.S. Government,
1994, Economic Report of the President, Washington, DC. , US Government Printing
Office,table 340, p. 314 and p. 341 respectively.

118

of success for the American negotiators. They make the hypothesis that the greater the
amount of US. trade deficit vis-a-vis the target (which is a negative value), the greater
support it will get from the domestic society. In turn, the greater the domestic support,
the greater the bargaining leverage of the US. negotiators. We should, therefore, expect
a negative sign (Le. a positive balance of trade would lower the bargaining leverage for
the American negotiators) for this variable (TBAL78) and we should read the hypothesis

as:

H9 The greater the trade balance deficit for the United States with the targeted
country (or group of countries), the more bargaining success for the
American negotiators when they use Section (Super) 301.

A common measure of societal support in the United States, used to test various
hypotheses in the domains of domestic politics and foreign policy, is the presidential
approval rating. With data provided by Gallup, this variable (APPROV) is included in
the statistical analysis. It is, however, with suspicion that I resort to this measure. We
can speculate that a high level presidential approval would bolster presidential confidence
for "big" foreign policy decisions, such has a decision to go to war or signing an
encompassing trade agreement. These initiatives receive wide media coverage and are
consequently well-known by the public. Cases involving the targeting of a foreign
country via the Section (Super) policy instrument, does not constitute a "big" case and,

in fact, rarely receive important media coverage. With this caveat in mind, presidential

 

73 The source for this variable is International Monetary Fund, Directions of Trade. ..
(various years).

119

approval is incorporated in the statistical analysis that follows in the next chapter, and
we may hypothesize that it will post a positive sign. The presidential approval hypothesis

can be read as follows:

H10 The higher the American presidential approval, the more bargaining
success for the United States’ negotiators when the use Section (Super)
301.

was

This chapter has presented the framework for an empirical analysis of the level
of success by American negotiators when resorting the Section (Super) 301. More
specifically, with the description of ten hypotheses, it indicates which variables will be
included in the statistical analysis of the next chapter. Each variable represents a
dimension of the structure of negotiation and a parameter of formal model presented in
the previous chapters. In other words, these variables are an indication of the bargaining
position of Washington’s delegates and should provide an explanation for their

negotiation success when they resort to Section (Super) 301 as a foreign policy tool.

CHAPTER 5

Empirical Test: Section (Super) 301

Introduction

The data set79 used in this analysis is partially comprised of a set of variables
taken from Bayard and Elliot’s (1994) study of US. trade policy of reciprocity and
retaliation. It also includes a certain number of variables not included in Bayard and
Elliot’s study. A detailed description of the sample was presented in the previous
chapter. This sample set was selected for several reasons. First, even though Bayard
and Elliot were testing different propositions, their results can be compared with the
results I get in this dissertation. Second, it keeps one of the actors, the United States,
constant for all the cases. This facilitates the analysis of different propositions. Indeed,
the data set indicates that the level of success achieved by the American negotiators
varies from case to case and is not related to a simple correlation with the economic size
of the target. In other words, it shows that in some instances, the US. was successful
in its negotiations with a target, while in some other instances they were unsuccessful
when negotiating with the same target. Hence, it is possible to test the hypotheses stated
in the previous chapter and it is possible to refute the model of this dissertation. Third,

it is possible to rely on Bayard and Elliot’s assessment of the level of success for

 

7" The entire data set is presented in Appendix D.

120

121

American negotiators. This has the advantage of bringing about some consistency
between the results achieved by their analysis and the results achieved in this chapter.
This chapter first examines some alternative propositions regarding the
explanations of outcomes of trade negotiations. It concentrates particularly on power,
interdependence and institutional constraint explanations. It is followed by an empirical

test of the theoretical and formal models proposed in the previous chapters.

1. Power and Dependence
A) Structural Power

The level of success achieved by the American negotiators with corresponding
target countries is demonstrated by Table 3. The most obvious finding when observing
Table 3 is that a reliance on structural power, especially if defined in terms of military
and economic power, is a weak predictor of success in international bilateral trade
negotiations. Indeed, if such was the case, we would expect the United States to have
much more success than shown by Table 3“". It is interesting to notice that a nation,
still considered by many as an hegemonic power, achieves success in its international

trade negotiations in only 13% of the cases (10 out of 75). Even if we also consider

 

8" Two factors, which we cannot control, affect the level of American bargaining
success. On one hand, there exists a certain sample selection bias in Bayard and Elliot’s
sample, namely that only cases in which actions were taken by the Office of the United
States Trade Representative (USTR) are included in the study. Cases that were resolved
prior to formal actions, for which we have no documentation, are not included.
Consequently, only highly controversial cases are studied. This affects the level of
American bargaining success. On the other hand, the USTR office is likely to take on
cases that have some chances to be solved through bargaining. All in all, we have two
sample selection biases pulling in different directions, and it is difficult to evaluate their
individual impact on the American bargaining success when resorting to Section (Super)
301.

122

"partial successes ", this figure is only 47% (35 out of 75). This means that in more than
50% of the cases, the US. negotiators were not even able to achieve at least a partial
success. Consequently, given the data at hand, we have a first hint that power may not
be fungible, which seems to signify that structural power is a poor predictor of
international trade negotiations.

Table 3 also demonstrates certain trends indicating the United States does not
automatically have more success when confronted with "small" powers rather than
"large" powers. It is possible to reorganize Table 3, to have a second look at this claim
on the fungibility of power. Table 4 indicates the mean score of the level of success
achieved by Washington vis-a-vis each target. The targets are ranked from the highest
to the lowest score. The second column shows the number of occurrence of the

utilization of Section (Super) 301 for each target.

123
TABLE 3

Level of Success for the American Negotiators When Using Section (Super) 301

 

Level of Successa

 

 

Failure (0) Nominal Success (1) Partial Success (2) Success (3)
Argentina (1) Argentina (2) Argentina (2) Brazil (1)
Canada (2) Brazil (1) Brazil (3) Canada (1)
EC (4) Canada (1) Canada (1) EC (2)
India (3) China (2) EC (5) Japan (4)
Portugal (1) EC (9) India (1) Korea (1)
Spain ( 1) Guatemala (1) Japan (5) Taiwan (1)

India (1) Korea (4)

Japan (3) Taiwan (4)

Korea (2)

Norway (1)

Taiwan (1)

Thailand (3)

USSR (1)

 

’ The number of occurrences for each case in indicated between parentheses.

124
TABLE 4

Level of Success for American Negotiators by Targets

 

Level of Success

 

 

Target (N) Score
Japan 12 2.08
Taiwan 6 2
Brazil 5 2
Korea 7 1 .86
EC 20 1 .25
Argentina 5 1 .2
Canada 5 1 .2
China 2 1
Guatemala 1 1
Norway 1 1
Thailand 3 1
USSR 1 1
India 5 0.6
Portugal 1 0
Spain 1 0
_A.lLta.tgets 75 1 44L

 

 

125

When we examine Table 4, we first notice that there is no direct correlation
between a target’s structural power and the level of bargaining success achieved by the
United States“. Indeed, American negotiators achieve their highest level of bargaining
success when dealing with Japan (2.08), better than with many "smaller" powers such
as Guatemala, Norway, Argentina. They also receive a generally good level of success
with the EC and Canada, ranked respectively fifth and seventh out of fifteen targets. In
fact, they obtain more success when confronted with another member of the G7 than
when confronted with a non-member: 1.51 versus 1.37. Even if this may not reveal a
significant difference, it certainly indicates that American negotiators do not receive a
bargaining advantage from their encounters with "smaller" powers.

Table 4 reveals two other tendencies: a good degree of success in their
confrontations with targets from Asia, and better success versus targets encountered on
numerous occasions. First, the regional concentration: three of the four targets (Japan,
Taiwan, and Korea) where American negotiators obtain the highest level of success are
from the Pacific region. Is this just a coincidence or is there a certain "regionalisation"
of their level of success? Could it be that the United States have better trade
representatives in this region of the planet? Or could there be some other underlying
factors explaining this phenomenon? I shall pay special attention to this phenomenon
when interpreting the results of the statistical analysis of the bargaining model that

follows. Second, we notice a much higher level of success with targets that the United

 

8‘ For simplicity, I use a target Gross National Product (GNP) as a measure of
power (Organski and Kugler 1980). These are indicated in Table 4. Even if we were
to use different measures of power, we must perforce recognize that the results would
be very similar.

126

States met at least five times. Indeed, with the exception of China, all targets which the
United States met five time or more are at the top of Table 4. This could be an
indication that a certain familiarity among trade representatives of different countries
breeds success. Still, we must be careful with this kind of interpretation. The overall
mean score (1.44) indicating U.S. level of success is still higher than the score it obtains
with the EC (1.25), Argentina (1.2), and Canada (1.2), despite a combined thirty
bargaining occurrence with these targets.

These descriptive statistics not only disclose the fact that structural power is a
poor indicator of bargaining success, but also serve as an interesting guide for the

interpretation of the ordered multinomial logit analysis that ensues.

B) Economic Power.

A second level of analysis brings us to consider only economic power. One
measure of economic power that is suited for this analysis is the Gross National Product
(GNP). Table 5 indicates the GNP of each target country”. If a simple reliance on
economic power was enough to suggest bargaining success in international bilateral trade
negotiations, we would expect American negotiators to obtain a higher level of success
with " small economic powers ", rather that "large economic power." If we consider
Tables 3, 4, and 5 we notice that cases of failures involve some of the countries with

very small GNP’s or with low score of American bargaining success. In fact, in three

 

32 When several negotiations occurred between the United States and a target, I
considered the target’s GNP for the last negotiation between the two parties. A similar
method is used for the data in Table 5.

127

cases (Argentina, Portugal and Spain) the GNP of the target is less than 200 billions”.
At the other end of the scale, we observe that 70% (7 out of 10) of the cases of
"successes" occurred with targets ranked in the top tier of the GNP classification
(Canada, EC, Japan). When we also take into account cases of "nominal successes", the
figure is still over 50% (18 out of 35). If we consider the American level of bargaining
success as indicated by Table 4, we uncover similar results. When confronted with the
"first tier" of targets in terms of their GNP (EC, Japan, USSR, Canada, China),
Washington’s combined score is 1.31. In comparison, the respective combined scores
when confronted with the second or third tiers of targets are 1.29 and .84. Based on
these figures, we are forced to admit that a strict reliance on economic power is a weak
indicator of the level of success that negotiators may have in a bilateral trade and

investment negotiations“.

 

83 Notice also cases of "nominal success" with Argentina (2 cases), Guatemala, and
Norway; which are all countries with relatively small GNPs.

8“ In order to appraise the impact of economic power on the United States level of
bargaining power, I also conducted a bivariate ordered multinomial analysis with the
targets’ GNP as the independent variable. The result indicates a level of significance
better than .1 (P < [2} = .089) with a positive sign for the regressor. We must use such
result with extreme caution, given the underspecification of the model, but the positive
sign of the regressor reveals that the United States’ negotiators obtain a higher degree of
success when meeting the representatives of a target with a large GNP. This could
denote the fact American negotiators avoid "playing the bully " with small economic
powers.

128
TABLE 5

Economic Power Hypothesis: Target’s GNPa

 

Target’s GNP (in billions)

 

 

 

Target GNP Target GNP
EC 6453.83 Spain 172.25
Japan 3108.96 Korea 168.79
USSR 1220.63 Thailand 102.55
Canada 564.83 Norway 93.12
China 526.99 Argentina 74.82
Brazil 382.83 Portugal 20.87
India 287.43 Guatemala 3.76
Taiwan 210.72

 

3 Sources: Carr (1982); CIA (1979/80); Republic of China, Taiwan Statistical... and
Monthly Bulletin...; World Bank, World Tables.

C) Dependence

The results of the economic power equation suggests that an answer may be found
in the economic dependence of the target on the United States. Even though the EC,
Japan, USSR, Canada, and China are considered in the first tier in terms of economic
power as calculated by their GNP, it could be the case that they are also the most
dependent on the United States for their foreign exchanges. Consequently, we could
believe that countries that have the largest amount of exports to the United States are put
at risk of endangering their economy in case of trade deterioration with the United States.
In a bilateral trade negotiation, this puts the United States in the driver’s seat. Hence,

we may suppose that the higher the targets’ exports to the United States (T EXP) are, the

129

higher success rate for the American negotiators“.

Table 6 shows an ordered ranking of the targets’ exports to the United States“.
We find a large degree of similarin when we compare Table 4 with Table 6. In both
tables we find Japan, Taiwan, Brazil, Korea, the EC, and Canada at the top of the scale.
The only slight anomaly is Argentina, which is ex-aequo with Canada in sixth place in
terms of American bargaining success (Table 4), but only twelfth in regards to its level
of exports to the United States. Conversely, Guatemala, Portugal, the USSR, Norway,
Spain and India are all at the bottom of the scale in both tables. All this indicates that
trade dependence of a target is an important factor for explanation of American

bargaining success when Washington resorts to Section (Super) 301‘”.

 

85 I tested this hypothesis with an ordered multinomial logit analysis. The
independent variable is the natural log of a target’s export to the United States (LTEXP).
I used the natural log in order to secure homoscedastic error terms in the estimation of
the model. Keeping in mind that the model suffers from underestimation, the result
indicates that the regressors have a positive sign and LTEXP is significant at the .05
level (P > [Z] = .033). This result reveals that American negotiators achieve a higher
degree of success with targets which are highly dependent on the United States for their
foreign trade.

8° When more than one case involve a single target, I used the value for the last
occurrence.

87 For this reason, I retained this variable when testing a second model based on
target’s compliance. The results are indicated further in this chapter.

130
TABLE 6

Dependence Hypothesis: Targets’ Exports to the United States8

 

Targets Exports to the United States (In billions)

 

 

 

Target TEXP ngget TEXP
Canada 102.318 India 3.228
EC 94.64 Spain 2.035
Japan 92.538 Norway 2.018
Taiwan 23 .572 Argentina 1 .201
Korea 20.84 USSR .587
Brazil 7.89 Portugal .42
China 7.398 Guatemala .205
Thailand 6.686

 

‘ Source: See chapter 4, footnote number 11.

However, these results are, in some ways, the flip side of the previous hypothesis.
Indeed, it is reasonable to believe that the largest economic powers are also those
exporting more of their resources to the United States. In fact, GNPs of target countries
(TGNP) show a high level of correlation of .61 with their exports to the United States
(TEXP). An alternative way to measure the effect of targets’ exports to the United States
is to consider their GNPs. That is, we need to measure the share of a target’s GNP that
is accounted for by its exports to the United States (TXGNP). This measure has shown

a very low level of correlation with TEXP (-.05).

131
TABLE 7

Dependence Hypothesis: Share of GNP Accounted for by Exports to the US.

 

TXGNP (in percentage)

 

 

Target TXGNP Target TXGNP
Canada 18.061 Portugal 2.012
Korea 12.347 Argentina 1.605
Taiwan 1 1 . 186 EC 1 .466
Thailand 6. 52 China 1 .404
Guatemala 5.452 Spain 1.181
Japan 2.976 India 1.123
Norway 2.167 USSR .048

Brazil 2 .061

 

132

Table 7 shows a ranking of target countries according to their share of GNP
accounted for by their exports to the United States“. When comparing Table 3 with
Table 7, we obtain mixed results. Three countries, Canada, Korea, and Taiwan, show
a very high dependence on the United States, with more than 10% of their GNP
accounted for by their exports to the United States. Nevertheless, out of 18 cases
involving these targets, the United States obtained a full success in only 3 cases, that is
a success rate of 17%. Furthermore, these constitute 30% of all cases of success (3 out
of 10), while they also represent 24% of all cases (21 out of 75). The difference is not
very large. However, if we also consider cases of "partial success", the test is more
conclusive: 12 of 18 cases (67%) involving these countries are classified as a " success"
or a "partial success" and they constitute 34% of all cases classified in these categories,
which is 10% more than would be expected by the null hypothesis. Thus, the difference
gets more significant and this seems to be an indication that there exists a relationship--
even though still a weak one--between trade dependence and level of success for US
negotiators.

If we also effectuate a cross comparison between Table 4 and 7, we also obtain
mixed results. The three countries for which the United States obtains on average at
least a partial success (Japan, Taiwan, Brazil) that is, at least a score of 2, are ranked
respectively sixth, third, and eighth in regards to their share of GNP accounted for by
their exports to the United States. This shows only a mitigated support for the

hypothesis that US negotiators obtained a higher level of success when negotiating with

 

88 Once again, I conducted an ordered multinomial analysis. The independent
variable, TXGNP is not significant (P > {Z} = .12), but sports the hypothesized sign
(positive).

133

countries that were highly dependent on them for their exports.

The results for the economic dependence propositions are somewhat surprising.
It appears that the net amount of exports to the United States is more important than the
share of GNP accounted for by exports to the United States. What this tells us is that
the United States’ negotiators seem to achieve more success when confronted with large
economic powers, given the high level of correlation between exports to the United States
and GNPs of targets.

The answer may not be that simple. We can gain some substantial insights by
testing a more fully developed model concentrating on trade interdependence, institutional
constraint, political necessity, and domestic support, as suggested in the previous

chapters. The next section evaluates such a model.

11. Empirical Results of Principal Model

This section tests the theoretical model of Chapter 2, the formal model and
propositions of Chapter 3, and hypotheses of Chapter 4. It starts with some descriptive
statistics and a bivariate analysis. It is followed by an ordered multinomial logit analysis.
Finally, it includes a study of the individual impacts of the independent variables on the

dependent variable.

134
TABLE 8

Summary of Level of Analysis, Elements of Analysis, Parameters, and Variables

 

 

 

 

 

Level Parameters Elements Variables
International Breakdown costs Interdependence Trade Interdep- (DEP)
" " " " " " lnterdep./GNP (DEPGNP)
" " " " Threat credibility Type of issue (ISSUE)
Domestic Information Institutional constraint U-S- th. Div. (DIVIDED)

 

N H

Tar. Constr. (INSCONST)

 

Costs for delays

Political necessity

U.S. Retaliation (RETAL)

 

Economic necessity

Goods Val. (DVCGOOD)

 

Misery Index (MISERY)

 

Societal support

Trade Balance (T BAL)

 

 

 

 

 

Pres. Approval (APPROV)

 

 

135

Table 8 recaps the essence of previous chapters. The level of analysis column
refers to the structure of negotiation as represented by Putnam’s (1988) two-level of
analysis metaphor. The second column is related to the parameters of the Stabl-
Rubinstein formal model. The third column indicates the elements of analysis invoked
by the propositions. Finally, the last column shows the variables included in the
empirical model. The dependent variable (Success), not indicated in Table 8, is the level
of success for the U.S. negotiators. It is a variable with the following four categories:
failure, coded as a "0"; nominal success, coded as an "1"; partial success, coded as a

"2"; and success, coded as a "3".

A) Descriptive Statisticmd Bivariate Analysis

Table 9 provides descriptive statistics of each variable used in the analysis that
follows. It is important to note that Success, ISSUE, DIVIDED, INSCONST, RETAL,
DVCGOOD are dummy variables, such that the significance of their descriptive statistics
is less meaningful. The descriptive statistics of the natural log of the targets’ exports
(LTEXP) are also included because this variable will be included in the compliance

model.

l 36
TABLE 9

Descriptive Statistics

 

 

N=75
Variable Mean Std. Dev. Minimum Maximum
Success (1nd.) 1.44 .919 0 3
DEP 1.031 1.483 .009 4.129
DEPGNP .386 .531 .003 2.308
ISSUE .4 .493 0 1
DIVIDED .333 .759 -1 1
INSCONST .387 .49 0 1
RETAL .253 .438 0 1
DVCGOOD .467 .502 0 1
MISERY 10.836 3.872 5.57 20.5
TBAL -7 .507 15 .096 -52.625 11.972
APPROV .529 .109 .32 .75
LTEXP 2.705 .179 -1.585 4.628

 

137

Table 10 represents a bivariate analysis indicating a crosstabulation of mean
values taken by the dependent variable (Success) according to different values taken by
independent dummy variables”. This type of bivariate analysis is highly
underspecified, but it provides a good, first estimation of the impact of those dummy
variables on the dependent variable. The first column represents the name of each
dummy variable. The second indicates a description of their category. The third shows
the number of occurrences for each category. Finally, the last column demonstrates the
mean value of the dependent variable for each category. The overall mean value for the
dependent variable (Success) is 1.44.

A general look at Table 10 reveals the realization that the impact of all dummy
variables goes in the hypothesized direction. Indeed, all changes in the mean value taken
by the dependent variable are influenced, to a different extent, by the categories of the
independent variables. Also, these impacts are all in the same direction indicated by the
hypotheses of the previous chapter. Let me provide an explanation of Table 10 by

discussing the impact of each independent variable separately.

 

89 This method can be used only with dummy variables. For continuous variable,
I will use a graphical representation of their marginal impact on the dependent variable.

Mean Value for Level of Success by American Negotiators

138
TABLE 10

Bivariate Analysis: Table of Means

 

 

 

 

 

 

 

 

 

 

 

 

 

Independent Variable Category (N) Mean
ISSUE Quota or Tariff (1) 30 1.77
"" Otherwise (0) 45 1.22
RETAL Threat or Use of Retaliation (1) l9 1
"" No Threat or Use (0) 56 1.59
DVCGOOD 100 Million or More (1) 35 1.31
"" Less Than 100 Million (0) 40 1.55
IN SCONST High Constraint (1) 29 1.07
"" Low Constraint (0) 46 1.67
DIVIDED Divided Government (1) 38 1.58
"" Divided Congress (0) 24 1.29
"" Undivided Government (-1) 13 1.31

 

 

 

 

 

139

It was hypothesized in the previous chapter that American negotiators achieve
more bargaining success when the issue on the agenda is related to a tariff or a quota.
We find a certain level of confirmation while observing Table 10: when the issue being
negotiated is a quota or a tariff the mean level of success goes from 1.44 to 1.77. Of
all dummy variables, this represents the largest positive impact. Conversely, when the
issue is about intellectual property rights or safety measures, their level of success drops
to 1.22. These results indicate that the type of issue under discussion influences the
bargaining success of American negotiators.

Measures of retaliation or reciprocity are tools made available to the U.S. trade
representatives when using Section (Super 301). However, when they decide to threaten
or directly use those tools, it may indicate some bargaining weakness on their part.
That’s what Table 10 reveals. It is especially true for the nineteen cases when they
decided to use these policy instruments. In fact, their level of success drops to 1
(nominal success) from 1.44 when using or threatening retaliation. Of all dummy
variables, this constitutes, the largest negative impact on the value taken by the dependent
variable. The impact is more moderate when they do not resort to retaliation, such that
their level of success only goes up to 1.59.

The higher the value of the goods or services being negotiated, the more difficult
it is for American negotiators to achieve their objectives. Table 10 shows some support
for this affirmation. The changes in their level of bargaining success goes in the
speculated direction, even though the impact is not impressive: a drop of .13 when the
value of the contested goods are 100 million dollars or more and an increase of .11 when

their value is less than that dollar figure.

140

The level of domestic institutional constraint of a target demonstrates an important
negative impact on the US bargaining success. Indeed, when a target is domestically
highly constrained for its decision-making process, American negotiators achieve much
less bargaining success. Then, their "mean score" plummets to 1.07. However, when
they face a target with low domestic institutional constraint, their level of bargaining
success increases to 1.67. An important explanation for this phenomenon is the number
of instances (20) when they must compose with the EC, a highly constraint target. We
have seen, in Table 4, that their overall level of success vis-a-vis the European Union
is only 1.25. Still, Table 10 also demonstrates that they are even less successful when
confronted to other highly constraint targets”.

Domestic institutional constraints do not have the same impact on the bargaining
success of American negotiators when the constraints come from their own domestic
institutions (DIVIDED). Nevertheless, DIVIDED has a consequential bearing on the
independent variable. Table 10 reveals an interesting facet. American negotiators do not
receive added bargaining power and success from a divided Congress. In fact, in those
circumstances, their level of bargaining success drops to approximately the same level
of success they achieve when the government is undivided (1.29 and 1.31 respectively).
The positive impact appears to come from a divided government, that is, when the
Presidency is controlled by one party, while the other party controls the Congress.
Then, the level of bargaining success achieved by Washington increases to 1.58. This

engaging outcome has a consequential bearing on the hypothesis that a divided

 

9° In fact, when a highly constrained target is not the EC, the level of bargaining
success achieved by American negotiators drops to only .67.

141

government gains some international bargaining momentum.

All the results of Table 10 should not be interpreted as conclusive. Due to the
underspecified nature of the analysis, those results are only exploratory. A multivariate
analysis should confirm or negate this "first cut" at the bargaining problem in

international trade and investment negotiations.

B) Multivariate Analysis

i) Method

Because the dependent variable (Success) is measured along an ordinal scale in
which the American negotiators level of success is categorized along a continuum going
from failure to success, the empirical model takes the form of an ordered multinomial
logit. An ordered multinomial logit is a generalization of the binary threshold model.
It has great semblance with a probit or a logit model utilized to evaluate binomial
dependent variables, except that in this case the dependent variable incorporates
information that is scaled and ordered in four categories (Greene, 1993; Maddala, 1993;

McKelvey and Zavoina, 1975). The general form of the equation to be estimated is:

Yi-a+XB+e (1)

Where Yi is a nxl vector of observed values for the dependent variable, X is an nxk

matrix of observed values for the regressors in the empirical model and 6 follows a

142
logistic distribution, such that the cumulative distribution of e is”:
CXP(Z,-)

. -—- 2
F(Zl) 1 + eXP(Zi) ( )

 

In the empirical model, the dependent variable (Success) represents a continuum
of unobserved variable Y,’. Hence, the relationship between the observed values of Yi

and the unobserved values of Y,‘ is:

y, = failure if y; s 0 (3)
=nominal success if 0 < y; s u,
= partial success if [1.1 < y; s u,

. n
= success lf “2 s y,

Underlying probabilities are assessed by a maximum likelihood procedure. The predicted
score is defined as Sj = x11131 + x2182 + + xijk (4). The ordered logit predictions
are then the probability that SJ- + it]. lies between a pair of cut points k,_] and k,. In the
specific case of this empirical analysis, the following probabilities indicate American

negotiators’ propensity to have a certain degree of success:

 

9‘ Unless we have a very large sample size, so that there are enough observations
at the tails, the logit and the probit results will be very similar. Indeed, since the
cumulative distribution used in a probit analysis, and the logit distribution are very close
to each other, except at the tails, we are not likely to get different results (Maddala,
1993: 328). Since we have a rather small sample size, both methods could have been
used alternatively.

143

Pr [0] - ms]. + p}. < k1) = 1 (5)

(1 + e’i‘“)

 

Pr [1] = Pr(k1< S]. + pl. < k2) = 1 — l

(1 + es}- - k2) (1 + 68’. — kl)

 

 

 

 

 

1 1
Pr[2]=Pr(k<S.+p..)<k= _.________
2 j j 3 (1+e‘i‘“) (1 0331'“) (1 +esi‘kt)
1 1 3
Pr[3]=Pr(k<S.+p.)=1- -
3 J I (1+e5i”"3) (1 +e‘I"‘2) (1 +es,-t,)

Equation 4 allows us to calculate the predicted score of the dependent variable (Success)
for all of the seventy-five cases. Equation 5 helps us unveil the probabilities that, for
each case, the American negotiators level of bargaining success is a failure (Pr=0), a
nominal success (Pr=1), a partial success (Pr=2), or a success (Pr=3). Generally, for
each case, the category that demonstrates the highest probability, also corresponds to the
predicted score, but this does not always happen. For the calculation of the "cases
correctly predicted (CCP)", I compared the predicted scores (unobserved scores), Y,‘,
and the coded scores (observed scores), Y,. It is, then, only a matter of calculating the
percentage of cases that indicate the same predicted and observed score.

Also, by holding all the variables constant at their mean, with the exception of
one, in equation 4, it is possible to examine the marginal effect of the variable not held
constant when sj is included in equation 5. For instance, let’s assume that our equation
contains only two independent variables, DEP and DEPGNP. To see the marginal
impact of the regressor of DEP on the dependent variable, Success, we can enter the

minimum and maximum value of DEP in equation 4, while holding DEPGNP constant

144
at its mean”. Next, we need to incorporate the predicted scores (S) of DEP in
equation 5, in order to evaluate its marginal effect on each category of the dependent
variable.

In addition to the marginal effects, a graphical method will be used to provide a
visual plot of the marginal changes in each regressor with respect to the dependent
variable”. In such case, not only do we incorporate the minimum and maximum value
of one of the regressors in equation 4, while holding other regressors at their mean, but
also each of its observed values. Hence, the graphical method isolates the effect of an
independent variable on the probability that American negotiators will achieve a certain
degree of negotiation success. Bennett Quifiones and Gates (1995: 72, n.27) lucidly
indicate the meaning of the marginal effects:

Note that these marginal effects are not equal to the coefficients.

Typically coefficients and t-ratios are reported with no discussion

of direction and magnitude of the effects of the regressors (see

Greene [1990]). The problem with interpreting only the

coefficients involves the lack of consideration of the effect of the

middle category for the dependent variable. By calculating the

marginal effects of each independent variable, it is possible to

determine the effects of a change in x on predicted probabilities for

different values of the dependent variable.

For instance, these calculations allow one to determine how an independent variable such

as DEP, the trade interdependence between the United States and target countries, affect

the negotiation success of American negotiators. It has the added advantage of indicating

 

92 When we have more than two independent variables, all variables, except one, are
held at their mean.

93 This graphical representation will not be used for dummy variables, because they
only take two or three different values.

145

the marginal effect, or importance, of one of the regressors, not only one the minimum
(0) and maximum (3) values taken by the independent variable (Success), but also its
middle categories (1 and 2). Another advantage of the graphical representation of the
marginal effect of the regressors is that each figure reduces the number of dimensions
to two: only the independent variable of interest and the predicted probabilities for the
dependent variable (Success) will be plotted.

Using the data and the methods described above, it is now possible to examine

the propositions and test the hypotheses of the theory and formal model of this thesis.

ii) Ordered Multinomial Logit Results

Estimates of the interdependence model used to evaluate the propositions of
Chapter 3 and test the hypotheses of Chapter 4, are located in Table 11. This empirical
test is used to determine which factors influence the degree of success for negotiators in
an international bilateral trade and investment negotiation. More specifically, Table 11
is an estimation, using ordered multinomial logit, of the level of success of American

negotiators when resorting to threat of retaliation.

Interdependence Model: Empirical Estimates

146

TABLE 11

94

 

 

 

Number of obs.: 75

chiz(10): 4395

Prob > CM": .0000

Pseudo R2: .2212

Log Likelihood: -75.692126

CCP: 55%
Success Coef. Std. err. z P > [z]
DEP .6255 .34 1.84 0.066
DEPGNP -.219 .746 -0.294 0.769
ISSUE 1.3453 .5137 2.619 0.009
DIVIDED .3124 .3758 0.831 0.406
INSCON ST 21381 .786 -2.72 0.007
RETAL -l .7548 .5672 -3 .094 0.002
DVCGOOD -1.04 .5047 -2.06 0.039
MISERY -.1144 .0767 -1.492 0.136
TBAL -.0502 .0199 -2.522 0.012
APPROV .1733 2.6017 0.067 0.947
_cutl -3.872135 2.027712
_cut2 -1. 167997 1.972112
_cut3 1 .399905 1 .985419

 

 

 

9“ The reader will notice that I used a z-score, instead of a Student-t statistics. The
reason is that the statistical package used, STATA, gives, by default, only one of the two
Thus, it appears that a sample size of 75
justifies the use of the z-score. To double check the correspondence, I asked Scott Gates
to run the same analysis with SST. SST reported Student-t statistics, and the results were
virtually the same as the results obtained with STATA.

statistics, depending on the sample size.

95 Chi2(10) is tantamount to -2LLRchi2

147

A general observation of the model indicates that some of the variables perform
well (DEP, ISSUE, INSCONST, DVCGOOD, RETAL, TBAL)96; another uncovers
nearly significant results (MISERY); while others indicate insignificant results
(DIVIDED, DEPGNP, APPROV). With the exception of DEP, all of the estimated
coefficients have the hypothesized sign. On average, my model’s bargaining success
predictions are correct in 55% of the cases”. This is a noteworthy improvement in
predictive success over the naive alternative model (null hypothesis) that predicts the
most frequent category continually. Note that the overall measure of fit (i.e., -2LLR)
is statistically significant at the .0000 level. There is, therefore, a meaningful support
for my general claim that domestic and international factors account for the level of
success of international negotiators.

An analysis of the significance of different independent variables in a logit
analysis may not expose all the relevance of a specific model. Table 12 provides the
marginal impacts of independent variables on the level of bargaining success
accomplished by American negotiators. The first row indicates the probabilities taken
by different categories of the dependent variable when all independent variables have
been kept at their mean. The table can be read in the following way: when all
independent variables have been held at their mean, the probability that American
negotiators’ level of success being coded as a failure is .028, .273 for a nominal success,

.548 for a partial success, and .151 for a full success. The following rows indicate the

 

96 Five of these are significant, in terms of a one-tailed test, at the .05 level.

97 The CCP for the null hypothesis is 37%. Appendix E indicates how the CCP
were obtained.

148

impact of only one independent variable when it takes its minimum and maximum value.
The values between parentheses indicate the marginal variations from the base values
(i.e. when all variables are held at their means). The calculations are based on equations
(4) and (5). See also further explanations in appendix B. What we expect to see is that
a variable that has a positive effect (i.e that sports a positive sign for the regressor) on
the level of success for American negotiators will increase the probability of success (or
partial success) when it takes its maximum value. Hence, the marginal effect of the
variable should be positive when Pr=2 and Pr=3. The reverse should also be true: the
marginal effect on Pr=0 and Pr=1 should be negative. When the same variable takes
its minimum value, we should expect a reverse trend: its marginal effect is negative on
Pr=2 and Pr=3, and positive on Pr=0 and Pr: 1. For a variable that sports a negative
sign, that is, a variable which, when it increases, decreases the chance of success for
American negotiators, should have the exact opposite effect: when at its maximum value,
the marginal effects on Pr=0 and Pr=1 should be positive, and its marginal effect on
Pr=2 and Pr=3 should be negative; while when at its minimum value, the marginal
effect on Pr=0 and Pr=1 should be negative, and its marginal effect on Pr=2 and Pr=3

should be positive.

149

TABLE 12

Interdependence Model: Marginal Impacts of the Regressors

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Pr (0) Pr (1) Pr (2) Pr (3)

All .028 .273 .548 .151

DEP (Min.) .052 (.034) .398 (.125) .465 (-.083) .086 (-.O65)
DEP (Max.) .004 (-.024) .054 (-.219) .389 (-.159) .553 (.402)
DEPGNP (Min.) .026 (-.002) .258 (-.015) .554 (.006) .162 (.011)
DEPGNP (Max.) .042 (.014) .354 (.081) .499 (-.049) .105 (-.046)
ISSUE (Min.) .047 (.019) .378 (.105) .481 (—.O67) .094 (-.057)
ISSUE (Max.) .013 (-.015) .149 (-.124) .554 (.006) .285 (.134)
DIVIDED (Min.) .042 (-014) .353 (.08) .5 (-.O48) .105 (—.046)
DIVIDED (Max.) .023 (-.005) .236 (-.O37) .561 (.013) .18 (.029)
INSCONST (Min.) .012 (-.016) .146 (-.127) .552 (.004) .289 (.138)
INSCONST (Max.) .097 (.069) .518 (.245) .339 (-.209) .046 (-.105)
RETAL (Min.) .018 (-.01) .198 (-.O75) .566 (.018) .217 (.066)
RETAL (Max.) .097 (.069) .518 (.245) .339 (-.209) .046 (-.105)
DVCGOOD (Min.) .017 (-.011) .192 (-.O81) .566 (.018) .224 (.073)
DVCGOOD (Max.) .048 (.02) .381 (.108) .479 (-.O69) .093 (-.058)
MISERY (Min.) .016 (-.012) .175 (-.O98) .564 (.016) .245 (.094)
MISERY (Max.) .08 (.052) .485 (.212) .379 (-.169) .056 (-.095)
TBAL (Min.) .003 (—.O25) .04 (-.233) .325 (-.233) .632 (.481)
TBAL (Max.) .071 (.043) .463 (.19) .403 (-.145) .063 (-.088)
APPROV (Min.) .029 (.001) .28 (.007) .545 (-.003) .147 (-.004)
APPROV (Max.) .027 (-.001) .266 $07) .551 (.003) .156 (.005)

 

“ Values between parentheses are marginal differences from the mean.

150

An evaluation of the overall importance of the variables, in terms of their impact
on the level of success of American negotiators, suggests that those variables which were
the most significant are also those having the most profound impact on the dependent
variable. In general, those variables affect especially the propensity for American
negotiators to get an outcome leading to a nominal or a partial success. This is not
surprising, given that the majority of outcomes fall into those categories.

The general assessment of the model can be supplemented by a consideration of
each of the individual coefficients and marginal impacts of the variables. I will first
comment briefly on the variables leading to insignificant results. Then, I will address
the variables indicating "nearly—significant" or significant results.

In the first section of this chapter, an analysis of economic dependence of target
countries on the United States, tested in terms of their level of exports to the United
States (TEXP), turned out being highly significant: Americans had more success with
countries with large exports to the United States, rather than with countries exporting less
to America. It could be expected that the flip side is also be valid: the United States will
have less success with countries where they ship large amounts of goods (USEXP). Such
is not the case: an ordered logit analysis on the variable USEXP indicates that the
regressors sports a positive sign, i.e., the United States obtained more success with
countries where they export a large amount of goods, and shows no significance (P >
[z] =0.695). What may at first be surprising, has a logical explanation. It is very
likely that the countries that export a great amount of goods to the United States are the
same countries that import many American goods. In fact, the level of correlation

between USEXP and TEXP is a whopping .8933. Thus, a better consideration of

151

economic power may be some measure of trade interdependence or balance of trade”.

To take into account some effects of trade interdependence, three variables were
created: TSHOUT, target’s share of United States trade output (US’s dependence);
USHOUT, United States’ share of target’s trade output (target’s dependence); and DEP,
the ratio of TSHOUT over USHOUT. Hence, the variable should post a negative sign,
such that a higher ratio is related to less bargaining leverage for the US negotiators. The
results are surprising: the variable has a positive coefficient and is significant at the .1
level (P > [z] = 0.066). This means that Americans seem to have more bargaining
success when a trade relationship with another country is to their disadvantage.

A possible explanation could be found in the level of resolve shown by the
American negotiators. In order to compensate for this asymmetry of trade dependence,
they might compensate by becoming more adamant in their demands. Very often,
adopting a tough stance leads to a successful outcome. When we analyze Table 12 we
find a strong support for this assertion. A first glance at the impact of the variable
clearly indicates that it has the most profound bearing on the level of bargaining success
for American negotiators when it takes its maximum value. The maximum impact,
when the variable takes its minimum value, is on the propensity of marginal success
(from .273 to .398), which is surpassed by the changes in the probability of nominal
success (from .273 to .054), partial success (from .548 to .389), and full success (from
.151 to .553) when the variable is at its maximum. This last figure is especially
interesting. It demonstrates that when Washington negotiates with a target which has a

definitive edge in terms of trade asymmetry, they might compensate by using a very

 

98 The balance of trade variable is discussed below.

152

tough stance, this leads to a very significantly increase in their level of full success. This
augmentation of .402 in the probability of success is, in fact, the most profound impact
on the dependent variable. Figure 13, which plots the marginal impact of the variable
on the degree of bargaining success, is also conclusive. It shows a rapidly increasing
slope for the curve representing the predicted probability of full success when DEP
becomes positive. This supports the claim that high trade dependence bolsters US
negotiators’ bargaining toughness, which, in turns, accentuates their probability of
reaching full negotiating success.

However, this "all or nothing" strategy has its drawbacks. As discussed in
chapter 2, by being too "tough" in any bargaining episode, a negotiator increases the risk
of a stalemate. Hence, his demands fall outside of other negotiators’ win-sets. In the
specific episodes of Section (Super) 301 utilization, American negotiators offered to a
target one "last chance " to reach a negotiated settlement. However, if they appear to be
too unyielding, it might be better for a target to retire from the negotiation table instead
of acquiescing to Washington’s requirements. In those instances, the resulting outcome
is coded as a "failure" in the empirical analysis. Table 12 discloses some support for
this phenomenon. While it is not the case that the probability of failure increases99
when DEP is at its maximum value, the respective probabilities of nominal success and
partial success indicate a much sharper drop‘o". Nevertheless, until we have an
adequate empirical measure of negotiators’ resolve, we can only speculate about the

connection between trade dependence and resolve.

 

99 In fact, the probability goes from .028 to .004.

‘00 This can also be clearly seen by consulting Figure 13.

153

When testing for the effect of economic dependence of the target on the United
States in the first section of this chapter, I indicated that when a target had a large share
of its GNP accounted for by its exports to the United States (TXGNP), it could be an
indication of some source of bargaining power for American negotiators. The variable
DEPGNP also considers the trade dependence of the United States towards the target
nation by considering its share of GNP associated with its exports to the target country
(USXGNP). DEPGNP represents the ratio of USXGNP over TXGNP, such that a
smaller value corresponds to a larger bargaining leverage for the US negotiators. Hence,
the variable should post a negative sign. Indeed, the variable has a negative sign, but
it shows almost no significance (P > [z] = 0.769).

Table 12 unveils similar results. The marginal impact of the variable is on the
hypothesized direction (i.e., the probability of success increases when the variable is at
its minimum value, while the probability of failure increases when it is at its maximum
value), but is rather weak. Still, the variable has a moderate impact on the probability
for American negotiators to get only some nominal success in their negotiations. Indeed,
when DEPGNP takes on its minimum value, it causes an increase in the predicted
probability of nominal success from .273 to .354. Figure 14 displays those findings: 1)
the greatest slope corresponds to changes in the predicted probability of nominal success
when the variable has small values; 2) in general the graphic has almost constant slopes,
which is an indication of the low impact of DEPGNP on the level of bargaining success
for American negotiators.

Substantially, this moderate impact on the predicted probability of nominal success

indicates, ceteris paribus, that American negotiators may often be able to reach an

154

agreement with targets on which they rely more for their foreign exchanges (rather than
the inverse), as controlled by GNP, but they are generally unable to get the target to
ratify the agreement to their satisfaction. Given this assertion, one should not be
surprised to find out that the United States exhibits this type of trade dependence mainly
with the European Union“. As suggested in previous chapters, this raises the specter
of involuntary defection. European negotiators may sign an agreement in good faith, but
are unable to get an approval or a consensus from their individual members.

Given the results obtained with the variables DEP and DEPGNP, it is not possible
to confirm the two hypotheses related to trade interdependence. It could be the case that
a dependence of the target country on the United States is a better indicator of bargaining
strength than a measure of trade interdependence. This hypothesis will be tested further.

The variable DIVIDED, an indication of division within the United States
government, behaves erratically. The hypothesis was that when American governmental
branches are widely divided, it constitutes a source of bargaining power, such that the
sign of the regressors should be positive. This means that American negotiators, acting
in the name of the president, can use the "good cop, bad cop " approach by arguing that
an agreement looks fine to them, but has no chance to pass Congress. A caveat is in
order here. This hypothesis may not be well-suited for a test on the use of Section
(Super) 301. When a decision is taken to initiate retaliation on a target, even in a
"Divided House", there already exists some sort of wide agreement on the course of

actions to be taken, and the strategy of stressing internal division may not work too well.

 

1‘“ Twelve out of fourteen cases in which dependence asymmetry was in favor of a
target involved the EC. The other two cases implicated Japan and the USSR.

155

The strategy of using internal political division in the United States may work better for
well-publicized trade agreements or security issues.

At first glance, the results indicate the poor showing of DIVIDED: even though
its sign goes in the right direction (positive), it shows no significance (P > [z] = .406).
Nevertheless, we should not conclude that a divided government is a source of weakness,
instead of power. It only demonstrates that the hypothesis may have a better chance to
be confirmed with cases involving a mutual desire for freer trade, instead of a threat of
retaliation.

Still, this variable (DIVIDED) shows some encouraging signs. We have seen in
the bivariate analysis that a divided government, when the Presidency and Congress are
controlled by different parties, has a salient impact on the bargaining strength of
American negotiators. On the basis of this preliminary result, the variable has been
recoded. First, the cases when at least one of the congressional institutions was
controlled by the same party as the presidency were regrouped. This new category
(coded as "0") corresponds to the categories "-1" and "0" of DIVIDED. Instances of
divided government were coded as a "1". Second, the same model indicated in Table
11 has been tested, but with the variable DIVIDED being recoded. The results are
interesting and have the same directional trend as the bivariate analysis. While other
independent variables remain unaffected, the recoded variable displays some encouraging
evidences. Even though it is still not significant (P > [Z] = .138), it displays a clear
improvement from its previous level (P > [2: = .406). Thus, a more intensive look
at this variable reveals that internal division in Congress has no bearing on bargaining

strength of American negotiators, but that a division between the Presidency and

156

Congress is an added advantage for Washington when implementing Section (Super) 301.
This would suggest that for other factors remaining unchanged, better bargaining
outcomes obtained by Washington during the second Reagan administration and Bush’s
term in office.

Presidential approval has been shown to have a significant effect on the propensity
of U.S. presidential use of force abroad (Ostrom and Job, 1986: 555). According to the
bargaining model of this dissertation, the variable does not have any significant impact
of the level of negotiation success for American negotiators when resorting to Section
(Super) 301“”. Indeed, we find little empirical support for stressing the importance
of presidential approval for cases of international economic negotiations. Even though
the sign of the regressors correspond to the hypothesized impact of the variable, it shows
relatively no significance (P > [z] = .947). Furthermore, Table 12 and Figure 15
display very compelling evidence. Table 12 convincingly shows that presidential
approval has virtually no impact on the predicted probabilities of any type of bargaining
success for American negotiators, a supported by the straight lines inside of Figure 15.
It does not matter if presidential approval is high or not; the degree of negotiation
success remains unaffected.

The dummy variable ISSUE was coded in such a way that it would take the value
of 1 if the trade dispute was related to a quota or a tariff, and 0 otherwise. Following
Bayard and Elliot’s (1993) argument, it was indicated that these types of issues were

easier to resolve to the advantage of American negotiators, while issues of intellectual

 

“’2 Dummy variables were developed to represent presidential approval to allow for
evaluation of some threshold effect. None of these have shown significant results.

157

property and standards (coded as 0) are much harder to solve. Hence, the sign of the
regressor is expected to take on a positive value. Table 10, containing a bivariate
analysis of the impact of independent variables on the level of American bargaining
success, has shown that this variable has a salient bearing on the dependent variable. It
is also exactly what Table 11 indicates: not only is the sign positive, but it is also very
highly significant (P > :2] = .009). Table 12 also demonstrates the importance of the
variable in the model. It shows that when ISSUE is at its maximum value, the predicted
probability of success for the American negotiators (Pr [2] and Pr [3]) increases, thus
indicating a positive relationship between ISSUE and Success. The inverse is also true,
when ISSUE is held at its minimum value, the predicted probability of American
negotiating success declines. The impact is especially strong on the probability of
nominal success (-.124) and probability of success (. 134) when the variable takes on its
maximum value; as well as the probability of nominal success (.105) when the variable
takes on its minimum value. Therefore, it can be concluded that the type of issue
involved in the use of Section (Super) 301 has a great impact on the propensity of
American negotiators to reach their goals. All these results confirm the hypothesis that
issues for which direct retaliation and reciprocity is possible, such as tariffs or quota, are
much easier to resolve to U.S. satisfaction than more complicated issues such as
intellectual properties or_health and security standards. In the latter instances, it is much
more difficult for the USTR to prove that injury has been made against a particular U.S.
industry. These results also confirm Bayard and Elliot’s (1993) earlier findings on the
subject.

The dummy variable INSCONST represents the level of institutional constraint

158

in the target countries. Countries with higher level of institutional constraint were coded
as an 1, while countries facing less institutional constraint where coded as a 0.
Following the logic developed in previous chapters, a high level of domestic institutional
constraint is seen as a bargaining advantage in international negotiations. In this
empirical test, this factor gives an advantage to countries facing U.S. retaliation and
reciprocity. Consequently, the coefficient of the variable is expected to bear a negative
sign, such that American negotiators have less chance of achieving their objectives when
confronted with a target country facing a high level of domestic institutional constraint.
The ordered multinomial logit analysis, as shown by Table 11 confirms the hypothesis:
the sign of the coefficient is negative and the variable is highly significant (P > [z] =
.007). This is in accordance with the bivariate analysis of the impact of the variable on
the changes from the mean value of the dependent variable (See Table 10).

The results of Table 12, indicating the marginal impacts of the regressors, go in
the same direction. They reveal that when INSCONST is at its minimum value, the
predicted probability of success for American negotiators, Pr (3), increases, thus
indicating a negative relationship between INSCON ST and Success. The inverse is also
true, when INSCONST is held at its maximum value, the predicted probability of
American negotiating success declines. Its influence is especially strong on the
probability of nominal success (-.127) and of success (.138) when the variable takes on
its minimum value; as well as the probability of nominal success (.245), partial success
(-.209), and success (-.105) when the variable takes on its maximum value. Therefore,
it can be concluded that the targets’ level of institutional constraint has an important

negative effect on the propensity of American negotiators to reach their goals when they

159

decide to use Section (Super) 301.

The aforementioned results depict a situation in which the largest impact of the
variable is on the predicted probability of nominal success when a target is highly
institutionally constrained. Going back to the theoretical section of this dissertation, we
are reminded that a highly constrained target may be unable to deliver on a promise
made when an agreement is signed (involuntary defection). Henceforth, it is not a
shocking revelation to learn that, out of twenty-nine cases in which a target is highly
constrained, twenty-two involve the EC. Given its complex ratification procedure, we
understandably see why it has a noticeable impact on the probability for American
negotiators to reach only some nominal success. However, we will see below that there
are other factors explaining why the negotiating success of Washington is lower than
their overall success‘03. The variable seems to have a more projecting impression when
the United States is confronted with other highly constrained targets. Generally, other
variables favor Washington when they are facing Argentina, Brazil, Spain and India, all
highly constrained targets. For instance, all those targets show some great trade
dependence towards the United States. Still, the US scores representing its level of
success are only 1.2 versus Argentina, 0.6 versus India, and 0 versus Spain. It is only
versus Brazil that US negotiators manage to get a better fortune than their overall
success, with a score of 2.0. Consequently, it is safe to assert that the level of domestic
institutional constraint is the main factor explaining why Argentina, India, and Spain

were able to deflect Washington’s pressure when confronted to Section (Super) 301.

 

‘03 The reader should be reminded that US overall success score is 1.44, while its
success versus the EC is only 1.25.

160

Section (Super) 301 is a tool made available to the USTR mandating him to
negotiate a settlement with a target before initiating sanctions. Hence, if the USTR
threatens sanctions or use of sanctions, it is a clear indication that bargaining attempts
have failed. In this respect, we have to expect that when no threat or use of sanctions
are implemented, it will only have some minor impact on bargaining success.
Conversely, if sanctions are set in motion, it symbolizes that US negotiators have failed
in their attempt to sway a target towards their demands through negotiations. For these
reasons, it is hypothesized that the threat of or use of sanctions by the United States is
an inept attempt that serves only to elevate a target’s resolve.

A first hint confirming this conjecture was found in Table 10. It shows that the
impact of retaliation on the mean value of the dependent variable is especially sharp
when it is implemented. On one hand, in the fifty-six cases when no retaliation or threat
of retaliation were put forth, the impact on the success score is minimal (from 1.44 to
1.59). On the other hand, when retaliation is implemented or a formal threat exists, the
success score plummets to only 1.0. Among all variables, this constitutes the most
profound impact. A second clue stems from the multinomial logit analysis. Table 11
displays that the variable is significant at the .005 level (P > [2} = .002) and the sign
of its coefficient is in the hypothesized direction. A third test confirming the role of
retaliation is constituted by the marginal impact of RETAL on the dependent variable.
As suggested, Table 12 clearly demonstrates that the marginal impact of retaliation (or
threat thereof) on the level of American bargaining success is limited. However, when
this tool is exercised, it is a good indication of a policy failure. It has a non-negligible

influence on all categories of the dependent variable. The predicted probabilities of

161

Success and Partial Success plunge by respective values of .105 and .209, while the
predicted probabilities of Failure and Nominal success soar to .097 and .518 from their
respective previous values of .028 and .273. It is interesting to notice that, among all
variables, the last two figures illustrate the largest marginal impact on the predicted
probability of Failure and Nominal Success.

It is also of interest that threat or use of retaliation have been used towards
specific targets. Not only did the United States use this policy instrument against their
three main trade partners, the EC (6 times), Japan (3 times), and Canada (twice), but
also in all cases involving China (twice) and Thailand (3 times). In the latter cases, can
we conclude that US negotiators have shown some exasperation indicating an
incomprehension of the way of "doing business" of these fairly recent trade partners“?
More studies on the specific trade relationship between these two countries and the
United States would be needed to reach a definitive conclusion.

The dummy variable DVCGOOD indicates the dollar amount of the goods and
services under dispute. An "important" case, for which the value of disputed goods and
services equals or exceeds 100 million dollars was coded a 1. Other cases were coded
as a O. The main assumption was that "important" cases were harder to solve to the
advantage of American negotiators. Therefore, we would expect a negative coefficient
for this dummy variable. The empirical estimate of the variable, as indicated by Table

11, corroborates the hypothesis: the coefficient of DVCGOOD is in the right direction

and it is highly significant (P > [z] = 0.039). Table 12, on the marginal effects of

 

‘04 The first case indicating the use of Section (Super) 301 against Thailand starts in
1989, while it only starts in 1991 for China.

162

regressors, also authenticates the hypothesis. It discloses results exhibiting that when
DVCGOOD equal its minimum value, the estimated probability of success for the
American negotiators, Pr (3), increases, thus indicating the negative relationship between
DVCGOOD and Success. The opposite is also valid, when DVCGOOD bears its
maximum value, the predicted likelihood of American bargaining success dwindles. The
consequence of this variable is not, however, as singularly robust on the probability of
American negotiating success as were the three previously examined variables
(ISSUE,INSCONST and RETAL). Its strongest marginal repercussion seems to be on
the propensity of achieving nominal success when the dummy variable takes on its
maximum value (.108). This indicates that when an issue under dispute involves a large
amount of money, an agreement is still possible, but its implementation presents many
difficulties. These cases are often complex and involve many players in the
implementation of the agreed upon outcome of an international negotiation. Despite a
moderate marginal impact of the dummy variable, we may reach the conclusion (mostly
because it is highly significant) that the value of the contested goods and services has a
consequential negative bearing on the propensity of American negotiators to attain their
aims.

To illustrate the economic necessity dimension of the theoretical model, I have
combined data on unemployment and inflation to create a misery index. It was
hypothesized in the previous chapter that in period of "bad times", US negotiators were
in dire need of international trade agreements in order to "ratchet up" their economy.
It is in fact the rationale behind the 1974 Trade Act. While the United States was losing

its hegemonic grip on the international economy, a new, more aggressive, international

163

trade policy was needed. The side effect of a new aggressive bilateral economic policy
was that it put the country at a certain disadvantage in international trade negotiations.
This desire of international measures to dampen "unfair trade" could also be perceived
by their foreign partners. Ergo, when a negotiator feels some sense of urgency in getting
a "quick fix" or an hasty agreement, he puts himself into a disadvantageous bargaining
position. This is, unfortunately, the situation in which American negotiators put
themselves in bad economic times, especially during periods of stagflation. In
consequence, we may suspect that the variable MISERY will not only be a significant
factor explaining US negotiators prowess, but also that it will have its most profound
impact when a combined measure of inflation and unemployment is high.

Table 11 shows that the variable is almost significant at the .1 level (P < [Z]
= .136). The marginal impact of the misery index uncovers more conclusive results.
By consulting Table 12, we can see that, as suspected, it has its most salient bearing on
the dependent variable when it takes on high values. It affects especially the propensity
for American negotiators to obtain the middle categories of success with a marginal
impact of .212 for nominal success and -. 169 for partial success. Figure 16 is even more
revealing. It shows that as the misery index increases, the respective predicted
probabilities of failure and nominal success soar, while the respective predicted
probabilities of partial success hit the floor. In sharp contrast the low values for the
misery index, where the predicted probabilities of nominal success and full success are
singularly similar, the gap between the two predicted probabilities of these two categories
of the dependent variable becomes extremely wide. These findings add to the perception

that in period of bad times, US negotiators are in a inconvenient bargaining position.

164

In light of these results, we may speculate that the mean level of success is low
for the first years covered by this data set, when unemployment and inflation were high.
Conversely, we may assume that American negotiators achieved more success in the mid-
eighties when the US economy was soaring. Table 13 indicates the correspondence
between yearly economic variables and yearly mean success score. The tables clearly
shows that when the MISERY index was at its highest level in between 1979 and 1981,
American negotiators achieved very few bargaining successes. On the other side of the
spectrum, American negotiators obtained their greatest success during the 1986—87 period
when MISERY fell to attain its nadir. Moreover, if we look at the data as a time series,
we see some definite trends. With very few cases being concluded in the first few years
covered by the data set (1976-78), American negotiators start out by obtaining a good
overall score of success. As the MISERY index converges towards its peak (1979-80),
a clear sign of stagflation, their level of success dwindles (from 3 to 1). The following
two years, even if the economy seems to slowly get a better, American negotiators seem
to still feel the pinch of the bad economic performances of the previous years and fail
in all of the cases concluded during those years. It is only from 1984, when finally the
MISERY index falls below the 10 points level, that, increasingly, Washington obtains
its best bargaining accomplishments, reaching its apex in 1987 (despite a good increase
in the MISERY index from the previous year). After an adjustment period in 1988
(when MISERY came down a little bit, while US negotiators obtain a little less success),
the dog years of 1989 and 1990, when the misery index once again climbed above the
10 points mark, had a slight negative impact on the bargaining fortune of the United

States. Finally, despite the fact that the MISERY index returned below the 10 points

165

mark (inflation was curbed while unemployment knew a slight increase), American
negotiators did not recover, in terms of their bargaining achievement, from the "bad
economic times" of 1989-90 for the rest of the period covered by this study.
According to those results, two conclusions can be reached in terms of the impact
of American economic performance on the US negotiators’ ability to settle for agreement
meting their objectives. First, the misery index has a salient effect on the middle
categories of the dependent variable, especially when the combination of unemployment
and inflation is particularly high. Second, there appears to be a lingering effect, such
that even when the US economy is recovering, it takes a few years before it can be
translated into bargaining success for American negotiators when resorting to Section

(Super) 301.

166

TABLE 13

Yearly Economic Data and Yearly Level of Success by American Negotiators

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Year (N) Unemployment Inflation Misery Index Yearly Score
1976 2 7.7 4.47 12.17 2
1977 1 7.1 6.41 13.51 3
1978 1 6.1 7.88 13.98 3
1979 3 5.8 11.17 16.97 1
1980 6 7.1 13.4 20.5 1
1981 2 7.6 9.2 16.8 1
1982 3 9.7 4.06 13.76 0
1983 1 9.6 1.6 11.2 0
1984 4 7.5 2.07 9.57 1.25
1985 8 7.2 .96 8.16 1.38
1986 6 7.0 -1.43 5.57 2.17
1987 2 6.2 2.3 8.5 2.5
1988 5 5.5 2.47 7.97 2
1989 6 5.3 5.18 10.48 1.5
1990 14 5.5 4.93 10.43 1.43
1991 3 6.7 2.1 8.8 1.33
1992 6 7.4 1.23 8.63 .83
1993 2 6.8 1.22 8.02 1
Mean - 7.0 4.4 11.4 1.44

 

 

 

 

 

 

 

167

The variable TBAL represents the trade balance between the United States and
the target. A positive balance for Washington is indicated as a positive value. It was
hypothesized that American negotiators would have more difficulty reaching their
objectives if they had a surplus of trade with the target country. The logical explanation
being that it is difficult for American negotiators to sway representatives of a target when
trying to convince them of unfair trade practice on their part, if the United States already
has a positive balance of trade with the target. American negotiators have a stronger
case to defend if their balance of trade with the target is negative. Hence, the sign of
the regressor in the statistical analysis should be negative: a positive balance of trade for
the United States is conducive to less bargaining leverage for US negotiators. Not only
does Table 11 displays a negative coefficient, but it also exhibits a high significance (P
> [z] = .012). Furthermore, among all independent variables, this variable
demonstrates the strongest impact on the predicted probabilities of American bargaining
success, as displayed by Table 12. The greatest impact is on the probability of success
(Pr = 3) when the variable is at its lowest value, with a marginal effect of .481. This
means that American negotiators have a very strong case to argue when they have a large
trade deficit with a target. We can even assert that the best predictor of full success for
the United States when using reciprocity and retaliation is the size of its trade deficit with
the target country (or group of countries). It is interesting to note, however, that this
does not apply to the prediction of partial success (Pr = 2) under the same
circumstances, as its marginal effect is negative (-.233). The main explanation for this
phenomenon is that without a large trade deficit, all else being equal, cases that are

predicted are predicted as a full success, would probably be coded as a partial success.

168

The variable is especially effective in assuring the transition between a partial success
and a full success. Nevertheless, the balance of trade has the hypothesized marginal
impact on the predicted probability of nominal success (-.233). All these results put
together, we have a good indication that the principal impact of a large trade deficit is
to decrease the possibility of failure, a nominal success, or a partial success in favor of
an all across board bargaining success. The variable has also an important impact when
it is at its maximum value with marginal effects of .19 and -.145 on the respective
predicted probabilities of nominal and partial success. Thus, even though its impact is
not as strong as a trade deficit, a trade surplus for the United States is a good indication
of the difficulty facing US negotiators in dealing with international actors when resorting
to Section (Super) 301. It affects particularly the ratification procedure of an agreement
in a target country, as suggested by the marginal impacts on the predicted probabilities
of nominal and partial success.

Figure 17 is a graphical representation of the marginal impact of the balance of
trade on the degree of bargaining success for the United States. It clearly indicates that
the predicted probabilities of failure and nominal success increases as the trade deficit of
the United States shrinks to a point of becoming a commercial surplus. The influence
is even greater on the predicted probability of success: the curve representing its
predicted probabilities has a very sharp drop, such that it is crystal clear that the chances
of total success are much higher as the trade deficit increases. The results regarding the
predicted probabilities of partial success are more puzzling: Figure 17 shows that the
probabilities of partial success are lower with large deficit than with small deficits, but

as the deficit goes towards a surplus the relationship is as expected, i.e., the predicted

169

probability of partial success decreases. Again, this is a possible indication a case that
would have been predicted as a partial success, without the impact of a large trade
deficit, is now predicted as a full success. Overall, the balance of trade between the
target and the United States may represent the best indicator of the fortune of Washington
when it goes ahead with instruments of reciprocity and retaliation.

If we look at the level of success of the United States with targets having a large
positive balance of trade with Washington, we also uncover interesting findings. It is
well acknowledged that, of all targets included in this study, the United States’ largest
trade deficit is with Japan. If we look back at Table 4, we can see that Japan is also the
target with which the United States has the best level of bargaining success. I will
discuss further the specificities of the structure of negotiation affecting bargaining
between the two countries, but we can already ascertain that the leading indicator of the
overwhelming level of success attained by American negotiators vis-a-vis Japan is the
balance of trade between the two countries. The United States also has large trade
deficits with Canada, Brazil, Korea, and Taiwan. Again, if we consult Table 4, we find
that the USTR achieves a reasonably good level of success when confronting those
targets. At the other end of the spectrum, we discover that the United States has a very
low level of success with countries with which it has a trade surplus. With none of those
targets, it has a degree of success higher than 1. If we include EC, which case is
discussed in the next paragraph, the top seven targets, with which the United States
enjoys the highest level of bargaining success, are also the exact same countries with
which it has large trade deficits! This constitutes a compelling evidence of the impact

of trade balance on US international trade bargaining fortune.

170

The EC represents the most engaging case. Prior to 1985 and after 1989“”, the
United States had a trade surplus with the European Union. During the years in
between, it had to contend with a very important trade deficit. We would therefore
suspect that the United States would have a better bargaining success with the
Community when it had a trade deficit. If we inspect the respective combined scores
indicating the level of US bargaining success when it had a trade surplus and compare
it to a similar measure when it has a trade deficit, we obtain a clear picture of the
situation: when the United States had a trade surplus with the European Union, it only
achieved a mean level of success of 1 (i.e. nominal success on average); while its level
of success climbs to 1.7 (i.e., almost partial success) when it has a trade deficit with its
European counterpart. These findings add to the confirmation of the prediction that trade
balances have a consequential influence on the degree of bargaining success in
international trade and investment negotiations.

The bivariate analysis, the ordered multinomial logit test, and the marginal impact
of the independent variables on the dependent variable largely confirming the theoretical
propositions, logical underpinnings, and hypotheses of the dissertation. Nevertheless,
two variables, presidential approval and trade interdependence as controlled by GNP,
turned out to be inappropriate elements of analysis. While it was highly significant when
it involved a target (IN SCON ST), the level of institutional constraint in the US domestic
institutions did not yield significant results, but we were still able to unearth an

interesting finding: in may be true that division within Congress has no real bearing on

 

‘05 More exactly, the EC and the United States had a roughly equal balance of trade
between them in 1989 and 1990.

171

international trade agreement, but division between Congress and the Presidency affects
the bargaining leverage of US international negotiators. A combined measure of inflation
and unemployment (Misery index) rate has been used to unveil the impact of the
American economy on its US international negotiators. It revealed two absorbing results:
results: when the misery index is high, it negatively affects the US bargaining
performance, and there exists a certain lag effect, such that it takes two or three years
for an improving economy to translate into better bargaining fortune. One of the most
engaging finding of the empirical analysis is the relationship existing between trade
interdependence and outright US bargaining success: when the asymmetry of trade was
to the disadvantage of US negotiators, their increased level of resolve helps them
accomplish more full bargaining success, but greatly deflates their chances of gaining
partial or even nominal successes. Their "all or nothing" strategy seem to work
adequately in most cases, but sometimes falls outside a target’s "win-set" and Washington
has no choice but to resort to other policy instruments than negotiations if they want to
achieve their international aims. Confirming results obtained by Bayard and Elliot
(1994), this analysis demonstrates that issues for which direct retaliation is possible, such
as tariffs, are more easily resolved to US satisfaction than more complex issues of
intellectual properties or governmental procurement. The analysis also divulges that a
threat or use of retaliation or reciprocity while negotiations were under way, it
constituted a net indication the American negotiators were unlikely to reach their
bargaining aims. As hypothesized, the value of the contested goods and services have
a negative impact on the propensity of bargaining success for Washington. Finally, the

analysis has revealed some compelling evidences regarding the balance of trade between

172

the United States and a target. Trade balance appears to be the single most important
factor explaining US bargaining success when using Section (Super) 301.

What does all of this have to say about the theoretical propositions and parameters
of the formal model? The next section tackles this question. It is followed by the
influence of these factors on the bargaining results obtained by the United States when
confronted with its three major trading partners, that is, Canada, Japan, and the
European Union. Then, an alternative model that assumes compliance instead of
interdependence, in an effort to uncovers the main elements explaining the fortune of
American negotiators when they exercise their reciprocity and retaliation options, is

tested.

iii- Elements of analysis of the theoretical model and parameters of the formal model

A closer look at the model reveals interesting results regarding the differential
impact of domestic and international factors.

One of the most striking result of the ordered multinomial logit analysis is the
impact of the interdependence elements of analysis (DEP and DEPGNP). When trade
interdependence was calculated according to the share of countries’ GNP accounted for
trade with a specific target (DEPGNP), it showed to have little impact on bargaining
leverage. It was not the case when we also consider respective share of exports in transit
between two countries (DEP). Indeed, this proved to be highly significant, but the sign
of the regressor was in opposition to what was hypothesized. It was speculated that the
reason for this puzzling result could be due to high resolve exhibited by negotiators when

suffering from trade asymmetries. Nevertheless, we must perforce admit that the

173

interdependence propositions did not pass the test. Still, it has uncovered unsuspected
and non obvious results, which, in terms of a research agenda, opens some brand new
doors.

Nevertheless, not all international variables behaved erratically. The variable
entailing the type of issue discussed (ISSUE) has shown to be significant at the .01 level
and had an important impact on the predicted probabilities of the dependent variable.
However, this variable did not directly test the proposition regarding the structure of
trade between the negotiating parties; more work remains to be done in this direction.
At least, this statistical analysis demonstrates that a threat of retaliation is much more
credible when it involves a tariff or a quota than when it is related to issues of
intellectual property, and health and environmental standards.

As for the institutional constraint proposition, it appears that, under the specific
context for which it was tested, the proposition is mostly reliable when applied to the
target of a threat of retaliation. In fact, the empirical analysis showed very weak
support, in terms of its significance, for the influence of institutional constraint on
American negotiators. Nevertheless, its marginal impact on the dependent variable was
relatively important and there are some good signs indicating that institutional division
between Congress and the Presidency affect US bargaining leverage. The results
regarding the level of domestic institutional constraint in the target countries are more
encouraging. It seems to be true, at least in this specific test, that, for a target, a large
level of domestic political constraint is an added leverage in international trade and
investment negotiations: American negotiators obtained, ceteris paribus, much less

success when facing highly divided political systems.

174

The two variables serving as indicators of economic necessity demonstrated
promising results. According to the empirical results of the test, the rate of success
obtained by Washington appeared to be influenced negatively by its misery index, the
value of disputed goods, and the use or threat of retaliation. First, in periods of high
unemployment rate and inflation, American negotiators may have been in a weakened
position because they counted on an agreement to act as a panacea for their economic dry
spells. Hence, under those circumstances, they preferred to sign an imperfect deal in a
short order, instead of waiting longer and go ahead with their measures of reciprocity
and retaliation. Second, results clearly indicate that the value of the goods under
contention have a profound impact on bargaining success. These results seem to indicate
that not only is it difficult to get a satisfying agreement when the dollar amount involved
in the dispute is immense, but also is an indication that American negotiators would often
be satisfied with a nominal success, instead of continuously negotiating for a better
agreement. As for the political necessity element of analysis, the use of or the threat of
retaliation is a good indicator that the negotiations are going nowhere for American
representatives, such that they are very unlikely to reach even some partial success.

The societal support element of analysis was the hardest to test empirically. The
two variables were chosen to represent the best proxies that could be used to unveil the
statistical force of the American societal support. The trade balance between the United
States and target countries is probably the best indicator of the bargaining fortune of
American negotiators. It seems to be true that large trade deficits attracts a lot of
attention from the public, which in turns accentuates the resilience of American

negotiators. It is with a high degree of skepticism that I included in the analysis the

175

proposition that presidential approval plays a role in international economic negotiations,
especially when issues discussed are not well-known by the general public. The
statistical analysis indicated indeed that this variable had no bearing on the level of
success attained by US negotiators. This mixed result regarding societal support warrants
some further research. More particularly, students of international trade negotiations will
need to use some imagination to determine which adequate, empirically testable,
variables could best represent this element of analysis.

Propositions 9, 10A, and 10B indicated the expected relationship between the
international costs for a breakdown in the negotiations and interdependence issues. There
is scant evidence in the empirical test to support the credentials of this parameter of the
formal model. This occurred, for different reasons. First, as indicated previously, the
data set used in the empirical test does not do justice to a test for interdependence. The
results obtained from the data set collected to test the success of American negotiators
for using Section (Super) 301 may not be translated directly to other instances of
international trade bargaining with much wider scopes. One can still suspect issues of
interdependence to be consequential in these last instances. Thus, the relationship
between interdependence and international costs for a break down in the negotiations
should not be entirely discounted. Second, in the specific case of the use of Section
(Super) 301 by the United States, the real relationship might be between the perceived
costs associated with a breakdown in the negotiations by the targets’ negotiators and their
level of trade dependence towards Washington. Indeed, when the United States Trade
Representative (USTR) opens a "case" there is already an acquiesced willingness to

suffer the costs involved with an impasse in the negotiations. Thus, the onus of a break

176

down is put on the other side. The compliance model will test this proposition. Third,
and finally, the empirical test did not consider propositions 10A and 10B. In the short
term, it was not possible collect data on complementarity of trade. Nevertheless, it could
easily be argued that the type of dispute has a important effect on the costs associated
with an deadlock in the negotiations. That is, at least, what the empirical results seem
to demonstrate. The international costs for a breakdown were much higher for the target
when the United States could use direct and commensurate trade retaliation if the
negotiations failed. That is precisely what the variable ISSUE was testing: when the
point of contention was a quota or a tariff, Washington could easily retaliate with
commensurate measures of reciprocity. This acted as a sword of Damocles hanging over
the head of the negotiators of target countries.

More positive comments can be made regarding the information parameter of the
bargaining model. Proposition 11 indicated that the more domestic institutional
constraint negotiators are faced with, the less information about their type they convey
to other negotiators. Furthermore, proposition 8 said the less information about their
type negotiators convey to the other side, more bargaining leverage they have. Seen
from this perspective, it is not surprising that the variable DIVIDED, a measure of
American institutional constraint, displayed mitigated results: the use of Section (Super)
301 is in itself a mandate to initiate retaliation. However, it leaves a door open for
further negotiations before the implementation of the measure. In the meantime, if the
American President and Congress do not see eye to eye on which general international
trade politics to hold, it sends some mixed messages to international targets. This

situation is more likely to occur when the Presidency and Congress are controlled by

177

different parties. There seems to be some international asymmetric information
advantage for American negotiators, not in a divided Congress. If US domestic
institutions are undivided, it leaves no doubts in the minds of opposing negotiators.
Hence, the latter have full information--even though they might not like the information
they receive-- about the type of negotiators they are facing. Conversely, when a target
is highly institutionally constrained, American negotiators do not have the luxury of
knowing the type of negotiators they are facing. This explains why the variable
representing a target’s institutional constraint was highly significant and had such an
impact on the predicted probabilities of the dependent variable. Americans, according _
to the test, have a much harder time reaching their goals when they are facing a target
which embodies a highly diffuse decision process. Henceforth, it is possible to conclude
a strategy based on the concealment of private information, especially when it is
institutionally induced, has a portentous impact on bargaining strategies and negotiation
outcomes.

With the exceptions of the effect of presidential approval (APPROV), all
dimensions (Political and economic necessity, and Societal support) and variables
(DVCGOOD, RETAL, TBAL, and MISERY) representing the domestic costs for delays
demonstrated good results in the empirical test. More particularly, these results indicate
that American negotiators were more inclined to settle for nominal or partial success
instead of holding out for a better outcome when the following conditions occurred: the
issue at stake was very high (DVCGOOD), their misery index was high (MISERY), they
had a large commercial surplus with the target, and they relied on threat or use

retaliation to attain their objectives. In all those situations they were facing high

178

domestic costs for delays in the negotiation and preferred to settle early instead of
holding out.

This section provided some general comments in regards to the effectiveness of
the elements of analysis of the theoretical model and the parameters of the bargaining
design. Now, we must turn to a more specific, more descriptive, type of analysis: the
role played by the independent variables of the statistical analysis in explaining the
bargaining success of Washington with its three main trade partners (i.e. Canada, Japan,

the European Union). That represents the task of the next section.

iv. From the General to the Specific: US Bargaining Success Versus Canada, Japan and
the European Union.

In exploring the bargaining relationship between the United States and its three
main international economic partners, three questions come to mind: 1) What is the
model’s performance in evaluating the level of success of American negotiators with
those targets? In other words, is there a high level of correspondence between the
"coded" and "predicted" scores" for each of those targets? 2) What are the main elements
of analysis that best explain the level of success obtained by American negotiators when
dealing with those targets? 3) How can we assess the American negotiators’ performance
with each of those targets, given the specific structure of negotiation affecting the
bargaining process between the United States and those targets? Each of these questions
will be dealt with in tums..

Tables 14, 15, and 16 represent the respective performance of the ordered

multinomial logit model with each of Washington’s three main international economic

179

partners. They exhibit the type of issue over which negotiators bargain, as well as coded
and predicted scores for the dependent variable. The last column indicates with an "X"
cases correctly predicted. The first striking revelation of these tables is that the model
performs much better in predicting the outcome of a negotiation when it involves Japan
(75 % correctly predicted) rather than Canada (40%) or the European Union (45 %). The
second finding is that in the great majority of the cases, the model’s prediction is never
"off the mark" by much (e.g. a predicted score of 1 is coded as a 0, or a predicted score
of 3 is coded as a 2)‘06. Finally, a third finding is that there is no skewness in the
distribution of the cases that were wrongly predicted, i.e., there is no tendency to err in
the same direction (no under or over-estimation). To add to this general interpretation,

we can look at the specific model’s performance regarding each target.

 

 

TABLE 14
Interdependence Model Performance for Cases Involving Canada
Casenum Issue Coded Predicted Identical
2 Eggs 3 2
14 Broadcasting 0 1
25 Subway Cars 0 2
42 Fish 2 2 X
66 Beer 1 1 X

 

 

‘06 In only one case out of thirty-seven involving those targets has the model entirely
erred in the wrong direction (case #25 involving subway cars with Canada). For the
entire data set, this situation occurred in only one other case (case #30 Soybean oil and
meal, Portugal).

180

Table 14 shows that the model does not perform well in predicting the level of
American bargaining success when the target of Section (Super) 301 is Canada. In only
40% of the cases does the model predict adequately the value taken by the dependent
variable”. This is less than the overall accuracy of the model (55%) and barely more
than the naive model (37 %). Also, there doesn’t seem to be any correspondence between
the type of issue involved and the model’s performance”)?

The interdependence model performs very well in predicting cases involving
Japan. Nine out of twelve cases are predicted accurately (75%) and those that are
wrongly predicted are close to predicted values”). This is far better than the overall
level of success of the model. It is interesting to note that all cases related to the service
or agriculture sectors are correctly predicted; compared to only four out of seven cases
related to the secondary sector. It is especially true of the high technology sector
(semiconductors and supercomputers), for which the model was too optimistic in
predicting American’s success. This might be an indication of the model’s
underspecification, such that a special dummy variable representing "high-tech" cases

could be included in the model.

 

‘07 The compliance model tested in the next section is even less accurate; predicting
only one case involving Canada correctly. Nevertheless, we should take these results
with a certain " grain of salt", given the size of the sample.

‘08 Even though it is true that the model has especially erred for the only case
involving the secondary sector.

“’9 Strangely enough the CCP for Japan in the compliance falls to 50%. This could
suggest that in the case of Japan, the interdependence model is more adequate.

181
TABLE 15

Interdependence Model Performance for Cases Involving Japan

 

 

Casenum Issue Coded Predicted Identical
11 Thrown Silk 3 2
12 Leather 2 2 X
16 Cigars 1 1 X
18 Pipe Tobacco 1 1 X
27 Footwear 2 2 X
35 Semiconductors l 2
37 Cigarettes 3 3 X
52 Citrus 3 3 X
55 Construction 2 2 X
60 Satellites 3 3 X
61 Supercomputers 2 3
62 Wood Products 2 2 X

 

182
TABLE 16

Interdependence Model Performance for Cases Involving the European Union

 

 

Casenum Issue Coded Predicted Identical

3 Egg Albumin 2 1
4 Canned Fruits and Vegetables 1 2
5 Malt 0 1
6 Wheat Flour 0 0 X
7 Canned Fruits 1 1 X
8 Soybeans 1 1 X
10 Steel 2 1
15 Wheat 1 l X
20 Sugar 0 1
21 Poultry 1 1 X
23 Pasta 2 1
24 Canned Fruits and Raisins 1 1 X
41 Corn, Sorghum, Oilseeds 3 2
46 Meatpacking 1 2
48 Beef 0 l
49 Soybeans 1 1 X
56 Fabricated C0pper 3 2
57 Canned Fruits 2 1
67 Corn, Sorghum, Oilseeds 2 2 X

w 1 L X

 

183
The model’s performance in cases involving the EC is a little worse than its

“0. We first notice that most

overall efficiency, predicting half of the cases adequately
cases (18 out of 20) involve agricultural products. This is not a striking revelation given
all the histrionics that went on between the US and the EC during the Uruguay round
negotiations. Farm subsidies, especially in France, is a "way of life" that most
cultivators are very reluctant to give up. In the two cases involving the secondary sector
(#10 steel, and #56 fabricated copper), the model underestimated the level of success for
American negotiators. Once again, this could warrant the creation of a new variable
discriminating between different sectors of the economy: agriculture, primary sector
(others), secondary sector (excluding hi-tech), hi-tech, services. The problem with this
potential variable is to theoretically determine an ordinal impact of its categories‘“.

In an attempt to answer questions 2) and 3) at the top of this section, a table
reporting a crosstabulation of the significant variables of the interdependence model with
different targets of Section (Super) 301 was constructed. The five columns of Table 17
represent the three main US trade partners (Canada, BC, and Japan) and separate

categories for the Asian/Pacific bloc (except Japan) and residual targets. The significant

variables are represented in the rows, as well as the number of cases involving each

 

”0 The compliance model does worse with a 35% CCP.

1” For instance, we could put forth a strong argument indicating that hi-tech cases
are harder to solve than those in other sectors, but how could we discriminate, for
instance, between the potential impact of agricultural products versus the service sector?
An alternative, would be to create dummy variable for each sector of the economy. The
problem with that approach is that we end up with a very cumbersome model in which
the number of cases coded as a one (for instance a "1" would indicate the hi-tech sector,
and a "0" would indicate otherwise) would be very small, not to mention an important
loss of degrees of freedom.

184

category of targets. A "plus sign" in the table marks an asymmetrical advantage for the
United States, given a specific variable. A "minus sign" indicates a structural bargaining
advantage for the target”. An "equality" sign symbolizes that there is no clear
bargaining advantage. The row named "overall" is an assessment of the impact of the
general structure of negotiation (i.e., considering all significant variables) on the
bargaining leverage of US negotiators vis-a-vis each targets. Once again, a "plus sign"
signify an overall bargaining advantage for the United States. The bottom row displays
the mean level of success achieved by American negotiators when facing each target.
Two more explanations regarding Table 17: 1) DEP’s (trade interdependence) impact is
assessed in terms of its statistical result, not in terms of its hypothesized impact. In other
words, cases in which the US was more dependent on a target, rather than the reverse,
were considered as an indication of US bargaining strength. 2) The impact of RETAL
was not entered (use of or threat of retaliation), as this variable is not a direct indicator

of the negotiation structure between the United States and a target.

 

”2 As an illustration, a minus sign at the intersection of the EC as a target and
INSCONST indicates that the EC is highly institutionally constrained, it constitutes a
bargaining handicap for the United States. A large asymmetry is indicated by two plus
signs or two minus signs.

185
TABLE 17

Assessment of US Bargaining Success Vis-a-Vis Different Targets

 

 

 

 

 

Variable/Target Canada EC Japan Pacific Others
(N) 5 20 12 23 15
DEP (-) (+) (-) (-) (9
ISSUE (=) (-) (-) (-) (-)
INSCONST (+) (-) (+) (+) (+)
DVCGOOD (+) (=) (=) (-) (+)
TBAL (+) (+) and (-) (++) (++) (=)
Overall (++) (=) (+) (+) (=)
Score 1.2 1.25 2.08 1.43 1.27

 

 

 

 

186

Table 17 contains a wealth of information. To reveal its richness I discuss below
each of its elements. But first, a few general comments are in order. The overall
negotiation structure appears to favor the United States, with the exception of its dealing
with the European Union or residual targets. Nevertheless, the connection between
bargaining leverage derived from an asymmetrical structure of negotiation does not lead
directly to successful outcomes. Indeed, on average the level of bargaining success
attained by Washington is low. Should we conclude automatically that the Office of the
USTR is "chasing phantoms" when resorting to Section (Super) 301 (Noland, 1997)?
Maybe not necessarily. This situation brings water to the argument that a sample
selection bias negatively affects US negotiators bargaining fortune‘”. Cases that are
easier to solve are dealt with informally; only "hard to crack" cases make it to the formal
stage. Despite the sample selection bias, it is feasible to make an assessment for the
level of US bargaining success in regards to different targets. For instance, given the
favorable structure of negotiation vis-a-vis Canada, one cannot help but be surprised that
the USTR would, on average, have so few successes with its Northern neighbor. Let’s
delve into the details of Table 17 to find some explanations to these engaging

phenomena.

0 Canada
The negotiation structure, according to the variables used in this study, endows

the United States with a stark asymmetric bargaining advantage. In Table 17, the only

 

“3 In all fairness to Noland, it must be indicated that he acknowledges this negative
bias and circumvents it in a research design that is different than the present.

187

variable that is indicated as giving a bargaining advantage to Canada is the trade structure
between the two countries. Still, it is noted that when the impact of this variable is
examined (see especially Table 12 and Figure 13) that it is only when the United States
is more dependent on the other side, rather than the contrary, that the variable has a
salient impact on the outcome of a bargaining session. In consequence, the lack of US
trade dependency is not transferred directly into a bargaining advantage for Canadians.

Three of the five issues involving negotiations between Canada and the United
States were related to tariffs or quotas. There is, therefore, no clear-cut advantage that
can be bestowed on either side according to the results displayed by this variable.

The level of institutional constraint in a target’s domestic political system has a
significant impact on an international trade bargaining outcome. In this case, a low level
of institutional constraint in Ottawa constitutes an added bargaining bonus for American
negotiators. The low level of institutional constraint in Canada (Parliament is controlled
by the Executive) implies that Canadian negotiators bargain in the name of the Prime
Minister, who in turns, can easily get ratification when a deal is struck with the
Americans. This negates the institutional veto power of Parliament. There is therefore
no uncertainty in the minds of American negotiators regarding Canadian ratification
process, and it constitutes, de facto, an international bargaining weakness for Canada.

Estimated values of contested goods and services in negotiations involving the two
North American trade partners never surpass the 100 million dollars mark. Those
"small" cases can more easily be resolved to Americans’ satisfaction than cases entailing
consequential amounts of money. Consequently, this adds to the probability for

American negotiators to sign an beneficial agreement.

188

For the entire period covered by the data set, Canada had a trade surplus with its
Southern neighbor. The US trade deficit with Canada was not of the magnitude it had
with Japan, but it grew significantly during the 1980’s. This drew significant attention
from Congress. This increased awareness logically translated into tougher, more
adamant, negotiating stances. Since the empirical results show that trade balance is the
single most important factor leading to American bargaining success, the trade deficit
with Canada endows the United States with a consequential bargaining advantage.

My general evaluation of the structure of negotiation between Canada and the
United States is that Americans have a very significant advantage. All important
variables point in that direction. It is therefore a striking revelation that American
negotiators did not obtain more success Canada. What could account for this discrepancy
between negotiation structure and American bargaining success with Ottawa? We can
investigate a few hypotheses. A long lasting friendship between the two neighbors may
suggest that there exists a "special relationship " between them. Close contacts between
them set up a routine in which trade partners on both sides of the table develop a high
level of trust for their counterparts. It is beneficial in an iterated relationship to make
small concessions in order to avoid a breach of trust. Also, throughout those years of
close friendship, in the eyes of American officials, the Canadian economy was seen as
an extension of the American economy. Thus, when we explore trade relations between
Canada and the United States, we might not be studying international trade politics, but
some kind of trade politics that falls in between domestic and international trade politics.
The synergy between Canadian and American business is another element suggesting a

special relationship between the two nations. Approximately 40% of Canadian

189

manufacturing output is owned by American companies (Williamson, 1993: 34). Hence,
when resorting to Section (Super) 301, the Office of the USTR is falling prey to
pressures from specific industries, but it cannot be too forceful in its demand in order
to avoid endangering an overall trade relationship, which would not only strain what is
considered a special friendship, but also ultimately affect numerous companies doing
business in Canada. By being too inflexible, American negotiators could simply shoot
themselves in the foot! Another element of explanation is that in multilateral trade
negotiations, American officials need strong support from their best ally, this ally being
Canada. For instance, there is more at stake on the international scene when haggling
over multilateral reduction of agricultural subsidies than a bilateral dispute with Canada
over beer distribution. By pushing too hard for a resolution of a bilateral issue with
Canada, even with a clear bargaining advantage derived from the structure of negotiation
between them, American officials put themselves at risk of losing some consequential
backing in the international trade arena.

In itself, the study of this "special relationship " requires more consideration and
could be the subject of an entirely new dissertation. Nevertheless, there are enough
evidences to suggest that it can account adequately for the discrepancy between the
structure of negotiation, which indicates asymmetrical bargaining leverage in favor of the

United States, and the relatively low level of success obtained by Washington.

0 The European Union
The structure of negotiation between the United States and the European Union

is one based on asymmetry. My empirical results showed that the most discriminating

190

and important factor affecting an outcome of international trade negotiation between them
is their trade balance. When the United States has a trade surplus, the advantage is in
the Union’s court. Conversely, in periods of trade deficit for Washington, American
negotiators accrued bargaining leverage‘”. When discussing the impact of trade
balance on bargaining outcome, I supplied a demonstration of its bearing on negotiations
between the European Union and the United States. The best American successes
happened between 1985 and 1989, when they had a trade deficit with the EC. Now that
they have redressed the situation, we may expect negotiating difficulties for American
representatives when dealing with their EurOpean analogues.

Other variables representing the negotiation structure between the two
international entities cancel each others out. The main American advantage when
resorting to Section (Super) 301 against Brussels stems from its trade dependency
towards the Union. This factor is significant enough to upset the disadvantage it gets
from the type of issue they haggle over and the Union’s level of institutional constraint.
When trade interdependence between a target and the United States indicates an
asymmetrical disadvantage for Washington, its negotiators adopt a "take it or leave"
bargaining position. Table 12 showed that when American negotiators utilize such a risk
acceptant strategy, they are more likely to achieve full bargaining success to the
detriment of partial or nominal success. On average, this endows them with more
bargaining leverage with the European Union, a target towards which they are highly

dependent for their international trade output.

 

“4 This explains why I entered a " + " and a "-" to describe the impact of the variable
on trade relationship between them in Table 16.

191

The counterbalance to the impact of trade interdependence is provided by the type
of issue negotiated with the EC and the latter’s level of institutional constraint. Almost
two-third of the disputes between the EC and the United States involve non quota or
tariffs related issues. There is no entrenched precedents set to help negotiators find a
suitable compromise for non-border problems such as health and security issues or
governmental procurement. The successive rounds of the General Agreements of Tariffs
and Trade (GATT) did not provide an adequate framework to steer bilateral negotiations
regarding those impediments to trade. This situation creates a void that impede a
successful conflict resolution. The consequence is that American negotiators cannot point
to a specific set of international standards when accusing an international target of unfair
trade policies. It is especially difficult with the European Union at the time when its
members are themselves trying to come to grip with a set of standards to resolve those
issues.

This leads to the second structural bargaining aspect for the EC. As a diffuse
decision-making community, the European Union sends a signal of division to American
negotiators. Skillful European negotiators can use this internal division on the
international stage. They can haggle for less concessions to American negotiators by
pointing to the shadow of a failed ratification. Facing the prospect of an agreement that
would not gain the favor of EC members, American negotiators have an incentive try to
settle for at least a partial success. However, this may not be the avenue taken by
American negotiators according to the results of Table 12. It shows that when a target
is highly institutionally constrained the predicted probability of partial success decreases

while the predicted probability of nominal success increases. This is an indication that

192

American trade officials may have not been inclined to make more concessions even
when facing the prospect of "involuntary defection".

The last significant variable included in this brief analysis is the value of the
contested goods and services. There is no aggregate bargaining advantage derived from
that variable. About 50% of the cases debated between the United States and the EC
surpass the 100 million dollars mark. It is therefore not possible to grant a bargaining
advantage to either of the international entities.

An overall assessment of the structure of negotiation between the EC and the
United States leads to the conclusion that one side gains a slight advantage when it has
a trade deficit with the other side. Given that in the majority of the cases (12 out of 20),
the United States has an important trade surplus, it is not surprising to find out that its
aggregate level of success vis-a-vis Brussels is slightly below its overall level of success
(1.25 compared to 1.44). Therefore, we can conclude that the level of success obtained
by American negotiators when facing the Union’s trade representatives is in accordance

with the structure of negotiation affecting their bargaining encounters.

0 Japan

The results of N oland’s analysis on the attention given by the Office of the USTR
to international trade targets indicate that Japan is "singled out, beyond what would be
expected on the basis of its economic characteristics" (1997: 366). Noland also
demonstrates that US trade actions were efficacious when Japan is targeted (Ibid: 384).
His results go side by side with the premisses of this dissertation. Not only is the

structure of negotiation is favorable to the United States, but also Japan is often the

193

perfect scapegoat for American economic woes in the eyes of the public. It is therefore
logic to believe that the Office of the USTR would single out Japan and would also
achieve a high level of success when targeting Tokyo. Indeed, a large of number of
cases in the data set implicates Japan and American negotiators to have on average their
best successes when confronted with the Pacific country. Let’s explore the structure of
negotiation between the United States and Japan in more details.

Two factors (trade interdependence and issue type) that play in favor of Japan
when facing American threat of trade retaliation are not as significant as two other
factors (institutional constraint and trade balance) affecting negatively its bargaining
power. The first of the two favorable factors, trade interdependence, displays a
relationship in which Japan depends more on the United States for its international output
than the reverse situation. We have seen, however, that in those circumstances, the
variable does not have a large impact on American bargaining success. Thus, we can
only confer a slight bargaining advantage to Japan on the basis of this variable. The
second factor leading to a certain bargaining advantage for Tokyo, the type of issue
involved in a dispute with the United States, has a more salient impact on the dependent
variable. The majority of trade disputes (8 out of 12) included in the data set between
Japan and the United States are not related to quotas or tariffs. As indicated before,
these issues are harder to solve to Washington’s satisfaction and, consequently, it confers
to Japan a certain bargaining advantage.

As was the case with Canada, the lack of institutional constraint in Japan
facilitates the bargaining situation for American trade representatives. Due to Japan’s

parliamentary system, its trade representatives cannot use a " good cop, bad cop " strategy

194

to encourage American negotiators to think that an agreement will not be ratified by the
Japanese Diet115 . This bestows American negotiations with a bargaining advantage.

In his recent study of the politics of the USTR, Noland (1997) argues that the
most important aspect attracting the Office’s attention is the balance of trade between the
United States and a foreign nation. This coincides with the empirical findings of this
dissertation. In fact, an American trade deficit with a target is the single most important
variable leading to a full American bargaining success (See Table 12). For the entire
period of the data set, trade balances between Japan and the United States were highly
skewed in favor of the former. This serves as a good explanation why Japan attracted
the attention of the Office of the USTR and that many actions were taken by this office
against Japan. More importantly, the American trade deficit served as a leitrnotiv for its
official when trying to redress a situation they considered as unfair trading by the
Japanese. This was the best argument Americans could use during a bargaining session
with Japan and it serves as the best indicator of the high level of success they have
achieved.

According to the structure of negotiation between Japan and the United States
described above, it is safe to say that the level of success achieved by USTR officials is
commensurate with the bargaining leverage they had. The argument is not that there was
prima facie evidences of unfair trade practices on the part of the Japanese, but that
Tokyo’s trade surplus with the United States was by itself enough to warrant all the

attention it received from American trade officials. Other variables affecting the

 

“5 We must be careful with this interpretation. A lot of institutional constraint in
Japan comes from within the ruling party. As constructed, the institutional constraint
variable does not account for intra-party dissension.

195

structure of negotiation also tend to favor American’s bargaining strength, such that it
is not surprising that the empirical results show that the Office of the USTR had its best

"success stories" when resorting to Section (Super) 301 against Japan.

0 Asian/Pacific Bloc

At the beginning of the chapter it was noticed that the United States had the
highest level of success when using Section (Super) 301 against Asian countries. They
have especially a high success rate when confronted with Japan, Taiwan, and Korea (See
Table 4). The answer to this phenomenon can be found in the negotiation structure
affecting bargaining leverages. The case of Japan was addressed above and I now turn
to the discussion of other Asian/Pacific targets”.

Table 17 shows that overall, the United States derives a bargaining advantage
from its structure of negotiation with the Asian/Pacific bloc. However, its overall level
of success does not correspond directly to its bargaining leverage. One should not jump
to any conclusions though before we inspect the data and results carefully. Particularly,
close attention must be paid to the results obtained with rapidly industrializing Asian
countries (Taiwan and Korea) during the period of the study. These two countries alone
constitute 56% of the cases involving the Asian/Pacific bloc. It can be suspected that the
negotiation structure between these two countries and the United States is very similar

to that between Washington and Tokyo. Consequently, it is expected that the resulting

bargaining success for American negotiators will be identical in both instances. The

 

“6 Because of the specificity of results implicating Asian/Pacific targets, the
discussion of these results are separated from the discussion of other residual targets.

196

success rate should drop dramatically for other Asian/Pacific targets (China, India,
Thailand).

In all cases involving the Asian/Pacific bloc the structure of trade dependence
suggests a bargaining edge for the targets of Section (Super) 301. Results differ for the
type of issues being negotiated. For Korea and Taiwan, there is an equal split between
cases involving tariffs and quota, and other issues. No clear bargaining advantages are
evident. For other Asian/Pacific targets, only 30% of issues are related to tariffs and
quotas. This indicates a bargaining advantage for those targets. For the institutional
constraint variables, results go in the same direction. Taiwan and Korea are not highly
institutionally constraint, which gives a bargaining advantage to the United States. As
for other Asian/Pacific targets only India is considered as institutionally constrained (5
cases), while it is not the case for China and Thailand‘”. Hence, there is no clear
bargaining advantage. Once again, similar results are obtained for the value of the
contested goods. In roughly 50% of the cases involving Korea and Taiwan, the value
exceeds 100 million dollars, while the figure is 80% when it implicates China, Thailand
or India. In the first instance there was no clear bargaining advantage, while in the
second instance, it confers an important advantage to the targets. The starkest difference
is in regards to trade balances between the United States and these targets. On average,
Washington’s trade balance with the Asian/Pacific bloc is -4.515 billion dollars. If we
dig a little deeper into the data, we find out that the mean deficit swells to -7.019 billion

dollars with Korea and Taiwan, while it falls to only -1.206 with the other countries.

 

”7 This could be only coincidental, but this could explain why the United States has
a slightly lower success rate with India (0.6) than it has with China and Thailand (1).

197

Hence, the trade balance factor confers to the United States a very consequential
bargaining leverage against Korea and Taiwan, and no real advantage against China,
Thailand, and India.

In summation, the type of issue being negotiated and its value do not confer any
bargaining advantage to either side, while there is a slight advantage conferred by the
trade interdependence variable to Korea and Taiwan. However, the most important
variables, institutional constraint and trade balance, confer to the United States a very
significant bargaining advantage. Overall, we can conclude that the Office of the USTR
is in the driver’s seat when the resort to Section (Super) 301 against Korea and Taiwan.
This is in accordance with the bargaining success obtained against these targets (1.92).
These results are very similar to the discoveries regarding the relationship between Japan
and the United States.

There exists an important reversal of the situation when the targets are China,
Thailand, and India. Two factors give either a slight advantage to the United States or
unveils a symmetrical relationship: trade balance (slight advantage) and institutional
constraint (symmetry). Other factors demonstrate a favorable bargaining advantage to
the targets: trade interdependence (slight advantage), type of issue (significant
advantage), and value of the contested goods and services (significant advantage).
Overall, the structure of negotiation plays against the United States. The empirical
results demonstrate that its disadvantageous bargaining leverage lead to poor bargaining
successes for the United States: on average the bargaining success score for the Office
of the USTR when using Section (Super) 301 against China, Thailand, or India is a

dismal .8. This is, on average, even worse than a nominal success.

198
O Other Targets

To complete the discussion of structure of negotiation between the United States
and different targets, I now turn to cases not already covered. The "residual" cases
involve Brazil (5 cases), Argentina (5 cases), Guatemala (1 case), Norway (1 case), the
Soviet Union (1 case), Portugal (1 case), and Spain (1 case). The overall structure of
negotiation with these targets and the level of success obtained by American trade
officials are very similar to the findings that were uncovered when discussing the results
obtained in regards to the European Union. There exists a slight advantage for the
targets when we consider factors such as trade interdependence and type of disputed
issues, while there is no clear advantage in terms of trade balance“. Two factors,
institutional constraint and value of contested goods and services, grant the United States
with additional bargaining resources”. An overall estimation of the negotiation
structure between the United States and residual targets suggest a symmetrical
relationship. The mean score of American success with them is 1.27. As was the case
with the European Union, one can advance an assessment indicating that the structure of
negotiation, in those instances, corresponds accordingly to the level of American
bargaining success when resorting to Section (Super) 301.

The significant variables of the empirical model, an indication of the structure of

negotiation, provide a very lucid explanation for the level of American success when

 

“8 As for all other non-EC cases, residual targets are more dependent on the United
States for their external outputs than the reverse situation; only 6 out of 15 issues are
tariffs or quotas related; and, on average, the United States only has a 793 million dollars
trade deficit with these targets.

“9 Only four out of fifteen targets are institutionally constrained, while in only six
cases the value of contested goods and services surpasses the 100 million dollars mark.

199

turning to the utilization of Section (Super) 301. In only one instance, that of Canada,
is there a discrepancy between the structure of negotiation and bargaining success. This
counterfactual evidence may suggest that there exists a certain "special" relationship
between the two North American partners, such that an analysis of trade relations
between the two countries may fall in a "middle of the road" category, not exactly
" international " , or " domestic " .

In the last few pages, I have highlighted the specific nature of the data set. I
started by providing some descriptive analysis, which was followed by the results of an
ordered multinomial logit analysis. Then I discussed the results according to different
targets. A common core of the demonstration is that most of the variables displaying
good empirical results concentrate on the behavior of targets’ negotiators or the American
domestic society. On this basis, I suggest an alternative model. For the remaining of
the dissertation I will refer to the previous model as the " interdependence" model, while

I will refer to the model of the next section in terms of the "compliance" model.

111. An Alternate Model of Compliance

I have discussed above the specific nature of the data set used to test the
propositions of this dissertation. In the first section of this chapter, I discussed the
importance of the trade dependence of targets on the United States as an indicator of the
ability demonstrated by American negotiators in reaching their objectives when using
policies of reciprocity and retaliation. This may suggest that what matters are not the

characteristics of American negotiators, but the characteristics of target nations as well

200

as the level of support gathered from the American population‘zo. Hence, a model
focussing on compliance might be superior to a model of interdependence when dealing
with Section (Super) 301 trade policy instrument. A similar model of compliance has
been tested by Noland (1997).

In a model of compliance, the onus is put on targets’ negotiators to react to an
American threat. Also, the resolve of American negotiators is influenced by the support
they receive from their population. Consequently, I chose to test a model that embodies
these two elements. The variables representing the targets "strength" are constituted by
their level of institutional constraint (INCONST), the natural log of their net value of
exports to the United States (LTEXP), the value of the contested goods and services
(DVCGOOD), and the type of issue under contention (ISSUE). The level of societal
support for the American bargaining tactics is represented by the trade balance between
the United States and target countries (TBAL). Table 18 portrays an ordered
multinomial analysis of the model, while Table 19 shows the marginal impact of the
regressors121 .

The model in its entirety indicates some very good results. Three variables
(LTEXP, DVCGOOD, and TBAL) are significant at the .1 level, while two others

(INSCONST and ISSUE) are significant at the .05 level. According to Table 18, all

 

12° Even though the results of the first section of this chapter indicate that there is no
correspondence between the level of US trade output towards a target and American
bargaining success, the compliance hypothesis should be tested in both directions. In
other words, another model should concentrate on the exclusive characteristics of the
United States (E.g., American institutional constraint, export to target...). This task is
left for future research.

121 I provide a more cursory discussion of the alternate model. I only point out to
its main empirical results and I indicate some comparisons between the two models.

201

variables have the hypothesized sign. Forty-five percent of the cases are correctly
predicted‘”. Table 19 demonstrates that all variables that were also included in the
interdependence model have less of an individual bearing on the dependent variable than
they had in the previous model. Due to the "strength" of other variables in the
compliance model, an individual variable does not have the same impact on the
dependent variable that it had in the interdependence model, where some variables had
almost no bearing on the American level of success.

As stated previously, a high level of domestic institutional constraint (INSCON ST)
is an added source of bargaining leverage in an international trade and investment
negotiation. Thus, a high level of domestic institutional constraint in the target has a
negative impact on the bargaining fortune of American negotiators. That is what is
shown in Table 18: the coefficient of the variable is negative and is significant at the .05
level (P > [z] = .037). Table 18 indicates that the variable has its most profound
impact on the predicted probability of nominal success (Pr=1) when INSCON ST is at
its maximum value, with a marginal influence of .132. This is tantamount to the results

obtained in the interdependence model. All this indicates that the level of domestic

 

‘22 If we compare the two models, we see that there is a trade-off between a model’s
overall significance and the cases it can predict correctly. The interdependence model
includes some variables that were not significant, but could predict 55% of the cases
correctly. All the variables of the compliance model are highly significant, but it can
only predict 45% of the cases correctly. The difference comes from the number of
variables included in those models. When we increase the number of variables (even if
they are not significant), especially dummy variables, we increase the "fit" between the
predicted and coded scores. Consequently, this increases the CPP. For instance, I tested
a "pot luck" model which included all tangible variables. The CCP went up to 66%, but
many variables were not significant and this increase in the CCP comes at the expense
of the loss of many degrees of freedom. Thus, in a model selection exercise, one must
often walk on a tightrope in order to find the adequate balance between significance and
accurate predictions.

202

institutional constraint is a major predictor of the level of bargaining success attained by
American negotiators in the compliance model.

Another salient proposition of the compliance model is that targets are more likely
to be swayed by an American menace when the issue debated involves a tariff or a quota.
The outcome of the ordered multinomial analysis, in Table 18, confirms the proposition.
The coefficient of the variable ISSUE is negative and ISSUE is significant at the .005
level (P > [z] = .002). Table 19 demonstrates that the impact of the variable is
especially significant on the predicted probability of nominal success (Pr=1), with
marginal impacts of .106 and -.127 when ISSUE is at its minimum and maximum values
respectively. However, the most meaningful impact is on the predicted probability of
full success (Pr=3) when the variable is at its maximum, with a marginal consequence
of .173. All this not only indicates that the variable is a momentous component of the
compliance model, but also that it is one of the best indicators of a full negotiating
success for Washington. These results are comparable to the findings of the

interdependence model.

203
TABLE 18

Compliance Model: Empirical Estimates123

Number of obs.: 75

 

 

 

chi2(10): 30.40

Prob > chiz: .0000

Pseudo R2: .1564

Log Likelihood: -81.993425

CCP‘“: 45%
Success Coef. Std. err. z P > [z]
INSCONST -1.2012 .5766 -2.083 0.037
LTEXP .2995 . 1786 1 .677 0.093
DVCGOOD -.7707 .4533 -l .7 0.089
ISSUE 1.478 .4796 3 .081 0.002
TBAL - .0341 .0204 -1.671 0.095
_cutl -1.3708 .5784
_cut2 .9746 .543
_cut3 3.2829 .6857

 

 

 

‘23 I have tested the same model with a binomial threshold model and the results of
this ordered multinomial logit analysis are much more conclusive. Appendix F presents
the logit analyses. The appendix also comports an ordered multinomial logit analysis
with three categories. In the latter, the two middle categories are collapsed.

‘24 Appendix E indicates how the percentage of cases correctly predicted was
obtained.

204
TABLE 19

Compliance Model: Marginal Impacts of the Regressors

 

 

 

 

Variable Pr (0) Pr (1) Fr (2) Pr (3)

All .039 .259 .512 .19

THREAT (Min-) .068 (.029) .365 (.106) .451 (-.061) .115 (-.075)
THREAT (Max.) .016 (-.023) .132 (-.127) .489 (-.023) .363 (.173)
INSCONST (Min.) .025 (-.014) .185 (-.074) .518 (.006) .272 (.082)
INSCONST (MaX-) .078 (.039) .391 (.132) .429 (-.083) .101 (-.089)
DVCGOOD (Min.) .026 (-.013) .201 (-.058) .52 (.008) .252 (.062)
DVCGOOD (Max.) .058 (.028) .332 (.073) .475 (-.037) .135 (-.055)
TBAL (Min.) .009 (-.03) .075 (-.184) .395 (-.117) .522 (.332)
TBAL (Max.) .073 (.043) .379 (.12) .441 (-.O71) .108 (-.082)
LTEXP (Min.) .128 (.089) .477 (.218) .334 (-.178) .061 (-.l29)
LTEXP (Max.) .022 (-.017) .173 (-.0_86) .514 (.002) .291 (.101)

 

a Values between parentheses are marginal differences from the mean.

205

If the type of issue has a positive impact on the faith of the American negotiation
team, its value has the opposite repercussion. The results shown in Tables 18 and 18 are
evidence that the higher the dollar amount associated with the dispute, the less successful
are the Americans in getting a target to comply. The coefficient of DVCGOOD is
negative and highly significant (P > [z] =.089). However, as was the case in the
interdependence model, it does not seem to have any particularly important impact on
the different predicted probabilities of the dependent variable, even though it has some
bearing on the predicted probabilities of nominal success (Pr=1) and full success
(Pr=3). Albeit not as portentous as other variables, the value of disputed goods and
services is still a notable element of the compliance model.

A variable that did not appear in the interdependence model is the natural log of
targets’ net value of exports to the United States (LTEXP). The argument behind the
inclusion of this variable is that American negotiators will achieve a higher degree of
success if a target is facing the prospect of putting a sizable portion of its economic
strength in jeopardy if its trade relation with the United States deteriorates. Hence, the
coefficient of LTEXP is expected to be positive: the larger amount of exports from the
target to the United States is, the more bargaining power for Washington. Indeed, the
coefficient is not only positive, but the variable is also significant at the .1 level (see
Table 18). Table 19 indicates that the marginal impact of the value of exports on the
dependent variable is especially prominent when LTEXP is at its minimum: it has a
marginal impact of .218, -.l78, and -.129 on the predicted probabilities of nominal
success (Pr=1), partial success (Pr=2), and full success (Pr=3) respectively. A

graphical representation of the phenomenon is exposed in Figure 18. It reveals very

206

clearly that as targets’ exports augment, predicted probabilities of failure and nominal
success take a plunge. Even though predicted probabilities of partial success levels off
with large values of exports, the inverse relationship is observed for predicted
probabilities of nominal success and full success. These evidences suggest that the net
value of a target’s exports to the United States is a notable predictor of the ability of
American negotiators to get the other side to comply with their requests.

A substantive analysis of these results reveals a surprising role for the variable
in explaining American bargaining success. The findings indicate that it would be
fallacious to believe that the United States derives its most important bargaining
advantage when a target is highly dependent on Washington for its trade output. It is
true that the results go in this hypothesized direction, but it is the absence of a target’s
trade dependency that has the most profound impact on American bargaining
success. The variable has a very salient negative impact on American bargaining
success, when a target is not highly dependent on the United States for its external
output. Hence, it is not trade dependence, but the lack of trade dependence that has the
largest impact on bargaining success. Also, even though trade dependence has a
considerable impact on the level of bargaining success, it is not the most important
indicator of success, as suggested by Noland (1997: 382). The palm goes to trade
balance in my study.

The team of negotiators representing a target country (or group or countries)
should feel less compelled to obey the demands of Washington if they already have a
trade deficit with the Americans. But more importantly, the American civil society will

ask for a tougher stance from their negotiators if the United States has a very large trade

207

deficit with the target. Since a positive value is associated with an American trade
surplus, we can expect the coefficient of TBAL to be negative. This is what Table 18
demonstrates. It also shows that the variable is significant at the .1 level (P > [z] =
.095). Among all variables, TBAL has the largest marginal bearing on the predicted
probability of success (Pr=3), as it was the case in the interdependence model. Indeed,
when TBAL is held at its maximal value, it has a tremendous marginal impact of .332
on the ability of American negotiators to get the other side of the negotiating table to
comply (see Table 19). Table 19 also indicates that the variable has a sizable implication
on the predicted probability of nominal success (Pr=1). A graphical representation of
the marginal impact of TBAL in Figure 19 demonstrates the trend. The predicted
probabilities of failure and nominal success increase as the balance of trade converges
towards a surplus for the United States. An inverse relationship is observed for the
predicted probability of success. Finally, similar to what occurred in the interdependence
model, the predicted probabilities of nominal success follow a convex trajectory. In
conclusion, it appears that the balance of trade has a vital impact on the dependent
variable in the compliance model.
some

Using a data set comprising Washington’s use of the Section (Super) 301 policy
instrument, this chapter started with some alternate explanations of bargaining
wherewithal and resulting outcomes. The main conclusions were that explanations based
on structural power, economic power, and dependence were poor predictors of trade
bargaining outcomes. It was followed by some descriptive, bivariate, and multivariate

analysis of an interdependence model. The main finding is that the most influential

208

variables affecting international trade bargaining outcomes are trade interdependence, the
type of issues involved in a negotiation, targets’ institutional constraint, use or threat of
retaliation by the United States, the value of the contested goods and services, trade
balance between a target and the United States. All significant variables, with the
exception of trade interdependence had a coefficient with the hypothesized sign.
Variables that were not significant include trade interdependence controlled by GNP,
United States’ institutional constraint, American’s economic misery index, and
presidential approval. Nonetheless, the level of American institutional (when coded
differently), and the misery index have shown some promising results. The next section
looked back at the elements of analysis of the theoretical model and parameters of the
formal model. The best correspondence between empirical results and theoretical
components of the dissertation is found for the bilateral trade structure, institutional
constraint, economic and political necessity elements of analysis, and information and
domestic costs for delays parameters. Less compelling findings accompany the
theoretical role of interdependence and societal support elements of analysis, and
international breakdown cost parameter. The next section showed that the
interdependence model provide a good explanation for the perspective of American
bargaining success when confronted with different targets. Finally, I introduced an
alternate model based on targets’ compliance. This model demonstrated very significant
results.

The concluding chapter recaps the main findings of the dissertation and links the
theory to the empirical results. It also provides a critical appraisal of what has been

accomplished in this study and some directions for future research in the field of

209

international negotiations.

210

0 Predicted Failure
n Predicted Partial Success

A PPSdiCth Nominal success
0 Predicted F011 SUCCBBB

 

 

 

 

.6 j
.5 y f -
.4 " .v,
.3 '-
.2 r
.1 "

M

W

o H

6 .ls I L's 21 235 5 335 l 435

Trade Interdependence
Figure 13: Marginal Impact of Trade Interdependence

211

0 Predicted Failure A Predicted Nominal Success
0 Predicted Partial Success . Predicted Full Success

.6n

'7
7"
‘v'
v

 

 

 

d .S l 1.'s 3 235
Trade IHtEPdependenCB by GNP

Figure 14: Marginal Impact of Trade Interdependence (By GNP)

0 Predicted Failure

212

:1 Predicted Partial Success . Predicted Full Success

 

 

 

 

.11 s E .1;

Presidential Approval
Figure 15: Marginal Impact of Presidential Approval

A Predicted Nominal Success

.51

0 Predicted Failure
0 Predicted Partial Success

 

s

213

A Predicted Nominal Success
0 Predicted Full Success

} 2’1 1'1 13 1‘3. 1'7 1'9 5'1
Misery Index

Figure 16: Marginal Impact of Misery Index

214

0 Predicted Failure A Predicted Nominal Success
1: Predicted Partial Success . Predicted Full Success

 

 

 

 

-éo 40 30 3 2'0

Balance Of trade
Figure 17: Marginal Impact of Trade Balance (Interdependence Model)

0 1

0 Predicted Failure

215

A Predicted Nominal Success

1:: Predicted Partial Success . Predicted Full Success

 

Figure 18:

H
H

Log of Targets’ Exports to US
Marginal Impact of Targets' Exports

216

 

 

0 Predicted Failure A Predicted Nominal Success

:1 Predicted Partial Success . Predicted Full Success
154
.5- = :2 z
.4- c 2: h
£3-
£2“
1 4 -,

- _1i—_——4re__°4-“-H-‘r..-ll"lr4l49

O 1 H-ee—

 

-60 -50 ~50 d 25

Balance Of trade
Figure 19: Marginal Impact of Trade Balance (Compliance Model)

CHAPTER 6

CONCLUSION

This dissertation started out with a set of ambitious tasks. It has reached most of
its initial objectives, but more work needs to be done if we are to have a better
understanding of the underpinnings of international trade and investment negotiations.
Let’s start with its main accomplishment, before turning to its shortcomings and setting
out some avenues for future research.

The main contribution of this dissertation is that it constitutes an innovative
attempt at combining a theory of negotiation and bargaining with formal and empirical
models. A great number of verbal models of negotiations and bargaining have been
proposed. The last two students of international negotiations to do so are chow (1996)
and Shambaugh IV (1996). While these constitute critical contributions to the uncovering
of the nature of negotiation and bargaining, they lack the aura of "scientificity" that is
bestowed upon an analysis on a large number of cases. Part of the reasons for writing
this dissertation was to provide a synthesis of existing theoretical propositions. But more
importantly, it has to be done in such a way that these propositions could be put in
relation in a logical fashion and be prone to statistical analysis. This difficult assignment
has not been accomplished to my full satisfaction, but I consider that with better

resources and more time, the Holy Grail may be in sight.

217

218

This dissertation starts out by pointing to the need to delve into general
negotiation theory in order to uncover an adequate theory of international trade
negotiations. A common core of those theories is that negotiation is time-dependent,
based on different type of interests (conflicting, complementary, and common), and based
on different levels of information held by negotiators. They also indicate that, in formal
language, the negotiation enterprise is not entirely cooperative, nor entirely non-
cooperative, but incorporates some elements of both visions of game theory. These
elements of analysis constitute the "structure" of negotiation. This structure of
negotiation affects the process of bargaining; a "give-and-take" convergence process
leading to an agreement. Thus, to study international trade negotiations and bargaining,
we need to unveil a specific structure of negotiation and a model to depict the process
of bargaining.

My argument is that a theory of international trade negotiations must be build on
a structure composed of domestic and international elements of analysis. A theory of
international trade negotiations must also look at the domestic/ international nexus, while
paying attention to the main components of more general negotiation theories. On this
basis the theory of international trade negotiations of this dissertation rested on the
following pillars: trade interdependence, bilateral trade structure and threat credibility,
domestic institutional constraint, political and economic necessity, and societal support.
These elements of analysis constitute the structure of international trade negotiations
affecting the process of international trade bargaining.

The logical underpinnings of the process of bargaining have been unveiled with

great amount of success by many economists, mainly during the 19805. Unfortunately,

219

the reader of these stylistic representations often gets lost in the technical details and has
a feeling that the connection with the real world is lost. These might be profane
criticisms uttered by many naysayers of formal approaches, but game theorists, must
attempt to generate some logical constructions that appeal by their simplicity to the rest
of the academic community. I am not the best judge of my performance in this regard,
but this issue was on my mind when I tried to convey the basic structure of the Sth-
Rubinstein bargaining model, as well as indicating how it could be related to real world
situations. More specifically, the bargaining model used in this dissertation builds on
three fundamental parameters: international breakdown costs, information, and domestic
costs associated with delays. Each of these parameters are logical, formal,
representations or approximations of the elements of analysis of a general negotiation
theory. These parameters also parallel the elements of analysis of the specific theory of
international trade negotiations.

Attempts to provide a statistical analysis of international trade negotiations
constitute a very recent adventure. It is not that there are no studies of trade flows or
any empirical international political economy studies. Inquiries of this sort have focused
on trade sanctions (Hufbauer, Schott, and Elliott, 1990), economic risk and protectionism
(Bennett Quifiones and Gates, 1995), and tariffs and partisan politics (Epstein and
O’Halloran, 1996) to name a few. Empirical tests to determine the nature of negotiations
outcome are a much rarer breed. The difficulty stems from the fact that we do not have
large data-gathering projects on international trade negotiations such as those developed
for the study of the causes of war. The closest we come to having such a data set stems

from the commendable effort by Bayard and Elliot (1994), who have collected data on

220

the American use of Section (Super) 301. I have attempted to push the barriers of our
empirical knowledge on the causes of trade bargaining success a little further through this
dissertation. As such, I have relied heavily on data provided in Bayard and Elliot study,
but I have also proceeded into a data collection project that goes beyond what they have
already used. Nevertheless, more effort needs to be done before students of negotiations
and bargaining can count on a data set including most of the variables needed to test a
more fully developed model of international trade negotiations.

Building on Bayard and Elliot’s ( 1994) data set I enumerated a series of
propositions and hypotheses connected to the elements of analysis of the theoretical
model of negotiations and the parameters of the formal of bargaining process. The
results of my empirical analysis demonstrate that domestic and international factors are
all important aspects of a model explaining the outcomes of international trade and
investment negotiations. These show that our analyses must go beyond issues of power,
be it structural or economic, and interdependence to fully account for such international
encounters. More specifically, trade interdependence (even though not in the
hypothesized direction), the level of institutional opposition, the type of issue being
negotiated, the use of or threat of retaliation, the value of the contested goods and
services, and trade balance, were all significant and important variables. This adds to the
linkages between domestic and international levels of analysis propositions.

In regards to the dimensions of the theoretical model and the parameters of the
formal model, it appears that none of them had a more salient than others. It is a
combination of all dimensions and their effects on the parameters that better represents

the nature of an international trade and investment negotiation. Indeed, most variables

221

demonstrated good results and all dimensions and parameters were represented by these
variables in the empirical models.

When we study the performance of American negotiators vis-a-vis different targets
we find out that it is commensurate with the bargaining leverage imparted to them by the
negotiation structure. Canada constitutes an anomalous case. The structure of
negotiation between the United States and its neighbor is highly advantageous to
Washington, but its trade representatives do not seem able to translate this bargaining
edge into correlated successful outcomes. This may be a good indication of a "special"
trade relationship between the two adjoining countries.

The analysis yields other surprising results. The most puzzling finding is that
American trade officials gain more from trade dependence on a target rather than the
other way around. This outcome is in sharp contrast with the hypothesized role that was
attributed to the variable. An explanation for this phenomenon could be the role played
by an intervening variable, that is, negotiator’s resolve. Just as the case of a negative
trade balance with a target, American negotiators may derive an increased level of
unyielding strength from trade dependency. Under such circumstances, they become
"tough" negotiators and adopt a " take it or leave " type of strategy, which often yields full
successes or failures. The chances of getting partial of nominal successes are greatly
diminished. Even though, this is what reveals the statistical analysis, until we find an
adequate way to measure "negotiator’s resolve", we can only speculate on its mediating
role between trade dependence (and/or trade balance) and bargaining success.

Another interesting result is that the level of targets’ exports to the United States

influences positively the United States’ bargaining success, while its own level of exports

222

to the targets does not affect negatively its bargaining fortune. On this basis, I tested an
alternate model based on targets’ compliance to American demands. The results indicate
that all variables included in a compliance model are significant, but that each variable,
individually, has less of an impact on American bargaining success and it does not
predict accurately as many cases as the main model of the dissertation.

Despite its accomplishments, this dissertation has its weaknesses. Allow me to

mention a few of them, with some elements of solution for future research:

1- Negotiators’ performance and type: An important aspect of negotiation not
accounted for in my theoretical model is negotiators’ effectiveness. This idiosyncratic
feature is often a leading variable in the explanation of any type of negotiation. The
results of the ordered multinomial logit analysis point to the fact that negotiators’ resolve
might constitute an important mediating variable between trade dependence and/or trade
balance and bargaining success. A formal model can include a parameter accounting for
a negotiator’s level of risk-attitude. We can observe logically the effect of this
parameter, but the difficulty is in assessing a priori a special type of risk perception to
a negotiator in a specific case of negotiation. Thus, I did not include this very important
variable in my model for the simple reason that I was not able to find a way to go from
a logical representation of its impact on an outcome of trade negotiation to a variable (or
a set of variables) amenable to statistical analysis. As for a solution to the problem, a
first step would be to include a parameter of risk attitude in the formal model, and do
more reading and thinking to see if there are some empirical representations of this

parameter.

223

2- Comparative statics: It would have been interesting to have done some comparative
statics analysis of the formal model and to have compared the results with the outcomes
of the empirical model. For the sake of simplicity, I have decided to let the statistical
results speak for themselves. Nevertheless, I intend to do some comparative statics in the
near future and to compare the results with the statistical outcome of this dissertation.
This comparison could be done by using some instrumental variables representing the

parameters of the formal model and test their levels of statistical significance.

3- Case studies: Even though I have mentioned a previous study (Duchesne and Clark,
1995) applying some aspects of the formal model to the Canada-United States Free Trade
Agreement (CUSFTA), more detailed analyses of the general model might be
needed”? Not only would it provide an added test of the propositions of the model,
but it could also uncover some elements of analysis that are not apparent in a statistical
or formal analysis. Two special cases, especially, need to be addressed, given the
empirical results of the dissertation: l) The "special" relationship existing between
Canada and the United States; 2) The high level of bargaining success attained by

American negotiators vis-a-vis industrialized Asian countries.

4- Trade interdependence: Another way to measure for trade dependence or
interdependence is to consider the concentration of production and consumption of the

specific goods and services under contention. This would add another level of

 

‘25 See also Cameron (1996) for an application of the Stahl-Rubinstein model to the
North American Free Trade Agreement.

224

verification for the proposition. It has the advantage of being very specific to each case.
This, in fact, also constitutes the main problem. It takes a time-consuming effort to
collect this type of data for each case. It is, however, a task that has to be tackled at
some point, because I suspect it to be the most important aspect of the interdependence

proposition.

5- Institutional constraint and Societal Support: Even though the results obtained by
the test of the institutional propositions were conclusive, I believe that Hagan’s (1993)
method of analysis of domestic opposition is a better, more detailed, representation of
the domestic forces (see Figure 7)”? The rewards associated with this important data
collection project might well be worth the effort. A similar effort could also be extended
to test the societal support dimension according to Figure 9. Presidential approval is not
significant in this study’s statistical analysis, but other variables more specific to trade
relations, such a industry concentration, could be tested for their significance. Once,

again, this demands a data collection project that is more specific to each case.

6- Nature of the data set: What is one of the major strong points of the dissertation
may also constitute its main weakness. The nature of the data set leads to unforeseen
conclusions. Indeed, by using a data set based on the use of Section (Super) 301 by
Americans, I was led to believe that issues of compliance were more important than

issues of interdependence. Thus, I have proposed and tested a new model accounting for

 

‘26 It could be especially the case with Japan, where the highest level of institutional
constraint may come from within the ruling party.

225

this discovery. Nevertheless, due to this data set, I was not able to test directly and
adequately some propositions of the theoretical model. This points to the need for a data
set specifically constructed to evaluate outcomes of international trade and investment
negotiations. Such a data set would take into account the difference between cases
including an explicit threat of retaliation, such as the use of Section (Super) 301, and
more conciliatory episodes of negotiations, such as NAFTA or MERCOSUR. Special
attention should be paid to better empirical definition of an outcome of trade negotiation.
Until then, we might still be in the dark ages of trade and investment negotiation and

bargaining theories.

7- Data collection project: The time is now ripe far a large data gathering project, such
as those we find for the study of war. With a small data set such has this one, it is
difficult to make any authoritative comments about the impact of individual variables,
especially when testing the level of success for American negotiators according to
different targets. Moreover, a standardized data set would help us test different, and
often contradictory, models and hypotheses. Also, with a larger data set, our analysis
could go beyond the study of American trade policies. Our future analyses must be
comparative in nature. Only then will we be able to make more discemable comments
about the impact of structural and economic power, trade dependence, compliance, and
reciprocity on bargaining leverages and outcomes.
***
It is only recently that international relations students have turn their attention to

international trade negotiation and bargaining. It is therefore not surprising that we did

226

not go a long way towards developing statistical analyses of this very important aspect
of nations-states’ interactions. With the end of the Cold War, and the dawn of a new
era, the study of international relations and, especially, international trade politics cannot
be complete without a better understanding of this "give-and—take" international strategy
played by our leaders; a strategy that, increasingly, affects our everyday life. This
endeavor will not be accomplished before we have the adequate tools, that is a larger
data set, and a common understanding that outcomes of international trade negotiations
have as much of an impact, if not more, than well-publicized security agreements on the

future of international interactions and domestic politics.

APPENDICES

APPENDIX A

Theoretical propositions

1. International Dimension of Negotiation

Pl

P2A

P2B

A) Interdependence

In a bilateral trade and investment negotiation, a nation-state that is less dependent
on another state for its commercial exchanges has more bargaining power than
the second nation-state.

B) Trade complementarity

When the nature of the trade relationship between two countries is
complementary, an international trade and investment agreement is more likely
to occur.

A threat of trade retaliation is more credible when the targeted product is also
produced in the targeting country and/or can be imported from a third country.

11. Domestic Dimension of Negotiation

P3

P4

P5

A) Institutional constraint

The higher the institutional domestic constraint faced by negotiators, the higher
their bargaining leverage in an bilateral international trade and investment
negotiation.

B) Political necessity

In a bilateral international trade and investment negotiation, the greater the
necessity for an agreement for one party, the less its bargaining leverage.

C) Domestic support

In a bilateral international trade and investment negotiation, the greater the
societal support for negotiators, the greater their bargaining leverage.

227

228

III. Parameters of Bargaining Process

P6

P7

P8

A) Negotiation breakdown costs

In a bilateral international trade and investment negotiation, the greater the cost
of negotiation breakdown for negotiator, the smaller their bargaining leverage.
B) Domestic costs for delays

In a bilateral international trade and investment negotiation, the higher the
domestic cost of delay for negotiators (or lower their discount factor), the less
bargaining leverage they have.

C) Information

In a bilateral international trade and investment negotiation, the more information
negotiators have about domestic costs of the other party (or its "strength"), the
more bargaining leverage they have.

111. Relationship Between Structure of Negotiation and Bargaining Process

P9

PlOA

PlOB

P11

A) Dependence and breakdown costs

In a bilateral international trade and investment negotiation, negotiator who
represent a country that is highly dependent on the other negotiating country for
its supply of international goods have higher costs for a breakdown in the
negotiations.

B) Complementarity of trade and breakdown costs

In a bilateral international trade and investment negotiation, a high
complementarity of trade leads to high international breakdown costs.

In a bilateral international trade and investment negotiation with a high
complementarity, an agreement is likely to occur.

C) Institutional constraint and information
In a bilateral international trade and investment negotiation, the more domestic

institutional constraint negotiators are faced with, the less information about their
type they convey to other negotiators.

P12

P13

229

D) Political necessity and costs for delays

In a bilateral international trade and investment negotiation, negotiators who have
great political need for an agreement face important domestic costs for delays in
the negotiations.

E) Societal support and costs for delays.
In a bilateral international trade and investment negotiation, the greater

negotiators’ societal support is, the less domestic costs for delays in the
negotiations they are faced with.

APPENDIX B

Description of Variables

I. Dependent Variable

The dependent variable is the level of success for the American negotiators when
they opt to use Section (Super) 301. It is divided in four ordered categories as coded by
Bayard and Elliot (1994):

Failures (coded as 0): The case was not implemented to US
satisfaction, or it was circumvented in some
other way.

Nominal success (coded as l): A case in which an agreement was reached
but not implemented to US satisfaction.

Partial success (coded as 2): A case in which only some of the US
objectives were implemented.

Success (coded as 3): A case in which the US objectives were

totally or largely implemented.

11. Independent variables

A) International dimension

1- Interdependence

The first independent variable included in the statistical test represents the trade
interdependence between the United States and the target country (or group of countries).
It is measured as the ratio of the target’s share of the United States’ trade output
(exports) over the United States’ share of the target’s trade output (exports). I order to
measure this variable (DEP) the following variables were created:
USTOUT: Annual United States’ Trade Output (in billions)
TATOUT: Annual Target’s Trade Output (in billions)

USEXP: Annual United States’ Exports to Target (in billions)
TEXP: Annual Target’s Exports to the United States (in billions)

230

TSHOUT:

USHOUT:
DEP:

231

Target’s Share of United States’ Trade Output (USEXP/USTOUT * 100)
(%)

United States’ Share of Target’s Output (TEXP/TATOUT *100) (%)
Ratio of TSHOUT over USHOUT

The second variable used to measure trade interdependence is the ratio of the United

States share of its gross national product that is accounted by its exports to the target

over the target’s share of its gross national product that is accounted by its exports to the

United States.

In order to measure this variable (DEPGNP) the following additional

variables were created:

USGNP:
TGNP:
USXGNP:
TXGNP:

DEPGNP:

Annual United States’ GNP (in billion)

Annual Target’s GNP (in billion)

United States’ Share of GNP accounted for by Exports to Target
(USEXP/USGNP * 100) (%)

Target’s Share of GNP accounted for by Exports to the United States
(TEXP/TGNP * 100) (%)

Ratio of USXGNP over TXGNP

2- Threat credibility

To assess the credibility of a threat issued by the United States when using the

instrument, I use Bayard and Elliot’s (1994) dummy variable ISSUE, which is coded as

a 1 of the issue under debate is a tariff and/or a quota, and coded as a 0 otherwise.

B) Domestic dimension

1- Institutional constraint

The first variable is related to the measure of the United States level of

institutional constraint. A variable called DIVIDED (Lohmann and O’Halloran, 1994)

accounts for the level of division among domestic institutions and it is calculated in the

232

following manner:

TABLE 20

American Institutional Division

 

President Senate Majority House Majority DIVIDED
DEM DEM DEM -l

DEM REP DEM O

DEM DEM REP O

DEM REP REP 1

REP DEM DEM l

REP DEM REP 0

REP REP DEM O

REP REP REP -l

A second variable (INSCONST) is related to the level of domestic institutional
constraint of the target country (or group of countries). It borrows from variables
included in the POLITY III project. I used Maoz and Russett (1993) five-point ordinal
scale measure of institutional independence of the chief executive (MONO), as well as
their three-point ordinal scale representing the degree of institutional centralization of the

state (CENT). For an added score of MONO and CENT varying from 2 to 4,

233
INSCONST was coded as a 0. For an added value of MONO and CENT varying from

5 to 8, INSCONST was coded as an 1.

2- Political necessity

A first variable accounts for an "economic" necessity for an agreement. To
measure the United States’ need for agreement, I have created a dummy variable
(DVCGOOD) coded as an 1 if the value of the contested good is 100 million dollars or
more, and coded as a 0 otherwise. This makes the distinction between "small" or large
cases.

Another measure of economic necessity is a misery index (Keech, 1995: 52-53),
calculated by adding the seasonally adjusted American unemployment rate (UNEMP) to
its inflation rate (MISERY). The inflation rate (INFLAT) is calculated the following
way:

ppi; -ppi(_|
ppit—l

* 100 (1982-100)

where ppi represents the production price index.
In order to account for political necessity, per se, a dummy variable taking a
value of 1 is included when American negotiators used a threat of retaliation or a direct

retaliation against an international target (RETAL) (Bayard and Elliot, 1994:20-21).

3- Societal support

A first measure of societal support is the trade balance (TBAL) between the

234
United States and the target (USEXP - TEXP).

Another, widely used, measured of societal support is presidential approval

(APPROV). Ergo, this variable is included in the empirical model.

APPENDIX C

List of Hypotheses

1. International Dimension

H1

H2:

H3:

A) Interdependence

When a larger share of the United States exports are going to the target country
(or group of countries) than the share of the target country’s (or group of
countries’) export going to the United States, the American negotiators are in a
weakened bargaining position when they use Section (Super) 301.

When a larger share of the United States’ gross national product (GNP) is
generated by its exports to the target country (or group of countries) than the
share of the target country’s (or group of countries’) gross national product
(GNP) generated by its exports to the United States, the American negotiators are
in a weakened bargaining position when they use Section (Super) 301.

B) Threat credibility

The American negotiators have more bargaining leverage in their use of Section
(Super) 301 when the issue debated is about tariffs and/or quotas than when the
issue debated is about intellectual property or technical, environmental and health
standards.

11. Domestic Dimension

H4:

H5:

H6

A) Institutional constraint

The more divided the United States’ government is, the higher rate of success for
the American negotiators in their use of Section (Super) 301.

The higher the level of institutional constraint for the targeted country, the less
bargaining leverage for the American negotiations in their use of Section (Super)
301.

B) Political necessity

The use of a threat of retaliation or a direct retaliation in cases involving the use
of Section (Super) 301 is an indication of bargaining weakness on their part.

235

H7

H8

H9

H10

236

The higher the value of the contested good or service leading the use of Section
(Super) 301, the less bargaining success for the American negotiators.

A higher misery index (Inflation + Unemployment rate) in the United States
leads to less bargaining success for American negotiators when they use Section
(Super) 301.

C) Domestic support
The greater the trade balance deficit for the United States with the targeted
country (or group of countries), the more bargaining success for the American

negotiators when they use Section (Super) 301.

The higher the American presidential approval, the more bargaining success for
the United States’ negotiators when the use Section (Super) 301.

237

APPENDIX D

Data Set

I. Legend

Casenum: Case number

Becase: Bayard and Elliot’s case number

Period: Time period of the negotiations

Target: Country (or group of countries) targeted by the United States
Success: Level of success of American negotiators (dependent variable)
Issue: Type of contested goods and services

Retal: Use of or threat of retaliation by American negotiators
Misery: Misery Index

Tbal: Trade balance

Dvcgood: Value of contested goods and services

Depgnp: Trade interdependence controlled by GNP

Dep: Trade interdependence

Insconst: Target’s institutional constraint

Divided: American’s institutional constraint

Approv: Presidential approval

Ltexp: Natural log of target's exports to the United States

II. Data set
TABLE 21

Data Set (Part A)

 

 

 

Casenum BECASE Target Period Success
1. 1 Guatemala 7—1-75/6-29-76 1
2. 2 Canada 7-17-75/3-14-76 3
3. 3 EC 8-7-75/7-21-80 2
4. 4 EC 9-22-75/1-5-79 1
5. 5 EC 11-13-75/6-19-80 O
6. 6 EC 12-1-75/1-83 O
7. 7 EC 3-30-76/6-18-80 l
8. 8 EC 3-30-76/1-5-79 1
9. 9 Taiwan 3-15-76/12-1-77 3

10. 11 EC 10-6-76/8-10-86 2
11. 12 Japan 2-14-77/3-3-78 3
12. 13 Japan 8-4-77/12-85 2
13. 14 USSR ll-lO-77/7-l2-79 l
14. 15 Canada 8-29-78/10-30-84 O
15. 16 EC 11-2-78/8-1-80 l
16. 17 Japan 3-14-79/1-6-81 1
17. 18 Argentina 5-25-79/7—25-80 1
18. 19 Japan 10-22-79/1-6-81 1
19. 20 Korea 11-5-79/12-29-80 1
20. 22 EC 8-20-81/6-28-82 O
21. 23 EC 9-17-81/12-84 1
22. 24 Argentina 10-9-81/11-16-82 O
23 25 EC lO~16-81/9-lS-87 2
24 26 EC 10-23-81/12-85 l

238

 

(Data set, Part A, continued)

Casenum Becase Target Period Success
25. 34 Canada 9-13-82/12-82 O
26. 35 Brazil 10-25-82/1985 2
27. 36 Japan 10-25-82/12-85 2
28. 37 Korea 10-25-82/1985 2
29. 40 Brazil 4-16-83/1985 2
30. 41 Portugal 4-16-83/1985 0
31. 42 Spain 4-16-83/1985 O
32. 43 Taiwan 9-29-83/3-22-84 2
33. 44 Argentina 9-21-83/5-25-89 2
34. 45 Taiwan 12-19-83/4-26-84 2
35. 48 Japan 6-14—85/6-4-91 1
36. 49 Brazil 9-16—85/10-6-89 2
37. 50 Japan 9-16—85/10-6—86 3
38. 51 Korea 9-16-85/8-14-86 3
39. 52 Korea 11-4-85/8-14-86 1
40. 53 Argentina 4—4-86/12-88 2
41. 54 EC 3-31-86/1-30-87 3
42. 55 Canada 4-1-86/6-1-90 2
43. 56 Taiwan 8-1-86/10-1-86 2
44. 57 Taiwan 10-27-86/12-5-86 2
45. 59 India 1-6-87/6-8-88 2
46. 60 EC 7-14-87/12-88 1
47. 61 Brazil 6-11-87/6-27-90 1
48. 62 EC 11-25-87/1-1-89 O
49. 63 EC 12-16-87/1-31-90 l
50. 64 Korea 1-22-88/5-31-88 2
51. 65 Korea 2-16-88/4—26—90 2
52. 66 Japan 5-6-88/7-5-88 3
53. 67 Korea 4-27-88/1-18-89 2
54. 68 Argentina 8-10-88/9—23-89 1
55. 69 Japan 11-21-88/7-31-91 2
56. 70 EC 11-14-88/2-26-90 3
57. 71 EC 5-8-89/10-1-89 2
58. 72 Thailand 4-10-89/11-23-90 1
59. 73 Brazil 6-16-89/5-21-90 3
60. 74 Japan 6-16-89/6-15-90 3
61. 75 Japan 6-16—89/6-15-90 2
62. 76 Japan 6-16-89/6-15-90 2
63. 77 India 6-16-89/6-14-90 O
64. 78 India 6-16-89/6-14-90 O
65. 79 Norway 7-11-89/4-26-90 1
66. 80 Canada 5-15-90/8-5—93 1
67. 81 EC 11-15-90/12-21-90 2
68. 82 Thailand 11-15-90/12-20-91 1
69. 83 EC 11-28-90/10-93 1
70. 84 Thailand 1—30-91/10-92 1
71. 85 India 5-26-91/2-26-92 1
72. 85 India 5-26—91/2-26-92 O
73. 86 China 5-26-91/1-17-92 l
74. 88 China 10-10-91/10-92 1
75. 89 Taiwan 4—29-92/6-5-92 1

239

 

Data set (Part B)
Casenum ISSUE RETAL MISERY TBAL

1. 1 O 15.66 .089
2. 1 O 19.15 -.6759987
3. 1 O 18.73 8.615999
4. 1 0 13.98 6.250999
5. O 0 18.73 8.615999
6. O 1 13.76 8.671997
7. 1 O 18.73 9.046999
8. l 0 13.98 6.199001
9. 1 O 13.51 1.456
10. 1 l 6.86 2.473003
11. O 0 13.51 -11.01
12. l 0 8.16 -19.666
13. 0 O 15.47 1.911
14. O 1 9.57 —9.916
15. 0 O 18.73 11.243
16. O 0 20.5 -12.345
17. 0 0 18.73 1.61
18. O 0 20.5 -12.345
19. O 1 20.5 -.0679998
20. O O 15.28 11.972
21. 0 O 9.57 4.792999
22. 1 l 13.76 .8
23. 0 1 8.33 -5.825001
24. O 1 8.16 .5559998
25 1 0 11.26 -12.809
26 1 O 8.16 -2.986
27. 1 1 8.16 -29.484
28. 1 O 8.16 —3.038
29. O O 8.16 -3.745
30. 1 O 8.16 .536
31. 1 O 8.16 .628
32. 0 0 11.2 -8.256001
33. O O 9.22 —.007
34. O 0 11.2 -8.256001
35. O 1 9.62 -49.207
36. O 0 10.48 -3.418
37. 1 0 5.57 -49.549
38. 0 O 6.86 —6.039
39. O 0 6.86 -6.039
40 O O 7.97 .0780001
41 1 O 5.57 -21.775
42. 1 O 10.46 -13.047
43. 1 0 5.57 -13.591
44. 1 O 5.57 -13.591
45. 1 O 8.15 -.3329999
46. O O 7.97 -15.89
47 O l 10.46 -3.413
48. 0 1 7.97 ~10.329
49. O 1 10.48 -7.357994
50. 1 0 8.15 -10.221
51. O 0 10.48 -7.330999
52 1 O 8.15 -52.625
53. 1 0 7.97 —8.472
54. 0 O 10.48 -.155
55. O 1 9.62 -47.146
56. 1 0 10.48 -2.230995
57. O 0 10.48 .8919983
58. 1 1 10.43 -1.424
59. 1 0 10.46 -2.96
60 O O 10.46 —45.954

240

 

 

(Data set, Part B, continued)

Casenum ISSUE RETAL MISERY TBAL
61. O O 10.46 -45.954
62. 0 O 10.46 -45.954
63. 0 O 10.46 -1.085
64. 0 0 10.46 -1.085
65. 0 0 10.48 -.8599999
66. 0 1 8.32 -12.705
67. 1 0 10.43 1.558998
68. 0 1 8.8 -2.279
69. 0 O 8.02 5.631004
70. 0 1 8.63 -2.816
71. O O 8.8 -1.27
72. 0 O 8.8 -1.27
73. 0 1 8.8 -.52
74. 1 1 8.63 -.52
75. O 0 8.71 -7.801001

Data set (Part C)

Casenum DVCGOOD DEPGNP DEP INSCONST

1. 0 .0031838 .0088909 0
2. 0 .1020394 .3254443 0
3. 0 1.260357 4.095727 1
4. 0 1.157693 3.878528 1
5. 0 1.260357 4.095727 1
6. 1 1.1915 3.800817 1
7. 0 1.261664 4.12925 1
8. 1 1.142119 3.887769 1
9. 0 .0188603 .1316274 0
10. 0 .8983911 3.243768 1
11. 0 .1760278 .3493065 0
12. 1 .1819206 .3384425 0
13. 0 2.308336 .5570821 0
l4. 0 .0580212 .2701695 0
15. 1 1.314437 4.090522 1
16. 0 .2513841 .373765 0
l7. 0 .0700449 .1367358 1
18. 0 .5544065 .373765 0
19. 0 .0223129 .0796183 0
20. 1 1.253086 3.592758 1
21. 1 .9501866 3.147549 1
22. 1 .0328685 .0694553 1
23. 0 .7186674 2.832983 1
24. 0 .8245333 2.92057 1
25. 0 .06716 .2432199 0
26. 0 .0324838 .0556887 1
27. 0 .1525367 .322349 0
28. 0 .0158307 .0814843 0
29. 1 .0260727 .0502814 0
30. 0 .0127456 .0557911 0
31. 1 .0604777 .1399229 1
32. 0 .0057882 .0491178 0
33. 0 .016084 .0313788 0
34. 0 .0057882 .0491178 0
35. 1 .2143768 .3447837 0
36. 1 .0264103 .0556063 0
37. 1 .1194046 .3002407 0
38. 0 .0123075 .074704 0
39. 1 .0123075 .074704 0
40. 1 .0182468 .0306271 0

241

 

 

(Data set, Part C, Continued)

Casenum DVCGOOD DEPGNP DEP INSCONST
41. 1 .5504177 2.694861 1
42. 0 .076493 .2985592 0
43. 1 .0052236 .0522499 0
44. 1 .0052236 .0522499 0
45. 0 .0481282 .0355617 1
46. 0 .7322114 2.862431 1
47. 0 .0377392 .0533383 0
48. 1 .8327505 2.960287 1
49. 1 .9011873 3.117041 1
50. 1 .0163311 .0995694 0
51. 1 .0239732 .1107411 0
52. 1 1.945434 .3458019 0
53. 0 .0199825 .1052621 0
54. 0 .012999 .023838 0
55. 1 .2782038 .3711877 0
56. 0 1.002439 3.230388 1
57. 0 1.04093 3.153629 1
58. 1 .0099332 .0401753 0
59. 1 .0444189 .0527498 0
60. 0 .290596 .373955 0
61. 0 .290596 .373955 0
62. 1 .290596 .373955 0
63. 1 .0389924 .0308392 1
64. 1 .0389924 .0308392 1
65. 0 .0099212 .046124 0
66. 0 .083274 .2670678 0
67. 1 1.091819 3.532042 1
68. 1 .0094952 .0380045 0
69. 0 1.151036 3.389171 1
70. 0 .0098103 .0408097 0
71. 1 .0288158 .0269102 1
72. 1 .0288158 .0269102 1
73. 1 .0809791 .1692211 0
74. 1 .0809791 .1692211 0
75. 1 .02324 .1218329 0

Data set (Part D)

Casenum DIVIDED APPROV LTEXP

l. l .5 -1.584
2. 1 .47 3.161
3. —1 .34 3.292
4. -1 .43 3.202
5. -1 .37 3.292
6. O .38 3.482
7. -1 .37 3.367
8. -1 .43 3.282
9. -1 .54 .632
10. 0 .63 3.762
11. -1 .51 3.123
12. 0 .62 3.66
13. —1 .32 -.533
14. 0 .6 3.836
15 -1 .34 3.545
16 0 .53 3.479
17. -1 .34 -.431
18. O .53 3.479
19. -1 .32 1.505
20. 0 .44 3.754

242

(Data set, Part D, Continued)

 

Casenum DIVIDED APPROV LTEXP
21. 0 .6 3.856
22. 0 .42 —.059
23. 1 .46 4.074
24. 0 .62 3.93
25. 0 .42 3.84
26. 0 .62 1.779
27. 0 .62 3.946
28. 0 .62 2.184
29. 0 .62 1.875
30. 0 .62 -.868
31. 0 .62 .71
32. 0 .54 2.573
33. 1 .57 -.035
34. 0 .54 2.573
35. 1 .75 4.453
36. 1 .61 2.007
37. 0 .61 4.308
38. 0 .63 2.501
39. 0 .63 2.501
40. 1 .57 -.05
41. 1 .41 4.362
42. 1 .63 4.385
43. 0 .55 2.945
44. 0 .55 2.945
45. 1 .47 .837
46. 1 .57 4.427
47. 1 .68 2.074
48. 1 .53 4.435
49. 1 .69 4.471
50. 1 .47 3.067
51. 1 .63 3.014
52. 1 .47 4.502
53. 1 .53 3.037
54. 1 .62 .183
55. 1 .65 4.52
56. 1 .69 4.488
57. 1 .61 4.451
58. 1 .52 1.568
59. 1 .63 2.066
60. 1 .63 4.528
61. 1 .63 4.528
62. 1 .63 4.528
63. 1 .63 1.269
64. 1 .63 1.269
65. 1 .63 .702
66. —1 49 4.628
67. 1 .52 4.569
68. 1 .53 1.732
69. -1 54 4.55
70. 1 36 1.9
71. 1 38 1.172
72. 1 38 1.172
73. 1 38 2.001
74. 1 36 2 001
75. 1 34 3.16

APPENDIX E

Calculation of Cases Correctly Predicted

I. Explanations

The predicted scores (Sj) are calculated according to the following
formula:

and the probabilities are:

 

 

 

 

 

1
Pr [0]=Pr(Sj+p.j<kl)= S—k
(1+e’ 1)
Pr [l]=Pr(kl<Sj+pj<k2)= ls — ls k (2)
(1+ei"‘2) (1+ei‘1)
Pr[2]=Pr(k2<Sj+p,j)<k3= 1 _____L___-
(1+esi"‘3) (1 +eSI”‘2) (1 +eSI‘"‘1)
Pr[3]=Pr(k3<Sj+pj).—.1— 1 __1___ _1__

.k1

(1 + 6’1"?) (1 +e‘i'k2) - (1 +eS’ )

or, put more simply:

Pr[xb +p < cutl] = Pr(0) (3)
Pr[cutl < xb + p < cut2] = Pr(1)
Pr[cut2 < xb + p < cut3] = Pr(2)

Pr[cut3 < xb + p] = Pr(3)
where cutl = -3.872135; cut2 = -1.167997; cut3 = 1.399905
Notice the following:
1- Cases correctly predicted (CCP) are maked by an "X"

2- Coded score for the dependent variable, as indicated in the data set,
is represented by "CS" below.

3- In the results below, "score" represents 'uda +p" in equation (3).
Therefore, "score" indicates the predicted score for each case. For
instance, case #1 indicates a score of "- 0475775" which lies between cut2
and cut3 in equation (3). This means that the predicted score for the

first entry is 2 (Partial success) and is indicated as such in the next to
last column named "PS".

243

244

4- Also note that for the first case, the highest probability (.56) also
suggests that the predicted score is a "2" (Partial success). It is not
always the case that the two values coincide. For instance, in case #15,
the highest probability (.5) suggests a predicted score of "0", while the
variable "score" is located between cut points 1 and 2, such that its
predicted score is 1.

5— The CCP is 55% (41/75)

II. CCP for the Interdependence Model.
TABLE 22

Interdependence Model: Cases Correctly Predicted

 

Target Score Fail. NoSuc. Psuc. Suc. CS PS CCP
1. Guatemala -0.047578 .021 .225 .564 .190 1 2
2. Canada -0.237407 .026 .257 .554 .163 3 2
3. EC «1.336829 .073 .469 .397 .061 2 1
4. EC -0.772172 .043 .359 .495 .102 1 2
5. EC -2.676891 .232 .587 .164 .017 0 1
6. EC -4.76077 .709 .265 .025 .002 0 0 x
7. EC —1.332599 .073 .468 .398 .061 1 1 x
8. EC —1.800238 .112 .541 .308 .039 1 1 x
9. Taiwan -0.414732 .031 .29 .54 .14 3 2
10. EC —1.515468 .087 .499 .363 .051 2 1
11. Japan -1.037186 .055 .412 .452 .08 3 2
12. Japan 0.6387447 .011 .130 .541 .318 2 2 x
13. USSR -2.280488 .169 .583 .223 .025 1 1 x
14. Canada —2.09163 .144 .576 .255 .03 0 1
15. EC -3.869029 .499 .438 .058 .005 1 1 x
16. Japan -1.455447 .082 .489 .374 .054 1 1 x
17. Argentina —4.545929 .662 .305 .03 .003 1 0
18. Japan -1.5218 .087 .501 .361 .051 1 l x
19. Korea -4.309597 .608 .351 .038 .003 1 0
20. EC ~3.478994 .403 .507 .083 .008 0 l
21. EC -2.649275 .227 .587 .168 .017 1 1 x
22. Argentina -5.093464 .772 .208 .018 .002 0 0 x
23. EC -2.54675 .21 .589 .182 .019 2 1
24 EC -3.100936 .316 .557 .115 .011 1 1 x
25. Canada 0.9102559 .008 .103 .509 .38 0 2
26. Brazil -1.441596 .081 .487 .377 .055 2 1
27. Japan 0.4134524 .014 .157 .558 .272 2 2 x
28. Korea 0.7189069 .01 .126 .532 .336 2 2 x
29. Brazil -1.652466 .098 .521 .336 .045 2 1
30. Portugal 0.5239633 .012 .143 .55 .294 0 2
31. Spain -2.616462 .222 .588 .173 .018 0 1
32. Taiwan -0.744051 .042 .354 .5 .105 2 2 x
33. Argentina -0.627502 .038 .331 .516 .116 2 2 x
34. Taiwan —0.744051 .042 .354 .5 .105 2 2 x
35. Japan -0.81252 .045 .367 .489 .097 1 2
36. Brazil -1.620482 .095 .516 .342 .047 2 1
37 Japan 2.424464 .002 .025 .237 .736 3 3 X
38. Korea -0.328545 .028 .274 .548 .151 3 2
39. Korea -1.368414 .076 .474 .391 .059 1 1 x
40. Argentina -1.529612 .088 .502 .36 .051 2 1
41. EC 0.5723392 .012 .138 .547 .304 3 2
42. Canada 1.395141 .005 .066 .43 .5 2 2 x
43 Taiwan 0.4775202 .013 .149 .554 .284 2 2 x
44 Taiwan 0.4775202 .013 .149 .554 .284 2 2 x
45. India -1.303328 .071 .463 .403 .063 2 1
46. EC -0.21064 .025 .252 .556 .166 1 2

245

 

(Table 22, continued)

Target Score Fail. Nosuc. Psuc. Succ. CS PS CCP
47. Brazil -2.325102 .176 .585 .216 .024 1 1 x
48. EC -3.252369 .35 .539 .101 .009 0 1
49. EC -3.578101 .427 .491 .076 .007 1 1 x
50. Korea 0.3386567 .015 .167 .561 .257 2 2 x
51. Korea —1.385417 .077 .477 .388 .058 2 1
52. Japan 2.200517 .002 .031 .277 .69 3 3 x
53. Korea 1.32442 .006 .071 .442 .481 2 2 x
54. Argentina —0.7597389 .043 .357 .497 .103 1 2
55. Japan -0.9308459 .05 .391 .47 .089 2 2 x
56. EC 0.3529504 .014 .165 .561 .26 3 2
57. EC «1.219505 .066 .447 .419 .068 2 1
58. Thailand —2.1466068 .151 .576 .245 .028 1 1 x
59. Brazil ~0.2982065 .027 .268 .55 .156 3 2
60. Japan 1.703313 .004 .05 .371 .575 3 3 x
61. Japan 1.703313 .004 .05 .371 .575 2 3
62. Japan 0.6634434 .011 .127 .538 .324 2 2 x
63. India -3.888287 .504 .434 .057 .005 0 0 x
64. India -3.888287 .504 .434 .057 .005 O 0 x
65. Norway -O.7079747 .041 .346 .505 .108 1 2
66. Canada -2.147396 .151 .576 .245 .028 1 1 x
67. EC -0.7319311 .041 .351 .501 .106 2 2 x
68. Thailand -3.261353 .352 .538 .1 .009 1 1 x
69. EC -1.689754 .101 .526 .329 .043 1 1 x
70. Thailand -2.20282 .159 .579 .236 .027 1 1 x
71. India —3.732554 .465 .463 .066 .006 1 1 x
72. India -3.732554 .465 .463 .066 .006 0 1
73. China -3.309284 .363 .532 .096 .009 1 1 x
74. China -1.948033 .127 .558 .28 .034 1 1 x
75. Taiwan -1.202381 .065 .444 .422 .069 1 1 x

 

246

III- CCP for Null Hypothesis
TABLE 23

Null Hypothesis: Empirical Estimates

Iteration 0: Log Likelihood =-97.193256

Ordered Logit Estimates Number of obs =
75
chi2(0) =
0.00
Prob > chi2 =
Log Likelihood = -97.193256 Pseudo R2 =
0.0000
Success { Coef. Std. Err. z P>}z{ [95% Conf.
Interval]
————————— +——.-——.———----—————-——-——————————-—-—————-———————-———-——————--—.
————————— +———-———-—-—-——————-————————-———-———————-———————————-————--—--—
_cutl : -1.658228 .3149704 (Ancillary parameters)
_cut2 : .1335314 .231455
cut3 1 1.871802 .3396831
scornul
1 to 75. 0

Pr[cutl < xb + p < cut2] => Pr (1) for all cases: the predicted score of
"0" is located. between "-1.658" and ".1335". This means that the
predicted score equals the coded score for 28 out of 75 cases, i.e., 37%
of the cases (CCP: 37%).

IV- CCP for the Compliance Model.

Legend: PS => Predicted Score
CS => Coded Score
CCP => Cases correctly predicted marked by an "X"

Predicted scores:

Pr[xb + p < —1.370799] => Pr(O)
Pr[-1.3707OO < xb + p < .9746096] => Pr(1)
Pr[.9746096 < Xb + p < 3.282908] => Pr(2)
Pr[3.282908 < xb + p] => Pr(3)

Compliance Model: Cases Correctly Predicted

Fail.

247

TABLE 24

NoSuc.

PSuC.

Succ.

Guatemala
Canada

EC

EC

EC

EC

EC

EC
Taiwan
EC

Japan
Japan
USSR
Canada

EC

Japan
Argentina
Japan
Korea

EC

EC
Argentina
EC

EC

Canada
Brazil
Japan
Korea
Brazil
Portugal
Spain
Taiwan
Argentina
Taiwan
Japan
Brazil
Japan
Korea
Korea
Argentina
EC

Canada
Taiwan
Taiwan
India

EC

Brazil

EC

EC

Korea
Korea
Japan
Korea
Argentina
Japan

EC

EC
Thailand

HOHNONWOHOOOOONNWHOOOU

2

0

.000214
.447955
.9690421
.022796
.5089093
.224504
.9768543
.2777954
.617579
.319283
.310781
.473805
.2247119
.486986
.293206
.462775
.385208
.462775
.4532568
1.
-0.
-0.
0.
-0.
3.
0.
3.
.23585
0.
l.
0.
1.
-0.
1.
2.
.0530283
.686565
.9550056
.1842879
.7884258
.554829
.236233
.05256
.05256
.5388017
.666428
.7375649
.2913507
.3820754
.974323
.382091
.849624
.676395
.0601462
.190304
.696945
.1014604
.225579

25547
9803149
5389073
2174751
0430708
064813
9114295
665023

0812783
199826
3026388
052014
0101231
052014
240316

.216
.071
.256
.081
.204
.081
.026
.211
.006
.089
.174
.358
.051
.01
.032
.032
.129
.115
.108
.254
.271
.034
.148
.005
.017
.193
.028
.044
.187
.069

.098
.423
.057
.194
.526
.373
.526
.399
.524
.399
.194
.525
.056
.416
.513
.495
.308
.084
.222
.222
.478
.461
.451
.526
.524
.235
.496
.048
.137
.521
.201
.282
.519
.368

.225
.445
.191
.422
.236
.422
.519
.229
.338
.406
.269
.13
.49
.417
.52
.52
.332
.355
.368
.193
.18
.518
.304
.309
.493
.248
.52
.503
.255
.449

.353
.038
.251
.17
.04
.113

l-‘NUJNHNwNNHOHHNNNNWNHWwwl-‘NNNOONNNNOHNOHOHPHHHOHNWNWHHOOHNUJH

NHNNHNWHNHHHHHNNNNHHHWHNNHNHNHNWHNHHHHHHNONHNHNNNNHNHHNHNN

X X

X><X X

 

 

(Table 24, continued)
Target Score
59. Brazil 1.426863
60. Japan 2.922531
61. Japan 2.922531
62. Japan 2.151814
63. India 51.554763
64. India —1.554763
65. Norway 0.2396255
66. Canada 1.819362
67. EC 0.821506
68. Thailand —0.174118
69. EC -0.0302645
70. Thailand 0.6651224
71. India -1.577698
72. India —l.577698
73. China -0.1535402
74. China 1.324411
75. Taiwan 0.4417549

CCP => 34/75

=> 4596

Fail.
.057
.013
.013
.029
.546
.546
.167
.04
.1
.232
.207
.115
.552
.552
.228
.063
.140

248

NoSuc.
.331
.111
.111
.207
.38
.38
.509
.261
.438
.527
.525
.461
.376
.376
.527
.35
.49

PSuc.
.476
.464
.464
.52
.066
.066
.279
.512
.383
.21
.233
.355
.065
.065
.213
.463
.315

Succ.
.135
.411
.411
.244
.008
.008
.046
.188
.079
.03
.035
.068
.008
.008
.031
.124
.055

CS

H*4PJO'JFJH1453914c30boniw£fl

PS

HbOPJOCDPJH+4P4N14c>orokiwt0

CCP

><X X><X

X><X

 

APPENDIX F

Logit Estimates
1. Bayard and Elliot’s (1994) Categories.

The following estimates presents a logit analysis in with two categories for the

 

dependent variable: Failure and nominal success, as well as partial success and success

are collapsed into one category. This classification for the dependent variable

 

corresponds to the one used by Bayard and Elliot (1994).

TABLE 25

Logit Analysis: Success/Failure

 

 

 

Number of obs = 75

Chi2(5) = 24.32

Prob > chi2 = 0.0002

Pseudo R2 = 0.2347

Log Likelihood = -39.658535

Success Coef. Std. err. z P > :2:
INSCONST -.9166 .7115 -l.288 0.198
LTEXP .2671 .2222 1.202 0.229
DVCGOOD —.5331 .5688 -.937 0.349
THREAT 1.7774 .5859 3 .033 0.002
TBAL —.0455 .0298 -1.529 0.126
_cons -l .3084 .6532 —2.003 0.045

 

 

249

250

11. Failure as a Separate Category.
This second analysis creates a special category for cases classified as failures.
Other three categories are collapsed.

TABLE 26

Logit Analysis: Failure

 

 

Number of obs = 75

Chi2(5) = 13.33

Prob > chi2 = 0.0205

Pseudo R2 = 0.2021

Log Likelihood = -26.312203
Success Coef. Std. err. z P > {2)
INSCONST -1.8885 .8729 -2.163 0.031
LTEXP .354 .2348 1.502 0.132
DVCGOOD - . 9504 .7281 -1.305 0. 192
THREAT .4194 .7446 .563 0.573
TBAL -.0179 .0435 -.411 0.681
_cons 2.0812 .8715 2.388 0.017

 

 

251

III. Success as a Separate Category

This third analysis creates a separate category for cases of full success. Other

 

three categories are collapsed.
TABLE 27

Logit Analysis: Success

Number of obs = 75
Chi2(5) = 12.03

Prob > chi2 = 0.0344
Pseudo R2 = 0.2185

Log Likelihood = -21.506364

 

 

Success Coef. Std. err. 2 P > :z}
INSCONST -1.4391 1.2804 -1.124 0.261
LTEXP .1631 .3533 .462 0.644
DVCGOOD -1.2511 .9304 -1.345 0.179
THREAT 1.6569 .8535 1.941 0.052
TBAL -.0384 .0331 -1.162 0.245
_cons -2.9669 1.0905 -2.721 0.007

 

 

252
IV. Ordered Multinomial Logit with Three Categories

Bayard and Elliot (1993:63) discuss the difficulty of classifying a cases into a
"nominal" or a "partial" success. This prodded me to collapse the these two middle

categories as see if the results are very different that a model with four categories.

TABLE 28
Ordered Multinomial Logit Analysis With Three Categories

Number of obs.: 75

 

 

 

Chi2(10): 20.88

Prob > chi2: .0009

Pseudo R2: .1724

Log Likelihood: —50.10218
Success Coef. Std. err. z P > (z:
INSCONST -1.5146 .7026 -2.156 0.031
LTEXP .3221 .2073 1.554 0.12
DVCGOOD —.8941 .5507 -1.624 0.104
THREAT 1 . 1224 .5705 l .968 0.049
TBAL -.0259 .0228 -1.133 0.257
_cutl -1.6689 .6855
_cut2 2.8849 .7669
_cut3 3.2829 .6857

 

 

The results are similar, but the model with four categories yields better results.
By collapsing the middle two categories, three variables are now not significant at the

.1 level.

 

 

 

LIST OF REFERENCES

REFERENCES

ALLISON, Graham T. , 1971, Essence of Decision: Explaining the Cuban Missile Crisis,
Boston, Little, Brown.

ALMOND, Gabriel A., and G. Bingham POWELL, Jr., 1978, Comparative Politics:
Systems, Processes, and Policy, Boston, Little, Brown.

ALMOND, Gabriel, and Sidney VERBA, 1963, The Civic Culture: Political Attitudes
and Democracy in Five Nations, Princeton, Princeton University Press.

ALT, James E. , 1987, "Crude Politics: Oil and the Political Economy of Unemployment
in Britain and Norway, 1970—1985, " British Journal of Political Science, Vol. 17,
pp. 149-99.

ALT, James E., and Barry EICHENGREEN, 1989, "Parralel and Overlapping Games:
Theory and an Application to the European Gas Trade, " Economics and Politics,
Vol. 1, pp.119-44.

ART, Robert J ., 1973, "Bureaucratic Politics and American Foreign Policy: A Critique, "
Policy Sciences, Vol. 4, pp. 467-90.

ARTIS, Michael, and Sylvia OSTRY, 1986, International Economic Policy Coordination,
Chatham House Papers, London, Routledge & Kegan Paul.

BALDWIN, David A., 1979, "Power Analysis in World Politics: New Trends Versus
Old Tendencies," World Politics, Vol. 31, pp. 161-94.

, 1980, "Interdependence and Power; A Conceptual Analysis," International
Organization, Vol. 34, No 4, pp. 471-506.

BARTOS, Otomar J., 1977, "Simple Model of Negotiation: A Sociological Point of
View," Journal of Conflict Resolution, Vol. 21, No 4, pp.565-79.

, 1974, Process and Outcome of Negotiations, New York, Columbia University
Press.

BAYARD, Thomas O. , and Kimberly Ann ELLIOTT, 1994, Reciprocity and Retaliation
in U. S. Trade Policy, Washington, DC, Institute for International Economics.

253

254

BENNETT QUINONES, Sherry, and Scott GATES, 1995, "Economic Risk and the
Politics of Protectionism," International Interactions, Vol. 21, No 1, pp. 63-83.

BHAGWATI, Jagdish, and Hugh Patrick, eds., 1990, Aggressive Unilateralism:
America ’s 301 Trade Policy and the World Trading System, Ann Arbor,
University of Michigan Press.

BINMORE and P. DASGUPTA, eds., 1987, The Economics of Bargaining, New York,
Blackwell.

BINMORE, K.G., A. RUBINSTEIN, and A. WOLINSKI, 1985, "The Nash Bargaining
Solution in Economic Modelling," ICERD Discussion Paper 85/112, London
School of Economics.

BOHARA, Alok, and William H. KAEMPFER, 1991, "A Test of Endogeneity in the
United States," American Economic Review, Vol. 80, pp. 952-6.

BRADY, Linda R, 1974, "Threat, Decision Time, and Awareness: The Impact of
Situational Variables on Foreign Policy Behavior, " Ph.D. Dissertation, Ohio State
University.

BUENO de MESQUITA, Bruce, 1990, "Multilateral Negotiations: A Spatial Analysis
of the Arab-Israeli Dispute," International Organization, Vol. 44, pp. 317-40.

, 1981, The War Trap, New Haven, Yale University Press.

BUENO de MESQUITA, Bruce, and David LALMAN, 1992, War and Reason:
Domestic and International Imperatives, New Haven, Yale University Press.

, 1990, "Domestic Opposition and Foreign War," American Political Science
Review, Vol. 84, pp. 747-65.

BUENO de MESQUITA, Bruce, D. NEWMAN, and A. RABUSHKA, 1985,
Forecasting Political Events: Hong Kong ’s Future, New Haven, Yale University
Press.

CALLAHAN, Patrick, 1982, "Commitment," in Patrick Callahan, Linda P. Brady, and
Margaret G. Hermann, eds. , Describing Foreign Policy Behavior, Beverly Hills,
CA, Sage Publications.

CAMERON, Duncan, 1988, ed., The Free Trade Deal, Toronto, James Lorimer &
Company, Publishers.

CAMERON, Maxwell, 1996, "From Free Rider to Free Trader: Modelling Trade
Negotiations with Mexico," in L. Perret, ed.,Vers une Amérique sans
frontiere/Towards a Borderless America, Montreal, Wilson & Lafleur.

255

CARLSNAES, Walter, 1992, "The Agency-Structure problem in Foreign Policy
Analysis," International Studies Quarterly, Vol. 36, pp. 245-70.

CARLSSON, H., 1986, "Perfect Equilibrium: A Paradox," Discussion Paper, Lund
University.

CHAN, Steve, "Mirror, Mirror on the Wall...: Are Freer Countries More Pacific,"
Journal of Conflict Resolution, Vol. 28, pp. 617-48.

CLAUDE, Innis L., Jr., 1962, Power and International Relations, New York, Random
House.

CLINE, Ray S., 1980, World Power Trends and U. S. Foreign Policy for the 1980’s,
Boulder, CO, Westview Press.

, 1975, World Power Assessement, Washington, DC, Georgetown Center for
Strategic and International Studies.

CONTINI, Bruno, 1958, "The Value of Time in Bargaining Negotiations: Some
Experimental Evidence," American Economic Review, Vol. 48, pp. 374-93.

COWHEY, Peter F., 1993, "Domestic Institutions and the Credibility of international
Commitments: Japan and the United States," International Organization, Vol. 47,
no 2, pp. 299-326.

CRAMTON, P.C., 1984, "Bargaining With Incomplete Information: An Infinite-Horizon
Model With Two-Sided Uncertainty," Review of Economic Studies, Vol. 51, pp.
579-93.

CRAWFORD, VP, 1987, "A Theory of Disagreement in Bargaining," in Ken Binmore
and Partha Dasgupta, eds., The Economics of Bargaining, New York, Basil
Blackwell, chapt. 7, pp. 119-53.

CRISPO, John, ed., 1988, Free Trade: The Real Story, Toronto, Gage Educational
Publishing Company.

CROSS, John G., 1991, "Economic Perspective," in Victor A. Kremenyuk, ed.,
International Negotiation: Analysis, Approaches, Issues, San Francisco, Jossey-
Bass Publishers, pp. 164-179.

, 1978, "Negotiation as a Learning Process," in I. William Zartman, ed., The
Negotiation Process: Theories and Applications, Beverly Hills, CA, Sage.

, 1969, The Economics of Bargaining, New York, Basic Books.

 

 

256

DAHL, Robert A., 1966, Political Oppositions in Western Democracies, New Haven,
CT, Yale University Press.

DALLEK, Robert, 1983, The American Style of Foreign Policy: Culture Politics and
Foreign Affairs, New York, Alfred A. Knopf.

De CALLIERES, Francois, translated by A.F Whyte, 1919 (1716), On the Manner of
Negotiating with Princes, Boston, Houghton Mifflin (Michel Brunet).

de FELICE, RB, 1976 (1778), "Negotiations or the Art of Negotiating," in I.W.
Zartman, The 50% Solution, New York, Anchor Books.

de RIVERA, Joseph, 1968, The Psychological Dimension of Foreign Policy, Columbus,
OH., C.E. Merrill Publishing Company.

DESSLER, David, 1989, "What’s at Stake in the Agent-Structure Debate? " International
Organization, Vol. 43, No 3, pp. 441-73.

DESTLER, I .M., 1995, American Trade Politics, 3rd ed., Washington, DC, Institute
for International Economics.

DESTLER, I.M., and Hisao MITSUYU, 1982, "Locomotives on Different Tracks:
Macroeconomic Diplomacy, 1977-1979," in I .M. Destler and Hideo Sato, eds.,
Coping With US-Japanese Economic Conflicts, Lexington, MA, Heath.

DESTLER, I.M., and John S. Odell, 1987, "Anti-Protection: Changing Forces in United
States Trade Politics," Policy Analyses in International Economics, Vol. 21,
Washington, DC, Institute for International Economics.

DEUTSCH, Karl W., et al., 1957, Political Community in North Atlantic area:
International Organization in the Light of Historical Experience, Princeton,
Princeton University Press.

DEUTSCH, Milton, 1973, The Resolution of Conflict: Constructive and Destructive
Process, New Haven, CT, Yale University Press.

DIXON, William J ., 1989, "Political Democracy and War: A New Look at an Old
Problem, " Paper presented at the annual meeting of the International Studies
Association, London, March-April.

DOBELL, Peter, 1980-81, "Negotiating with the United States," International Journal,
Vol. 36, No 1, pp. 17—69.

DOWNS, George W., and David M. ROCK, 1990, Tacit Bargaining, Arms Races, and
Arms Control, Ann Arbor, University of Michigan Press.

 

 

257

DOYLE, Michael W., 1986, "Liberalism and World Politics," American Political
Science Review, Vol. 80, No 4, pp. 1151-69.

DUCHESNE, Erick, and William Roberts CLARK, 1995, "A Formal Model of the
Canada-U.S. Free trade Agreement, " Prepared for presentation at the annual
conference of the Canadian Political Science Association, June.

DRUCKMAN, Daniel, 1978, "Boundary Role Conflict: Negotiation as Dual
Responsiveness" in I. William Zartman, ed., The Negotiation Process: Theories
and Applications, Beverly Hills, Sage, 1978.

DUPONT, Christophe, and Guy-Olivier Faure, 1991, "The Negotiation Process," in
Victor A. Kremenyuk, ed. , International Negotiation: Analysis, Approaches,
Issues, San Francisco, Jossey-Bass Publishers, pp. 40-57.

EASTON, DA, 1965, systems Analysis of Political Life, New York, Wiley.

EICHENBERG, Richard C., 1993, "Dual Track and Double Trouble: The Two-Level
Politics of INF, " in Peter B. Evans, Harold K. Jacobson, and Robert D. Putnam,
Double-Edged Diplomacy: International Bargaining and Domestic Politics,
Berkeley, University of California Press, pp. 45-76.

EPSTEIN, David, and Sharyn O’Halloran, 1996, "The Partisan Paradox and the U.S.
Tariff, " International Organization, Vol. 50. No 2.

EVANS, Peter B., 1993, "Building an Integrative Approach to International and
Domestic Politics: Reflections and Projections," in Peter B. Evans, Harold K.
Jacobson, and Robert D. Putnam, Double-Edged Diplomacy: International
Bargaining and Domestic Politics, Berkeley, University of California Press, pp.
397-430.

, 1979, Dependent Development: The Alliance of Multinational, State, and Local
Capital in Brazil, Princeton, Princeton University Press.

EVANS, Peter 8, Harold K. JACOBSON, and Robert D. PUTNAM, Double-Edged
Diplomacy: International Bargaining and Domestic Politics, Berkeley, University
of California Press.

FERRIS, Wayne H., 1973, The Power Capabilities of Nations States, Lexington,
Lexington Books.

FUDENBERG, D., D.K. LEVINE, and J. TIROLE, 1987, "Incomplete Information
Bargaining with Outside Opportunities," The Quarterly Journal of Economics, pp.
37-50.

258

FUDENBERG, D., and J. TIROLE, "Sequential Bargaining with Incomplete
Information, " Review of Economic Studies, Vol. 50, pp. 221-47.

GADDIS, John Lewis, 1982, Strategies of Containment: A Critical Appraisal of Postwar
American National Security Policy, New York, Oxford University Press.

GEORGE, Alexander L., 1980, "Domestic Constraints on Regime Change in U.S.
Foreign Policy: The Need for Legitimacy," in Ole R. Holsti, Randolph M.
Siverson, and Alexander L. George, eds., Change in the International System,
Boulder, CO, Westview Press.

, 1969, "The ’Operational Code’: A Neglected Approach to the Study of Political
Leaders and Decision-Making, " International Studies Quarterly, Vol. 13, pp. 190-
222.

GILPIN, Robert, 1981, War and Change in World Politics, Cambridge, Cambridge
University Press.

GOERTZ, Gary, Contexts of International Politics, Cambridge, Cambridge University
Press.

GOLDMAN N, Kjell, 1988, Change and Stability in Foreign Policy: The Problems and
Possibilities of Detente, Princeton, NJ, Princeton University Press.

, 1979, "The International Power Structure," in Kjell Goldmann and Gunnar
Sjostedt, Power, Capabilities, Interdependence, London, Sage.

GOUREVITCH, Peter, 1986, Politics in Hard Times: Comparative Responses to
International Economic Crises, Ithaca, NY, Cornell University Press.

, 1978, "The Second Image Reversed: The International Sources of Domestic
Politics", International Organization, Vol. 32, No 4, pp. 881-911.

GREENE, William H., 1993, Econometric Analysis, 2nd ed., New York, MacMillan.

HAAS, Ernst B., 1958, The Uniting of Europe: Political, Social, and Economic Forces,
1950-1957, Stanford, Stanford University Press.

, 1953, "The Balance of Power: Prescription, Concept, or Propaganda," World
Politics, Vol. 5, pp. 442-77.

HABEEB, William Mark, 1988, Power and Tactics in International Negotiation: How
Weak Nations Bargain With Strong Nations, Baltimore, The Johns Hopkins
University Press.

 

 

259

HAGAN, Joe D., 1993, Political Opposition and Foreign Policy in Comparative
Perspective, Boulder, CO, Lynne Rienner Publishers.

, 1986, "Domestic Political Conflict, Issue Areas, and Some Dimensions of
Foreign Policy Behavior Other Than Conflict," International Interactions, Vol.
12, pp. 291-313.

HAGGARD, Stephen, and Beth SIMMONS, 1987, "Theories of International Regimes, "
International Organization, Vol. 41, No 3, pp. 491-517.

HALPERIN, Morton H., 1974, Bureaucratic Politics and Foreign Policy, Washington,
DC, Brookings Institution.

HARSANYI, John C., 1977, Rational Behavior and Bargaining Equilibrium in Games
and Social Situations, New York, Cambridge University Press.

, 1967, "Games with Incomplete Information Played by Bayesian Players, I-III, "
Management Science, Vol. 14, 159-183, 320-334, 486-502.

HART, Jeffrey, 1976, "Three Approaches to the Measurement of Power in International
Relations," International Organization, Vol. 30, pp. 289-305.

HART, Michael (with Bill Dymond and Colin Robertson), 1994, Decision at Midnight:
Inside the Canada- US F ree-T rade Negotiations, Vancouver, University of British
Columbia Press.

HERMANN, Charles F., 1990, ”Changing Course: When Governments Choose to
Redirect Foreign Policy," International Studies Quarterly, Vol. 34, No 1, pp. 3-
21.

HERMANN, Margaret G., 1982, "Independence/Interdependence of Action, " in Patrick
Callahan, Linda P. Brady, and Margaret G. Hermann, eds. , Describing Foreign
Policy Behavior, Beverly Hills, Ca, Sage.

HERMANN, Margaret G., and Nathan KOGAN, 1977, "Effects of Negotiators’
Personalities on Negotiating Behavior, " in Daniel J. Druckman, ed. , Negotiations:
Social-Psychological Perspectives, Beverly Hills, CA, Sage.

HILSMAN, Roger, 1987, The Politics of Policy-Making in Defense and Foreign Policy:
Conceptual Models and Bureaucratic Politics, Englewood Cliffs, NJ, Prentice—
Hall.

HIRSCHMAN, Albert, 1945, National Power and the Structure of Foreign Trade,
Berkeley, University of California.

 

 

260

HOFFMANN, Stanley, 1975, "Notes on the Elusiveness of Modern Power,"
International Journal, Vol. 30, pp. 183-206.

, 1968, "Balance of Power, " in David L. Sills, ed., International Encyclopedia
of the Social Sciences, Vol. 1, New York, Mcmillan, p. 507.

HOLSTI, Kal J., 1982, "Restructuring Foreign Policy: A Neglected Phenomenon in
Foreign Policy," in Kal J. Holsti, ed., Why Nations Realign: Foreign Policy
Restructuring in a Postwar World, London, Allen Unwin.

, 1978, "A New International Politics? Diplomacy in Complex Interdependence, "
International Organization, Vol. 32, No 2, pp. 513-30.

HOMANS, GO, 1961, Social Behavior, New York, Harcourt, Brace & World.

HOPPMAN N, P. Terrence, 1978, "Asymmetrical Bargaining in the Conference on
Security and Cooperation in Europe, " International Organization, Vol. 32, No
1, pp. 141-77.

HUDSON, Valerie M., with Christopher S. VORE, 1995, "Foreign Policy Analysis
Yesterday, Today, and Tomorrow," Mershon International Studies Review, Vol.
39, Supp. 2, Pp. 209-38.

HUDSON, Valerie M., and Susan M. SIMMS, "Calculating Regime Response to
Domestic Opposition: Anti-Americanism Among U.S. Allies, " in David Skidmore
and Valerie Hudson, eds., The Limits of State Autonomy: Societal Groups and
Foreign Policy Formulation, Boulder, CO, Westview Press.

HUELSHOFF, Michael, 1993, "Domestic Politics and Dynamic Issue linkage. A
Reforrnulation of Integration Theory," Unpublished paper.

HUFBAUER, Gary Clyde, Jeffrey J. SCHOTT, and Kimberly Ann ELLIOTT, 1990,
Economic Sanctions Reconsidered, 2 vols., 2nd. ed., Washington, Institute for
International Economics.

IIDA, Keisuke, 1992a, "Do Negotiations Matter?: The Second Image Reversed in Two-
Level Games," Mimeo, Princeton University.

, 1992b, "Involuntary Defecton in Two-level Games," Mimeo, princeton
University.

, 1991, "Two-Level Games with Uncertainty: An Extension of Putnam’s Theory,"
Mimeo, Princeton University.

IKLE, Fred, 1964, How Nations Negotiate, New York, Harper and Row.

261

International Monetary Fund, Direction of Trade Statistics Yearbook, Washington (DC . ),
International Monetary Fund.

JAMES, Patrick, 1988, Crisis and War, Kingston and Montréal, McGill-Queen’s.

JAMES, Patrick, and John R. Oneal, 1991, "The Influence of Domestic and International
Politics on the President’s Use of Force," Journal of Conflict Resolution, Vol. 35,
No 2. PP. 307-32.

JOHNSTON, R., A. BLAIS, H. E. BRADY, and J. CRETE, 1992, Letting the People
Decide: Dynamics of a Canadian Election, Stanford, Stanford University Press.

JONES, Stephen B. , 1954, "The Power Inventory and National Strategy, " World Politics,
Vol. 6, PP. 421—52.

KATZENSTEIN, Peter J., 1985, Small States in World Markets: Industrial Policy in
Europe, Ithaca, NY, Cornell University Press.

, ed. , 1978, Between Power and Plenty: Foreign Economic Policies of Advanced
Industrial States, Madison, University of Wisconsin Press.

, 1976, "International Relations and Domestic Structures: Foreign Economic
Policies of Advanced Industrial States," International Organization, Vol. 30, No
1, PP. 145.

KEECH, William R., 1995, Economic Politics: The Costs of Democracy, Cambridge,
Mass., Cambridge University Press.

Keesing’s Contemporary Archives, London, Keesing’s Limited.

KEMPER Robert E., and Daniele Renee KEMPER, 1994, Negotiation Literature: A
Bibliographic Essay, Citations, and Sources, Metuchen, N.J., The Scarecrow
Press, Inc.

KEOHANE, Robert 0., 1989, International Institutions and State Power: Essays ion
International Relations Theory, Boulder, Westview Press.

KEOHANE, Robert O. , and Joseph S. NYE, 1977, Power and Interdependence, Boston,
Little, Brown.

KIERANS, E., and W. STEWART, 1988, Wrong End of the Rainbow: The Collapse
of Free Enterprise in Canada, Toronto, Collins.

KILGOUR, Mark D., 1991, "Domestic Political Structureand War Behavior: A Game-
Theoretic Approach," Journal of Conflict Resolution, Vol. 35, pp. 266-84.

262

KNOPF, Jeffrey W., 1993, "Beyond Two-Level Games: Domestic-international
Interaction in the Intermediate-Range Nuclear Forces Negotiations," International
Organization, Vol. 47, No 4, pp. 599-628.

KNORR, Klaus, 1975, The Power of Nations: The Political Economy of International
Relations, New York, Basic Books.

KRASNER, Stephen D., ed., 1983, International Regimes, Ithaca, NY, Cornel
University Press.

, 1978, Defending the National Interest: Raw materials Investments and U. S.
Foreign Policy, Princeton, Princeton University Press.

, 1972, "Are Bureaucracies Important? (or Allison Wonderland), " Foreign Policy,
Vol. 7, pp. 159-79.

KRAUS, Ellis S. , 1993, "U.S. Japan Negotiations on Construction and Semiconductors,
1985-1988: Building friction and relation-Chips," in Peter B. Evans, Harold K.
Jacobson, and Robert D. Putnam, eds., Double-Edged Diplomacy; International
Bargaining and Domestic Politics, Berkeley, University of California Press,
chapter 9, pp. 265-99.

KREMENYUK, Victor A., 1991, International Negotiation: Analysis, Approaches,
Issues, San Francisco, Joey-Bass Publishers.

KREPS, David M., 1990, A Course in Microeconomic Theory, Princeton, Princeton
University Press.

KRUEGER, Anne 0., 1995, American Trade Policy: A Tragedy in the Making,
Washington, DC, The AEI Press.

LAKOS, A., 1989, International Negotiations: A Bibliography, Boulder, CO, Westview
Press.

LALL, Arthur S., 1966, Modern International Negotiation: Principles, and Practice,
New York, Columbia University Press.

LAMBORN, Alan C., 1991, The Price of Power: Risk and Foreign Policy in Britain,
France, and Germany, Boston, Allen, Unwin.

, 1985, "Risk and Foreign Policy Choice," International Studies Quarterly, Vol.
29, pp. 385-410.

LAVERGNE, Réal P., 1983, The Political Economy of U.S. Tariffs: An Empirical
Analysis, Toronto, Academic Press.

263

LAX, DA, and J.K. SEBENIUS, 1986, The Manager As Negotiator, New York, Free
Press.

LEBOW, Richard Ned, 1996, The Art of Bargaining, Baltimore, The Johns Hopkins
University Press.

, 1981, Between Peace and War: The Nature of International Crisis, Baltimore,
Johns Hopkins University Press.

LEVY, Jack 8., 1989, "The Diversionary Theory of War: a Critique," in Manus
Midlarsky, ed., Handbook of War Studies, Boston, Unwin Hyman.

, 1988, "Domestic Politics and War," in Robert I. Rothberg and Theodore K.
Rabb, eds., The Origin and Prevention of Major Wars, New York, Cambridge
University Press.

, 1982, "Historical Trends in Great Power Wars, 1495-1975," International
Studies Quarterly, Vol. 26, pp. 278-300.

LIJPHART, Arend, 1977, Democracy in Plural Societies, New Haven, CT, Yale
University Press.

LIPSEY, Richard G., and Robert C. YORK, 1988, Evaluating the Free Trade Deal: A
Guided Tour Through the Canada— U. S. Agreement, Toronto, C. D. Howe Institute.

LISKA, George, 1957, International Equilibrium: A Theoretical essay on the Politics and
Organization of Security, Cambridge, MA, Harvard University Press.

LOCKHART, Charles, 1979, Bargaining in International Conflicts, New York,
Columbia University Press.

LOHMANN, Susanne, and Sharyn O’HALLORAN, 1994, "Divided Government and
U.S. Trade Policy: Theory and Evidence," International Organization, Vol. 48,
No 4, pp. 595-632.

LUEBBERT, Gregory M., 1986, Comparative Democracy: Policymaking and Governing
Coalitions in Europe and Israel, New York, Columbia University Press.

MADDALA, GS, 1993, Limited Dependent and Qualitative Variables in Econometrics,
New York, Cambridge University Press.

MAGEE, Stephen P., William A. BROCK, and Leslie YOUNG, 1989, Black Hole
Tarifi‘s and Endogenous Policy Theory, Cambridge, Cambridge University Press.

264

MAOZ, Zeev, and N asrin ABDOLALI, 1989, "Regime Types and International Conflict,
1916-1976," Journal of Conflict Resolution, Vol. 33, No 1, pp. 3-35.

MAOZ, Zeev, and Bruce RUSSETT, 1993, "Normative and Structural Causes of
Democratic Peace, 1946-1986," American Political Science Review, Vol. 87, No
3, pp. 624-38.

McCLINTOCK, Charles G., Frank S. STEACH, and Linda J. KEIL, 1983, "The
Influence of Communication Upon Bargaining," in Paul B. Paulus, ed., Basic
Group Processes, New York, Springer-Verlag.

McCRAE, Donald M., and Debra P. STEGER, eds., 1988, Understanding the Free
Trade Agreement, Ottawa, The Institute for Research on Public Policy.

MCDONALD, Ronald H. , 1971, Party systems and Elections in Latin America, Chicago,
Markham.

McGINNISS, Michael D., and John T. WILLIAMS, 1993, "Policy Uncertainty in Two-
level Games: Examples of Correlated Equilibria, " International Studies Quarterly,
Vol. 37, pp. 29-54.

, 1992, "A Model of Domestic Coalitions and International Rivalry," Paper
Presented for presentation at the conference on " Linkage Politics and International
Conflict," University of California, Davis, May 8-9.

McKELVEY, Richard, and William ZAVOINA, 1975, "A Statistical Model for the
Analysis of Ordinal Level Dependent Variables," Journal of Mathematical
Sociology, Vol. 4, pp. 103-20.

McMILLAN, John, 1990, "Strategic Bargaining and Section 301," in Jagdish Bhagwati
and Hugh T. Patrick, Aggressive Unilateralism: America ’s 301 Trade Strategy and
the World Trading system, Ann Arbor, University of Michigan Press.

MO, Jongryn, 1993, "The Logic of Two-Level Games with Endogenous Domestic
Coalitions," Mimeo, University of Texas at Austin.

, 1990, "International Bargaining and Domestic Political Competition," Mimeo,
Stanford University.

MORAVCSIK, Andrew, 1993, "Introduction: Integrating International and Domestic
Theories of International Bargaining," in Peter B. Evans, Harold K. Jacobson,
and Robert D. Putnam, eds. , Double-Edged Diplomacy: International Bargaining
and Domestic Politics, Berkeley, University of California Press.

265

MORGAN, Clifton T. , and Kenneth N. BICKERS, 1992, " Domestic Discontent and the
External Use of Force, " Journal of Conflict Resolution, Vol. 36, No 1, pp. 25-52.

MORGAN, T. Clifton, and Valerie L. SCHWEBACH, 1992, "Take Two Democracies
and Call Me in the Morning: A Prescription for Peace," International
Interactions, Vol. 17, pp. 305-20.

MORGAN, T. Clifton, and Sally Howard CAMPBELL, 1991, "Domestic Structure,
Decisional Constraints, and War: So Why Kant Democracies Fight?, " Journal of
Conflict Resolution, Vol.35, pp. 187-211.

MORGENTHAU, Hans J ., revised by Kenneth W. Thompson, 1985, Politics Among
Nations: The Struggle for Power and Peace, 6th ed., New York, Alfred A.
Knopf.

MORROW, James D., 1994, Game Theory for Political Scientists, Princeton, Princeton
University Press.

, 1991, "Electoral and Congressional Incentives and Arms control,’ Journal of
Conflict Resolution, Vol. 35, pp. 245-65.

, 1985, "Moving Forward in Time: Paths Toward a Dynamic Utility Theory of
Crisis Decisions," in Urs Luterbacher, and Michael D. Ward, eds., Dynamic
Models of International Conflict, Boulder, Lynne Rienner.

MUELLER, John, 1989, Retreat From Doomsday: The Obsolescence of Major War, New
York, Basic Books.

NASH, J.F., 1953, "Two-Person Cooperative Games," Econometrica, Vol. 21, pp. 128-
40.

, 1950, "The Bargaining Problem," Econometrica, Vol. 18, pp. 155-62.

NOLAND, Marcus, 1997, "Chasing Phantoms: The Political Economy of USTR,"
International Organization, Vol. 51, No 3, pp. 365-87.

NYE, Joseph S. , 1988, "Old Wars and Future Wars: Causation and prevention, " Journal
of Interdisciplinary History, Vol. 18, pp. 581-90.

ODELL, John S. , 1993, "International Threats and Internal Politics: Brazil, the European
Community, and the United States, 1985-1987," in Peter B. Evans, Harold K.
Jacobson, and Robert D. Putnam, eds., Double-Edged Diplomacy: International
Bargaining and Domestic Politics, Berkeley, University of California Press, pp.
233-264.

266

Office of Technology Assessment, 1991, Competing Economies: America, Europe and
the Pacific Rim, OTA-ITE-498, Washington, US Government Printing Office.

ORGANSKI, A.F.K., 1968, World Politics, 2nd ed., New York, Alfred A. Knopf.

ORGANSKI, A.F.K., and Jacek KUGLER, 1980, The War Ledger, Chicago, The
University of Chicago Press.

OSBORNE, Martin J ., 1994, A Course in Game Theory, Cambridge, MA, MIT Press.
, 1990, Bargaining and Markets, Toronto, Academic Press, Inc.

OSTROM, Charles W., and Brian D. JOB, 1986, "The President and the Political Use
of Force," American Political Science Review, Vol. 80, No 2, pp. 541-66.

OSTRY, S., 1990, Government and Corporations in a Shrinking World: Trade and
Innovation Policies in the United States, Europe and Japan, New York, Council
on Foreign Relations Press.

PAGE, Glenn D., 1968, The Korean Decision, New York, Free Press.

PAHRE, Robert, 1993, "Who’s On First, What’s On Second, I Don’t Know’s On Third:
Actors and Institutions in Indeterminate Two-Level Games," Paper prpared for
the annual meeting of the Political Science Association, Washington, DC,
September 2-5.

, 1992, "Reactions, Reciprocity, and Retaliation: A Model of Tariff Policy
Linking State and Domestic Interests, " Mimeo, Institute of Public Policy Studies,
The University of Michigan.

PATTERSON, Thomas G., 1988, Meeting the Communist Threat: Truman to Reagan,
New York, Oxford University Press.

PAUL, T.V. , 1994, Asymmetric Conflicts: War Initiation by Weaker Powers, New York,
Cambridge University Press.

PERLMUTTER, Amos, 1974, "The Presidential Political Center and Foreign Policy; A
Critique of the Revisionist and Bureaucratic-Political Orientations," World
Politics, Vol. 27, pp. 87-106.

PRUITT, Dean G., 1981, Negotiation Behavior, New York, Academic Press.

PUTNAM, Robert D., 1988, "Diplomacy and Domestic Politics: The Logic of Two-
Level Games," International Organization, Vol. 42, No 3, pp. 427—60.

267

PYE, Lucian W. , and Sidney VERBA, eds., Political Culture and Political Development,
Princeton, Princeton University Press.

RAIFFA, Howard, 1982, The Art and Science of Negotiation, Cambridge, MA, Harvard
University Press.

RAPOPORT, A., 1960, Fights, Games, and Debates, Ann Arbor, University of
Michigan Press.

Republic of China, Taiwan Statistical Data Book, Taipei, Economic Planning Council,
Executive Yuan.

, 1993, Monthly Bulletin of Statistics of the Republic of China, Vol. 20, No 1
(Jan.), Taipei, Directorate-General of Budget, Accounting & Statistics, Executive
Yuan.

RODRICK, Dani, 1994, "The Rush for Free Trade in the Developing World: Why So
Late? Why Now? Will it Last?," in Stephen B. Haggard and Steven B. Webb,
eds., Voting for Reform, Oxford, Oxford University Press.

ROMER, Thomas, and Howard ROSENTHAL, 1978, "Political Resource Allocation,
Controlled Agendas, and the Status Quo," Public Choice, Vol. 33, No 4, pp. 27-
44.

ROSECRANCE, Richard, 1963, Action and Reaction in World Politics, Boston, Little,
Brown.

ROSEN, 1972, "War Power and the Willingness to Suffer," in Bruce M. Russett, ed.,
Peace, War and Numbers, Beverly Hills, Sage Publications.

, 1970, "A Model of War and Alliance, " in Julian Friedman, et al, eds., Alliance
in International Politics, Boston, Allyn and Bacon.

ROSENAU, James N., 1973, Theorizing Across Systems: Linkage Politics Revisited,"
in Jonathan Wilkenfeld, ed. , Conflict Behavior and Linkage Politics, New York,
David McKay.

, 1969, Linkage Politics: Essays on the Convergence of National and International
Systems, New York, Free Press, 1969.

, 1966, "Pre-theories and Theories and Foreign Policy," in R. Barry Farrell, ed.,
Approaches to Comparative and International Politics, Evanston, 11, Northwestern
University Press.

268

ROTH, AB, 1985, Game-Theoretic Models of Bargaining, New York, Cambridge
University Press.

RUBIN, Jeffrey Z., and Bert R. BROWN, 1975, The Social Psychology of Bargaining
and Negotiation, New York, Academic Press.

RUBINSTEIN, Ariel, 1985, "A Bargaining Model with Incomplete Information About
Time Preferences," Econometrica, Vol. 53, September 1985, pp. 1151-72.

, 1982, "Perfect Equilibrium in a Bargaining Model," Econometrica, Vol. 50, pp.
97-107.

RUMMELL, Rudolph J ., 1983, "Libertarianism and International Violence, " Journal of
Conflict Resolution, Vol. 27, No 1, pp. 27-71.

RUSSETT, Bruce M ., 1990, Controlling the Sword: The Democratic Governance of
National Security, Cambridge, MA, Harvard University Press.

SALMORE, Barbara G., and Stephen A. SALMORE, 1978, "Political Regimes and
Foreign Policy," in Maurice A. East, Stephen A. Salmore and Charles F.
Hermann, eds. , Why Nations Act: Theoretical Perspectives Foreign Policy Studies,
Beverly Hills, CA, Sage.

SAMPSON, Martin, 1987, "Cultural Influences on Foreign Policy," in Charles F.
Hermann, Charles W. Kegley, Jr., and James N. Rosenau, eds., New Directions
in the Study of Foreign Policy, Boston, Allen and Unwin, pp. 384-408.

SARTORI, Giovanni, 1976, Parties and Party Systems: A Framework of Analysis,
Cambridge, Cambridge University Press.

, 1966, "European Political Parties: The Case of Polarized Pluralism, " in Joseph
LaPalombara and Myron Weiner, eds., Political Parties and Political
Development, Princeton, NJ, Princeton University Press.

SCHELLING, Thomas C., 1966, Arms and Influence, New Haven, Yale University
Press.

, 1960, The Strategy of Conflict, Cambridge, MA, Harvard University Press.
SCHNEIDER, Gerald, 1993, "Getting Closer at Different Speed; Strategic Interaction

in the Widening of European Integration, " in Pierre Allan, and Christian Schmidt,
eds. , Qualitative Games and International Relations, Cheltenharn, Edward Edgar.

269

SCHOPPA, Leonard J ., 1993, "Two-Level Games and Bargaining Outcomes: Why
Gaiatsu Succeeds in Japan in Some Cases and not Others," International
Organization, Vol. 47, No 3, pp. 353-86.

SCHRODT, Philip, 1995, "Event Data in Foreign Policy Analysis," in Laura Neack et
al., eds., Foreign Policy Analysis: Continuity and Change in Its Second
Generation, Englewood Cliffs, NJ, Prentice-hall, pp. 145-66.

SEBENIUS, James K., 1991, "Negotiation Analysis," in Victor A. Kremenyuk, ed.,
International Negotiation: Analysis, Approaches, Issues, San Francisco, Jossey-
Bass, pp. 203-15.

SHAMBAUGH IV, George E., "Dominance, Dependence, and Political Power:
Tethering Technology in the 19805 and Today," International Studies Quarterly,
Vol. 40, No 4.

SHAPIRO, R.Y., and B. PAGE, 1988, "Foreign Policy and the Rational Public,"
Journal of Conflict Resolution, Vol. 32, No 2, pp. 21147.

SHEPSLE, Kenneth A. , and Barry R. WEINGAST, 1987, "The Institutional Foundations
of Committee Power," American Political Science Review, Vol. 81, No 1, pp. 85-
104.

SHORTELL, A. 1991, Money Has No Country: Behind the Crisis in Canadian Business,
Toronto, MacMillan Canada.

SIEBE, Wilfred, 1991, "Game Theory," in Victor Kremenyuk, ed., International
Negotiation: Analysis, Approaches, Issues, San Francisco, Jossey-Bass, pp. 180-
202.

SINGER, Eric, and Valerie M. HUDSON, eds. , 1992, Political Psychology and Foreign
Policy, Boulder, CO, Westview Press.

SINGER, J. David, Stuart BREMER, and John STUCKEY, 1972, "Capability
Distribution,, Uncertainty, and major Power War, " in Bruce Russett, ed. , Peace,
War, and Numbers, Beverly Hills, Sage Publications.

SJOSTEDT, Gunnar, 1991, "Trade Talks," in Victor A. Kremenyuk, ed., International
Negotiation: Analysis, Approaches, Issues, San Francisco, Jossey-Bass, pp. 315-
30.

SMALL, Melvin, and J. David SINGER, 1976, "The War-Proneness of Democratic
Regimes, 1816-1965," Jerusalem Journal of International Relations, Vol. 1, No
1, pp. 50-69.

270

SNYDER, Glenn H., and Paul DIESING, 1977, Conflict Among Nations: Bargaining,
Decision Making, and System Structure in International Crises, Princeton, NJ,
Princeton University Press.

SNYDER, Jack, 1991, Myths of Empire: Domestic Politics and International Ambition,
Ithaca, NY, Cornell University Press.

SNYDER, Richard C., H.W. BRUCK, and Burton SAPIN, 1963, Foreign Policy
Decision-Making: An Approach to the Study of International Politics, Glencoe,
ILL., Free Press.

, 1954, Decision-Making as an Approach to the Study of International Politics,
Foreign Policy Analysis Project No 3, Princeton, Princeton University Press.

SOBEL, J ., and I. TAKAHASHI, 1983, "A Multistage Model of Bargaining," Review
of Economic Studies, Vol. 50, pp. 411-26.

SPECTOR, Bertram I., 1978, "Negotiation as a Psychological Process," in I. William

Zartman, ed. , The Negotiation Process, Beverly Hills, Sage Publications, chapt.
3, Pp. 55-66.

SPROUT, Harold, and Margaret SPROUT, 1971, Toward a Politics of Planet Earth,
New York, Van Nostrand.

, 1965, The Ecological Perspective on Human Afiairs: With Special Reference to
International Politics, Princeton, Princeton University Press.

, 1957, "Environment Factors in the Study of International Politics, Journal of
Conflict Resolution, Vol. 1, pp. 309-28.

, 1956, Man-Milieu Relationship Hypotheses in the Context of International
Politics, Princeton, Princeton University Press.

STaHL, Ingolf, 1994, "The Rubinstein and Stahl Bargaining Models: A Comparison and
an Attempt at a Synthesis," EFI Research Paper 6535, Stockholm, Stockholm
School of Economics.

, 1988, "A Comparison Between the Rubinstein and Stahl Bargaining Models,"
EFI Research Paper 6347, Stockholm, Stockholm School of Economics.

, 1972, Bargaining Theory, Stockholm, Stockholm School of Economics.

271

STEIN, Janet Gross, 1993, "The Political Economy of Security Agreements: The Linked
Costs of Failure at Camp David," in Peter B. Evans, Harold K. Jacobson, and
Robert D. Putnam, eds. , Double-Edged Diplomacy: International Bargaining and
Domestic Politics, Berkeley, University of California Press, pp. 77-103.

STEINBRUNER, John D., 1974, The Cybernetic Theory of Decision, Princeton, NJ,
Princeton University Press.

STEPAN, A., and C. SKACH 1993. "Constitutional Frameworks and Democratic
Consolidation: Parliamentarianism versus Presidentialism," World Politics, Vol.
46, pp. 1-22.

STOHL, Michael, 1980, "The Nexus of Civil and International Conflict," in Ted Gurr,
ed., Handbook of International Conflict, New York, Free Press.

SUTTON, J. , 1986, "Non-Cooperative Bargaining Theory: An Introduction," Review of
Economic Studies, Vol. 53, pp. 709-24.

THOMPSON, William R., 1988, On Global War: Historical-Structural Approaches to
World Politics, Columbia, University of South Carolina Press,

TOLLISON, RD, and TD. WILLETT, 1979, "An Economic Theory of Mutually
Advantageous Issue Linkages in International Negotiations," International
Organization, Vol. 33, pp. 425-49.

TROUT, B. Thomas, 1975, "Rhetoric Revisited: Political Legitimation and the Cold
War," International Studies Quarterly, Vol. 19, pp. 251-84.

TSEBELIS, George, 1990, Nested Games, Rational Choice in Comparative Politics,
Berkeley, University of California Press.

TYSON, Laura D’Andrea, 1992, Who ’3 Bashing Whom? Trade Conflict in High-
Technology Industries, Washington, Institute for International Economics.

UNDERDAL, Arild, 1991, "The Outcomes of Negotiation," in Victor A. Kremenyuk,
ed., International Negotiation: Analysis, Approaches, Issues, San Francisco,
Jossey-Bass, pp. 100-115.

United States Central Intelligence Agency, 1979/80, National Basic Intelligence Fact
Book, Washington, DC, Central Intelligence Agency.

U.S. Government, 1994, Economic Report of the President, Washington, DC, US
Government Printing Office.

272

U.S. House of Representatives, Committee on Ways and Means, 1973, Hearings, 93rd
Congress, lst Session, HR. 6767 (9 May).

VASQUEZ, John A., 1987, "Foreign Policy, Learning, and War," in Charles F.
Hermann, Charles W. Kegley, and James N. Rosenau, eds., New Directions in
the Study of Foreign Policy, Boston, Allen and Unwin.

VIOTTI, Paul R. , and Mark V. KAUPPI, 1987, International Relations Theory: Realism,
Pluralism, Globalism, New York, Macmillan.

WAGNER, R. Harrison, 1988, "Economic Interdependence, Bargaining Power, and
Political Influence," International Organization, Vol. 43, No 3.

WALT, Stephen M., 1985, "Alliance Formation and the Balance of World Power,"
International Security, Vol. 9, pp. 3-43.

WALTZ, Kenneth N., 1979, The Theory of International Politics, Reading, MA,
Addison-Wesley.

, 1959, Man, the State, and War: A Theoretical Analysis, New York, Columbia
University Press.

WARD, Michael Don, and U. WIDMAIER, 1982, "The Domestic-International Conflict
Nexus: New Evidence and Old Hypotheses," International Interactions, Vol. 9,
No 1, pp. 75-101.

WEEDE, Erich, 1984, "Democracy and War Involvement," Journal of Conflict
Resolution, Vol. 28, pp. 649-64.

WENDT, Alexander, 1987, "The Agent-Structure Problem in International Relations
Theory," International Organization, Vol. 41, No 3, pp. 335-70.

WILKENFELD, J., and Dina A. ZINNES, 1973, "A Linkage Model of Domestic
Conflict Behavior," in J. Wilkenfeld, ed. , Conflict Behavior and Linkage Politics,
New York, David McKay, pp. 325-56.

WILKINSON, B. W, 1993, "Trade Liberalization, the Market Ideology, and Morality:
Have We a Sustainable System? " in Richardo Grinspun, and Maxwell A.
Cameron, eds., The Political Economy of North American Free Trade, New York,
St. Martin’s Press, pp. 27-43.

World Bank, World Tables, Baltimore, Johns Hopkins University Press for the World
Bank.

273

WRIGGINS, H., 1987, "Up for Auction, " in I. W. Zartman, ed., The 50% Solution:
How to Bargain Successfully with Hijackers, Strikers, Bosses, Oil Magnates,
Arabs, Russians, and Other Worthy Opponents in This Modern World, New
Haven, Yale University Press.

WRIGHT, Quincy, 1942, A Study of War, Chicago, University of Chicago Press.

YOUNG, Oran, 1975, Bargaining: Formal Theories of Negotiations, Urbana, IL,
University of Illinois Press.

ZARTMAN , I. William, 1991, "The Structure of Negotiation, " in Victor A. Kremenyuk,
ed., International Negotiation: Analysis, Approaches, Issues, San Francisco,
Jossey-Bass Publishers, pp. 65-77.

, 1977, "Negotiation as a Joint Decision-making Process," Journal of Conflict
Resolution, Vol. 21, pp. 619-38.

, 1971, The Politics of Trade Negotiations Between Africa and the EEC,
Princeton, NJ, Princeton University Press.

ZARTMAN, I. William, and M. BERMAN, The Practical Negotiator, New Haven, Yale
University Press.

ZINNES, Dinna A., 1980, "Why War? Evidenc on the Outbreak of International
Conflict, " in Ted Robert Gurr, ed. , Handbook of Political Conflict: Theory and
Research, New York, Free Press, pp. 331-60.

ZINNES, Dina A., et al., 1961, "Capability, Threat and the Outbreak of War, " in James
Rosenau, ed., International Politics and Foreign Policy: A Reader in Research
and Theory, New York, Free Press.