,f_-'

llIllllllllIllIll!IllHill||||llllHllllllllllHlllllllllllllll

31293 01782 5005

LIBRARY
Michigan State
University

 

 

 

 

This is to certify that the

dissertation entitled

THE SYSTEMIC DEMOCRATIC
PEACE

presented by

Sara McLaughlin

has been accepted towards fulfillment
of the requirements for

Ph.D. Political Science

 

 

degree in

gflffi

 

Major professor

12/11/47

Date

 

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

 

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

WI! mas-nu

THE SYSTEMIC DEMOCRATIC PEACE
By

Sara McLaughlin

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
THE SYSTEMIC DEMOCRATIC PEACE
By

Sara McLaughlin

In this study, I develop a systemic level theory of the relationship between
democracy and war based on Kant’s writings on perpetual peace. I argue that the
relationship between the proportion of democracies in the international system and the
proportion of nations fighting interstate wars is both endogenous and evolutionary. War
creates conditions that favor the development of democracies and a rule of law among
nations because the experience of warfare improves the internal institutions of states
over time and because democracies are more likely to win the wars they fight, allowing
them to promote democracy in the aftermath of war. Peaceful relations between
democracies creates greater peace and furthers the spread of democracy and democratic
norms in the international system. The relationship between democracy and war at the
systemic level is also evolutionary; the pacific impact of democracy on war increases
over time as the system moves closer to Kant’s federation of free states. In addition, I
assert that while war acts as a force for the development of democracy, its substantive
impact will decline over time as the democratic rule of law expands internationally and
as the frequency of war declines over time. I also examine the impact of three
exogenous variables on war and democracy at the systemic level: major power
capability concentration, systemic trade, and world production. I argue that increases in

concentration and global trade make war in the system less likely, whereas increases in

world production are positively related to the growth in the proportion of democracies.
I test these theoretical propositions with time series data from 1816-1988. I utilize two
estimation techniques: 1) Full Information Maximum Likelihood estimation of two
simultaneous equations models, and 2) Sliding window estimation, where I split the
sample into 30, 40 and 50-year overlapping sub-samples. The latter technique is
employed to examine the evolutionary relationship between democracy and war in the
international system.

My empirical analysis demonstrates that the systemic level relationship between
the proportion of democracies and the proportion of nations fighting interstate wars is
indeed endogenous, with democracy having a significant and negative impact on war;
war also has a significant and positive effect on the proportion of democracies. This
empirical finding of a positive relationship between war and democracy contradicts
many other studies in the democratic peace literature that posit a negative relationship
between these variables. Also, the largest wars in the sample time period produce the
greatest structural breaks in the models. For example, sliding window estimation of the
impact of democracy on war reveals that the relationship does not become statistically
significant until World War I, and the degree of the pacific impact of democracy
becomes strongest in the post-World War II era, as I predicted theoretically. I also find
empirical support for the power preponderance hypothesis, i.e., as the concentration of
power in the international system increases, the proportion of nations fighting war
decreases. Finally, the results do suggest that systemic trade is negatively related to
systemic conflict, while world production has little or no effect on trends in global

democracy.

For my mother

ACKNOWLEDGMENTS

The completion of this dissertation would not have been possible without the
support and guidance of many people in my life. I would like to thank foremost my
committee members: Scott Gates, Gretchen Hower, Jim Granato, and Paul Abramson.
This project has its genesis in a conversation I had with Scott about his masters thesis
on the democratic peace, a conversation which resulted in numerous co-authored papers.
I am incredibly grateful to Scott for always pushing me to make my work better, and for
helping me to understand the importance of the broader theoretical debates in
international relations. We have developed a very special fiiendship in the last six
years, and I owe much of my success to Scott’s constant support and guidance. Scott
always makes me laugh, and that light-hearted Spirit is what makes it such a joy to work
with him.

I am also grateful for Gretchen Hower’s support and encouragement throughout
graduate school. In my opinion, she was the most outstanding professor in the Political
Science Department at Michigan State, going well above and beyond the call of duty to
meet with students and to provide detailed comments on written work. She is also a
good friend and I will always cherish the memories of playing softball and poker

together.

I would like to thank Jim Granato for his encouragement and guidance. He is an
excellent teacher and mentor, and he helped me to increase my understanding of the
time series techniques utilized in this study. Jim is also a good friend, and I appreciate
the advice he has given me in the past few years. I would like to thank Paul Abramson
for his advice and support in my graduate career. He always finds a way to make me
laugh and he keeps me on my toes.

I am also grateful to numerous colleagues at Michigan State for their friendship
and support, especially Renee Agress, Philip Alderfer, Shelly Arsenault, Misha Barnes,
Sherry Bennett, Chris Butler, John Davis, Kristin Dawley, Cooper Drury, Erick
Duchesne, Michelle Kuenzi, Kimberly Ludwig, Jon Monroe, Bob Orme, Brandon Prins,
Malia Reddick, and Chris Sprecher. Thanks to Renee for keeping me happily distracted
from my work and for being a great friend. Thanks to Erick for always challenging me
intellectually, believing in me, and giving me a good challenge on a pool table. Thanks
to Shelly for always bringing a smile to my face. I will always remember the weekend
trips to Southfield.

Several colleagues outside Michigan State provided support throughout my
graduate career. I thank John Williams for giving me the opportunity to be his teaching
assistant at ICPSR and for aiding in the development of my time series training. I also
thank Nils Petter Gleditsch for giving me the opportunity to work in Norway. I
completed a large majority of my dissertation in Trondheim and 0130, and I am grateful
for the wonderful comments Nils Petter gave me on my work. I would also like to
thank Dina limes and Bob Muncaster for giving me the opportunity to work at the

Merriam Lab for one year. Several others deserve recognition as well for their support

vi

and encouragement throughout the past six years: Chris Achen, Kathy Barbieri, Mark
Crescenzi, Andrew Enterline, Havard Hegre, Paul Hensel, Kelly Kadera, Kevin Murrin,
Dennis Patterson, Bill Reed, Brian Sala, Harold Spaeth, Cameron Thies, Richard
Tucker, and Ken Williams.

Finally, I would like to thank my family members for always being there for me.
I am the strong, independent, and successful woman that I am today because of their
love and support. Special thanks to Mom, Dad, Becky, Bob, Jenny, Jill, Jodi, Justin,

Denise, Jay, Amanda, Emily, Thomas, Lisa, Grady, Paula, Jane, and Mike.

vii

TABLE OF CONTENTS

LIST OF TABLES ........................................................................... x
LIST OF FIGURES .......................................................................... xi
INTRODUCTION ............................................................................ 1
CHAPTER 1
THE DEMOCRATIC PEACE ............................................................. 7
Three Levels of Analysis ........................................................... 8
The Dyadic Democratic Peace ........................................... 9
The Monadic Democratic Peace ......................................... 13
The Systemic Democratic Peace ......................................... 14
Systemic Theorizing vs. Aggregation Across Levels of Analysis ........... 18
Conclusion ........................................................................... 21
CHAPTER 2
A SYSTEMIC THEORY OF WAR AND DEMOCRACY ........................... 26
Neorealism and its Critics ......................................................... 27
Beyond Neorealism ................................................................. 31
A Systemic Theory of Democracy and War .................................... 35
Conditions for Perpetual Peace .......................................... 35
War: A Force for Democracy ............................................. 39
A Comparison of Waltz and Kant ....................................... 42
The Diffusion of Democratic Rules and Norms ....................... 44
Testable Propositions ...................................................... 49
Power Concentration and Systemic War ......................................... 51
Trade and Systemic War ........................................................... 53
Production and Global Democracy ............................................... 54
Conclusion ........................................................................... 55
CHAPTER 3
GATHERING EVIDENCE ................................................................. 57
Endogenous Variables .............................................................. 57
Systemic War ............................................................... 58
Systemic Democracy ...................................................... 59
Exogenous Variables ............................................................... 62

viii

Systemic Capability Concentration ...................................... 62

Systemic Trade ............................................................. 63
World Production .......................................................... 63
The Issue of Stationarity ........................................................... 64
Conclusion ........................................................................... 65
CHAPTER 4
TESTING THE SYSTEMIC DEMOCRATIC PEACE: A STRUCTURAL
EQUATIONS APPROACH ................................................................ 86
Hendry’s Approach to Time Series Estimation ................................ 87
Model One: Doyle’s Measure of Democracy ................................... 88
Model Two: Polity III Measure of Democracy ................................. 93
The Issue of Structural Stability .................................................. 96
Conclusion .......................................................................... 99
CHAPTER 5
TESTING EVOLUTIONARY RELATIONSHIPS: SLIDING WINDOW
ANALYSIS .................................................................................. 110
Sliding Window Analysis ......................................................... 112
Conclusion ........................................................................... 1 17
CHAPTER 6
CONCLUSION .............................................................................. 135
The Systemic Level Relationship between Democracy and War ............ 137
The Evolution of Rules and Norms in the International System ............. 142
Economic Relations, Democracy, and War ..................................... 146
Systemic Power Concentration and War ........................................ 147
APPENDIX A ................................................................................ 149
APPENDIX B ................................................................................ 151
BIBLIOGRAPHY ........................................................................... 159

Table 1:

Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

Table 7:

Table 8:

Table 9:

Table 10:

Table 1 1:

Table 12:

Table 13:

LIST OF TABLES

Democratic Peace Research at the Monadic Level of Analysis
Democratic Peace Research at the Dyadic Level of Analysis ........
Democratic Peace Research at the Systemic Level of Analysis
Operationalization of Theoretical Concepts ............................
Summary Statistics ........................................................

Summary of the Integrated Order and Transformations for
each Series ..................................................................

Variable Legend ............................................................
FIML Estimates of Model One, 1816-1982 ............................
Summary of Model One Reduction .....................................
FIML Estimates of the Final Reducted Model One, 1816—1982
FIML Estimates of Model Two, 1816-1988 ...........................
Summary of Model Two Reduction .....................................

FIML Estimates of the Final Reduced Model Two, 1816-1988

23

24

25

66

67

68

101

102

103

104

105

106

107

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:

LIST OF FIGURES

The Proportion of System Members Fighting Interstate
Wars, 1816-1992 ........................................................... 70

ACF of the Proportion of System Members Fighting
Interstate Wars ............................................................. 71

PACF of the Proportion of System Members Fighting
Interstate Wars .............................................................. 72

The Proportion of Democracies in the International

System, 1816-1992 ........................................................ 73
ACF of the Proportion of Democracies (Doyle’s Measure) ......... 74
PACF of the Proportion of Democracies (Doyle’s Measure) ........ 75
ACF of the Proportion of Democracies (Polity III Measure) ........ 76
PACF of the Proportion of Democracies (Polity III Measure) ...... 77
Major Power Capability Concentration, 1816-1990 .................. 78
ACF of Major Power Capability Concentration ....................... 79
PACF of Maj or Power Capability Concentration ..................... 80
British Exports (logged), 1816-1988 .................................... 81
ACF of British Exports (logged) ......................................... 82
PACF of British Exports (logged) ....................................... 83
US Gross National Product (logged), 1816-1988 ..................... 84
ACF of US Gross National Product (logged) .......................... 85

xi

Figure 17:
Figure 18:
Figure 19:
Figure 20:
Figure 21:
Figure 22:
Figure 23:
Figure 24:
Figure 25:
Figure 26:
Figure 27:
Figure 28:
Figure 29:
Figure 30:
Figure 31:
Figure 32:
Figure 33:
Figure 34:
Figure 35:
Figure 36:

Figure 37:

Figure 38:

PACF of US Gross National Product (logged) ......................... 86

One Step Ahead Chow Tests for Model One .......................... 108
One Step Ahead Chow Tests for Model Two ......................... 109
The Evolving Impact of Democracy on War ........................... 119
The Evolving Impact of Democracy on War ........................... 120
The Evolving Impact of Democracy on War ........................... 121
The Evolving Impact of War on Democracy ........................... 122
The Evolving Impact of War on Democracy ........................... 123
The Evolving Impact of War on Democracy ........................... 124
The Evolving Impact of Concentration on War ........................ 125
The Evolving Impact of Concentration on War ........................ 126
The Evolving Impact of Concentration on War ........................ 127
The Evolving Impact of Concentration on War ........................ 128
The Evolving Impact of Concentration on War ........................ 129
The Evolving Impact of Concentration on War ........................ 130
Comparison of 30, 40, and 50 Year Sliding Windows ................ 131
Comparison of 30, 40, and 50 Year Sliding Windows ................ 132
Comparison of 30, 40, and 50 Year Sliding Windows ................ 133
Comparison of 30, 40, and 50 Year Sliding Windows ................ 134
Kalman Filter Estimates of the Impact of Democracy on War ....... 155

Kalman Filter Estimates of the Impact of War on Democracy,
Lag One ...................................................................... 156

Kalman Filter Estimates of the Impact of Concentration on
War, Lag Three ............................................................. 157

xii

Figure 39: Kalman Filter Estimates of the Impact of Concentration on
War, Lag Four .............................................................. 158

xiii

INTRODUCTION

The Cold War has ended, and we now have a chance to forge a
democratic peace, an enduring peace built on shared values—
democracy and political and economic freedom. The strength of these
values in Russia and the other new independent states will be the
surest foundation for peace—and the strongest guarantee of our
national security—for decades to come.

Secretary of State James Baker, February 5, 19921

The twentieth century saw the developed world descend into a
paroxysm of ideological violence, as liberalism contended first with the
remnants of absolutism, then bolshevism and fascism, and finally an
updated Marxism that threatened to lead to the ultimate apocalypse of
nuclear war. But the century that began full of self-confidence in the
ultimate triumph of Western liberal democracy seems at its close to be
returning full circle to where it started: not to an ‘end of ideology’ or a
convergence between capitalism and socialism, as earlier predicted, but
to an unabashed victory of economic and political liberalism
(Fukuyama, 19893).

The quotes above highlight one of the most striking changes in the international
system in the last two centuries, the tremendous growth of democracy in the world. The

number of democracies expanded from one in 1816 to 76 in 1992, and the proportion of

democracies increased substantially from 4% in 1816 to 42% in 1992.2 Today democracies

 

' This quote is taken from Russett (1993:128-9).

2 These figures are calculated with data from the Polity 111 data set (Jaggers and Gurr, 1995). I count as a
democracy any nation who scores six or higher on the Polity III democracy scale (ranging from 0-10).
The Polity III ordinal democracy measure combines information from several institutional characteristics
of a polity: the competitiveness of political participation, the level of executive constraints, and the
openness and competitiveness of chief executive recruitment. (Jaggers and Gurr, 1995:471) I also utilize

2
constitute a majority of nations in the international system (Gleditsch and Hegre, 1997:304-

5). The potential impact of the spread of democracy on international relations between
nations is profound. Many scholars and policy makers believe that the end of the Cold War,
coupled with the spread of democracy and liberalism, signals a new era of international
relations.

Perhaps the most significant consequence of the spread of democracy is the potential
for peace in the international system. “To the degree that countries once ruled by autocratic
systems become democratic, a striking fact about the world comes to bear on any discussion
of the future of international relations: in the modern international system, democracies have
almost never fought each other (Russett, 1993 24).” This notion of a democratic peace finds
support in a plethora of empirical studies in international relations. Not only do democracies
not fight wars against each other, they are also less likely to engage in militarized disputes
amongst themselves, and they are less likely to escalate militarized disputes between
themselves to war. In fact, peace between democracies has been called “the closest thing we
have to an empirical law in international relations (Levy, 1988:662)” and “one of the
strongest nontrivial or nontautological generalizations that can be made about international
relations (Russett, 1990: 123).”

Given the incredible growth in the number of democracies in the international system
and the overwhelming peace between democracies, one may posit that the very nature of

international relations will be transformed. Ray (1995:201) ponders this possibility:

 

Doyle’s (1986) measure of democracy, which is based largely on Kant’s conceptualization of a
republican government. See Chapter 3 for a more detailed description of these measures.

3
If relationships among democratic states are fundamentally different

from those among combinations of democratic and undemocratic states,
as well as those among uniformly undemocratic states, then a
significant trend towards democracy, even if it is restricted to the most
powerful states in the international system, could transform
international politics.
Chan (1997:59) also argues that the increasing trend in global democracy could alter the
nature of international interactions between states.
Democracies constitute for the first time in history a majority of the
states in the international system. Therefore, the norms governing their
relations have a better chance now than earlier to become the dominant
mode of interaction in world politics.

Ideas about the possible extension of the democratic peace to the international system
as a whole, stand in stark contrast to the views of neorealism, an influential theoretical
approach in international relations research in the last few decades.’ Neorealists assert that
the defining characteristic of the international system is anarchy, and that the distribution of
power in the international system is a major determinant of the level of systemic conflict.
The relationships between states have been and will continue to be characterized by anarchy,
the pursuit of power, and the security dilemma Contrary to advocates of the democratic
peace proposition, neorealists argue that the domestic structure of states in the international
system has little or no effect on outcomes in world politics. In addition, neorealists present a

theoretical picture of the international system that is enduring and unlikely to change in the

future.

 

3 I am interested in primarily the systemic level arguments stemming from the realist tradition, thus I
choose to focus on Waltz’s (1979) rendition of neorealism and subsequent research in the spirit of Waltz.

4
The primary goal of this thesis is to address the debate between advocates of the

democratic peace and defenders of the neorealist tradition,4 focusing on the systemic level
consequences of the long term growth in democracy, the outcomes of wars in the
international system, and peaceful relations between democracies. In particular, I address the
following questions: 1) What is the systemic level relationship between democracy and war,
and how is this relationship evolving over time? 2) Are the traditional rules and norms of the
international system described by neorealists being transformed as the international system
becomes more democratic? 3) How do economic relations between states in the system
influence the systemic level relationship between democracy and war over time? and 4) How
is the level of war in the international system related to the concentration of power between
nation-states, and how is this relationship changing over time?

I argue that the best way to answer these questions is to develop a systemic level
theory of the relationship between democracy and war. In Chapter 1, I review previous
work on the relationship between regime type and war at three levels of analysis (monadic,
dyadic, and systemic). Scholars have a tendency to view the systemic level relationship
between democracy and war as a puzzle that emerges from the somewhat contradictory
findings at the monadic and dyadic levels of analysis: 1) democracies do not fight wars
against each other (dyadic result), but 2) democracies participate in as many wars overall as
nondemocracies (monadic result). The systemic level relationship between democracy and

war is often deduced from empirical findings at other levels of analysis. The evolving,

 

‘ Several scholars have explicitly criticized the democratic peace proposition from a realist perspective
(e.g. Mearsheirner, I990; Layne, l994; Spiro, l994; Farber and Gowa, 1995).

5

theoretical systemic relationship between democracy and war is an extremely interesting
t0pic, but it has been given rather short shrifi in the current democratic peace literature.

In Chapter 2, I address this void in the literature by developing a systemic theory of
the relationship between democracy, war, trade, production, and power concentration. My
systemic theoretical model contributes to our understanding of the democratic peace
proposition in several ways. First, I argue that democracy and war are endogenously related
to each other over time. War creates conditions that favor the development of democracy and
a pacific rule of law among nations, while peaceful relations between democracies creates
greater peace and furthers the spread of democracy and democratic norms in the international
system. A vast majority of studies on the democratic peace study only the impact of
democracy on war, without considering how war affects the development of democracy.

Second, I model the relationship between democracy and war dynamically, going
beyond the tendency in international relations research to assume that a single model explains
the relationship between variables of interest over time. My theoretical model predicts that
the pacifying (and negative) impact of democracy on war in the international system will
become stronger over time. I also expect the positive impact of war on democracy to get
weaker over time as the number of democracies in the system increases. The third primary
contribution of my theoretical approach is the added emphasis on rules and norms of
behavior in the international system, particularly how these are evolving as a consequence of
the dynamic relationship between democracy and war.

Finally, the theory is developed at the systemic level, going beyond previous
tendencies to aggregate empirical findings fi'om the monadic and dyadic levels of analysis to

the systemic level. I argue that the empirical results on the democratic peace proposition at

6

the monadic level are questionable, and that a theory developed at the systemic level avoids
the problems of aggregating potentially faulty findings across levels.

In Chapter 3, I introduce data collected from 1816-1988 that is utilized to test the
theoretical propositions delineated in Chapter 2. In addition to presenting the basic time
series plots for each data series, I also run diagnostic tests for stationarity to determine
what transformations (if any) are needed to ensure that each series is characterized by a
constant mean and variance over time.

In Chapter 4, I test the systemic democratic peace theory with two simultaneous
equation empirical models that are estimated for the entire time period. Both models
provide support for my theoretical propositions, although the parameters for both models
are not structurally stable, i.e., they vary over time. To determine how the substantive
relationships between my variables are changing over time, I present the results of sliding
window analysis in Chapter 5 where the data is divided into 30, 40, and 50-year
overlapping samples. I find that the impact of democracy on war in the international
system has grown stronger over time, supporting my theoretical argument that democratic
rules and norms of interaction are diffusing throughout the international system over time.
I also find that war has a positive impact on democracy, and that this relationship is getting
somewhat weaker over time. I conclude in Chapter 6 with a summary of the major

findings in this thesis and I discuss several avenues for fiiture research and extensions.

CHAPTER 1

THE DEMOCRATIC PEACE

In this chapter, I review previous research on the democratic peace at three main
levels of analysis: the monadic level, the dyadic level, and the systemic level. Some scholars
assert that peace between democracies translates into greater peace in the international system
as the number of democracies increases. Others regard the systemic level relationship
between democracy and war as a puzzle that emerges from two significant empirical
findings at the monadic and dyadic levels: 1) democracies do not fight wars against each
other (dyadic result), but 2) democracies are just as war-prone in general as
nondemocracies (monadic result). While the results obtained at the systemic level through
aggregation across levels of analysis are interesting, they are limited theoretically. I argue
that our understanding of the evolving relationship between democracy and war in the
international system has been hampered by the lack of theorizing at the systemic level, and I
conclude the chapter with a discussion of why a systemic level theory of the democratic peace

is needed.

Three Levels of Analysis

One useful way of understanding international relations is to distinguish between
levels of analysis. Singer’s (1961) classic discussion of two levels of analysis, the
international system and the nation state, along with Waltz’s (1959) three images (man, the
nation state, and the state system), have provided international relations scholars with a
strong tradition of conceptualizing theoretical questions in terms of the level of analysis
adopted. Organizing studies of the democratic peace along these analytical lines has proven
to be just as fruitful.

Research on the relationship between regime type and international conflict has
typically been conducted at one of three primary levels of analysis: 1) the monadic or nation-
state level, 2) the dyadic or relational level, and 3) the systemic level. Monadic level research
of the democratic peace proposition examines the relationship between a state’s regime type
and its foreign policy behavior, asking the question “do democracies more frequently
maintain peace overall? (Gleditsch and Hegre, 1997283).” Dyadic level research examines
the interactions between pairs of states, posing the question “do democracies usually keep the
peace among themselves? (Gleditsch and Hegre, 1997283).” Finally, systemic level
research of the relationship between global democracy and systemic conflict focuses on the
following question: “is an international system with a high proportion of democratic states
more peaceful? (Gleditsch and Hegre, 1997:283)”

Tables 1-3 provide a brief summary of the research on the democratic peace

pr0position at each of these three levels of analysis.’ I briefly review the most significant

 

5 The tables and figures in this study are presented at the end of each chapter. It is not my intention to
provide a comprehensive survey of the democratic peace literature in this chapter. I focus instead on the

9
theoretical approaches and empirical findings at each level of analysis in the next three

sections. This is followed by a comparison of theoretical aggregation across levels of
analysis, versus the development of a theoretical model of the democratic peace at the

systemic level.

The Dyadic Demom Peace

The strongest evidence for peace between democracies stems from the dyadic level
research, i.e., the finding that democracies do not fight each other (Babst, 1964; Rummel,
1983; Chan, 1984; Weede, 1984; Maoz and Abdolali, 1989; Bremer, 1992; Bueno de
Mesquita and Lalman, 1992; Maoz and Russett, 1992; Morgan and Schwebach, 1992;
Weede, 1992; Bremer, 1993; Maoz and Russett, 1993; Russett, 1993; Rousseau, et a1 1996;
Oneal, et a1 1996; Gleditsch and Hegre, 1997; Oneal and Russett, 1997; Raknerud and Hegre,
1997). Studies in this tradition typically focus on the relationship between joint regime type
and three aspects of conflict: 1) the outbreak of dyadic militarized interstate disputes,6 2) the

escalation of dyadic militarized interstate disputes to war,7 and 3) the outbreak of interstate

 

empirical work most relevant to my research question, the relationship between regime type and
international conflict (including interstate wars and militarized disputes). Some of the important research
excluded from Table 1 includes the work on democratization and war (Mansfield and Snyder, 1995;
Enterline, 1996; Thompson and Tucker, 1997), democracy and internal conflict (e.g. Rummel, 1984;
Gelpi, 1997), regime type and foreign policy behavior (Haas, 1965; Rosenau, 1966; East and Gregg,
1967; Rummel, I968; Salmore and Hermann, 1969; Zinnes and Wilkenfeld, 1971; East and Hermann,
1974), democracy and foreign intervention (Kegley and Hermann, 1995; Hermann and Kegley, 1996),
democracy and international crisis (Hewitt and Wilkenfeld, 1996), and democracy and alliances
(Siverson and Emmons, I991; Farber and Gowa, 1995; Gaubatz, 1996; Simon and Gartzke, 1996),
among others. More comprehensive surveys of the democratic peace literature are contained in Ray
(1995) and Chan (1997).

6 Gochman and Maoz (19842586) define a militarized interstate dispute as “a set of interactions between or
among states involving threats to use military force, displays of military force, or actual uses of force. To be
included, these acts must be explicit, overt, nonaccidental, and government sanctioned.” I frequently adopt
the acronym MID in this study as a short hand for militarized interstate dispute (refening to the data
collected by the Correlates of War project).

7 The most common data set on war analyzed in the democratic peace literature is the Correlates of War

10

wars. Research at the dyadic level demonstrates that democracies are less likely to fight
militarized disputes amongst themselves, democracies are less likely to escalate disputes to
wars, and that interstate war between democracies is rare or nonexistent.8

In his fourth volume of Understanding War and Conflict (1979), Rummel developed
one of the first and most influential theoretical models of the democratic peace. His well
known “Joint-freedom proposition” (Rummel, 1983:29), posited that “Libertarian systems
mutually preclude violence (violence will occur between states only if at least one is
nonlibertarian; Rummel, 1979:277—279).” Rummel’s conceptualization of libertarianism was
broader than most definitions of democracy in the literature; it included a political (civil
liberties and political rights) and economic dimension (based on the level of government
decentralization). Analyzing data on foreign conflict collected from 1976-1980, Rummel
(1983240) discovered that “empirically there is no instance of violence between politically
free or free states.” Rummel’s work on the democratic peace inspired great interest and
debate in the relationship between regime type and conflict at both the monadic and dyadic
levels of analysis.

Evidence of interest in testing the dyadic relationship between democracy and conflict
inspired by Rummel is found in Stuart Bremer’s (1992) article, “Dangerous Dyads:
Conditions Affecting the Likelihood of Interstate War, 1816-1965.” Bremer compares seven

predictors of the outbreak of dyadic conflict: geographical proximity, power parity, power

 

(COW) data on interstate wars, i.e., wars between system members that exceed a threshold of 1000 battle
deaths (Small and Singer, 1982).

' Although some scholars would argue that democracies have never fought wars against one another (e.g.
Babst, 1964; Rummel, 1979; Doyle, 1986; Levy, 1988; Maoz and Russett, 1993), others have discussed
the possible “exceptions” to the rule (Ray, 1993; Gleditsch and Hegre, 1997). Examples of these
“exceptions” include the Spanish-American War (1898), World War 11 (Finland vs. Western democratic
allies), and Peru vs. Ecuador (1981). See Ray (1993, 1995) for an insightful discussion of these cases

ll

status, alliance membership, economic development, militarization, and democracy. His
expectation was that democratic dyads would be less likely to engage in war than dyads
involving at least one nondemocracy. Bremer’s multivariate analysis supports the
proposition of peace between democracies; dyads involving one or more undemocratic states
resulted in a 3.2% increase in the likelihood of war onset.

Maoz and Russett (1993) also test a dyadic theory of the relationship between regime
type and conflict, comparing two theoretical explains of the dyadic democratic peace, a
structural (or institutional) theory and a normative (or cultural) theory. The normative model
is derived largely from Immanuel Kant’s writings on “perpetual peace”,9 and it is based on
two assumptions (Maoz and Russett, 1993:625).

Normative Assumption 1: States, to the extent possible, extemalize the norms of

behavior that are developed within and characterize their domestic political processes

and institutions.

Normative Assumption 2: The anarchic nature of international politics implies that a

clash between democratic and nondemocratic norms is dominated by the latter, rather

than the former.
Based on these assumptions, Maoz and Russett argue that militarized disputes between states
with strong democratic norms should not only be less likely to break out, they should also be
less likely to escalate to higher levels of violence. The authors argue that the strength of

democratic norms will be positively related to how long democratic institutions have been

operating effectively.

 

historically.

9 Kant’s 1795 essay, “Perpetual Peace”, provides an explicit link between republican forms of
government and war. Kant’s ideas became much more popular in the democratic peace literature after
the publication of three articles by Michael Doyle (1983a, 1983b, 1986). Kant’s ideas are discussed in
much greater detail in Chapter 2.

12
An alternative theoretical model, the structural model, is discussed at length in War

and Reason (Bueno de Mesquita and Lalman, 1992).'0 Focusing on the constraints
democratic institutions place on leaders, the structural model makes two key assumptions
(Maoz and Russett, 19932626).
Structural Assumption 1: International challenges require political leaders to mobilize
domestic support to their policies. Such support must be mobilized from those
groups that provide the leadership the kind of legitimacy that is required for action.
Structural Assumption 2: Shortcuts to political mobilization of relevant political
support can be accomplished only in situations that can be appropriately described as
emergencres.
The main conclusion derived from these assumptions is that militarized disputes should be
less likely to occur between two states (and less likely to escalate), the higher the political
constraints in both countries. Bueno de Mesquita and Lalman (1992) show that negotiation
and status quo are the only two possible outcomes of the international interaction game with
complete information. Even in environments of incomplete information, democracies signal
their dove-like qualities to each other much more effectively, thus leaders in democracies are
unwilling to bear the domestic costs associated with the use of force.
Maoz and Russett’s (1993) empirical analysis of dyadic conflict in the post World
War 11 period garners support for both the structural and normative models of the democratic
peace. They find, however, that “the relationship between institutional constraints and
measures of dispute and war occurrence is not as robust as the relationship between measures

of democratic norms and the dependent variables. This suggests that the normative model

may be a better overall account of the democratic-peace phenomenon than the structural

 

'0 Other advocates of the structural model include Rummel ( 1979), Morgan and Campbell (1991), and

13
model (Maoz and Russett, l993:636).” This conclusion has been the subject of controversy

in recent years, especially when one examines the research at the monadic level of analysis.
Advocates of structural and normative models compete to explain the primary (but debatable)
finding at the monadic level, i.e., that democracies are just as conflict-prone as

nondemocracies.

The Monadic Democratic Peace

While the empirical findings of the democratic peace proposition at the dyadic level
of analysis are incredibly robust, the empirical results at the monadic level are mixed. Early
work on the link between regime type and general foreign policy behavior concluded that
nations with democratic forms of government were more cooperative and peaceful than
nondemocratic states (Haas, 1965; Rosenau, 1966; East and Gregg, 1967; Rummel, 1968;
Salmore and Hermann, 1969; Zinnes and Wilkenfeld, 1971; East and Hermann, 1974).“
Rummel’s (1983 :29) freedom proposition makes a similar claim: “Freedom inhibits violence
(the more libertarian a state, the less it tends to be involved in violence; Rummel, 1979:292-
293).” His early empirical analysis provides support for the freedom proposition, finding a
positive link between libertarianism (or democracy) and peaceful foreign policy behavior.

The majority of the conflict community has not accepted the claims made by Rummel
and others due in large part to two critical studies of Rummel’s work on the freedom

proposition (Chan, 1984; Weede, 1984).l2 In fact, it has been much more common to

 

Morgan and Schwebach (1992).

" Ray (1995:l l) argues that scholars in the conflict community have largely ignored this work.

'2 Chan (1984) and Weede (1984) found no significant relationship between regime type and war
involvement. Small and Singer (1976) reached a similar conclusion, i.e., that democracies were just as
violent at the monadic level as nondemocracies.

14

conclude that there is no difference in the war-proneness of democracies and
nondemocracies, i.e., that democracies are just are likely to engage in international conflict as
nondemocracies overall (e. g. Bueno de Mesquita and Lalman, 1992; Maoz and Russett,
1993; Gleditsch and Hegre, 1997). More recent work (Ray, 1995; Rummel, 1995; Benoit,
1996) has called this conclusion into question, coming back full circle to the idea that
democracies are more peaceful as a whole. My purpose here is not to address this debate
fully, but rather to point out that the vast majority of the systemic democratic peace work
operates with the assumption that democracies are no less war-prone than nondemocracies
(the standard assumption). I argue below that a theory developed at the systemic level avoids
the problem of aggregating potentially faulty findings from the monadic and dyadic levels to

the systemic level.

th Systemic Democratic Peace

The final level of analysis, the systemic level, has been addressed theoretically in the
democratic peace literature, but it has received very little empirical attention.13 This
perspective usually correlates the number of democratic states in the international system
with the incidence of global war or conflict. It is often assumed that peace in the
international system follows logically from peace at the dyadic level (Singer and Wildavsky,
1993) or the monadic level (Small and Singer, 1976), especially as the number of

democracies increases over time (Starr, 1992). This perhaps explains why so few rigorous

 

'3 The lack of empirical systemic analysis is not limited to the democratic peace literature, despite
extensive theorizing at the systemic level in international relations (e.g. Waltz, 1959, 1979). The few
areas characterized by extensive empirical tests at the systemic level of analysis include the distribution
of power and war studies, status discrepancy studies, and long cycle analysis.

15
empirical tests of the systemic proposition have been attempted in recent years. As Gleditsch

and Hegre (1997) note, many researchers claiming systemic level results in the democratic
peace literature often depend on monadic and dyadic level empirical evidence. In addition,
the few empirical studies of the systemic democratic peace analyzing system level data have
produced mixed results. Some studies find a positive empirical relationship between
proportion of democracies in the system and systerrric conflict, while others find a negative
relationship.'4

Maoz and Abdolali (1989) undertake one of the earliest empirical studies of the
systemic democratic peace. They test the relationship between regime type and conflict at
all three levels of analysis (monadic, dyadic, and systemic). The question they address at
the systemic level is (198924) “does the level of conflict in the international system decline
as the number of politically free states increases?” They test three system-level
hypotheses: 1) The higher the proportion of democracies, the lower the number of disputes
in the system, 2) The higher the proportion of democracies, the smaller the proportion of
interstate disputes that escalate to war, and 3) The more homogenous the interstate system
(in terms of regime type), the lower the number of disputes and the smaller the proportion
of disputes that escalate to war.

These hypotheses are tested with autoregressive moving average (ARMA)
regression models from 1817-1976. Maoz and Abdolali (1989) compare several measures
of systemic conflict: the number of disputes begun, the number of disputes underway, the

number of dyads in disputes begun, the proportion of dyads underway, the number of wars

 

" Maoz and Abdolali (1989), for example, find a positive relationship between the proportion of
democracies in the system and general measures of conflict (militarized disputes), but a negative

l6

begun, and the proportion of wars begun. Independent variables include the proportion of
jointly democratic dyads in the system, the proportion of jointly autocratic dyads in the
system, and the heterogeneity of regime types in the system.

Their primary findings can be summarized as follows: 1) In general, the proportion
of jointly democratic dyads in the system had a positive effect on the number of disputes
begun and underway, but a negative effect on the frequency of wars, 2) The proportion of
jointly autocratic regimes had similar effects on disputes and wars, 3) These results hold
for the 20th century, whereas the proportions of jointly democratic and jointly autocratic
dyads negatively affected most of the conflict variables in the 19th century, and 4) The
degree of heterogeneity in the system had a positive effect on both dispute and war
frequency. In conclusion, Maoz and Abdolali find a positive relationship between global
democracy and general measures of systemic conflict (disputes), contradicting the
expectation of a peaceful system with an increasing proportion of democracies. Their
results on war, however, do support the democratic peace proposition: “while more
democracies made for more disputes and more dispute involvements (number of dyads
involved in disputes), they tended to make for less war, and they tended to reduce the
probability that a low-level militarized dispute would escalate to an all-out war (Maoz &
Abdolali, l989:29-30).”

One attempt to formalize the systemic level relationship between the level of global
democracy and the level of systemic conflict is undertaken by Gleditsch and Hegre

(1997).'5 Their formal model is the first to provide a systematic and logical link between

 

relationship between the proportion of democracies in the system and systemic war.
'5 The formal model developed in the Gleditsch and Hegre (1997) paper is an extension of the model in

17

the three levels of analysis in the democratic peace literature. They criticize previous
tendencies to assume either that an increase in the proportion of democracies globally will
decrease the level of systemic conflict (which follows from the dyadic level) or that the
spread of democracy will have no impact on the overall level of conflict in the system
(which follows from the monadic level).

Gleditsch and Hegre (1997) start with an assumption that nondemocratic dyads
have a lower propensity for war than mixed democratic-nondemocratic dyads.“ They
model the relationship between regime types in the international system and systemic
conflict as a function of 1) the probability of war in jointly autocratic dyads, 2) the
probability of war in democratic-nondemocratic dyads, 3) the proportion of democracies in
the system, and 4) the total number of nations in the system. If the probability of war is
greater in the mixed dyads than in the jointly autocratic dyads, then the relationship
between the proportion of democracies in the international system and the frequency of
international conflict is bell or parabola-shaped. In other words, if the proportion of
democracies in the system is low, then the initial effect of new democracies entering the
system is an overall increase in the level of global conflict. This increases up to a threshold
point, where the proportion of democracies is large enough to produce greater levels of
systemic peace. The threshold point is a function of the probability of war in mixed and
jointly autocratic dyads. In conclusion, the Gleditsch and Hegre (1997) model predicts an
increasing historical level of democracy to have produced more war. Their empirical

analysis provides some support for this predicted theoretical relationship, with the

 

Raknerud and Hegre (1997).
'6 The authors note that this assumption is supported empirically for the incidence of war, the onset of

18

pre-World War 11 time period being associated with more democracy and more war, and
the post-World War 11 time period being characterized by more democracy and less war in

the international system.

Systemic Theorizing vs. Aggregation Across Levels of Analysis
As I noted earlier, the differences between the dyadic and monadic level findings in

the democratic peace literature have been treated as a puzzle by many scholars (e. g. Bueno de
Mesquita and Lalman, 1992; Maoz and Russett, 1993) The articles by Maoz and Abdolali
(1989) and Gleditsch and Hegre (1997) are good examples of theoretical aggregation from
the monadic and dyadic levels of analysis in the democratic peace literature to the systemic
level. Gleditsch and Hegre (1997), for example, view the systemic level relationship
between democracy and conflict as a puzzle that emerges from the somewhat contradictory
findings at the monadic and dyadic levels. They argue that the systemic level:

has rarely been subjected to empirical investigation, but it is

commonly assumed that it can be answered by a simple deduction

from one of the two other levels. The most common conclusion is

that if democracies do not fight each other, an increasing number of

democracies in the system will produce a more peaceful system.

Others have argued that if democracies are as war prone as

nondemocracies, it makes no difference at the system level if the

number of democracies increases. Both of these system-level

statements cannot be true at the same time, so there must be

something wrong either with the deductions or with the empirical

regularities. (284)

Maoz and Abdolali (1989) also regard the systerrric level relationship between democracy

and war as a logical extension of the monadic and dyadic level results:

 

new dyadic interstate war, and militarized disputes (300).

19

The argument that democracies are less conflict prone than
autocracies has a normative implication that can—and ofien has
been—extended to the systemic level of analysis. This is also a
logical extension of the ‘joint freedom’ proposition. If politically free
states do not fight one another, then the more democracies, the less
international conflict. Extended ad infinitum, Kant’s notion of eternal
peace—a peaceful world composed of a federation of politically free
republics —gets a new look according to this logic. It is therefore
important to examine whether the joint freedom proposition has
significant system-level implications. (8-9)

The theoretical relationship between democracy and war at the systemic level has
not been fully elaborated in previous studies, in large part because scholars either assume
that the systemic level relationship can be deduced from other levels of analysis, or they
discount the importance of the democracy-conflict relationship at the systemic level.‘7
This lack of theorizing at the systemic level presents several problems in assessing the
relationship between democracy and war in the international system. As Huntley
(1996:45) argues succinctly, “to view the significance of the liberal peace as a test of
opposing ‘levels of analysis’ misses deeper issues.” The most significant problem with
aggregating findings across levels of analysis is that the empirical results, especially at the
monadic level of analysis, are debatable. For example, Gleditsch and Hegre (1997)
assume that the probability of war participation at the monadic level is the same for
democracies and nondemocracies. Whether one should accept this “conventional
wisdom” given the contradictory findings at the monadic level is highly questionable.

A larger issue is the need for a systemic theoretical understanding of the relationship

between war and democracy. Theories developed at the systemic level have several

 

'7 Wade Huntley (1996) develops one of the only systemic level theories of democracy and war in the literature,
and his theory is derived largely from the works of Kant.

20
advantages over their “reductionist” counterparts (Waltz, 1979). First, decisions made by

agents (nations, leaders, etc.) occur in the context of the international system. The structure
of the system has an impact on the units in the system “through socialization of the actors and
through competition among them (Waltz, 1979:74).” A theoretical understanding of the
democratic peace at the systemic level will provide insight into the context and rules of the
international system, and how these are changing over time.18 Second, “systerrric theories are
generally more parsimonious than their first-image and second-irnage competitors
(Mansfield, 1994:6).” One reason for the lack of parsimony of first and second image
theories is that it is “difficult to trace the effects of foreign policy behavior on global and
regional systems (James, 199726).” The “puzzle” that has been discussed in the democratic
peace literature, i.e., the contradictory findings between the monadic and dyadic levels of
analysis, demonstrates the difficulty in drawing systemic level conclusions from
“reductionist” theories. Finally, some scholars have argued that systemic or third-irnage level
theories have fared better in explaining international politics. Waltz (1979:39) makes the
point succinctly.

...a systems approach will be needed, if outcomes are affected not only

by the properties and interconnections of variables but also by the way

in which they are organized. If the organization of units affects their

behavior and their interactions, then one cannot predict outcomes or

understand them merely by knowing the characteristics, purposes, and

interactions of the system’s units. The failure of the reductionist

theories. . .gives us some reason to believe that a system’s approach is

needed.

All of these arguments demonstrate the necessity of developing a systemic level

theory of the democratic peace. Contrary to what some scholars have argued, the

 

'8 Keohane (19862193) makes a similar argument: “a systemic theory is important because we must

21

significance of the liberal peace does not make systemic level approaches to the debate
obsolete.” Rather, there are systemic level consequences of the relationship between
democracy and war that must be examined. And perhaps the most significant consequence
of the democratic peace at the systemic level is the potential for fundamental change in the

rules and norms of interaction in the international system.

Conclusion

In this chapter, I have briefly reviewed the literature linking regime type and conflict
at three levels of analysis: monadic, dyadic, and systemic. I have argued that in order to
understand the systemic level relationship between democracy and war, we need to go
beyond theoretical aggregation across levels of analysis. A systemic level theory of the
democratic peace phenomenon will provide a greater understanding of the dynamic
relationship between democracy and war and how this relationship is affecting the context,
structure, and rules of interaction in the international system. Given the significant policy
implications of the democratic peace, it is important to determine if the growth in democracy
and the outcomes of wars over time have altered the nature of the international system. If, as
Rosenau (1990) argues, world politics has entered a state of turbulence and change that could
alter the basic “parameters” of the system (James, 1997), an understanding of one source of

that change, the democratic peace, will be extremely useful. Only a theory of the

 

understand the context of action before we can understand the action itself.”

'9 Huntley (1996245) states a good example of this conclusion: “For neorealists, no peace dependent on
only the internal pacific disposition of liberal republics can endure. Supporters of the Kantian
interpretation respond that properly constituted republics can, in fact, overcome the anarchy among them,
and that the present liberal peace therefore challenges the adequacy of ‘systemic’ theories of international
politics.”

22

relationship between democracy and war developed at the systemic level can address these

larger political and scholarly implications.

23

Table 1 - Democratic Peace Research at the Monadic Level of Analysis

 

Authors (Year)

Dependent Variable(s)

 

Small and Singer (1976)

Frequency, average duration, average severity,
and battle deaths of COW wars by regime type

 

Rummel (1979, 1983)

Foreign conflict scale (sum of negative
communications, negative sanctions, warning and
defensive actions, and military violence)

 

 

 

 

 

 

 

Chan (1984) Total democratic COW war participations per
year divided by total system membership years

Weede (1984) Frequency of war involvement per year

Maoz and Abdolali (1989) Frequency of participation and initiation of
militarized interstate disputes and COW wars

Morgan and Campbell (1991) Dichotomous measure of dispute (MID)
escalation to a COW interstate war

Benoit (1996) Frequency count of war involvement per year
(using Weede’s 1984 data)

Gleditsch and Hegre (1997) Percentage of country years with conflict (COW

 

interstate wars, militarized interstate disputes, and
armed conflict)

 

 

24

Table 2 - Democratic Peace Research at the Dyadic Level of Analysis

 

Authors (Year)

Dependent Variable(s)

 

Babst (1964, 1972)

Dyadic war involvement (Wright’s (1942) definition
of war)

 

Rummel (1979, 1983)

Dyadic directed foreign conflict

 

 

 

 

 

 

 

 

Maoz and Abdolali (1989) Dyadic involvement in militarized interstate disputes
(MIDS)

Bremer (1992) Dichotomous dyadic measure of COW war
outbreak/no war outbreak (limited to war participants
on the first day of a war)

Bueno de Mesquita and Dichotomous measure of negotiation or status quo

Lalman (1992) outcome (one) versus any other outcome of the
International Interactions game (for 238 randomly
generated European dyads)

Maoz and Russett (1992) Dyadic involvement in militarized interstate disputes

Morgan and Schwebach Escalation of a militarized dispute to a COW

(1992) interstate war in a dyad

Weede (1992) Dyadic war and militarized dispute involvement given
territorial conflict or strategic interdependence

Bremer (1993) Annual dichotomous measure of dispute and war
outbreak between initial protagonists

Maoz and Russett (1993) Dichotomous measure of occurrence of dyadic

militarized dispute/no militarized dispute for
politically relevant dyads

 

Rousseau, et a1 (1996)

Initiation of force and highest level of force used in a
crisis (testing monadic and dyadic effects in one
model)

 

 

 

 

 

Oneal, et a1 (1996) Dichotomous measure of dyadic dispute involvement
(MID) annually for politically relevant dyads

Gleditsch and Hegre (1997) Percentage of dyad years at war (using COW
definition of interstate war) and armed conflict
(Wallensteen and Sollenberg, 1996)

Oneal and Russett (1997) Dyadic measure of dispute occurrence/no dispute
occurrence per year (for politically relevant dyads)

Raknerud & Hegre (1997) Dyadic outbreaks of COW interstate wars (measured

 

on the first day of the war)

 

 

25

Table 3 - Democratic Peace Research at the Systemic Level of Analysis

 

Authors (Year)

Dependent Variable(s)

 

Maoz and Abdolali (1989)

Annual measures of militarized disputes
begun, disputes underway, dyads in
disputes begun, proportion of dyads
underway (disputes), number of COW
interstate wars begun, proportion of COW
interstate wars begun

 

 

Gleditsch and Hegre (1997)

 

Annual measure of the proportion of
country years at war (COW interstate wars)

 

 

CHAPTER 2

A SYSTEMIC THEORY OF WAR AND DEMOCRACY

Woodrow Wilson. . .saw clearly one of the essential elements of a
third-image analysis, that everyone’s policy depends upon everyone
else’s. With many authoritarian states in the world, he realized that
even the non-authoritarian state, must on occasion be prepared to use
force to defend its interests. But, convinced that democratic states are
peaceful because their governments reflect the aspirations of the
people, he foresaw a day when the internal condition of all states
would mean not the constant possibility of war, but rather the
assurance of perpetual peace. (Waltz, 1959:226-227)

In Chapter 1, I argued for the development of a systemic level theory of the
relationship between democracy and war. While advocates of the democratic peace view the
spread of democracy as a potential force for change in international relations, defenders of
neorealism characterize the intemational system as immutable, i.e., international politics have
been and will continue to be determined by anarchy and the distribution of power. In this
chapter, I outline the most important features of the neorealist theoretical approach to
international relations. Next I develop my theoretical model of the democratic peace,
drawing heavily from the work of Immanuel Kant (1970a; 1970b; 1970c) and Wade Huntley
(1996). I consider first the endogenous and evolutionary relationship between democracy

and war. This is followed by a discussion of three other factors that have important

26

27
influences on democracy and war in the international system: power concentration, trade, and
production. To reiterate, my theoretical model attempts to address the following questions:
0 What is the systemic level relationship between democracy and war, and how
is this relationship evolving over time?
o Are the rules and norms of the international system being transformed as the
international system becomes more democratic?
- How do economic relations between states in the system influence the
systemic level relationship between democracy and war over time?
o How is the level of war in the intemational system related to the
concentration of power between nation-states?

I derive propositions to be tested and I conclude with a general discussion of the novel

insights offered by my theoretical model.

Neorealism and its Critics

The theoretical approach termed “realism” dates as far back to Thucydides’ study
of the Peloponnesian Wars.20 Hans Morgenthau is one of most influential modern realist
writers, in addition to Kenneth Waltz, whose systemic level theory of realist politics (1959,
1979) has been influential in the study of international relations in recent years. Realism in
its original or structural (neorealist) form, makes three key assumptions about international
politics (Keohane, 1986b: 164-165):
0 Nations are the most important actors in international relations.
0 States behave according to rational, utility maximizing principles.
0 States seek power and calculate their interests in terms of power.

Many early theories in the realist tradition focused on individual state behavior, or a

nation’s foreign policy (the monadic level). Waltz (1979) criticizes these “reductionist”

approaches, arguing that the organization of states in the international system has a

28

profound impact on state behavior. To understand state behavior at the monadic (or
dyadic) level, it is first necessary to understand the structural features of the international
system, the forces that produce structural change, and how the structure of the system
influences and constrains agents’ actions.

Waltz’s systemic theory of international politics is rooted in the basic ideas of
system and structure; thus it is worthwhile to clarify each of these terms.21 In general, a
system “refers to a group of parts or units whose interactions are significant enough to
justify seeing them in some sense as a coherent set. . . [and a system] therefore comprises
units, interactions, and structure (Buzan, et a1, 1993). A broader definition of a system is
provided by Brecher and Ben Yedhua (1985217): “a set of actors who are situated in a
configuration of power (structure), are involved in regular patterns of interaction (process),
are separated from other units by boundaries set by a given issue, and are constrained in
their behavior from within (context) and from outside the system (environment)22.” For
Waltz (1979:88), the defining characteristic of the international system is the way in which
the system is structured, with structure being determined by l) the principle by which the
units are organized, 2) the differentiation of the units, and 3) the distribution of capabilities
across units.

While structure does not directly influence behavior in the system, Waltz (1979:74)
asserts that its impact occurs indirectly through “socialization of the actors and through

competition among them.” Socialization creates norms of behavior that constrain and

 

2° See Thucydides (400 B.C./1951).

2' Another key concept in neorealist thought is “process” which can be defined as “the patterned relations
among units that go on within a system—relations that reflect in varying degrees the constraints imposed
by the system’s structure (Ruggie, 19862134)”

22 James (1997) includes a summary table of various definitions of “system” in the literature. He adopts

29

shape states’ decisions, in addition to encouraging conformity among nations in the
system. Competition, on the other hand, “spurs the actors to accommodate their ways to
the socially most acceptable and successful practices (Waltz, 1979:77).” Thus structure
both constrains behavior, and encourages the adoption of successful strategies in world
politics.

Waltz (1979) argues that systems or structures are transformed by changes in one of
the three components of structure (ordering principle, differentiation of units, distribution
of capabilities). The two primary ordering principles of the international system are
anarchy and hierarchy, and change fiom one ordering principle to another would
fundamentally transform the system. Like most international relations scholars, Waltz
believes that the modern international system is best characterized by anarchy (a
decentralized system lacking a central authority) as opposed to hierarchy (one dominant
state). Changes in the second characteristic of structure, differentiation of units, can also
produce changes in the system’s structure. Waltz, however, asserts that nations are “like
units” motivated by similar concerns, namely the pursuit of power and security. Thus he
does not consider changes in the units comprising the system a fundamental source of
structural change in the international system.23 Finally, changes in structure can occur if
the distribution of capabilities across units changes. The vast majority of Waltz’s
conclusions are based on comparisons of different distributions of power in the
international system precisely because the other two characteristics of structure, anarchy

and the differentiation of units, are constant or assumed unimportant in producing systemic

 

the Brecher and Ben Yehuda (1985) definition in his study of elaborated structural realism.
’3 “Since no functional differentiation of states exists apart from that imposed by relative capability, the

30
or structural change. In addition, much of the work building on Waltz’s neorealist

approach emphasizes the distribution of power as a primary determinant of conflict and
peace in the international system.24
Waltz’s systemic theory of international politics has been the recipient of both

widespread support and spirited criticism, testimony to the profound impact of Waltz’s
ideas on international relations research. One of the most fundamental criticisms of
structural realism is that it cannot adequately account for structural change (e. g. Buzan and
Jones 1981; Ashley, 1986; Keohane, 1986a, 1986b; Ruggie, 1986). Waltz (1979)
emphasizes only one source of structural change, the distribution of capabilities among
states, and he minimizes the impact of changes in the differentiation of units on structural
change. Ruggie (1986) criticizes Waltz on this score, asserting that differentiation of units
in the international system provides an essential source of structural change.

The problem is that a dimension of change is missing from Waltz’s

model. It is missing because he drops the second analytical

component of political structure, differentiation of units, when

discussing international systems. . ..If anarchy tells us that the political

system is a segmented realm, differentiation tells us on what basis the

segmentation is determined. The second component of structure,

therefore, does not drop out; it stays in, and serves as an exceedingly

important source of structural variation. (Ruggie, 19862142)

Ruggie focuses on the shift from the medieval to the modern international system, a

system transformation unexplained by Waltz’s framework, primarily because both systems

are anarchic. Ruggie asserts that changes in the distribution of capabilities in the system

cannot account for the striking systemic change in the relevance of property rights and state

 

second component of political structure is not needed at the international level (Ruggie, 1986: 135).”
2‘ See, for example, the debate on systemic polarity and war.

31

sovereignty, concepts that were almost nonexistent in the medieval period, yet crucial in
the modern system.25
Even Waltz (1986:343) admits the potential significance of the differences among

nations as a source of structural change.

Changes in, and transformations of, systems originate not in the

structure of a system but in its parts. Through selection, structures

promote the continuity of systems in form; through variation, unit-

level forces contain the possibilities of systemic change. . .Systems

change, or are transformed, depending on the resources and aims of

their units and on the fates that befall them. [emphasis added]
We will see below that systemic war has a significant impact on the dynamic changes in
the differentiation of states (units), especially in terms of their domestic institutions, or
regime types. Increasing numbers of democratic states, in turn, have systemic level
consequences, especially for the rules and norms of behavior in the international system.
In the next section, I discuss various extensions of Waltz’s theory that add the notions of

rules, regimes, and norms. These concepts will be crucial in the theoretical model

developed in the following section.

Beyond Neorealism

The neorealist (or structural realist) theoretical approach has expanded considerably
since the publication of Waltz’s Theory of International Politics. Numerous academics
accept Waltz’s basic systemic approach and his emphasis on the importance of structure

(and its three components) in explaining international relations. Many of the extensions of

 

2’ Buzan, Jones, and Little (1993:65) also discuss the importance of unit differentiation as a force for
structural change. They emphasize the relevance of the distributional structure, or “systemic patterns in
the distribution of unit attributes.”

32

structural realism have concentrated on the importance of rules and norms of behavior in
the system.

Rules and norms play an important role in regime theory, one of the most
prominent extensions of neorealism. Regime theory finds its roots in work by Robert
Keohane and Joseph Nye (1972, 1977) on transnational relations and work by John Ruggie

26

(1975) on regimes. The concept of an international regime broadly encompasses both
rules and norms, which is made clear in the following definition of a regime: “sets of
implicit or explicit principles, norms, rules, and decision-making procedures around which
actors’ expectations converge in a given area of international relations (Krasner, 198322).”
Regime theorists argue that the international system is characterized by a set of

unwritten rules, even though the system is anarchic.27 Rules in the international system
have been conceptualized in a variety of ways.

Nations must have available rules, the media through which they

communicate with one another and coordinate their actions. . .A rule

is, in its most basic sense, an understanding about how to proceed or

“go on” in given social circumstances. (Dessler, 19892454-455)

Rules are general imperative principles which require or authorise

prescribed classes or groups to behave in prescribed ways. (Bull,

1995:52)

Rules are specific prescriptions or proscriptions for action. (Krasner,
198322)

 

2" For further work on regime theory, see Young (1980), Puchala and Hopkins (1982), Stein (1982),
Strange (1982), Haas (1983), Krasner (1983), Ruggie (1983), Young (1983), Keohane (1984), Haggard
and Simmons (1987), Young (1991).

27 Dessler (1989:458-459) elaborates this point: “Scientific realism insists that all social action depends
on the preexistence of rules, implying that even under anarchy, rules are an essential prerequisite for
action. . .Rules are. . .both logically and praxiologically necessary for social action.”

33

Examples of rules generally practiced in the international system include the exchange of
diplomats, acceptance of the right of neutrality, the recognition of state sovereignty, and
the use of treaties to formalize international agreements.

The concept of international norms, another key element of regime theory, is
closely related to the idea of rules. Norms often involve a moral dimension or are “morally
binding regardless of considerations of narrowly defined self-interest (Keohane, 1984257).”
Even if one adds a moral element to the conceptualization of norms, it is still difficult to
distinguish norms from rules. Rules usually entail specific rights or obligations, and they
can usually be altered more easily than norms (Keohane, 1984258).

Goertz and Diehl (1992) apply the theory of international norms to the study of
. international conflict. They argue that an international norm is characterized by four key
elements. First, norms are identified by their “regularity and consistency of behavior
(Goertz and Diehl, 19922636)” Second, adherence to norms is often in conflict with a
nation’s self-interest. Third, those who violate international norms must face some form of
sanction or punishment. Finally, there is a normative element, or “issues of justice and
rights of a moral or ethical character (Goertz and Diehl, 1992:638-639).” The authors
demonstrate the relevance of each of these properties of norms in their historical analysis
of the decolonization norm that has developed over the last two hundred years.

The importance of rules and norms in world politics is also stressed in the theory of
international society, an approach developed by British scholars such as E.H. Carr, C.A.W.
Manning, Hedley Bull, and Martin Wight, among others. Bull (1995213) asserts that an
international society exists when “a group of states, conscious of certain common interests

and common values, form a society in the sense that they conceive themselves to be bound

34

by a common set of rules in their relations with one another, and Share in the working of
common institutions.” A focus on international society stresses the importance of common
institutions, economic interactions, ideologies, and transnational ties in promoting
cooperation between states, an approach that goes beyond the neorealist tendency to
emphasize conflict and the distribution of power in the international system.

The theoretical concepts of rules, norms, and regimes discussed above are
extremely important in the cultural or normative theoretical argument of the democratic
peace proposition. Maoz and Russett (1993), for example, contend that democracies do not
fight each other (the dyadic proposition) because they extemalize the norms of behavior
that characterize their domestic political processes. These norms or rules include the
resolution of conflicts by nonviolence and respect of political opposition. It is common in
the democratic peace literature to emphasize the peaceful relations or peaceful community
among democracies.28 What has not been explored at the systemic level, however, is how
changes in the relationship between democracy and war over time influence the types of
rules and norms of behavior practiced in the international system or “international society”
at large. I demonstrate below that Kant’s cultural argument for a democratic (perpetual)
peace has a systemic component that has been largely ignored in the literature. In addition,
Kant emphasizes the important role wars play in promoting republican forms of
government in the system. The systemic theoretical model developed in the next section

emphasizes both this endogenous aspect of the relationship between democracy and war,

 

2’ A nice example is Singer and Wildavsky’s (1993) distinction between “zones of peace” among
democracies and “zones of turmoil” between nondemocracies (or a mixture of democracies and
nondemocracies).

35

and the evolutionary changes in the rules and norms this relationship produces in

international relations.

A Systemic Theory of Democracy and War

Waltz’s neorealist theory of international politics, and the extensions discussed
earlier, contribute to our understanding of war at the systemic level. As Keohane
(1986b2159) notes, however, “Realism is particularly weak in accounting for change,
especially where the sources of that change lie in the world political economy or in the
domestic structure of states.” In this section, I argue that the domestic regime type of
individual states has a significant impact on war in the international system, and that war
both creates changes in these domestic structures and provides the impetus for structural
change. I first summarize Kant’s theoretical ideas for a perpetual peace, focusing on the
relationship between democracy (republicanism) and war at the systemic level. Next, I
analyze the reverse relationship, i.e., how wars in the system affect the level of democracy
internationally. The crucial elements of the theoretical argument are gleaned from
Immanuel Kant’s work on perpetual peace and Wade Huntley’s (1996) third image

explication of Kant’s ideas.

Conditions for Perpetual Peace
Many theoretical explanations of the democratic peace proposition, especially at the
dyadic level, stem from Kant’s (1795) essay on “Perpetual Peace”.29 Yet the “systemic”

aspects of Kant’s writings have been largely ignored. Wade Huntley (1996) fills this

36
theoretical gap by elaborating the systemic (or third image) sources of the liberal peace

developed by Kant. Huntley’s analysis of Kant’s writings on the liberal peace is quite
comprehensive, therefore I refer frequently to his interpretation of Kant’s philosophy in the
sections that follow.

Huntley (1996) begins by discussing Kant’s moral philosophy that serves as a
foundation for perpetual peace. Kant is concerned with the conditions for acting morally
and he distinguishes between two types of awareness or experience, noumenal and
phenomenal. “Phenomenal experience informs us of our more ‘selfish’ needs and desires,
including our quests for security and happiness (Huntley, 1996247).” Noumenal
experience, on the other hand, refers to morality and the recognition of rights, especially
“the basic equality of all humans, and the priority of human freedom, most especially the
freedom to be moral (Huntley, 1996247).” Noumenal awareness is essential for creating a
moral law or “categorical imperative”. One of the most important categorical imperatives
discussed by Kant is the elimination of warfare between individuals and states (Kant,
1970c2174; Huntley, 1996248). Kant (1970bz99) contends that only a republican form of
government, founded on freedom for individuals, a single common legislation, and legality
equality for all, “establishes the possibility of freedom as well as security under the rule of
law. The key characteristics of the republic are the retention of sovereignty by the citizens
who compose it and the incumbent separation of legislative and executive powers.

(Huntley, 1996248)”

 

29 See, for example, Maoz and Russett (1993) and Doyle (1986).

37

The republican form of government plays an essential role in Kant’s vision of a
perpetual peace. Kant (1970b299-105) identifies three key conditions for such peace, or
“an international rule of law (Huntley, 1996249)”:

0 The Civil Constitution of Every State shall be Republican.
0 The Right of Nations Shall be based on a Federation of Free States.
0 Cosmopolitan Right shall be limited to Conditions of Universal Hospitality.

The first condition establishes the supremacy of republican constitutions for
establishing peace. Kant (1970b2100) argues that republican governments are inherently
inclined towards peace because the government requires the public’s consent before it
becomes embroiled in warfare. War is costly in terms of human lives, economic resources,
and the accumulation of debt incurred throughout the course of a war. Thus the public will
act as a constraining force on the ability of the state to engage in warfare."0

The second condition implies that the federation of free states envisioned by Kant can
support a perpetual peace only if its members are republics. The federation extends the “rule
of law” inherent in the domestic structure of republican constitutions to an international “rule
of law”.31 Huntley (1996250) elaborates this point: “Kant is holding that establishing freedom

and the rule of law domestically and internationally are mutually dependent, symbiotic

processes.” Kant does not support the idea of a world government that enforces an

 

3° Kant recognizes that states must be prepared to defend themselves against non-republican
governments, but he believes that the public in republican states will prevent their leaders from initiating
war. (Huntley, 1996249)

3' Many supporters of the democratic peace proposition use this aspect of Kant’s argument to assert that
while democracies will be peaceful amongst themselves, there is no reason to believe that they will be
more peaceful overall, or that a world filled with a greater number of democracies will not necessarily
translate into greater peace. I show below, though, that the “rule of law” between democracies extends to
the international system. In particular, the rules or norms of behavior between democracies diffuse
throughout the international system over time as the proportion of democracies in the system increases.

38

international rule of law. Rather nations must maintain their sovereignty, evidence that
Kant’s perpetual peace is firmly grounded in the concept of federalism.

Finally, Kant insists that universal hospitality, or the right for people to travel to other
countries, is also an important condition for perpetual peace between nations. “Originating in
the priority of human freedom, Kant identifies a ‘world community’ manifesting moral duties
beyond the state and common to all (Huntley, 1996251).” Kant’s emphasis on universal
hospitality follows logically from his moral philosophy, particularly his categorical
imperative that emphasizes the respect for freedom. “The idea of a cosmopolitan right. . .is a
necessary complement to the unwritten code of political and international right, transforming
it into a universal right of humanity (Kant, 1970b2108).”

To summarize, Kant believes that perpetual peace depends on republican forms of
government domestically, an international federation of free states, and a principle of
cosmopolitanism, or universal hospitality. The categorical imperative to end warfare can
only be reached through the spread of a domestic and international “rule of law”, whose
principles are founded on individual freedom, legal equality, and separation of executive and
legislative powers. A republican form of government built on these principles creates the
freedom to act morally. “The history of the human race as a whole can be regarded as the
realisation of a hidden plan of nature to bring about an internally—and... externally, perfect
political constitution as the only possible state within which all natural capacities of mankind
can be developed completely (Kant, 1970a250).” Kant concludes that democracy leads to
peace in the international system by decreasing the uncertainty that arises in a state of

anarchy (or state of war).

39

War: A Force for Democracy
One of the novel aspects of the systemic dimension of Kant’s theoretical approach is
that he considers the relationship between democracy and war to be endogenous, i.e., he
argues that both democracy and war cause changes in each other across time.32 Not only do
republican constitutions promote peace in the international system, “the propensity for war
itself sows the seeds of war’s end (Huntley, 1996:56).” War serves as a catalyst for the
development of republican constitutions and an international “rule of law” among nations.
Several scholars have discussed the importance of the endogeneity between democracy and
war (James, Solberg, and Wolfson, 1995; Midlarsky, 1995; Gates, Knutsen, and Moses,
1996; Thompson, 1996), but few have examined this relationship at the systemic level.33
According to Kant, war is the strongest force creating more democracy in the

international system, pushing us ever closer to a federation of fiee states, one pillar of
perpetual peace. War justifies the development of democratic governments and creates a
more widespread peace, which is essential to the survival and improvement of republican
constitutions. (Huntley, 1996261)

Nature has thus again employed the unsociableness of men, and even of

the large societies and states which human beings construct, as a means

of arriving at a condition of calm and security through their inevitable

antagonism. Wars. . .are the means by which nature drives nations

to. . .abandon[ing] a lawless state of savagery, and enter[ing] a

federation of peoples in which every state, even the smallest, could

expect to derive its security and rights not from its own power or its

own legal judgement, but solely from this great federation. (Kant,
1970a247)

 

’2 A technical discussion of the issues surrounding endogenous relationships is contained in Appendix A.
3’ The most notable exception is James, Solberg, and Wolfson (1995).

40
According to Kant, war both creates and destroys nations, and brings about new relations

between states in the system. The experience of civil and international war improves the
internal institutions of states over time, creating “a state of affairs. . .which,. . .can maintain
itself automatically (Kant, 1970a248).” And to protect this improvement in their domestic
republican constitutions (the domestic rule of law), states seek the same improvements in
their international relations (the international rule of law). Huntley (1996256) elaborates:
“The necessary elements of the rule of law—civil law, international law, and the law of world
citizenship—all emerge from. . .war.”

War also promotes the development of democracy because democracies tend to win
the wars they fight more often than nondemocracies.34 Lake (1992231) finds that of the 26
wars fought since 1816 between democracies and autocracies, democracies have won 81%
and lost 19%. After a war, democratic victors are better able to influence the development of
democracy in individual countries, establish liberal economic institutions, and promote
democratic rules and norms of behavior in the international system.

Lake (1992) offers three theoretical explanations for why democracies are more likely
to win wars they fight as opposed to nondemocracies. Lake’s hypotheses are derived from
the literature on rent-seeking. He argues that democratic states earn fewer rents overall,
making it possible to 1) “devote more resources to security, (2) to enjoy greater societal
support for their policies, and therefore a greater extractive capacity; and (3) to form

overwhelming countercoalitions against expansionist autocracies (Lake, 1992230).” The last

 

3‘ Gaubatz (19972147) reaches a similar conclusion: “history shows that democratic victory in war has
been the most effective instrument for expanding the democratic sphere.” Likewise, Huntington (1991)
emphasizes the importance of World War [I outcomes on subsequent waves of democratization.

41
part of the argument implies that democracies have historically allied together in large-scale

wars, and that these alliances have increased the probability of victory for democracies.”

Democracies are more likely to win the wars they fight for other reasons as well.
First, democracies have historically mobilized resources more effectively in wartime than
nondemocracies. Second, the public in a democracy will be willing to bear greater costs in
war because “in legitimate states, the mass publics are less likely to believe that they will be
excluded from the gains from victory and more likely to believe that the costs they must bear
will be fairly distributed throughout society (Starn, 1996265).” Also, because democratic
leaders are more likely to be punished for engaging in unsuccessful wars (Bueno dc Mesquita
and Siverson, 1995), democracies will be less likely to enter into wars that will entail great
material and political costs. Given that democracies are more likely to win the wars they
fight, I expect war to have a positive impact on democracy at the systemic level.

Gleditsch and Hegre (19972308) reach a similar conclusion: “War would seem to
promote more democracy in the longer run, and-«at least above a certain level-«more
democracy in turn leads to a reduction of war.” The success of democracies in war over time
has both increased the overall proportion of democracies and encouraged peaceful rules of
engagement in the international system.

Kant also alludes to the strength of democratic governments. “For Kant, the rule of
law within a state does not merely signify a regime type, but is itself an aspect of state

power. In the long run, the liberal republic is also the stronger state (Huntley: 1 996255).”

 

3’ Reiter and Stam (1997) criticize the alliance explanation of democratic success in warfare, pointing out
that 80% of the wars analyzed by Lake (1992) were fought by a single democracy without allied support.
Their empirical analysis of war outcomes shows that “the tendency for democracies to win the wars they

fight is most likely due to the greater legitimacy of democratic regimes rather than because of lower rent-
seeking or democratic bandwagoning (26).”

42
This provides further support to his argument that war serves as an impetus for the

development of republican constitutions, and the democratic “rule of law” in the international
system.

In conclusion, Kant shows us that the historical relationship between democracy and
war is an endogenous and evolving one. In the next two sections, I discuss how Kant’s
theory of the systemic democratic peace differs from Waltz’s systemic theory of international
politics. I argue that the most significant consequence of the movement towards a democratic
“rule of law” internationally, is that the norms and rules that characterize behavior between
democracies are gradually diffusing throughout the international system over time. This idea
casts doubt on the neorealist contention that the international system is constant and

unchanging.

A Comparison of Waltz and Kant

Kant and Waltz come to very different conclusions about the systenric level
relationship between the domestic structure of states and war. While Waltz views the
differentiation of units as relatively unimportant for explaining structural change, Kant
envisions a dynamic relationship between democracy and war; war promotes republican
constitutions in individual states, which in turn promote the spread of the domestic “rule of
law” in the international system.

Huntley (1996) identifies three other key differences in the Kantian and neorealist
theoretical approaches. First, both Kant and Waltz stress the importance of anarchy in the
international system, but they differ in their conceptualizations of anarchy. For neorealists,

anarchy is simply the lack of central authority in the international system. Kant, on the other

43
hand, views anarchy as “the state of war”, a state that impedes the categorical imperative to

act morally. At the other extreme is the rule of law embodied in the perpetual peace, a law
that is based on legitimacy rather than coercive power. (Huntley, 1996260)

The source of structural change is also a source of disagreement between Kant’s
theory of international politics and neorealism. Waltz argues that changes in the distribution
of capabilities has the largest impact on structural change, due to the persistence of anarchy
and the “likeness” of units in the system. For Kant, war is the greatest force creating
structural change through its impact on the domestic structure of the units in the system. The
structure and the units “are always interacting, defining a process of evolutionary change that
continues perpetually (Huntley, 1996261).” Thus Kant, like Ruggie, views the differentiation
of units in the system as an important source of structural change. His theoretical approach
specifies how the structure of the system (particularly the state of war) produces changes in
the units, and how the internal make-up of the units results in changes in the structure
(moving the system closer to the “rule of law” internationally).

The third primary source of disagreement between Waltz and Kant lies in their
interpretation of the impact of the ordering principles (anarchy, hierarchy) on the
international system. Waltz treats the structure’s ordering principles, anarchy and hierarchy,
as diametric opposites, whereas Kant places them on a continuum. The state of war, or
anarchy is at one end of the continuum, and the rule of law is at the other. “This continuity
between war and peace follows from the dialectical conception of their interaction. History
may be viewed as the story of erratic (and never ending) progress along this continuum

(Huntley, 1996262).”

44
Both Kant and Waltz recognize the importance of the socialization among states and

the competition between them. In the next section, I discuss the competitive and socializing
forces that produce a dynamic relationship between democracy and war over time. One
important consequence of this dynamic relationship that has been overlooked is its impact on

the norms and rules of engagement in the international system.

The Diffusion of Democratic Rules and Norms

Huntley ( 1996) argues convincingly that the processes of socialization and
competition are equally important in Kant’s and Waltz’s systemic theory of international
politics.

In neorealism, competition and socialization are the two mechanisms by
which the structure of anarchy influences international outcomes. For
Kant, put in neorealist terms, the violence inherent in the structure of
anarchy actuates a transformational dynamic independent of the
intentions of states themselves. Anarchic competition compels states to
extend the rule of law internally, and externally in their relations with
other states ruled by law, while socialization reinforces both these
republics’ peaceful relations and other states’ competitive incentives to
“join the party.” In neorealism, competition and socialization produce
convergence in state behavior and international outcomes. Kant’s
thought adds the idea that this process works dynamically, promoting
convergence towards the rule of law. This is the core of the systemic
dimension of Kant’s formulation. (Huntley, 1996:58-59).

The greater strength and success of republics over time encourages other states to liberalize
(competition), while the “rule of law” extemalized by democracies in their relations with

each other (socialization) creates a more democratic international “rule of law” over time.“5

 

3" Ray (1995) also discusses the importance of competition on global trends in democracy. “Nation-
states in the global system see themselves as facing fundamentally. . .sirnilar goals, such as political
autonomy, territorial integrity, and economic well-being. It is only natural, or rational, that political
leaders in the great majority of. ..states would watch and see who among them best achieves these goals
and how. (53)”

45

One of the most important consequences of this process of socialization and
competition is its impact on the rules and norms of international interaction. The more
democratic the international system, the more likely that the norms of interactions between
states will change. In particular, the greater the number of democracies in the system, the
more likely the traditional “rules” of the international system (e. g. power politics) will
become replaced by more cooperative, conciliatory, and peaceful rules of interaction like we
see between democracies. In Kant’s language, the greater the number of republican
governments in the world, the more widespread their international “rule of law”, which is
both created from their domestic rule of law and perpetuated by the peace among themselves.
War serves to promote the spread of republican forms of government, and the overall strength
and success of republics serve to further their growth and influence over time. Cooperation
in economic and cultural spheres serves to strengthen the peaceful and cooperative systemic
norms of behavior. Russett (1993:24) makes a similar argument about the spread of
democracy: “It may be possible in part to supersede the ‘realist’ principles (anarchy, the
security dilemma of states) that have dominated practice to the exclusion of ‘liberal’ or
‘idealist’ ones since at least the seventeenth century.”

Hegemonic leadership plays an important role in promoting democratic norms in the
international system. Advocates of hegemonic stability theory (e. g. Kindleberger, 1978;
Keohane, 1984; Gilpin, 1987; Lake, 1988) claim that cooperation in the international political
economy is facilitated by a powerful liberal state (hegemon) that creates and maintains liberal
economic institutions and regimes. Also a hegemon aids in the development of a world
market economy. (Gilpin, 1987) While scholars have employed numerous definitions of

hegemony, most agree that a hegemon is the most powerful state in the international system

46
that dominates economic and political affairs.37 Keohane (1984232) relies on a more specific

definition: “Hegemonic powers must have control over raw materials, control over sources of
capital, control over markets, and competitive advantages in the production of highly valued
goods.” Also, in order to be considered a hegemon, a state must be committed to liberal
values, especially the rules and norms of a liberal trading order. The most common historical
examples of hegemons based on these criteria are Great Britain and the United States.
Hegemons facilitate the growth of democratic norms in several ways. First, they
create liberal regimes and institutions whose norms and rules stem from their domestic
political processes. A good example is the General Agreement on Tariffs and Trade (GATT),
an institution established by the United States in the aftennath of World War II to promote
open trade policies. One of the key principles of the GATT is the Most Favored Nation
(MFN) policy of “unconditional reciprocity” where concessions made to one member are
automatically extended to all other members. (Gilpin, 1987274) A second way that a
hegemon aids in the promotion of democratic values is through its support of open trading
policies that allow its goods and services to flow abroad freely. One consequence of such
global trade is that the hegemon’s domestic cultural values and norms are often transmitted to
other states. Deudney and Ikenberry (19912112), for example, discuss the spread of the
American business culture “in which the norms of openness, predictability, and performance
prevail” and the American commodity culture “created by globally standardized and mass-

marketed products such as blue jeans, Coca Cola, and television.” They argue that the spread

 

’7 The preponderance of military power is another typical characteristic of hegemony (e.g. Organski and
Kugler, 1980; Modelski and Thompson, 1988).

47
of American and Western culture was a significant international factor leading to the breakup

of the Soviet Union.

Finally, the hegemon requires support from other major states in the system in order
for its leadership to be effective, what Gramsci refers to as “ideological hegemony”
(Keohane, 1984244-45). Gilpin (1987273) elaborates this point: “other powerful states must
also have an interest in the growth of market relations; the hegemon can encourage but
cannot compel other powerful states to follow the rules of an open world economy.” It is
easiest for democratic states to support each other in such a system, due to their common
domestic “rule of law” that becomes translated into common norms and rules for the
institutions and regimes they establish. The prevalence of these liberal economic institutions
globally promotes the Spread of the rules and norms underlying their structure, furthering the
Kantian international “rule of law”.

What types of rules and norms are characteristic of the interactions between
democracies, i.e., what types of behavior characterize the intemational “rule of law”
promoted by democracies? In democracies, political conflicts are more likely to be resolved
peacefully through compromise and negotiation. In addition, decision-makers are expected
to respect “the rights and continued existence of opponents (Russett, 1993235).” Political
success is not achieved through the elimination of political opponents; all parties agree to the
nonviolent transfer of power and they rely on compromise to resolve debates over

contentious issues.38 These norms of peaceful conflict resolution prevail when democracies

 

3’ Dixon (1994) calls this the norm of bounded competition. “All modern democracies are openly
competitive systems of governance where conflicting material interests and basic political values
routinely clash over the proper course of public action. . .Just as competition is a constant of democratic
governance, so too is the presence of rules, procedures, or guidelines for setting its boundaries (Dixon,
1994215).” Dixon argues that one of the most important rules of bounded competition is to refrain from

48
interact with other democracies. Democracies are also more likely to agree to third-party

arbitration for dispute resolution (Dixon, 1993, 1994; Raymond, 1994, 1996), and to settle
disputes by stalemate or agreement, as opposed to relying on an imposed solution (Russett,
1 993 2625).

There are several other examples of norms and rules that stem from the international
and domestic republican “rule of law”. Russett (1993231) observes that “a norm of equality
operates both as voting equality and certain egalitarian rights to human dignity” between
democratic states. He argues further that democracies are more likely to uphold the
intemational agreements they Sign “because of the visible nature and public costs of breaking
comrrritrnents39 (Russett, 19932137)” Democracies are also more likely to respect and
promote human rights internationally, a principle that follows naturally from Kant’s
noumenal awareness, especially respect for the freedom to be moral. (Deudney and
Ikenberry, 1991; Huntley, 1996) Finally, there is some evidence that democracies are less
likely to make nationalist or territorial claims outside their boundaries, promoting a norm of
respect for the territorial status quo. (Kacowicz, 1995:266)

In conclusion, as the proportion of democracies increases, the rules of engagement
between democracies will start to diffuse throughout the international system, particularly if
the largest states are liberal democratic regimes. My expectation is that as the proportion of
democratic states increases, the level of systemic war will decrease. Also, the pacifying

impact of democracy on war should get stronger over time, as more and more states become

 

the use of violent force to obtain political goals or advantage over political adversaries.
’9 See also Fearon (1994) and Gaubatz (I996).

 

 

 

49

democratic and as the rules of the international system shift away from “neorealist” rules

towards Kant’s republican rule of law.

Testable Propositions

Several propositions can be derived from the theoretical arguments discussed above.
The first is the basic democratic peace proposition at the systemic level (Pl), which was
derived from Kant’s notion of perpetual peace.

Pl: As the proportion of democracies in the system increases, the
proportion of nations fighting war in the international system decreases.

An alternative hypothesis can be derived from previous theoretical work that
aggregates findings at the monadic and dyadic levels to the systemic level. I use the
Gleditsch and Hegre (1997) model for comparison. If their model is accurate, then the
relationship between democracy and war could be positive if the proportion of democracies
in the international system is small initially. I test the first proposition against this
alternative:

P2: Periods of global democratization can be followed by periods of

higher or lower systemic conflict, depending on the total number of

democracies in the system. If the overall number of democracies is

small, the relationship should be positive.
Testing these competing hypotheses provides a nice comparison of a theory of democracy
and war developed at the systemic level versus a theory derived logically from findings at
other levels of analysis.

Also I expect the impact of democracy to get stronger over time. As the proportion of

democracies in the world increases, the domestic “rule of law” that is characteristic of

republican forms of government will diffuse throughout the international system, moving us

50

closer to Kant’s international “rule of law”, and making the rules and norms of behavior
between democracies more prevalent in international relations at large. Thus, the substantive
significance of democracy should be increasing historically.

P3: The degree of the negative impact of democracy on war in the

international system will become stronger over time as the proportion of

democracies increases.

A key feature of the systemic theoretical relationship between democracy and war
discussed above is that the relationship between these variables is endogenous. More
democracy in the system makes for less war, but war also promotes republican forms of
government, due to the success of democracies in wars and the improvement in domestic and
international institutions that war brings about over time. This reverse causal relationship is

tested with the following proposition:

P4: An increase in the proportion of nations fighting war in the
international system will increase the proportion of democracies.

As the number of democracies grows over time, however, the international “rule of law” will
move ever closer to Kant’s vision of perpetual peace. Thus the frequency of international
conflict should be declining as the system becomes more democratic. Also, the positive
influence of war on the creation of more democratic forms of government should decline
over time as the incidence of war becomes less frequent and as the federation of fi'ee states
enlarges.

P5: The degree of the positive impact of war on democracy will

decrease over time as the proportion of democracies in the international
system increases.

51

Having elaborated the relationship between democracy and war at the systemic level, I turn
now to a discussion of three other factors that influence both democracy and war in the

international system: power concentration, world trade, and world production.

Power Concentration and Systemic War

Recall that one important source of structural change in the neorealist theoretical
model of international relations is the distribution of power in the system. While I have
shown that the differentiation of units in the system (especially with regards to their
domestic regime type) has a significant influence on the likelihood of systemic war, it is
also possible that Waltz’s other primary source of structural change, the distribution of
capabilities, is important as well. Work on the relationship between the distribution of
power and war at the systemic level can be grouped into two broad theoretical approaches:
the power preponderance school and the balance of power school.

Both power preponderance and balance of power theorists argue that systemic war
is largely determined by the distribution of power between states and the uncertainty
associated with the relative position of states and changes in that distribution over time.
Uncertainty refers simply to “the difficulty which foreign policy elites experience in
discerning the stratification’s and clusters in the system, and predicting the behavior of the
other members of that system (Singer, et al, 1972271)” According to Singer, Bremer, and
Stuckey (1972), three factors increase the level of uncertainty in the system: 1) an equal
distribution of capabilities, 2) a change away from high concentration of power towards a

more equal distribution of power, and 3) relatively fast changes in the concentration of

52

capabilities. Balance of power theorists argue that war in the system is less likely when
there is a relatively equal distribution of capabilities between major powers, when power
concentration changes towards parity, and under conditions of “a relatively fluid power
hierarchy (Singer, et a1, 19722293)”, where it is easy for nations to change places in the
rank order of states. Power preponderance theory, on the other hand, predicts that war is
less likely if the concentration of power in the system is high (concentrated in the hands of
very few states), if concentration changes towards greater concentration, and if the rank
order of states in terms of power is relatively stable. Singer, Bremer, and Stuckey (1972)
find support for the power preponderance model in the 20th centtuy, but find that the
balance of power model does a better job of predicting systemic war in the 19‘h century.
Later analyses of power concentration and war (e. g. Doran and Parsons, 1980; Thompson,
1983a, 1983b) find further support for the power preponderance position, i.e., that systemic
war is less likely the higher the concentration of global capabilities. Many of these studies
also question the inter-century differences in the relationship between power concentration
and war uncovered by Singer, Bremer, and Stuckey (1972). I test the following power
preponderance proposition.40

P6: The proportion of nations fighting war in the international system
will decrease as the concentration of power in the system increases.

 

‘° Mansfield (1992, 1994) argues that the relationship between systemic capability concentration and war
is nonlinear. He finds empirical support for an inverted U-shape relationship, i.e., “both the highest and
lowest levels of concentration may be associated with the lowest incidence of warfare while intermediate
levels of concentration may be associated with the highest incidence of warfare (Mansfield, 199227).” I
do not test this relationship in my analysis, but it would be an interesting avenue for fiiture research.

53
Trade and Systemic War

I argued earlier that a hegemon facilitates the growth of democratic norms and rules
in the international system at large, especially through the liberal trading regimes it
establishes and maintains. Through its global trading network, the hegemon promotes
greater cooperation and less conflict in the international system, while at the same time
transmitting its domestic cultural values and norms to other states. In this sense, increases in
the level of hegemonic trade can decrease the frequency of war in the international system
by promoting the Kantian international “rule of law”. More generally, “economic
interdependence reinforces constitutional constraints and liberal norms by creating
transnational ties that encourage accommodation rather than conflict. Thus material
incentives add their force to law and morality (Oneal, et al, 1996212).”

Mansfield (1994) provides an alternative theoretical approach, where he also
examines the systemic relationship between trade and war.“ Siding with the commercial
liberals, Mansfield (1994) argues that increased trade in the international system makes
war less likely. Trade increases dependence between states and “provides a more efficient
and less costly means than war for states to gain access to resources and markets
(Mansfield, 19942122)” He points out, however, that the distribution of capabilities also
has a strong influence on the likelihood of systemic war in addition to international trade,
consistent with the work on interdependence (e. g. Keohane and Nye, 1977, 1987).

Mansfield finds strong evidence for a negative relationship between trade and war, i.e., as

 

4' As Mansfield (1994) points out, the majority of work on the relationship between trade and conflict has
been conducted at the dyadic level. These studies are tangential to the systemic theory developed here,
thus I do not include them in my discussion. Good examples of dyadic studies of trade and conflict
include Polachek (1980), Gasiorowski and Polachek (1982), Gasiorowski (1986), Pollins (1989a, 1989b),
Barbieri (1996), Oneal, et al (1996), and Oneal and Russett (1997).

54

the level of world trade increases, the amount of war in the international system declines. I
test this proposition:

P7: The higher the level of trade in the international system, the lower
the proportion of nations fighting war.

Production and Global Democracy
Thus far I have focused on only one factor that promotes the spread of democracy

globally, war. Other factors may also influence the processes of democratization in the
international system. Much of the work in comparative politics has focused on domestic
political and economic factors conducive to the development of democracy. (e.g.
O’Donnell, et a1, 1986) In his analysis of the “third wave” of democratization, Huntington
(1991) identifies a few global or regional influences on the spread of democracy, in
addition to numerous domestic factors that promote democratization. These include
experience as a British colony, occupation by a prodemocratic foreign power, influence by
a prodemocratic foreign power, and elite desire to emulate democratic nations, As Ray
notes,

Huntington himself (1991:45-46) posits that the most recent. . .wave of

democratization in the global system was a result of (l) legitimacy

problems of authoritarian systems on the global level, (2) global

economic growth, (3) changes in the policies of the Catholic Church,

(4) changes in the policies of the European Community, the US, and

the Soviet Union, and (5) snowballing or demonstration effects from

the earlier democratizers of this time period. All of these factors are

basically external to the states being analyzed. (Ray, 1995:51-52)

I focus here on the impact of global economic growth on the development of

democracy in the international system. I expect higher levels of economic production in

55

the system to be associated with greater levels of democracy. I test the following
proposition:
P8: Increases in world production will increase the proportion of
democracies in the international system.
Conclusion
The theory developed in this chapter makes several contributions to the existing
literature on the democratic peace. First, like Kant, I argue that democracy and war are
endogenously related to each other over time. War creates conditions that favor the
development of republican constitutions and a rule of law among nations, while peaceful
relations between democracies creates greater peace and furthers the spread of democracy and
democratic norms in the international system. “The perfected republic is then a consequence
as much as a condition of peace—a dependent as much as an independent variable (Huntley,
1996259).” A vast majority of studies on the democratic peace study only the impact of
democracy on war, without considering how war affects the development of democracy."2
My approach is also unique because I model the relationship between democracy and
war dynamically. It is typical in international relations research to assume that a single model
explains the relationship between the variables of interest over time. To be true to Kant’s
ideas, cultural theories of the democratic peace should take into account the evolving
relationship between democracy and war. I demonstrate that the systemic component of

Kant’s theory predicts an increasing (negative) impact of democracy on war in the system,

 

‘2 I show clearly in Appendix A that the parameter estimates in the estimation of a single equation are
biased if the true theoretical relationship between two variables is endogenous.

56

whereas the (positive) impact of war on democracy should be getting weaker as the system
moves closer to Kant’s vision of perpetual peace.

The third primary contribution of my theoretical approach is the added emphasis on
rules and norms of behavior in the international system, particularly how these are evolving
as a consequence of the dynamic relationship between democracy and war. Other scholars
have discussed the potential influence of the spread of democracy on the norms and rules of
behavior in the international system. For example, Bruce Russett (1993) concludes his book,
Grasping the Democratic Peace, with the following observation.

Perhaps major features of the international system can be socially

constructed from the bottom up: that is, norms and rules of behavior

internationally can become extensions of the norms and rules of

domestic political behavior. . .if enough states become stably democratic

in the 1990’s, then there emerges a chance to reconstruct the norms and

rules of the international order to reflect those of democracies in a

majority of interactions. (Russett, 1993:137-138)
I have demonstrated that this evolution towards systemic democratic norms and rules stems
from the dynarrric and endogenous relationship between democracy and war over time.

Finally, the theory is developed at the systemic level going beyond previous
tendencies to aggregate empirical findings from the monadic and dyadic levels of analysis to
the systemic level. Given that the empirical results on the democratic peace proposition at
the monadic level are debatable, a theory developed at the systemic level is preferred to a
systemic result derived from aggregation across levels.

In the next chapter, I describe the operational measures used to test the propositions

derived from my systemic theory of the democratic peace. I test these propositions in

Chapters four and five with structural equation and sliding window estimation.

CHAPTER 3

GATHERING EVIDENCE

In this chapter, I provide a detailed description of the operational measures used
in the statistical analysis in Chapters four and five. I also present the time series plots,
descriptive statistics, and Augmented Dickey Fuller (ADF) tests for each variable. The
present study’s propositions are tested with annual systemic data from 1816-1988.1 The
data used to test my theory is derived from several data sets. Table 4 contains a summary
of each variable and its source and Table 5 presents the summary statistics for each

variable.

Endogenous Variables

The two endogenous time series variables analyzed in this study are the level of
systemic war and the level of global democracy. Measuring these two series is challenging
at the systemic level because they both depend to some degree on the total number of states

in the system. To minimize this problem, I adopt proportional measures of war and

 

’ In future research, I plan to extend the temporal domain. My time series end in 1988 due to the lack of
availability of two economic series (from the sources used), US Gross National Product and British
Exports. Also, the Correlates of War national capability data, the source for the measure of power
concentration, ends in 1991.

57

58

democracy in the system that control for the total number of nations in the international

system per year.44

Systemic War

The level of war in the international system is measured as the total number of
Correlates of War interstate war participants per year divided by the total number of
system members (Propwar). The COW project classifies a nation as a member of the
interstate system if 1) its population exceeds 500,000 and it receives diplomats from any
two major powers,45 or 2) the nation is a member of the League of Nations or the United
Nations any time in its existence (Small and Singer, 1982:41). A COW interstate war is
any violent conflict fought between two or more system members that exceeds a
minimum battle death threshold of 1000. A nation is counted as a participant in a war if
it either incurs 100 battle deaths, or it commits at least 1000 armed personnel to active
combat (Small and Singer, 1982255).

I begin by plotting the first endogenous variable in this study, the proportion of
system members fighting interstate wars (Propwar), in Figure 1. The autocorrelation
and partial autocorrelation functions for this time series are plotted in Figures 2 and 3.
The autocorrelation function (ACF) of a time series, y,, is the covariance of yt and

subsequent values of M divided by the variance of y,:

 

4‘ Raknerud and Hegre (199727) discuss the impact of increasing numbers of states in the system on
dyadic level analysis of the democratic peace proposition. They demonstrate that the “probabilities at the
dyadic level must depend on the size of the system. ..if the war probabilities in each dyad-year are
constant, the probability of war on the national level approaches one as N increases.”

‘5 The Correlates of War project (Small and Singer, 1982:47-50) identifies nine major powers from 1816-
1988 (US 1899-1988, UK 1816-1988, France 1816-1940, 1945-1988, Germany/Prussia 1816-1918,
1925-1945, Austria-Hungary 1816-1918, Italy 1860-1943, USSR 1816-1917, 1922-1988, China 1950-

59

EIIY: - #l(Yr-k -#)1
Pk = Var(Yr)

 

Each pk falls between -1 and +1. (Judge, et a1, 19802227) Each spike in the partial
autocorrelation function represents the partial derivative for each y,_k with respect to y,. In
other words, each value in the PACF corresponds to the regression coefficients (it’s) for

the following equation:

Yr : ler-l + szr-Z + ° ° ° + AkYr-k + 8I

As expected, the proportion of system members fighting wars is highest during the
Seven Weeks War, World War I, World War II, and the Korean War. Examination of
the ACF and PACF functions indicates that the measure of systemic war (Propwar) is
stationary, i.e., the sample moments such as the mean and variance are invariant with

respect to time.“

Systemic Democracy

The other endogenous variable, systemic democracy, is more difficult to measure
due to the wide variety of conceptualizations and measurements of democracy in the
international relations and comparative politics literature. The three dominant sources for
empirical measures of democracy in the intemational relations literature include: Polity III

(Jaggers and Gurr, 1995; Gurr, J aggers, and Moore 1990), the Freedom House Indicator

 

1988, Japan 1895-1945).
’6 I present the results of a more formal test for stationarity, the Augmented Dickey-Fuller test, later in
this chapter.

60

(Gastil 1990), and Doyle's (1986) dichotomous measure of democracy. Numerous scholars
have created separate indices with these data sets including Rummel (1983), Chan (1984),
Weede (1984), Maoz and Abdolali (1989), Bremer (1992), and Maoz and Russett (1993).

The systemic theory developed in Chapter 2 is derived largely from Kant’s theory of
perpetual peace between republics. Doyle’s (1986) empirical measure of “liberal regimes”
comes the closest to Kant’s definition of a republican form of government. Doyle
(1986:1164) identifies democratic states on the basis of four institutional features that Kant
considered firndamental characteristics of republican governments: 1) market and private
property economics, 2) external sovereignty, 3) juridical rights for citizens, and 4)
“republican” representative government (republican or parliamentary monarchy). I created a
measure of the proportion of democracies (Doyledem) by dividing the number of “liberal
regimes” per year by the total number of Correlates of War system members. This measure,
which covers the time period 1816-1982, is plotted in Figure 4.

I also analyze an alternative measure of democracy using the Polity 111 data set. A
large portion of the empirical work in the democratic peace literature ad0pts the Polity III
ordinal measure of democracy, thus including it provides both a comparison to previous
analyses and an alternative conceptualization of democracy. In addition, the Polity 111 data
set covers a longer time period (1800-1994) than Doyle’s measure of democracy (1800-
1982). ). The Polity 111 project created an ordinal measure of democracy that combines
information from several institutional characteristics of a polity: the competitiveness of
political participation, the level of constraints on the chief executive, and the openness and

competitiveness of chief executive recruitment. (Jaggers and Gurr, 19952471) This

61

measure of democracy is based primarily on institutional features of democracy, and it has
been used frequently to test the structural theory of the democratic peace.

Using the Polity 111 data, I define a democracy as any nation whose democracy
score is greater than five."7 For each year, I divide the number of democracies in the
system by the total number of Correlates of War system members to produce a second
measure of the proportion of democracies in the system (Politydem). Any countries whose
democracy scores are missing or who are not system members according to the Singer and
Small (1982) criteria are excluded from the proportional calculation. This measure of the
proportion of democracies is plotted in Figure 4 along with the indicator derived from
Doyle’s list of liberal regimes.

The most interesting thing to note is the similarity between the measures, even
though the conceptualizations of democracy in the two data sets are quite different. Also,
there is a slight upward trend in both series over time, indicating the growth of democracy
in the international system over time. The Pearson product moment correlation between
these measures is fairly high (r=.939). The autocorrelation and partial autocorrelation
functions for these two systemic democracy measures are plotted in Figures 5-8. Both
series have strong persistence in the autocorrelation function, indicating that the series

could be nonstationary.

 

‘7 Some scholars utilize a more stringent definition of democracy. The cutoff value of six makes sense in
the present study because 1) a lower threshold is probably more appropriate for analysis in the 19"I
century and 2) a value of six is more likely to capture any changes towards democracy after war which
might be precluded by a more stringent criteria. I also tested the second model in Chapter 4 using a
Polity III democracy threshold of nine, and the results are similar to those reported here.

62

Exogenous Variables

The focus in this study is on the endogenous relationship between democracy and
war. As I noted in Chapter 2, I also examine the impact of three additional variables on
democracy and war: systemic capability concentration, systemic trade, and world

production. I describe how each of these variables is measured in the sections that follow.

Systemic Capability Concentration

The indicator of systemic capability concentration in the international system
employed is Similar to the concentration measure developed by Singer and Ray (1972), and
analyzed in many studies including Singer, Bremer, and Stuckey (1972) and Mansfield
(1992, 1994). The capability concentration measure in these studies is computed on the
basis of each major power’s share (S) of total composite capabilities in the international

system."8 Concentration (Concen) is measured as:

 

N 1
s.2——
2() N

i=1
1
1__
N

 

Concen =

The number of major powers in the system is denoted as N, and Si refers to each nation’s
percentage share of total composite capabilities. While many previous scholars analyze
only five-year aggregate measures of concentration, I compute the capability concentration

in the system for each year from 1816-1988.

 

‘8 Composite capabilities are calculated by summing a major power’s values for military personnel,
military spending, total population, urban population, energy consumption, and iron and steel production
(COW National Material Capabilities Data). 1 adopt the same criteria for major power status as the

63

The systemic capability concentration measure (Concen) is plotted in Figure 9 and
the ACF and PACF are plotted in Figures 10 and 11. The concentration of power among
major powers has increased over time, which is indicated by the persistence in the

autocorrelation function (Figure 10).

Systemic Trade

I measure the level of systemic trade as the total amount of annual British exports
per year (Mitchell, 1988). Given that the hegemon serves to promote democratic norms, it
makes sense to rely on British data. Also, British exports are highly correlated with
general measures of world exports such as Goldstein’s (1988) general indicator of world
trade.49 I take the natural log of the British export series in order to account for increasing
variance in the series over time; the logged data is plotted in Figure 12 and the ACF and
PACF are plotted in Figures 13 and 14.

As is common with most economic series, British exports Show an increasing trend
over time. This can be seen both in the time series plot (Figure 12) and in the persistence

present in the autocorrelation function (Figure 13).

World Production
The final exogenous variable analyzed in this study is the level of world

production. I argued that the level of democracy in the system is positively related to the

 

Correlates of War project (Small and Singer, 1982247-50); see list above.

‘9 The Pearson bivariate correlation between world exports (Goldstein, 1988) and British exports from
1816-1975 is 0.953. The correlation between the natural log of these two measures is also high (r =
0.933).

64

level of production in the system. I measure systemic production using the annual
measure of US. Gross National Product. This correlates highly with more general
measures of world production (Goldstein, 1988).’0 The series is logged in order to
account for increasing variance over time. The logged series is presented in Figure 14,
and the ACF and PACF are plotted in Figures 15 and 16. Like systemic trade, world
production is also nonstationary, with an increasing mean over time. This can be seen in

the raw data plot and in the persistence of the spikes in the autocorrelation function.

The Issue of Stationarity

The first step involved in any time series analysis is to determine the integrated
order of each series to ensure that the data are stationary and that each equation is balanced
(i.e., both sides of an equation are 1(0)). A stochastic process is stationary if the sample
moments such as the mean and variance are invariant with respect to time. An example of
nonstationarity in time series data is an upward trending series that is characterized by an
increasing mean over time. Inferences made from models with nonstationary data are usually
invalid due to an increase in variance as the sample size increases."

I utilize Augmented Dickey Fuller (ADF) unit root tests to determine the integrated

order of each variable. A variable is said to be integrated of order d, or I(d), where d refers

 

5° The Pearson bivariate correlation between world industrial production (Goldstein, 1988) and US Gross
National Product from 1816-1975 is 0.976. The correlation between the natural log of these two
measures is also high (r = 0.988).

5' One is also more likely to encounter a spurious regression problem when two (or more) upward-trending
series are regressed on each other. Thus we will tend to find significant relationships between two
nonstationary variables, "even if the only thing they have in common is the upward trend. In fact, the R2 for
a regression of y, on x1 and a constant will tend to unity as n -> oo whenever both series can be characterized
by (nonstationarity)...even if there is no correlation at all between the stochastic part of yt and x,." (Davidson
& MacKinnon, 19932671) The basic result is that we are unable to make valid inferences from models that

65

to the number of times a series must be differenced to become stationary. The ADF test is
a more formal method for determining if a series is nonstationary, as opposed to
examination of the ACF and PACF plots that I discussed above. The augmented version of
the Dickey-Fuller test accounts for the possible presence of serial correlation. The null
hypothesis for the ADF test is that 7 equals zero in the following model:
Ari = 11 + r ya + 2¢1Aytj+ 81

Rejection of the null hypothesis implies that y, is stationary, or 1(0). A failure to
reject the null indicates the presence of a unit root, or that Ayt is stationary. Table 6
summarizes the results of the Augmented Dickey-Fuller tests and the transformations
necessary to achieve stationarity for each variable analyzed.52 Each time series that
contains a unit root is first differenced, which is true for every variable except the

proportion of system members fighting interstate wars (Propwar).

Conclusion

In this chapter, I discussed the operationalization of each key concept in my
theoretical model. I presented the basic descriptive time series plots for each series and
I also presented the results of Augmented Dickey-Fuller tests for the presence of unit
roots. I first differenced any variables that contained a unit root; these transformed

series will be employed in the structural equation estimation in Chapter 4.

 

use nonstationary data. Even in cases where we find significant results, they may be artifacts of the
nonstochastic properties of the series.
52 The ADF tests are run with two lags of Ayn.

66

Table 4 - Operationalization of Theoretical Concepts

 

 

 

 

 

 

 

Concept Variable/Measurement Sources
Systemic War Annual proportion of Correlates of War Interstate
system members fighting War Data, System
interstate wars Membership Data
(Propwar)
Democracy Annual proportion of Doyle (1986)
democracies relative to total Jaggers and Gurr (1995)
number of system members (Polity III)
(Doyledem, Politydem)
Systemic Trade Annual British exports Mitchell (1988)
(logged)
(Lnukexp)
World Production Annual US Gross National Mitchell (1988)
Product
(Lnusgnp)
Systemic Capability Annual major power Singer and Ray (1972)
Concentration capability concentration Singer, Bremer, and
score (ranging from 0-1) Stuckey (1972)

 

(Concen)

 

 

 

67

 

 

 

 

 

 

 

 

84.2 8:3 SSA 33a 32: $2.22 A: $25
an: 8:; 8o: 33 SM d e as 3:: m: aacafi
88d 826 m2 3 $85 a: 3 $2-2M: m: 8280
Sad 38¢ ES 083 802 32.22 m: asagoa
Eds $86 $8.9 52 $3 $2.22 E 803:8
893 88¢ :33 ~23 $86 32.22 m: aaaea
newsman
.5852 SEE: aaeasm names: at: was; 2 Dane;

 

 

 

 

 

 

 

 

mosmufim begm I m 035.

 

68

 

3&3ch

 

 

 

 

 

 

 

 

 

8:20me “BE 2: Rewmscd 8355 3:232 «.35 m3 mo ms 18:32
axsuafiv
00:20me “SE C: axoxzcd 3.8qu 555 we wo_ 1.532
@3503
monocota «RE 2: 38:0an coumbcoocoo 5:533 526a 5.38 28393
Sagas
8:08me EE 2: AEocbzomv 88?? 3:059:85 05 5 86800838 mo :oEogoE
Exceeds
oocouoba “RE 2: AEoonoQV 83m? EcoumEBE 2: E 86885.net MO 22:0qu
282 8: ASBQOHAV m8? 88285 macaw—m £382: 58?? mo :oEomoE
cosmetommcflh EEO wBfiwBE oEmcm>

 

motom some cow mcogficofiem; can EEO wBSwBE 23 mo inseam - o 2an

 

69

Proportion War

 

 

 

   

 

  

»

 

 

  
 

ll .. —- a. . . :ll—i

 

1J1.

a; -5: :mm
_mNm swam Lma_ _mde amen _eum _ewu

4.9:.

~3de A

 

laws 63 .- -

$3-

‘1.

 

-madm

3%

125 3360303 om mama 33:ch Emwmsm 388»me €39 $3-38

 

70

Autocorrelation Function

i . . . . .1. .1 .11-
obi _ -1-11I.| .. . 1

Oosmaasoa C35

 

I 083083

A. N a s 1m m dim. e113 : 5.: E a a
752:5an

Emma N

>94 ow :8 338030: om 3883 288an Emrmsm 38888 $58

71

Partial Autocorrelation Function

 

 

 

 

 
 

 

 

. n
-.m 1 _ Oosmqosnm 5.138

kg 003399:

 

H 1N w -, A M1 a 1N 1%-:10 5-3-131 3. 1.: a a:

bum 25:32
Emma w

Tram 0m :5 303030: o», mwmfia Zchma Emrmsm 38368 [Ema

72

 

 

 

>
.m- \K«>2
A. 1_<> / f}; 7.9g \
é .\ Z 7 \
w I >|\|\/\//\Z 3. , H .> \r, \K
. 1 / \ I I I
r er \ K 2 \ < \ /\f\/\I\
RU / <\ Uofinw 308:8
.N; . ‘ ma
I : 2 0 35080
L / / J R .\ \{N/\.\ II v~
T n a .
. I I (Fl 2 Hue—HQ E 30850
< /IIL
ob . I ow aoBooSow

 

53 5.3 53 5.3 5mm Goa EMA SAN Goo Gum
5mm 53 :3— 53 53 $3 $3 53 Saw 53

J83

Ema?“ A

125 38030: O». 00800398 5 :6 5835303: mvfioB. Hm _ 958

73

Autocorrelation Function

 

 

 

 

 

 

re
.m.
Pf
-.m i Oosmmosoa 582m
1; . -I I 0830on
wwwwmwwmoszsazaa
rum 2:333

Ewan m

>Om om :5 38038: on Dan—00398 Eofiamw 332:3

74

Partial Autocorrelation Function

 

 

S 1, I11. ..I .

 

 

-.m , 003.5058 :8me

 

 

L .c I Ooammowaa

— ~ _ _

_Noammwmoa:533_wIa

 

fimm 2539
Emma 9

H399“ om So 32530: 0». 02500398 Quofiqm 80858

 

75

Autocorrelation Function

 

 

 

-.m i . nonmagoo 555

 

 

I Ooowmomna

 

Lb

q _

uamoomoa:aaiaa

 

rum 2:869.
Emma q
>9“ 0». 38038 om 0038898 coo—5N a 3858.3

 

76

Partial Autocorrelation Function

_.o

 

 

ob - III-II-III'II..I I -

-.m 1 091.588 :85.

 

 

L .o l Ooommomaa

«IIIqIIlluIIII— . Id q A «

.awwaimmqmos:szzaa

.-
—1
-1
_1
-
.-

rmm 2:369
Ema?" m
599.. o». Ea 32530: om 0958898 QuomQ E 25353

Concentration

77

 

2-111-! . n

 

 

 

 

 

P

 

 

Sr --
a;

 

_

_ _

_ _ u «I a q _

53 - RS. 5.3 a? 3.8 3f 65 3.8 3%
ammm Hmaw _mo_ _ch umou He_m chw _omu _ooo new“

5%:
Emma c
38. ca woéaa 038ng nonowfiafio:v Hm _ Q. ~ coo

78

Autocorrelation Function

 

5o

 

 

 

-.m 1

Lb

 

 

 

‘
CI!

Hmwamaqmoa:aaz

rum 2:802
Emfia 5

>9“ 0». 38.9 @029. 00009:? 0000038300

0950008 335

I Congomwa

79

Partial Autocorrelation Function

Bo

 

 

ob 1

 

Lb

 

 

 

_~winmmqmc_.o:asr

bum 2:36am
Emcnm 5
Exam cm 33.3 Hue/20H Ombmng 00:85:30:

3

3

003.5058 :35

I 08393:

80

Natural Log of British Exports

Hm -

 

_o1

N

 

 

Hmwo
_mmm

<8.

H mm;

5.

Hmpw

 

 

 

 

 

#wwm

«

Hmofl

_ u — .1

5.3 8.8 6.:
Hmoc ammo Hoam Howw

5.3

Emfia 5
W13: mxwonm commons. 53-3mm

as

_

HomH

6.8

Home

a;

Hcmq

81

Autocorrelation Function

 

 

 

 

 

—

_Nwwwaowca:&

.1
d
-q

rum 2:362
Ema—8 5
>9“ 0». was? mxnonm Common:

3

E

d

E

a

 

nonmagoa :8me

I Ooommomoa

82

Partial Autocorrelation Function

 

be

 

ob- IllI-l

 

 

q 1 _ «

#Nwawaomoa:fiaiaa

fimm 2:553
Emma 3
599» o», wing mxuonm commmav

 

nonmaaaom :35.

I 000399:

83

all: - ,

 

 

 

 

 

 

 

E-
m 51
S
U
c1
00° :7
O
L
m
m T
a
N
0.
3|--. ll . - . _ _
_m~m Hmwa ammm _moo _mmm Hooa _owp _oau ueoo _oom d
Hmmm Hmcw umo_ _moc ammo uo_m new“ _cm~ Home Homo
<03
3%:on

Gm 083 Zmfimosm: 38:2 commons. 53-3%

84

Autocorrelation Function

 

 

 

 

 

Lb

 

rum 25:53
#4me 3
>Om. om Cm 03mm 230:»: 3.09:: Commaav

nonmagon :85

I 000392:

85

Partial Autocorrelation Function

 

 

 

 

 

 

foil
.m.
OoOI llllillllilll'lllllllllllll'l'lllll]
-.m-
Lbil
wahmaqmoSZHNSEGE
fimmZEddQ.

Ewan 3
From om Cm 08mm 2395— wnoacofi Cowman:

003.538 :85

I 000399:

CHAPTER 4

TESTING THE SYSTEMIC DEMOCRATIC PEACE:
A STRUCTURAL EQUATIONS APPROACH

In Chapter 2, I developed a systemic theory of the relationship between democracy
and war based on Kant’s writings on perpetual peace. Two key features of my theoretical
model are 1) that the relationship between democracy and war in the system is endogenous,
and 2) this relationship is evolving over time. I derived several testable propositions that are
tested in this chapter with structural equations time series analysis. The methods of analysis
utilized in the next two chapters are well suited to capture both the endogenous and dynamic
properties of my theoretical model. In this chapter, I analyze several propositions derived in

Chapter 2:

0 P1: As the proportion of democracies in the system increases, the proportion of nations
fighting war in the international system decreases.

- P4: An increase in the proportion of nations fighting war in the international
system will increase the proportion of democracies.

0 P6: The proportion of nations fighting war in the international system will
decrease as the concentration of power in the system increases.

0 P7: The higher the level of trade in the international system, the lower the
proportion of nations fighting war.

86

87

0 P8: Increases in world production will increase the proportion of
democracies in the international system.

The remainder of this chapter is organized as follows. First, I discuss the approach to
time series estimation that I adopt in my analysis. Next, I present results from an empirical
model utilizing Doyle’s measure of systemic democracy. Third, I present results from a
second model with the Polity III measure of democracy. Fourth, I discuss the issue of
structural stability, which is crucial given the evolutionary relationships predicted by the
theoretical model. I present the results of Chow tests for structural stability before turning to

a more dynamic approach to estimation in Chapter 5.

Hendry’s Approach to Time Series Estimation

The time series analysis in this paper follows David Hendry's approach to
econometric modeling.53 The approach is general to specific in nature, and it is concerned
with issues of model specification and validation. Hendry's general to specific modeling
strategy is predicated on the argument that if two researchers start with small models, and
then reparameterize, they are unlikely to end up with the same model, making it difficult to
know which one is the correct specification. Similarly, if diagnostic tests or residual
correction procedures are applied (such as GLS), it is hard to know if the new specifications
are accurate. Hendry's methodology involves three basic steps (Pagan, 1990:99): 1) specify a
general empirical model that is consistent with the relationships between variables specified
by the theory, 2) simply the general model “to the smallest version that is compatible with the

data (congruent)” using a variety of statistical tests, and 3) evaluate the residuals and

 

’3 Hendry’s methodology is described in a variety of sources, including his (1995) book entitled Dynamic

88

predictive power of the reduced model, making sure that no relevant information has been
lost.54 As Pagan (1990:99) notes, “theory and data continually interplay in this
methodology.” Hendry argues that this type of "progressive" research strategy increases
modeling efficiency. In the structural equations estimation below, I start with a general
model based on theory and then test a series of restrictions that leads to a final “reduced”
model, consistent with Hendry’s approach to time series. The first model estimated in the

following section employs the measure of democracy taken from Doyle (1986).

Model One: Doyle’s Measure of Democracy
Using the transformed stationary series, I begin estimation with the following general

model. I include a legend for the variables and their full names in Table 7.

Propwart = 50 + 5,Propwar,,, + 62Propwar,_2 + 63Propwar,,3 + 8,,PropwarM +
iSSADoyledemH + 56ADoyledem,_2 + 57ADoyledern,_3 + 68ADoyledemM +
89ALnukepo + 5,0ALnukexp,_2 + 61 lALnukexp,_3 + 5,2ALnukexpM +
6,3AConcenH + 5,.,AConcen,,2 + 5,5AConcerr,,3 + 51,,AConcenM + a1t

ADoyledem, = [30 + BIPropwarH + [31,Propwart,2 + [33Propwar,_3 + B4PropwarM +

[3,ADoylederrr,_l + B,,ADoyledem,_2 + [37ADoylederrr,,3 + [3,;ADoyledemM +
BgALnusgnp.-1 + BmALnusgnp..2 + BuALnusgnp..3 + BtzALnusgan + 2,,

I include a reasonable number of lags for each variable (four) in order to capture the

dynamics in each time series.55 In addition, all variables are lagged in both equations to

 

Econometrics.

5“ Hendry’s use of the term “loss of relevant information” is in the classic sense. “Classic methods
discard information if they do not meet some prior criteria (standard errors, p-values, Schwarz criterion,
residual variance, and the like).” (Granato, 1992: 125)

5’ The use of four lags is reasonable because the data is annual, and the typical number of lags employed
in the Hendry methodology is five for quarterly data (which contains a much greater number of time
points). See Pagan (1990), page 99.

89

ensure that each right hand side variable is predetermined. The two equations are jointly
estimated with a full information maximum likelihood (FIML) estimator.56 The parameter
estimates for the general model are presented in Table 8.

Table 8 presents estimates for both equations listed above, with the proportion of
system members fighting interstate wars (Propwar) estimated in Equation 1 and the
proportion of democracies in the international system (ADoyledem) estimated in Equation 2.
The first thing to note in Equation 1 is that the dependent variable (Propwar) has a strong first
order autoregressive component, which is evident in the size of the parameter for Propwar1H
(0.67007); the remaining lags of Propwar are statistically insignificant. Doyle’s measure of
democracy (ADoyledem) is significantly and negatively related to the proportion of system
members fighting wars (Propwar), although the impact is somewhat lagged; only the fourth
lag is negative and significant (p<.05). The measure for world made, British exports
(ALnukexp), is significant only for the first lag and in the predicted direction (0089302).
Also, systemic power concentration (AConcen) has a negative impact on the proportion of
system members fighting war; the third and fourth lags are statistically significant (p<.05).

Turning to Equation 2 in Table 8, we can see that all four lags for Propwar are
statistically insignificant, although the first three are in the predicted direction (I expected war

to have a positive impact on the proportion of democracies in the system). This can be

 

’6 Estimation procedures such as ZSLS and 3SLS estimate each equation in a simultaneous equations
model separately. “A more general approach is to specify a similar equation for every endogenous
variable in the system, calculate the joint density of the vector of all of the endogenous variables
conditional on the predetermined variables, and maximize the joint likelihood function. This is known as
full—information maximum likelihood estimation, or FIML (Hamilton, 1994247).” If a model is just
identified, i.e., the number of excluded exogenous variables is no less than the number of included
endogenous variables minus one, then ZSLS, 3SLS, Instrumental Variables (IV), and FIML produce
identical estimates. The models estimated in the next section are over-identified (the number of excluded
exogenous variables is greater than the number of included endogenous variables minus one). In this

90

explained by the high level of correlation across these lags.57 The endogenous lags of
democracy have mixed signs (lags 1 and 2 are negative, lags 3 and 4 are positive), and none
of the parameters is statistically significant. Finally, the fourth lag for world production (as
measured by ALnusgnp) is significant (p<.10) and in the predicted direction (0.033735).

It is clear from Table 8 that several of the independent variables are insignificant in
the first general model. I test a series of zero-order restrictions on each (seemingly)
insignificant variable with a Wald statistic. The null hypothesis for each restriction is that a
given coefficient(s) is zero. Acceptance of the null hypothesis permits the omission of that
particular variable in the re-estimation of the model, given that no significant loss of
information occurs. Table 9 provides a summary of each reduction stage, including the
variable dropped from estimation, the parameter and t-statistic for the omitted variable, and
the overall standard errors for each equation (0,,02) that are obtained with estimation of the
model excluding the variable in question. The general to specific reduction produces the
final model (Table 10) describing the relationship between systemic war and the proportion
of democracies in the international system.

Table 10 includes information on the range of parameters from the general to the
final reduced model. For example, in Equation 1 the parameter on the fourth lag of
capability concentration (AConcen1 4) varies from -0.229 to -0.180 in the reduction
process from the full model to the reduced model, while the parameter for democracy
(ADoyledemM) varies from —0.73616 to -0.53261. Presenting the range of parameters

across the reduction process provides a sensitivity test for model specification,

 

case, FIML produces more efficient estimates than ZSLS, 3SLS, or IV estimators (Hamilton, 1994:250).
5’ The correlations were as follows: r(Propwar,,,, Propwar,,2)=0.724, r(Propwar,,,, Propwar,,,)=0.54l,

91

something similar in spirit to Leamer’s extreme bounds analysis (EBA).58 The final
reduced model in Table 10 is extremely robust across the reduction stages, considering
that no parameters in the final model switch signs from the full to the reduced model.
As Granato (1992:134-135) notes, “reduction does involve a change in the original
parameters, but for valid reduction, the parameters must. . .maintain the same sign as in
the original model.” In addition, the range of betas is not very large, indicating that the
parameters in the final model are representative of the more general theoretical model.

The endogenous relationship between the proportion of democracies in the
international system (ADoyledem) and the proportion of nations fighting interstate war
(Propwar) is clearly significant. The results confirm my expectations about the sign of the
endogenous variables. I hypothesized that increases in the proportion of nations fighting war
would increase the proportion of democracies in the system (P4) because war creates
conditions that favor the development of republican constitutions and a rule of law among
nations (an argument derived from Kant), and because democracies are more likely to win the
wars they fight (promoting and establishing democracy in the aftermath of war). The
relationship is positive and significant (p<.05) for the second lag of Propwar in Equation 2
(02:0.035324), meaning the larger the proportion of states fighting war, the greater the
increase in the change in the proportion of democracies in the system. The negative and
significant (p<.05) parameter on the proportion of democracies (ADoyledenr,_4, 58: -0.5 3261)

in Equation 1 (Table 10) supports the democratic peace proposition (P1). In other words, the

 

r(Propwar,,2, Propwar,,3)=0.726, r(Propwar,_3, Propwar,4)=0.725.

5“ Leamer’s extreme bounds analysis (EBA) involves l) formulating a general family of models, 2)
identifying prior distributions for the parameters of interest, 3) analyzing the sensitivity of inferences on
the parameters of interest to the choice of the prior distributions, and 4) obtaining a narrower range for

92

greater the increase in the proportion of democracies in the system, the lower the proportion
of states fighting interstate wars. Peaceful relations between democracies create greater peace
and further the spread of democracy and democratic norms in the international system. Thus
the empirical relationship between democracy and war at the systemic level is clearly
endogenous as I argued earlier.

Another interesting result in the model is the fairly large impact of systemic capability
concentration (AConcen,_3, AConcen1 4) on the proportion of system members fighting war
(Equation 1). The third and fourth lags of concentration have a strong negative impact on
war, which implies that positive changes in systemic capability concentration make war less
likely in the system, confirming P6. This empirical finding provides support to the power
preponderance hypothesis, or the notion that war is less likely when concentration of power
in the system changes towards greater concentration. Positive changes in concentration
(moving closer to total preponderance or AConcen=1) make war less likely because the level
of uncertainty that leaders have about the relative power rankings of nations in the system
decreases making it easier to predict the behavior of opponents. This also indicates that the
key explanatory variable for war in the neorealist framework, the distribution of power, is
important even afier controlling for the pacifying impact of democracy at the systemic level.

The two economic variables, British exports (ALnukexp) and US Gross National
Product (ALnusgnp), are also statistically significant and in the predicted direction. I
expected the higher the level of world trade (British exports), the less likely war in the
international system (P7) because trade increases dependence across states, making war a

more costly means for states to gain access to valuable resources. This is confirmed by the

 

inferences (Pagan, 1990:104-105). See also Learner (1983) and Leamer and Leonard (1983).

93
negative sign on ALnukepo in Equation 1 (Table 10), although the parameter is marginally

significant (p < .10). In addition, world production is positively related to the growth in
global democracy (P8), supporting Huntington’s (1991) argument that global economic
growth promotes the adoption of democratic forms of government in the world. This
relationship is confirmed by the positive and significant (p < .10) sign on ALnusgan in
Equation 2 (Table 10). In summary, the structural equations model tested with Doyle’s
measure of democracy confirms each of the theoretical propositions derived in Chapter 2. I
turn now to an alternative measure of democracy, the Polity III measure, to determine if the

findings in the first model are robust.

Model Two: Polity HI Measure of Democracy
Model two is similar to model one in that I utilize the transformed stationary series in
both, although I use the Polity III Measure of Democracy (discussed in Chapter 3) instead of
Doyle’s democracy indicator. I begin estimation with the following general model:
Propwart = 50 + 8,Propwar,_l + 82Propwar,,2 + 63Propwar,,3 + 8.,PropwarM +
8,APolitydem,,l + 86APolityderrr,_2 + {‘37APolitydem,_3 + SgAPolitydernM +
8,ALnukexp,_, + 6,0ALnukexp,_2 + 6, ,ALnukexp,,3 + lileLnukexp,_4 +
SBAConcenM + 5.4AConcen,_2 + 8,5AConcen,,3 + 6,6AConcenM + 3,,
APolitydemt = [30 + [.’)1Propwar,_l + B,Propwar,,2 + [i3Propwar,_3 + B4Propwar,_4 +
[SSAPolityderrr,_l + [3,,APolitydem,2 + [37APolitydern,_3 + [3,,APolitydernM +
BgALnusgan + BroALnusgnPr-z + B1 rALnuSgnPt-s + BrzALnusgan + 82:
Again, I include four of lags of each variable to capture the dynamics in each individual time
series. The F IML parameter estimates for the general model are presented in Table 11.

The parameter estimates for the first equation in Table 11 (Propwar) are similar to the

results in Table 8. The proportion of nations fighting interstate wars (Propwar) again exhibits

94

a strong first order autoregressive process, which is evident by the positive coefficient on
Propwar“ in Equation 1 (85068880, which is close to the value of 0.67 presented in Table
8). As predicted, the Polity measure of democracy has a negative impact on the proportion of
states fighting war (Propwar); the first, third, and fourth lags (APolitydemM, APolitydem,_3,
APolitydemM) have negative parameters, although the parameter for the first lag is the only
one that is statistically significant (p<.05). The measure for world trade, ALnukexp, exhibits
three positive lags (contrary to theoretical expectations), although the coefficient for the first
lag is statistically significant (p<.10) and in the predicted direction (-0.075). The results for
the systemic concentration measure in Equation 1, Table 11 are extremely similar to the
estimates obtained in Model 1. The first, third, and fourth lags (AConcenM, AConcen,,3,
AConcen1 4) are negative as expected, and the last two are significant (p<.05).

The results in Equation 2, Table 11 for the proportion of democracies in the system
(APolitydem) are slightly different from the results obtained in Model One (Table 8). The
impact of war on the proportion of democracies is in the predicted direction for the first lag
(B1 = 0.075) and this parameter is statistically significant (p<.01). As was the case in Model
1, none of the autoregressive lags for democracy (APolitydem) are statistically significant and
they produced mixed signs (lags 1 and 4 are positive, lags 2 and 3 are negative). Finally, the
indicators of world production (ALnusgnp) have no significant impact on the proportion of
democracies in the international system, although 3 of the 4 lags are in the predicted direction
(positive).

Many of the independent variables in Model two are insignificant. Again, I test a

series of zero-order restrictions on each (seemingly) insignificant variable with a Wald

95

statistic. The null hypothesis for each restriction is that a given coefficient(s) is zero.
Acceptance of the null hypothesis permits the omission of that particular variable in the re-
estirnation of the model, given that no significant loss of information occurs. Table 12
provides a summary of each reduction stage, including the variable dropped fi'om estimation,
the parameter and t-statistic for the omitted variable, and the overall standard errors for each
equation (01,62) that are obtained with estimation of the model excluding the variable in
question. The general to specific reduction produces the final model (Table 13). To reiterate,
Table 13 includes information on the range of parameters from the general to the final
reduced model. The final parameters in Model two are robust, given that no parameters in
the final model change signs across the reduction stages.

Model two also provides evidence of an endogenous relationship between the
proportion of democracies in the international system (APolitydem) and the proportion of
system members fighting interstate wars (Propwar). The positive and significant (p < .001)
sign on PropwarH in Equation 2 provides support for P4, i.e., as the level of war in the
international system increases, the change in the proportion of democracies in the system also
increases. Democracy (APolitydemM, APolitydemH) in turn has a negative and significant
impact on the amount of war in the system, which I predicted theoretically (PI). The greater
the increase in the proportion of democracies in the system, the lower the proportion of states
fighting interstate wars.

I discovered a significant and negative lagged relationship between systemic
capability concentration and systemic war in Model one, which is confirmed in Model two as

well (AConcen,_3, AConcenM). The third and fourth lags of concentration have a strong

96

negative impact on war, which imply that positive changes in capability concentration make
war less likely in the system, supporting P6 (the power preponderance hypothesis). Systemic
trade also has a negative impact (ALnukepo) on the proportion of system members fighting
war, which confirms my theoretical expectation (P7). While world production had a positive
impact on global democracy in model one, it drops out of model two in the reduction
process.59

The two models estimated above confirm my theoretical expectations about the
relationship between war, democracy, power concentration, trade, and world production at
the systemic level. I expect, however, that the relationship between democracy and war is
evolving over time (P3, P5). In particular, I anticipate that the negative impact of democracy
on war in the international system will become stronger over time as the proportion of
democracies increases (P3). Also, the positive impact of war on democracy should decrease
over time as the proportion of democracies in the international system increases (PS). If the
relationship between democracy and war at the systemic level is an evolving relationship, the
models estimated above should show evidence of non-structural stability, an issue I address

below.

The Issue of Structural Stability
The two equation models tested in the previous section support my argument about
the endogenous relationship between democracy and conflict in the international system.

One important issue in model specification is structural stability, i.e., whether the parameters

 

’9 Although, the parameter for world production in Model one (Table 10) was significant only at

p==0.0586.

97

in a model are constant across one or more subsamples (such as different time periods). I
have theoretical reasons to believe that the models estimated above may contain time-varying
parameters. For example, I expect the pacifying impact of democracy on war to increase
across the sample period. As the proportion of democracies in the world increases, the
domestic “rule of law” characteristic of democracies will diffuse throughout the international
system, moving us closer to a federation of free states. Also the rules and norms of behavior
between democracies should become more prevalent over time as the number of democracies
increases, and as democratic hegemonic states create and promote liberal institutions and
regimes. As the system moves closer to the federation of free states, the pacifying impact of
democracy on war will become stronger over time as democracies become greater in number
and as their norms of behavior spread throughout the international system as a whole.

I also argued in Chapter 2 that the relationship between the proportion of nations
fighting war and the proportion of democracies in the system is evolutionary as well. Higher
levels of warfare in the system promote the development of democratic states due to the
success of democracies in wars and the improvement in domestic and international
institutions that war brings about over time. However, the substantive impact of war on the
creation of democracy should decline over time, as war becomes less frequent in the
international system (due to the expansion of the international “rule of law”).

If the impact of the democratic peace is getting stronger over time relative to other
factors that affect the level of war in the international system (and vice versa), then estimation
techniques must permit dynamics in the parameter estimates. The F IML estimation
discussed in the previous section assumes that the parameters are constant. Because I have

theoretical reasons to believe the parameters in Models One and Two are changing over time,

98

I conducted one-step ahead Chow tests for both models to determine if the parameters are
characterized by structural change. The results of the Chow tests are presented in Figures 18
and 19. The one step ahead Chow test is conducted under the null hypothesis of constant
parameters and the test statistic is distributed as F(1, t-k-l) for t = M,. . ..T (l < M < t < T).
Any values of the test statistic that exceed the critical F value (points that lie about the 5%
cutoff line in the graphs) indicate the presence of structural change in the model.

It can be seen in Figures 18 and 19 that the parameter estimates are not stable across
the estimation period, as I hypothesized. There are several “break” points in the Chow tests
for both models, indicating the likely presence of time varying parameters. Model One is
characterized by five significant structural changes, or break points (Figure 18): 1919, 1941,
1946, 1950, and 1960. The first break point, which is also the largest, coincides with the end
of the First World War. The next two significant spikes occur towards the beginning (1941)
and at the end of World War II (1946). Given that war has a significant impact on the
proportion of democracies in the international system, it seems plausible that the greatest
wars in history have the largest impact on the international system, producing structural
changes in the relationships between democracy and war. Model one also contains two
structural breaks in the post-World War 11 period, 1950 and 1960. This first spike can be
accounted for by the large increase in the number of democratic states established after
WWII. The second spike can be explained by the large increase in the number of system
members in 1960. Many Afiican states gained independence in 1960, increasing the number
of system members from 89 in 1959 to 107 in 1960. Doyle (1986) indicates no changes in
the number of democracies between 1959 and 1960, whereas the Polity III data adds four

democracies to the system in 1960 (from 37 to 41). This could explain the lack of structural

99
change (or significant spikes) after 1946 in the second model (using the Polity III measure).

The structural breaks in Model Two (Figure 19) are similar to the break points for Model
One: 1914, 1917, 1919, 1940, and 1946. In this case, all five significant spikes are associated

with World War I or World War 11.

Conclusion

In this chapter, I presented the results of two structural equation models designed
to test several of the theoretical propositions delineated in Chapter 2. My results can be
summarized as follows:
1) The relationship between democracy and war in the international system is endogenous.
As the proportion of democracies increases, the level of systemic war decreases, and this
holds for both measures of democracy (ADoyledem, APolitydem). Also, the greater the
proportion of nations fighting war in the system, the greater the growth in global
democracy. These results support the theoretical propositions I derived from Kant’s
systemic argument of the relationship between democracy and war.
2) Systemic capability concentration has a negative impact on the proportion of system
members fighting wars. This provides support to the power preponderance theoretical
approach that predicts increasing power concentration to lead to greater systemic peace.
3) Systemic trade decreases the overall level of war in the international system, consistent
with Mansfield’s argument that trade increases dependence across states, and that war is a

more costly means for states to gain access to valuable resources.

100

4) World production increases the level of democracy in the international system, although
this result is significant in only one model (Model one) and the marginal impact is fairly
small.

The most significant contribution of this chapter is that I provide the first
empirical test of the endogenous relationship between democracy and war at the
systemic level. I have demonstrated that scholars need to consider how war affects the
development of democracy in addition to emphasizing peaceful relations between
democracies. Although the FIML empirical results support several of my theoretical
propositions, these models assume that the relationships between variables are constant
across time. I presented the results of Chow tests for both models, and these tests
confirmed my theoretical expectations that the relationship between democracy and war
at the systemic level is changing over time (or evolutionary). The models contain
several break points, most of which are associated with the largest wars in history (WWI
and WWII).

As a first step in determining how the parameter estimates vary over time, I split
the entire time period into 30, 40, and 50-year intervals starting at 1821. I estimated the
final reduced model for Model two in each sub-sample, increasing the year by one across
the series. This produces a series of “sliding windows” that permit me to see how the

parameters change over time. I analyze these sliding windows in Chapter 5.

101
Table 7 — Variable Legend

 

 

Variable Name Variable Description

Propwar Proportion of System Members Fighting
Interstate Wars

Doyledem Proportion of Democracies in the
International System Measured with
Doyle’s (1986) measure of democracy

Politydem Proportion of Democracies in the
International System Measured with the
Polity III measure of democracy
(democracy > 5)

Lnukexp Natural log of British Exports

Concen Major Power Capability Concentration

Lnusgnp Natural log of US Gross National Product

 

 

102

Equation 1 for Proportion of War (Propwar)

Variable

Propwart- 1
Propwart-2
Propwart-3
Propwart-4

ADoyledemt- 1
ADoyledemt-2
ADoyledemt-3
ADoyledemt-4
ALnukexpt- 1
ALnukexpt-2
ALnukexpt-3
ALnukexpt-4

AConcent- 1
AConcent-2
AConcent-3
AConcent-4
Constant

o1= 0.0612357

Equation 2 for Proportion of Democracies (ADoyledem)

Coefficient

0.67007
0.1 1944
-0.019987
0.021355
0.024961
0.40060
-0.31933
-0.73616
-0.089302
-0.005 8248
0.051321
0.029437
-0.0094043
-0.016361
-0. 19435
-0.21738
0.016030

S;
0.080169
0.097176
0.095617
0.081673
0.27532
0.27665
0.27573
0.27588
0.041433
0.041487
0.040019
0.03 8181
0.090797
0.10038
0.098605
0.084957

0.0070731

t-value
8.358
1.229
-0.209
0.261
0.091
1.448
-l.158
-2.668
-2.155
-0. 140
1.282
0.771
-0.104
-0.163
-1.971
-2.559
2.266

Table 8 - F IML Estimates of Model One, 1816-1982

t-probability

0.0000
0.2210
0.8347
0.7941
0.9279
0.1497
0.2487
0.0085
0.0328
0.8885
0.2017
0.4419
0.9176
0.8707
0.0506
0.01 15
0.0249

Variable Coefficient i1; t-value t-probability
Propwar“ 0.01 1 199 0.024994 0.448 0.6548
Propwart-2 0.012521 0.030195 0.415 0.6790
Propwart-3 0.018149 0.030090 0.603 0.5473
Propwart_4 -0.0081068 0.025136 -0.323 0.7475
ADoyledemH -0.045964 0.081878 -0.561 0.5754
ADoyledeth -0.016292 0.081684 -0.199 0.8422
ADoyledemt_3 0.087951 0.082096 1.071 0.2858
ADoyledemt_4 0.090592 0.081221 1.1 15 0.2665
ALnusgnpt-1 -0.0034249 0.019833 —0.173 0.8631
ALnusgnpt-2 0.021572 0.020857 1.034 0.3027
ALnusgnpt-3 -0.00052256 0.020658 -0.025 0.9799
ALnusgnpt-4 0.033735 0.019754 1.708 0.0898
Constant —0.00377 1 3 0.0022331 -1.689 0.0934

02=0.0191646

Log-likelihood = 1109.3809 T = 162

103

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

858 :8 88888 83 88:8 .8385 N.,.__§8_8%
8:8 :8 88888 83 888 .8385 N.8885
828 :8 88888 83 8888 .8385 8,5888%
«88 :8 88888 88. _- M888 H8385 2.888%
8888 8888 :8: 8888 5.838% a88.48%
888 :8 88888 88.: 88:8 “8385 8-588825%
8888 88888 :3 28:8 .aoB_8% :8828%
8888 88888 88 8888 H8385 858854
885:8 5888 88- 8888- 8.828% 88.288%
88:8 8888 88 82 :8 838% 28385
8 _ 8 :8 88888 68 t _ _ :8 5.828% 28385
888 :8 $888 888- 8888- .8828% 18385
8888 88888 88 88:8 H8885 18885
88:8 88888 8.8- 8888- H8385 a8385
88:8 8888 88- 8888- 5.828% 758.88%
8888 2 888 :88- 88:8- é828% N5.828%
55:8 1888 88- 82.88- H8385 8-58854
888:8 8888 R _ 8- 8.: :8- .8385 8-588%
888:8 8888 88- 8888- 8385 288%
888:8 5868 :88 88888 .8385 7.838%
88858 8888 88- 88888- 8.828% M888%
888:8 888:8 -.. ... ..- 882
vote—=0.— 602.50.-
AN .58 No C .35 6 38:53 @9689. a am: .8533— 933:5

 

 

 

 

 

 

:28:qu 0:0 382 .«o ng - o 033-

 

104

3386 - gamed
Nmmmod - mmde
mfiom mo vac—«Im-

 

8883 - 8888-
:8823 - 8888-
888.3 - 888.
288.3 - £888.
838 - £88
mSom mo own—MM

$3.33

3826
owned
386

 

8:888;

good
good
336
owned
@086
ooood

 

8:883

83.
mag
888
884-4

NEH-.-

8386
News :3
mmmgod

dum-

388: n 8888883

wmwNood-
omammod
vmmmmod

2865000

32% 36 H Nb

88880
19885-3

38335
Baum-58>

EBB—3&3 memos-80:85 mo 8anon 8m N cosmscm

va.m

Ogd-
m 8&-
SK. T
who.”
wmw.m_
028?“

2386
covoood
wwownod
m 530.0
8836
mmcmmod

fl

3386
©3328
332.c-
$38.?
2336.
©8986

EBB-meow

Swooood H 5

38880
flavoaoo<

2:00.804
75:83:44
3:83an
2.83885
3

c.8385 83 mo 8:535 8m _ :osmscm

.25 882 8885 8:5 2:8 88885 355 - S 038

105

Equation 1 for Proportion of War (Propwar)

Variable Coefficient §§= t-value t-probabilitx
Propwart_1 0.68880 0.079516 8.662 0.0000
Propwart_2 0.1471 1 0.10025 1.467 0.1443
Propwart-3 -0.051 126 0.097290 -0.526 0.6000
Propwart-4 0.028918 0.081880 0.353 0.7244
APolitydemH -0.60418 0.23409 -2.581 0.0108
APolitydemt-2 0.30144 0.23577 1.279 0.2030
APolitydemt_3 -0.074835 0.24405 -0.307 0.7595
APolitydemt_4 -0.39245 0.23783 -1.650 0.1010
ALnukepo -0.075009 0.0401 12 - l .870 0.0634
ALnukexpt-2 0.040844 0.041282 0.989 0.3240
ALnukexpt_3 0.021568 0.038696 0.557 0.5781
ALnukexpt-4 0.010873 0.036061 0.302 0.7634
AConcent_1 -0.053751 0.087556 -0.6 14 0.5402
AConcent-2 0.013600 0.096572 0.141 0.8882
AConcent_3 -0.18565 0.096141 -1.93 1 0.0553
AConcent-4 -0.20427 0.084707 -2.41 1 0.0171
Constant 0.0141 13 0.0068431 2.062 0.0409
61 = 0.0602191
Equation 2 for Proportion of Democracies (APolitydem)
Variable Coefficient fl; t-value t-probabilig
Propwart-1 0.075435 0.027704 2.723 0.0072
Propwart_2 -0.028874 0.034184 -0.845 0.3996
Propwart-3 -0.01 1640 0.034038 -0.342 0.7328
Propwart_4 0.032556 0.028275 1.151 0.2514
APolitydemt- 1 0.024008 0.080646 0.298 0.7663
APolitydemt-2 -0.040843 0.081785 -0.499 0.6182
APolitydemt-3 -0.1 1958 0.082538 -1.449 0.1494
APolitydemt_4 0.053362 0.080941 0.659 0.5107
ALnusgan 0.021966 0.022106 0.994 0.3220
ALnusgnpt_2 0.00080819 0.023134 0.035 0.9722
ALnusgnpt-3 0.0010501 0.022760 0.046 0.9633
ALnusgnpt-4 -0.0010263 0.021607 -0.048 0.9622
Constant -0.0039329 0.0024528 -1.603 0.1 109
02 = 0.0211797
Log-likelihood = 1135.4457 T = 168

Table 11 - FIML Estimates of Model Two, 1816-1988

106

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

88888 2:888 885 8888 .8385 N88382
88888 88888 82- 8888- 883882 258382
88888 88888 888 88:88 3:83:82 1.3385
88888 888888 82. 8888- 88382 N.8385
88888 888888 83 8888 88382 2888.2
88888 882888 888 88888 8838.2 188382
88888 82888 888- 88888- 8385 2:882
88 .888 888888 888 8888 8385 8-58582
82888 882888 888- 8888- 88382 N88382
88888 8.8888 888- 28888. .8385 M.8385
83888 88888 888- 8288- 88382 m.8385
3888 88888 888 8888 8385 18385
5888 888888 888- 8888- 8385 288382
28888 88888 888 83.88 8838.2 288382
8.8888 8888 888 8888 8385 58852
88288 $8888 838 88288 .8385 28882
88288 88888 888 888888 8883:82 28882
8_ 5 88 888888 888- 88 5 888- 88382 18882
8_ _ 88 888888 888 8 _ 88888 w88382 N8888.2
8: 88 88888 ... ... ... 882
6025—5.—
3 .88 8 a ...a 6 83:8. 88853.-. 88.8.. 8 8 8 882.5 8.8:;

 

 

 

 

 

 

cote—60¢ 03,—. 3602 no 588% .. m5 oEaH

 

107

mnowod - Eoood
880m .8 owcmm

2228-8 - 8828-
8828-8 - 8888-

8888 - 8888
58888-8 - 8888-
3888.8 - 8888-
$888-8 - 8888-

858 - 8888

2.936 .. waved
£30m mo owamm

 

3:6
wooed

 

3838883

mde
mvood
3.86
35.0
omvod
$2.0
2 5d
wowed
ooood

 

38888.3

82-
888
812%

88 u 5. 88.8: n 888388..

88888
85%
.58

mgmcod-
mammood

820E850

vonwomod H Nb

E9200
23>»qu
03th>

AEoubzomg momofioofioa mo 523.5qu 8m N 20:83.”.—

NOON
QWN-
and-
mmmA
Sued-
vmc. T
wmmN-
omwé
39.x
028?“

©2086
8586
chomwod
Nmmwmod
91am mod
omommmd
omenmmd
3%de
vmavnod

4w.

momm 5d
owwni .o-
253 _ .o-
wwmnvod
:menod-
ooonofio-
onnowmd-
omfimid
3386

3205000

22336 .I. 6

88280
Ecoocoo<

Ecoocoo<
29883544
29883544

188382
288382
$5335
23335
a

CmBaoEV .83 no 525305 5o.“ _ :osmscm

$2-33 63,—. 352 @33qu 3:5 05 mo mofififimm A25 - 2 2an

108

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

w 1 . ill-.11! . - -. . 1-5.-- .
m. :8 0502.32-11-9 l . {Mm-1|
. 8.8.
. 3.:
Pm -
Eco
N 1
1 JWG
. _
fm
H l
.m I
li/\/>I.E_ __._ _ _t__+_.__._3§(/\/r<kL
00 we we so 3o :o 50 Go :0 go

 

Emca :w .. Ono maw >385 0:02 H88 mg 23%— 0:0

30

 

3o _

109

 

3L0

 

 

 

 

 

 

 

 

 

   
 

 

 

 

  

 

 

A m a :6 nmmsw.--I| we;
_ - iii: .. _-!..-,$
.3“- ,2)
m
m,-
Now
N-I _,
G,- _
H .31 3&5
<- _ >
fl
a»
h
TLg»»_ _ __.____ .___»_p% __
mo 3 mo 8 :5 :o so so :5 Go .8 So
- -IL

 

 

 

Emcna E - Ono man >393 Q52 Hama m9 Zona— Héo

CHAPTER 5

TESTING EVOLUTIONARY RELATIONSHIPS: SLIDING WINDOW ANALYSIS

In Chapter 4, I analyzed two structural equations models to test several propositions
derived in Chapter 2. I demonstrated that while the empirical results support my theoretical
propositions, the models show evidence of structural change. Such structural change was
expected given the evolutionary relationships discussed in Chapter 2. In particular, I derived
two propositions that predict both an endogenous and an evolutionary relationship between
democracy and war at the systemic level:

P3: The degree of the negative impact of democracy on war in the
international system will become stronger over time as the proportion of
democracies increases.

P5: The degree of the positive impact of war on democracy will decrease
over time as the proportion of democracies in the international system
increases.

The domestic “rule of law” characteristic of democracies will diffuse throughout the
international system as the proportion of democracies increases, moving the world closer to a
federation of free states. Also as the proportion of democracies in the system increases, the

rules and norms of behavior that stem from democratic forms of government will become

more prevalent over time. Democratic hegemonic states facilitate the spread of democratic

llO

lll

norms through the creation of liberal institutions and regimes. As the system moves closer to
the federation of free states, the pacifying impact of democracy on war will become stronger
over time as democracies become greater in number and as their norms of behavior spread
throughout the international system as a whole. Warfare, on the other hand, promotes the
development of democratic states due to the success of democracies in wars and the
improvement in domestic and international institutions that war brings about over time.
However, the degree of the impact of war on the creation of democracy should decline over
time, as war becomes less frequent in the international system, and as the democratic
international “rule of law” expands.

To determine how the relationship between the proportion of democracies in the
system and the proportion of system members fighting interstate wars is evolving, I break the
full sample (1816-1988) into 30, 40, and 50-year overlapping samples, creating a series of
“sliding windows”. I estimate the final reduced model (Table 13) for Model Two in each
sub-sample, increasing the year by one across the series.‘50 Recall that the final reduced
model in Table 13 was the following:

Propwar, = 50 + 5,Propwar,,, + 82Propwar,,2 + 53APolityderr1,,I 54APolityden'1,_4 +
BSALnukepo + 236ALnukexp,_2 + 57AXConcen,_3 + 58AConcenM + a"

APolitydern1 = [50 + [31Propwar,_l + 82,

The 30-year sliding window samples begin with estimation of the model above with a

sample from 1821-1850.61 The start year of the sample is increased by one, and the model is

 

6° Because the empirical results of the two models are similar, I analyze sliding window estimates for
only one model (Model Two).

6' The first year of estimation is 1821 due to five degrees of freedom consumed by the lagged variables in
the reduced Model 2 (Table I3).

1 l2
re-estimated for the second sample, 1822-1851. This continues up through the last sliding

window sub-sample, 1959-1988. The 40-year windows range from 1821-1860 to 1949-1988,
and the 50-year windows range from 1821-1870 to 1939-1988. This produces a series of
parameter estimates that can be plotted across time (for each sub-sample). I select three
different window sizes (3 O, 40, and 50 years) to determine if the results obtained are robust to
the size of the sample. A 30-year window makes sense for the smallest sample, given that
statistical significance tests can reasonably be conducted with a sample size of 30 or greater.

I present the results of the sliding window analysis in this chapter and conclude the study in

Chapter 6 with a discussion of my results and avenues for future research.

Sliding Window Analysis

The sliding window parameter estimates for the impact of democracy (APolitydemH)
on war (Propwar) are presented in Figures 20, 21, and 22 (30, 40, and 50 year windows,
respectively). The years displayed in the graph represent the start year for each sub-sample.
For example, the parameter estimate for year 1821 is estimated from the sub-sample 1821-
1850 for the thirty-year windows (Figure 20), 1821-1860 for the forty-year windows (Figure
21), and 1821-1870 for the fifty-year windows (Figure 22). The first data point in Figure 20
represents the parameter for APolitydemM estimated in the model from 1821-1850 (83,
Equation 1 above), which is 1.1572. I include 90% confidence intervals around the
parameter estimate plots to indicate when the parameter is statistically significant from zero.62

If zero is contained within the confidence intervals, then the parameter is not statistically

 

”2 I adopted a larger confidence interval (a=.10) due to the relatively small sample size of each
“window”.

113

significant in that sub-sample. I include a horizontal reference line of zero to help indicate
when the value of the parameter is statistically significant.

As I expected, the relationship between the proportion of democracies in the system
and war is getting stronger over time (P3), as can be seen in Figures 20-22. The level of
democracy in the international system has the largest pacifying affect in the samples
including and following World War I,63 and this negative parameter gets progressively larger
across the sliding window estimates. This shows that as the proportion of democracies
increases, as the federation of free states enlarges, and as democratic norms become more
prevalent in the international system, the pacific impact that democracy has on war gets larger
over time. Contrary to some arguments that the democratic peace is primarily a post-World
War II phenomenon (Russett, 1993), I find that the relationship is statistically significant
prior to that period, particularly after World War I. This provides ammunition in response to
realist criticisms of the democratic peace that attribute the long peace in the post-WWII era to
superpower relations between the United States and the Soviet Union (Mearsheimer, 1990).

The democracy parameter is positive in some earlier time periods for all three
window sizes, which is what Gleditsch and Hegre’s (1997) model (derived from the monadic
and dyadic level findings) predicts, i.e., that an increase in the proportion of democracies
would initially increase the frequency of war (P2). These parameters, however, are never
statistically significant, providing support for the first pr0position that as the proportion of
democracies increases, the proportion of states fighting interstate wars will decrease in the

international system. The empirical findings in Chapter 4 provide support for PI as well.

 

6’ The relationship between democracy and war first becomes negative and statistically significant in
1914-1943 for the 30-year windows, 1905-1944 for the 40-year windows, and 1894-1943 for the 50-year

114

Another interesting thing to note is that the impact of democracy on war is negative in the
earliest samples (although not statistically significant), becomes slightly positive in the latter
part of the 19th century, and then is strongly negative in the most recent sample periods.64
This trend holds across all three window sizes (30, 40, and 50), although the estimates with
thirty-year windows show the greatest short term variance.

The sliding window parameter estimates for the impact of war (Propwar,,,) on the
proportion of democracies (APolitydemJ are presented in Figures 23, 24, and 25. The effect
of war on democracy is almost always positive (supporting P4) and fairly constant across
time. I argued that the positive impact of war on democracy in the system would decline
over time (P5), because as the proportion of democracies increases, the democratic
international “rule of law” will become more prevalent, and the international system should
experience less war as a result (i.e., become closer to Kant’s perpetual peace). Thus the
strength of the positive influence of war on the creation of more democratic forms of
government should decline over time as the incidence of war becomes less frequent and as
the federation of free states enlarges. Contrary to my expectations, however, the impact of
war on democracy is fairly constant across the sliding window samples, declining only
slightly in the most recent periods. The impact of war on democracy at the systemic level
might decline once we account for the most recent wave of democratization following the
collapse of the Soviet Union; the parameters presented in Figures 23-25 are estimated only

with data through 1988.

 

windows.

6‘ In the last few samples, the size of the democracy parameter becomes smaller and is not significant.
This could be due to the end point of my sample being 1988, which does not include the most recent
wave of democratization. In future research, I plan to update my data set to at least 1994 to obtain a

115

The proportion of system members fighting war in the system (PropwarH) increases
the proportion of democracies most profoundly in the years close to or following World War
I and World War II."5 This makes sense theoretically; the largest wars in history have the
biggest positive impact on the change in the proportion of democracies, acting as “shocks” to
the overall increasing trend in the proportion of democracies in the system. This is confirmed
by an examination of Figure 4 (Chapter 3), where it can be seen that the largest increases in
both measures of the proportion of democracies occur after World War I and World War II.
The general trends in the systemic war parameters are very similar across all three window
sizes, with the 50-year windows again showing the least amount of variance in the movement
of parameters (30, 40, and 50).

Figures 26-31 contain the sliding window parameter estimates for the impact of
capability concentration (AConcen,_3, AConcenH) on war (Propwar) for 30, 40, and 50-year
sample sizes. The parameter for concentration is negatively related to war in every period
except from the 1860’s to the early 1920’s. The parameter for power concentration (lag four)
is strongly statistically significant in both the earliest and most recent periods, although the
third lag of concentration is not statistically significant in earlier periods. And while
concentration is positively related to war in a few time periods, it never reaches standard
statistical significance levels.

The power preponderance hypothesis is supported (P6) given that the parameter for

concentration is negatively and significantly related to the proportion of system members

 

more accurate picture of the recent impact of democracy on war.

6’ Note the significant spikes in all three graphs (Figures 23-25) in the samples that start in the late 1800’s
and contain World War 1. The 30 and 40-year windows (Figures 23-24) also show significant positive
spikes in the war parameter in samples containing World War II.

116
fighting interstate wars, especially in the first 50 years of the sample and in the 20‘h century as

well. This pacifying influence of concentration on war has been extremely stable in the 20th
century; the parameter has shown very little temporal fluctuation in this time period.
Contrary to Singer, Bremer, and Stuckey (1972), I do not find any evidence of a positive
relationship between systemic capability concentration and war in the 19th century. In fact the
strongest negative relationship between concentration and war in the system is observed
during the first half of the 19th century. For example, in Figures 30 and 31, the negative
parameter for power concentration takes on its largest statistically significant values in the
samples starting in 1821 through the early 18608. While the parameter remains negative and
statistically significant in the 20th century samples, its substantive impact is much smaller.

Another interesting thing to note in Figures 26-31 is that the parameter for systemic
capability concentration fluctuates substantially in samples starting in the 18605 up through
samples ending in the early 19205. In fact, the parameter becomes sharply negative and then
positive prior to World War 1, although the standard errors for the concentration variable
increase making the relationship insignificant in this time period in all 6 graphs. When one
examines the overall trends in major power capability concentration (Figure 9), it is clear that
the level of concentration does not change very much from approximately 1861-1915. This
constant level of concentration translates into a differenced score of zero (AConcen), which
helps to explain both the wild movements in the parameter and the lack of statistical
significance of the concentration parameters in this time period.

In Figures 32-35, I present the sliding window parameters estimates for each

parameter discussed above for the three window sizes. For example, Figure 32 compares the

117

sliding window estimates of the impact of democracy on war (lag one) across 30, 40, and 50
year samples. The parameters follow the same overall trend, regardless of the window size,
although the 30-year samples contain more variation in the parameters than the 50-year

samples as I noted above. Comparisons across the different sliding windows sizes increases

confidence in the overall trends in the parameters discussed in this chapter.66

Conclusion

One key feature of the theoretical model developed in Chapter 2 is not only the
endogenous relationship between democracy and war at the systemic level, but also the
evolutionary relationship between these variables. I argued that the pacifying impact of
democracy on war should get stronger over time as the federation of free states enlarges,
and as democratic norms become more prevalent in the international system. Also, while
war in the system promotes the development of democratic forms of government, I argued
that its substantive impact should decline over time as the proportion of democracies
increases and as war becomes less frequent overall. I tested these evolutionary arguments
in this chapter by splitting the 1816-1988 sample into 30, 40, and 50-year overlapping
sliding window sub-samples. I then estimated Model Two from Chapter 4 in each sub-
sample, plotting the resulting parameter estimates for each “window”. Not only is this
estimation approach unique in that it does not assume a constant relationship between

variables over time, it also produces many interesting substantive results.

 

“5 The sliding window results presented in this chapter represent snapshots of the parameters at
overlapping points in time. An alternative method for analyzing models with time varying parameters is
Kalman filter analysis. Unlike sliding window techniques, the Kalman filter updates information across
the entire time period. In Appendix B, I present some preliminary results of Kalman filter estimates for
Model 2. The Kalman filter analysis provides support for the empirical findings discussed in this chapter.

118

Sliding window estimation of the impact of the proportion of democracies in the
system on the proportion of system members fighting wars garnered support for
Proposition 3. The pacifying impact of democracy has grown stronger over time, and its
largest effect occurs after World War I. I did not find such strong support for Proposition
5. While I expected the positive impact of war on the growth of democratic forms of
government to decline over time, I found that the impact has remained fairly constant and
significant.

Finally, I presented sliding window analysis for the relationship between major
power capability concentration and the proportion of system members fighting wars. The
results in the chapter lend greater support to the power preponderance theoretical approach
that predicts increasing power concentration to lead to greater systemic peace. One
interesting finding was that the negative relationship between concentration and war is
strongest in the first 50 years of the sample (19th century), which contradicts earlier studies
that found a positive relationship between power concentration and systemic war in the
19th century. Also, this relationship is fairly constant across time, contrary to Singer,
Bremer, and Stuckey’s finding of an inter-century (19th, 20‘“) difference.

I turn now to the final concluding chapter in this study where I summarize the
major contributions and findings of my work and I offer several suggestions for future

research and extensions.

119

 

 

008838.

rem Ono

H.029. OH:

-m C2540?
5.2.8. $3.8 53.8 $3.8 $3.8 $3.8 Sombo $3.8 Guwbo 33.8

5225 Samba $398 2398 #598 $3.8 $5.8 Emobo $3.8 GmAbO

 

 

 

 

m8: M83

Emfia No - 1:5 m<o_<msm 56mg om Coupon-mow. on $3

mo J33 mmawnm $5335

120

 

 

r023 OH:

5080088.

fimm Ono

 

 

 

-w €83 0%.
$5.8 38.8 $8.8 $3.8 $3.8 $3.8 $8.8 $5.8 $8.8 33.8

58.8 53.8 58.8 $8.8 $8.8 $8.8 $5.8 $8.8 $8.8

 

m8: <03

388 B .. 1:5 m<o§=m 56mg om UaBooEov~ 0: «<3

8 <03 255m $3395

121

 

..u

 

 

 

 

 

:88 . 88.8
83.8 88.8 82.8 88.8 88.8 83.8 :88 3:8 $8.8 888

was.” <03

u

8.08 _ 80.0.8 . 8.0.8 _ 08.08 . 8.0.8 . 08.0.8 . $3.8. $8.8 _

3.830 mm - .200 m<0_<8m H880.” 0». 6080088. 08 $3

8 <03 m=Q8m 388020

5080088.

bum 080
H1023 OH:

C800. OH:

122

 

 

23 310m 080v

 

. .\ H.033 0%.

 

 

 

m8: <03
3.830 8 - 1:00 9.073% H8000” 0m $3 08 U080088.

8 <03 988m $809.8

123

.01.: , ,t-w:-| - -- ,-

 

 

 

 

 

A
_, A
a
.N- _ , \,
fill. \)../l.l \.\/.\./x; __ Hills .\ // \.l,lf \ I
\u\/ (11 .1 v». /\ x 1.1 .\/( \ #
x \ l r p
H 1 < x
n ...
..l-/\-/,/\..rl-./\\\, : .\’
\ . . .s \/ t I .. .
o.o . # _ r Il\r H, \ r u\.l / \ \Ui. I . K j ,
\l.L . A W _.\. . ..\ -
K:
-.H - 0‘
...N
588° Hmwmbo Sheba :3wa 8.3.09 ngbo Hmowbo $8.90 Howwbo Hob-Foo
HmNmbo Hmhwbo gumbo quobo Ewaroo wmowbo $580 chmbo $55.28
m8: <03

$1830 N8 - .200 m<0_<8m H880» 0». $3 08 6080088.
8 <03 258m £88080

$3 98 080v

H.083 OH:

€803 OH:

124

 

 

 

 

 

 

 

«<3 28m 080v
H. K. H1023 OH:
- fl r _ _ . _ _ . _ _ . _ . . _ _ _ - . _ l :83 OH:
xomox. AWN x8e. x80 86. \o.,..v. \un 88 8&0 x0.»- \%%x. 8%» saw. 30 800., 80 x00 Amok swab 8......

m8: <03

3830 Nm .. .200 m<0_<8m H880” 0». $3 08 U080088.

m o <03 msamsm $88020

125

 

 

 

 

 

w -
_ 3 .8 7 v
N .. _ fl \. ’ ~ /.
_ z x. , . ,
_ . , . 2.. . ,
O H\ IJKuI/NI! >> /\ I/Iv\. l» . \:/.__ .\ l
7 . O08008830:
fiwm .3300
s /
. ~ .\ H.023 OH:
-w 1 . I I .-
,, \
L $83 OH:

 

 

:88 _ 881.8. 38.8. 88.8 . 8.0.8. 838. 8.0.8 _ 8.0.8. 388 _ 68.8.
888 8.0.8 888 88.8 88.8 88.8 85.8 88.8 $8.8 888

m8: <03

3930 mm - .80 m<0_<8m H880” 0». O080088308 08 $3

8 <03 258m $88020

126

 

 

 

005003330:

fimm H300

 

-N - . ,\\ r0200 OH:

-u C3000?
55.8 . Hmwwbo 3398 33:5 53.8 53.8 $3.8 $3.8 $3.8 $3.8

Emmbo 53.8 53.8 53.8 Egbo $3.8 $5.8 Swabs $3.8

 

 

 

mam: <08.

Emma mu - ‘30 m<o§=m 5602 S: 098055—30: on 23

go <03 mzawsm $5320

127

 

 

 

 

 

 

N- .
a;
\ I; f/
T x ,
. ,.
8‘
. / )|/)\.
o/ NC?) 0 107‘ < llllllll 1!!
E lllll \/ II/ {\IIIIIIJ/III’.
:l|.\- .....
-7. V \.-/\ \Z.
I I(lal‘lfilgl 1x .
/.>. u
N s L

 

 

58.8 _ Ego q Ego _ 23.0.8 _ 53% 53.8 _ 5&8 _ 33.8 . 33.8. 30.98.
53.8 53.8 52.8 53.8 598 33.8 58.8 3:8 600.8 5098

m5: <03

Emca um - H30 m<o_<5m :3qu 0». 080053030: 0: $3

mo <08. 255m $3820

0280:5555

rmm H300

H6200 0%.

C300 OH:

128

 

 

0800:0330:

 

 

 

 

hmm moca
..m . .r\ F0200 0H:
-5 _ . . . . . . _ _ . _ . _ d _ . q _ . C0000 0H:
\nmux. Am» 0,009 A000 x000 @000, «.000 \%»0 A?» $00 xoobx x0100 x000, A00 A00 A000» 009 x040 90v \090

Emca we .. 0:0 m<0_<w:m 3602 0», 00:00::830: 0: $50

We <00~ 935m 3.50020

129

 

 

 

00:00:00,030:

5:0 008.

 

-0 - H0200 0.0.

-m C0090?
38.8 Swmbo 330.00 230.8 5.3.00 50:5 Scuba $30.8 3000.8 33.8

Emmbo :KNbo 5398 5.3.8 333.8 Ecmbo 305.8 300.8 33.8

 

 

 

m6: <08.

0550 we - 0:0 m<0_<3:m 5:000: 03. 00:00::830: 0: $3

0o <08. 03330 $53020

130

 

 

 

 

00:00::030:

bum 00:0

H.028. 0H:

-m C0008. 09..
58.8.5598 53.8 53.8 58.8 553.8 505.8 58.8 $3.8 38.8

53.8 550.8 53.8 535.8 5.3.8 550.8 500.8 82.8 805.8 88.8

 

 

 

 

m8." <03

05:00 .9: - .030 m<0_<3:m 38000: 03. 00:00:5030: 0: $8.

8 <03 2338 23:30:20

131

 

 

 

 

T
\/\._
o \ \\.LP.\,9- \ 0080000000 90m 0:3
1 |\ . /\ I. .I\ . I. . .0 .
\ ../(>:/..>>r 3.. 0. f. . wo<000€530e<m
. > I/ kw?! I-I
‘ Vs..//\~\J‘Xa\ \I.\ I/ v
-3 - , r A. r . 0030003~ 90m 0:3
V. /\ 0 1
(ML. 8 <00~ «5:00:00
.0. III
003000000N 90m 0:3
.0 mo <08. $53020
N N N N N N N N N N N N N N N N N N N N
oo oo 0 .0 oo .0 .0 .0 oo .0 oo .0 0 0 0 0 0 0 .0 ,0
00,000.0009000v0000000000000 .0. 00......
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
m8.” <000

0300000 50 - 00808300: 0». 8. 8. 0:3 8 <00H m:&:m $83020

.0300 3:000» 03. 0080000000 0: $8.

132

 

 

 

 

:
a : ..
.. _ _ \ ll
_ \ J \(6 ' _ _ ,.
\lfiwi/r ...zl).\qn\\ \ ; .._\.l/_. ..J I J €8.9mm0:3
v.l\ . \JI/u\ .. . .
_ “\f \— _ ‘ f s . \I I \\ \z . ”we <03 gmaaogm
\ 3 :1. (x /« 94/3 , ll:
:1 \
$8. Pam 0:3
ob-
Ac <08. (5:935
$8. 9% 0:3
L we <03 cfiaoim
N N N N N N N N N N N N N N N N
a. Q. Q. Q. Q. oo % % Q. % Q. m. b b ..o b
abNQ$E9fin®9fiflwMQV$bfi

m8; <08

Emca mm - 02:33:: om mo. 3. m5: we <mma m:&:m Eggim

1:5 56:3 om $3 0: 0080088.

133

_.m

 

 

 

 

 

 

 

0800:2030? rum u

we <0»: 350020

00:83:50? fimm .0.

k5 <08. $50020

0030:2000? fimm w

mo <08. £50020

0.0.

.m-

ob,

-.m-

Lb,

L.m......wl_._.._4.
......m0.m0.m.m.m...mfi0M...M...M...M..\.M....M...M.¢M...M..%... .. ......

QQQQQQQQQQQQQQBQQB

m8: <03

0550 00 - 00:60:00: 00. mo. 00. 0:0 mo <03 m:&:m £50020

1:0 H8000” ow 00:00::030: 0: <5:

134

 

 

 

 

 

— .a: a q q _ q — q q q — a q u — n a «

m8: <0“:

Emma mm - 00300100: 0», mo. 00. 0:0 mo <03 m:&:m 290020

.90 0:009 0». 00:00:9020: 0: 20:.

Ill

0800:0030? fi0m 0

mo <03 «5:00.20

0300:0800? b0m 0

00 <03 $50020

0000020030? r0m 0

mo <03 €300.20

CHAPTER 6

CONCLUSION

One of the most striking changes in the international system in the past two
centuries has been incredible growth in the number of democratic states. Today
democracies constitute a majority of states in the international system, and the absence of
war between democracies (the democratic peace) paints an optimistic view of future
peaceful international relations as the trend towards greater democratization in the world
continues. Many scholars and policy makers have posited that the very nature of
international relations could be transformed as a result of the growth in democracy and
peace between democratic states. Russett (1993:137-138) expresses this sentiment well,
“if enough states become stably democratic. . .then there emerges a chance to reconstruct
the norms and rules of the international order to reflect those of democracies in a majority
of interactions.” The notion that domestic characteristics of states have the potential to
alter the rules and norms of interaction in the system is an idea that challenges the leading
systemic theoretical approach in international relations, neorealism. Neorealists such as
Waltz argue that states are like units, being motivated by the pursuit of power and security
concerns. The differentiation of states in the system (such as regime type), according to

this approach, has no impact on outcomes in world politics, most notably war.

135

136

In this dissertation, I compare these two contrasting views of international politics
by developing a systemic theory of the relationship between democracy and war.
Contrary to neorealist theories of international politics, my theoretical approach is
evolutionary, derived from the writings of Immanuel Kant. In Chapter 1, I assert that
scholars often assume that peace between democracies will translate into greater systemic
peace as the proportion of democratic states increases (Small and Singer, 1976; Starr,
1992; Singer and Wildavsky, 1993). Others have treated the systemic level relationship
between democracy and war as a puzzle that emerges from two paradoxical findings in the
literature (e.g., Maoz and Abdolali, 1989; Bueno de Mesquita and Lalman, 1992; Gleditsch
and Hegre, 1997): 1) democracies do not fight wars against each other (the dyadic finding),
but 2) democracies are just as war-prone overall as nondemocracies (the monadic finding).
I criticized the tendency to aggregate findings across levels in the democratic peace
literature, advocating instead the development of a systemic theory of the relationship
between democracy and war.67

In Chapter 2, I develop a systemic theory of war and democracy, addressing four
primary questions: 1) What is the systemic level relationship between democracy and war,
and how is this relationship evolving over time? 2) Are the traditional rules and norms of the
international system described by neorealists being transformed as the international system
becomes more democratic? 3) How do economic relations between states in the system

influence the systemic level relationship between democracy and war over time? and 4) How

 

(’7 One of the biggest problems with aggregation is that the assumptions made depend on the validity of
empirical results at each level. The monadic level finding that democracies fight in as many wars as
nondemocracies has been challenged recently (e.g., Benoit, 1996), casting doubt on the validity of
systemic conclusions drawn from such assumptions.

137

is the level of war in the international system related to the concentration of power between
nation-states, and how is this relationship changing over time? In this concluding chapter, I
discuss each of these questions focusing on my theoretical approach for answering each

question, the results obtained in empirical analysis, and extensions for fiJture research.

The Systemic Level Relationship between Democracy and War

The systemic theoretical model of democracy and war developed in Chapter 2 is
based largely on Kant’s philosophy of perpetual peace. My theory has two key features: 1)
I argue that the systemic level relationship between democracy and war is endogenous and
2) I contend that this systemic level relationship between democracy and war is an
evolutionary process.

First, I argue that the proportion of democracies and the proportion of nations
fighting interstate wars are endogenously related, i.e., both variables cause changes in each
other across time. Like the vast majority of scholars in the democratic peace literature, I
argue that as the proportion of democracies in the system increases, the proportion of states
fighting interstate wars will decrease because the “rule of law” that democracies promote
moves the system closer to a federation of free states and decreases the level of uncertainty
that arises in a state of war (or anarchy). The additional insight that Kant offers is that
democracy promotes peace, but war itself serves as the greatest force for the creation of
democratic or republican forms of government. War creates and destroys nations, and the
experience of war improves the internal institutions of states over time. War is also

positively related to democracy at the systemic level because democracies tend to win the

138

wars they fight, making them better able to promote democracy and democratic
institutions/regimes in the aftermath of war.

In Chapter 4, I test these theoretical arguments by estimating two simultaneous
equations time series models from 1816-1988; the proportion of democracies in the system
and the proportion of states fighting interstate wars are the two endogenous variables in the
models. The results confirm my theoretical propositions (Pl, P3). I find that democracy
has a strong negative impact on the proportion of nations fighting war and that war in turn
has a positive and significant effect on the proportion of democracies in the international
system. The conclusion that war is positively related to the development of democracy is a
novel empirical finding in the international relations literature, and it provides support to
Kant’s theoretical arguments about war serving as an impetus for democracy. While other
scholars have acknowledged that democracy and war are endogenously related (Gates, et
al, 1996; Thompson, 1996; Maoz, 1997), they contend that the relationship between war
and democracy is negative (the reverse of what I argue). In other words, the greater
number of wars that states participate in, the less likely they are to develop democratic
institutions because war tends to concentrate power in the hands of the state. Also,
personal freedoms tend to be restricted during warfare. One important difference between
my study and the others cited above is that I examine the relationship at the systemic level
while Thompson (1996) and Maoz (1997) examine the relationship at the dyadic and
regional levels. One important extension of the analysis presented in this study is to test
for endogeneity between war and democracy at other levels of analysis, most notably the

monadic level. Analyses conducted below the systemic level of analysis have the

139

advantage of taking into account regional or geographic effects of war and economic
development on the development of democracy.

It might also be important to distinguish between democracy and democratization
in future research. Both measures of democracy employed in this study (Doyle’s measure,
Polity III measure) code a nation as democratic when it has reached a fairly advanced stage
of the democratization process. It would be interesting to empirically examine the
relationship between war and the process of democratization. This could lend insight into
some of the recent debates on whether newly democratizing states are more war-prone than
autocratic states (Mansfield and Synder, 1995; Enterline, 1996; Thompson and Tucker,
1997)

Not only does Kant view war as an impetus for the creation of republics, he also
contends that war is the greatest force creating structural change. Neorealists, on the other
hand, view changes in the distribution of capabilities among nations as the biggest force for
structural change. According to Kant, the state of war produces changes in the units that
comprise the system, and the internal makeup of states in turn changes the system as
nations move closer to a federation of free states. I found strong empirical support for
Kant’s argument that war is the biggest force for structural change. In Chapter 4, I
conducted Chow tests for structural stability, and these tests confirm that the largest
structural breaks in the last two centuries occur after World War I and World War II. For
example, the sliding windows analyses in Chapter 5 demonstrate that the negative impact
of democracy on war at the systemic level becomes significant after World War I. I also
find that the positive impact of war on the proportion of democracies in the system is

largest during the two world wars.

140
One of the most insightful aspects of the systemic theory developed in Chapter 2 is

that I not only expect the relationship between democracy and war to be endogenous, but I
also expect the relationship to be characterized by an evolutionary process. Very few
scholars in the international relations community consider the possibility for evolutionary
relationships in the system, and none have been tested empirically at the systemic level
with appropriate statistical techniques that can account for time-varying parameters.68 I test
two evolutionary propositions (P3, P5). The first proposition states that the degree of the
impact of democracy on war at the systemic level will become stronger over time as the
proportion of democracies increases, as the system moves closer to a federation of free
states, and as the norms that are characteristic of democracies become more widespread in
the system at large. I also hypothesize that the positive impact of war on democracy
should decline over time as the international “rule of law” extends making war less
frequent overall in the system. Chow tests for structural stability conducted at the end of
Chapter 4 confirm the presence of structural breaks in the models estimated for the whole
time period. The Chow tests imply what is expected theoretically, i.e., that the relationship
between the proportion of democracies and the proportion of system members fighting
interstate wars varies over time.

The evolutionary relationship between democracy and war presented in Chapter 2
provides a nice extension of Waltz’s ideas on competition and socialization. According to
Waltz, socialization creates norms of behavior that constrain states’ decisions, whereas

competition encourages nations to adopt successful strategies in world politics. For Kant,

 

6“ One exception is Paul Hensel’s work on the evolution of interstate rivalry, although his analysis is
conducted at a different level of analysis (dyadic) than the present study.

141

the state of war or anarchy is at one end of a continuum and the rule of law is at the other.
International relations may be viewed historically as making erratic progress along this
continuum. The processes of competition and socialization produce convergence in state
behavior in the neorealist framework. Kant adds the notion that this process is
evolutionary, promoting convergence towards the international rule of law.

To test the evolutionary propositions discussed above, I adopt a novel empirical
estimation procedure, sliding window analysis. I split the entire sample (1816-1988) into
several overlapping sub-sambles or “windows”. I estimate one of the empirical models
from Chapter 4 in each window for sample sizes of 30, 40, and 50-year windows. I then
plot each parameter over time to determine how the relationships between variables in the
model are changing. I found empirical support for P3, namely that the pacific impact of
democracy on war at the systemic level has increased over time. Also this relationship
becomes statistically significant afier World War I, which supports Kant’s contention that
wars produce the largest structural changes in the system. While I anticipated that the
substantive impact of war on democracy would decline over time, I discovered that the
relationship has remained significantly positive and fairly constant from 1816-1988. One
limitation of the results presented in Chapters 4 and 5 is that the data ends in 1988,
excluding the most recent wave of democratization. Future research will extend the time
period to at least 1994 to determine if the degree of the impact of war on democracy has
remained constant or has indeed declined in the last ten years.

The significant changes in the parameters estimated in Chapter 5 point more
broadly to a shortcoming in many empirical papers in the international relations literature.

It is fairly common practice to report temporal differences between the 19th and 20th

142

centuries, especially among the Correlates of War research community.69 This temporal
break point is extremely arbitrary, having no real basis in international relations theory.
The time series techniques utilized in this study help to identify these structural breaks
more systematically through the use of Chow tests. I find, for example, that both World
War I and World War 11 represent important structural break points in the systemic model
tested in Chapter 4. In addition to identify these break points empirically, the sliding
window estimation procedure in Chapter 5 and the Kalman filter estimation in Appendix B
can be used to model relationships that are evolving or changing over time. To assume that
a single model captures the whole Correlates of War time period (1816-1992) is not

plausible for most theories of international politics.

The Evolution of Rules and Norms in the International System

In Chapter 2, I contend that the primary consequence of the endogenous,
evolutionary relationship between democracy and war at the systemic level is that the
norms and rules characteristic of democratic institutions will become more prevalent in the
international system as a whole. I expect democratic norms to become more prevalent in
the international system over time for two primary reasons. First, the greater strength and
success of democracies over time encourages other states to liberalize (competition), while
the rule of law extemalized by democracies in their relations with each other (socialization)
creates a more democratic international rule of law. Thus the greater the proportion of

democracies in the international system, the more likely the traditional “rules” of the

 

(’9 For example, Small and Singer (1982) make numerous inter-century comparisons throughout Resort to
Arms.

143

system described by realists will become replaced by more liberal and cooperative rules of
interaction like we see between democratic nations. Second, the hegemon as a liberal,
democratic state serves an important role in promoting democratic norms by creating and
maintaining liberal regimes and institutions, such as the World Trade Organization. The
hegemon also furthers the transmission of democratic norms to the system as a whole by
engaging in high levels of trade and other economic interactions with the vast majority of
nation-states. Cooperation in economic and cultural spheres strengthens the peaceful and
cooperative systemic norms of behavior that stem from the domestic and international
“rule of law”.

I empirically test the endogenous and evolutionary relationship between the
proportion of democracies and the proportion of system members fighting wars, although I
do not directly measure the extent to which democratic norms are prevalent in the
international system. Rather, I make an assumption that the larger the proportion of
democratic states, the more extensive the international “rule of law” and thus the more
widespread democratic norms and rules. One interesting extension of this research would
be to examine the historical evolution of systemic norms and rules more systematically.
The first step in that process is to determine what types of norms and rules democracies
transmit from their domestic rule of law to the international rule of law. Several examples
of democratic norms were discussed in Chapter 2, such as respect for political opposition
and respect for human rights. Many scholars in the democratic peace literature are testing
the normative or cultural theoretical approach by examining the types of norms
democracies utilize in their interactions with other democracies. Dixon (1993, 1994) and

Raymond (1994, 1996) find that democracies are more likely to use third-party arbitration

144

for dispute settlement, although the use of arbitration does not necessarily translate into
greater success for dispute resolution (Raymond, 1996). It would be interesting to
determine the extent to which such democratic norms and rules are being adopted by
nondemocratic states, or are becoming the “norm” in the international system as a whole.
One norm that democracies may transmit to the international system, for
example, is a respect for territorial boundaries. Kacowicz (1995:265) finds support for
this contention, asserting that: “well-established democracies do not fight each other
since they are conservative powers, usually satisfied with the territorial status quo
within and across their borders.” He argues that well-established democracies are
satisfied with the territorial status quo for two primary reasons. First, as democracy
becomes more institutionalized (established), the number of nationalist and irredentist
claims outside homeland territorial boundaries will decrease. His second argument
stems from work on power transition theory; democracies have tended to be the most
powerful states in their regions, often creating the regional status quo. Thus they are
more likely to be satisfied with the status quo, creating a situation where democracies
have few or no incentives to fight each another. Lemke and Reed (1996) also attribute
peace between democracies to their satisfaction with the international status quo.70 An
examination of the temporal trends in territorial claims and territorial disputes would
help shed light on the question of whether this democratic norm is becoming a systemic

norm as well.

 

7° Lemke and Reed (19962145) define the status quo in the international system as “the rules, norms, and
accepted procedures that govern international relations.”

145
I argued in Chapter 2 that Kant has a different view on anarchy than Waltz. For

Waltz, anarchy is simply the lack of central authority in the international system, whereas
Kant views anarchy as the state of war, a state that impedes the categorical imperative to
act morally. At the other extreme for Kant is the rule of law embodied in the perpetual
peace, a law that is based on legitimacy rather than coercive power. One change in the
norms of the international system that could be expected based on Kant’s argument is that
anarchy should become less important over time. Supranational actors, such as
international organizations, might become more powerful and legitimate as a result of this
process. As states extend their domestic rule of law to an international rule of law, they
have more incentives to respect the principles embodied in the federation of free states. A
good example of this argument is provided by Russett (1993), where he contends that
democracies are more likely to uphold international agreements because of public costs for
breaking commitments (stemming from the domestic rule of law).

If the role of anarchy does become less pronounced in the system, the norm of
adjudicating disputes through international organizations might become more prevalent
over time. International courts might be utilized more frequently to resolve international
disputes. For example, many militarized disputes between democracies in the post-World
War II period involve disputes over fisheries stocks and maritime resources/boundaries.
Iceland and Great Britain fought four militarized disputes in this time period over Iceland’s
claim to a 12-mile maritime boundary; Great Britain recognized only a 6-mile boundary.
International organizations such as the European Union and the United Nations have
played a prominent role in recent years in resolving these types of disputes between

nations. The maritime boundary dispute between Great Britain and Iceland was brought

146

before the International Court of Justice. Future historical and empirical research will help
determine the extent to which democratic norms such as these have become system-wide

norms.

Economic Relations, Democracy, and War

In Chapter 2, I also examined the relationships between two important economic
variables and democracy and war: systemic trade and world production. I tested
Mansfield’s (1994) theoretical argument that systemic trade is negatively related to war in
the international system because trade increases dependence and provides a less costly
means than war for gaining access to valuable markets and resources. Trade also serves to
promote the spread of democratic norms or the international rule of law because the
hegemonic state trades with a large number of nations. One consequence of such global
trade is that the hegemon’s domestic cultural values and norms are often transmitted to
other states. As a result, we would expect to find a negative relationship between trade and
war at the systemic level. The empirical models estimated in Chapter 4 provide some
support for this relationship, with the first lag of systemic trade (as measured by British
exports) being negatively and statistically significant in both models. I found, however,
alternating signs on the various lags of systemic trade in Model One and Model Two
indicating that the relationship between trade and war may be more complex than it is
currently modeled (i.e., not linear). One alternative specification would be to draw from
the research on long cycle theory (e.g., Goldstein, 1988), which predicts a cyclical

relationship between multiple economic variables and war. More generally, we need to

147

think explicitly about how economic interdependence, especially in the form of trade,
operates at the systemic level.

The other key economic variable analyzed in this study is world production and its
impact on global democracy. Drawing from the literature on democracy in comparative
politics, I hypothesize that the higher the level of world production, the greater the growth
in the proportion of democracies in the system (P8). I found very weak empirical support
for this proposition only in Model One (with Doyle’s measure of democracy). The
relationship between economic growth or production and democracy might be more
important at lower levels of analysis. It would be interesting to compare, for example, the
importance of war versus economic development on the process of democratization in

individual states, taking into account regional and geographic factors (Maoz, 1997).

Systemic Power Concentration and War

The final relationship examined in this study was between systemic power
concentration and war. In Chapter 2, I summarized the two competing theoretical
arguments on concentration and war at the systemic level: balance of power and power
preponderance theories. According to power preponderance theory, increasing
concentration of power among major powers in the system will make systemic war less
likely because high levels of concentration decrease the level of uncertainty associated with
the relative position of nations and changes in that distribution over time. I tested this
proposition (P6) and found strong empirical support in both models in Chapter 4. The third
and fourth lags of the change in major power capability concentration are negatively and

significantly related to the proportion of system members fighting interstate wars.

148
I analyzed the changes in this relationship in Chapter 5 through sliding window

analysis. One prominent study of concentration and war at the systemic level (Singer,
Bremer, and Stuckey, 1972) finds an inter-century (19m, 20‘“) difference in the
concentration-war relationship; higher levels of systemic power concentration are
negatively related to war in the 20th century (consistent with power preponderance theory),
yet positively related to war in the 19th century (consistent with balance of power theory).
My sliding window analysis contradicts this finding, demonstrating that the power
preponderance model is the best model for the entire time period (1816-1988), with the
strongest negative relationship between concentration and war observed in the first 50
years of the sample. What is nice about the sliding window estimates is that they can show
any structural breaks in the relationships between variables of interest without imposing
artificial cut-off points such as 1899/1900.

In conclusion, the growth in the number of democracies in the world and the
overwhelming peace between democracies offers an optimistic view of the future of
international relations. The theoretical and empirical results contained in this study
indicate that the pacific impact of democracy on war at the systemic level is getting
stronger over time, and that the international system is moving ever closer to Kant’s vision
of a perpetual peace. It is quite likely that the very nature of the international system will
be transformed in the future as more states become democratic and as the cooperative and
peaceful norms that characterize democratic interactions become more pronounced in the
international system as a whole. The extent to which liberalism prevails in the fiature of

international relations remains to be seen.

APPENDIX A

APPENDIX A

As I noted in Chapter 2, an important feature of Kant’s theoretical model is that
democracy and war are endogenously related to each other over time. This appendix
describes the technical problems associated with endogeneity if it is not modeled properly.

If endogeneity is present, the relationship posited in a single equation analysis is
biased; the explanatory variable (proportion of democracies in the system) will be correlated
with the residuals and the estimates will be biased. If parameters are biased, we have no way
of knowing in what direction. In other words, the effects of democracy on systemic war could
be greater or smaller than we think. King, Keohane, and Verba (1994:188) elaborate this
point:

If we have endogeneity bias, we are estimating the correct

inference plus a bias factor. Endogeneity is a problem because

we are generally unaware of the size or direction of the bias.

This bias factor will be large or small, negative or positive,

depending on the specific empirical example.
If two variables are endogenously related and the relationship is modeled in only one
direction, the estimator is not consistent, i.e., the parameter estimates are not near their true
values in large samples. Suppose, for example, that one endogenous variable Y, is regressed

on one explanatory variable, X.

Y=BX+8

Now suppose that X also depends on Y, i.e., X and Y are endogenously related to each other:

1300 = W

149

l 50
APPENDIX A

If we estimate the first equation with OLS assuming F0, then it can be shown that OLS is
not consistent, in other words, the probability limit of (b - [3) does not converge to zero as the

sample size goes to infinity because the covariance between X and a does not equal zero.

."‘[Z(Xia)]
p11

Z(x.)2

More formally:

 

2

 

¢O

 

pli

Z

This introduces a systematic bias in the estimates of B where E(b) = B + bias.

The most important implication of endogeneity for the democratic peace is that we
have no way of knowing how robust previous findings at the systemic level are, because they
are based on underspecified models. If in fact, the true theoretical relationship between
democracy and war at the systemic level is endogenous as I have argued, then any models
that fail to account for this relationship will be biased. In addition, endogeneity can have a
serious impact on the size of the coefiicients, which affects both the substantive and statistical

significance of the parameters.

APPENDIX B

APPENDIX B

In this Appendix, I present the preliminary empirical results obtained with Kalman
Filter estimation of Model Two. The Kalman Filter was developed in the engineering
literature (Kalman, 1960) to estimate state space models consisting of two key parts: 1) the
transition equation, which describes the evolution of a set of state variables, and 2) the
measurement equation, which describes how data is generated from the state variables. More

formally, the state space model can be written as:

Yr = Zr at + 8r
at = T (1H + m
2. ~ N(0,0'2)
m~ N002 Q.)

Olo ~ N(ao, GZPO)

where there are m independent variables A and y is the dependent variable. The second
equation is the transition equation that describes the evolution of the parameters. The
parameters (1.0 and P0 in the last equation represent the mean and variance of the initial state
vector. Once a model is placed in state space form, it can be estimated via the Kalman filter,
which is “a recursive procedure for computing the optimal estimator of the state vector at time
I, based on the information available at time t. This information consists of the observations
up to and including yt. (Harvey, 1989:104).” The Kalman filter is a powerful tool for

estimation primarily because any of the parameters described above can be allowed

151

l 52
APPENDIX B

to vary over time (including the parameters and the variance of the parameters). It is well
suited to the analysis of time varying parameter models, making it useful in the present
context.

Recall that the sliding window technique provides overlapping “snapshots” of the
parameters across time. The Kalman filter, on the other hand, updates the parameters in a
Bayesian fashion, utilizing all information from the past. I estimated the final reduced form
of Model Two with the Kalman filter, using OLS estimates of the individual equations in the
first 20 years of the sample as initial starting values. The remainder of the sample was then
estimated with the filter, and the procedure generated evolving state vectors for each
parameter in the model for every year in the sample.

The Kalman filter estimates of the impact of the proportion of democracies
(APolitydemH) on the level of war in the system (Propwar) are plotted in Figure 34. Each
point in the figure represents the parameter value for that year, which is updated across the
entire series with increasing amounts of information. The general trend in the parameter is
what I predicted theoretically (P3), i.e., the pacific impact of democracy on war is getting
stronger over time. The sign, however, is strongly positive in the early years of the sample,
and becomes negative only during World War II. This could be attributed to the very high
and positive initial starting value in the 1840’s. The overall trend in the Kalman filter

estimates, however, supports the results obtained in the sliding window estimation. The

l 53
APPENDIX B

negative impact of democracy on war in the system has grown stronger over time, as the
international “rule of law” promoted by republican governments has expanded.

Figure 35 contains the Kalman filter estimates of the impact of war (Propwar.-.) on
democracy (APolitydem). The results in this graph generally confirm the results obtained in
the sliding window estimation. The relationship is positive across the majority of the sample
years, i.e., as the proportion of system members fighting interstate wars in the international
system increases, the proportion of democracies in the system increases (supporting P4).
Also, the impact of war on democracy increases in the 19th century, drops off in the early 20th
century, and then increases again following World War 11. While I expected a declining
impact of war on democracy (PS), the results of the Kalman filter suggest that the relationship
is positive, and fairly stable over time, a result obtained in the sliding window estimation as
well.

Figures 36 and 37 present the Kalman filter estimates of the relationship between
capability concentration (AC0ncen,,,, AConcen,_4) and systemic war (Propwar). In the
sliding window estimation, 1 found a strong, negative relationship between concentration
and war in the earliest and most recent subsamples. The Kalman filter estimates show a
positive relationship between the third lag of concentration and war in the 20th century,
whereas the relationship between the fourth lag of concentration and war is negative from
1860 to the present. These results support the general conclusion drawn from the sliding

window estimation, namely that there is more support for the power preponderance

154

APPENDIX B

hypothesis, i.e., that the greater the concentration of capabilities between major powers, the

less likely war is in the international system.

155

Democracy parameter (Politydem, Lag One)

mo ll

 

-.N 1

 

-.0 - ..
:Kmbo

—

_ q q — — u _

03.0.8 50.0.8 20.0.8 50.0.8 $0.98 30.8 60.0.8

230.00 230.00 53.00 3300 $3.00 $3.00 $00.00 33.00

<03

Ech 00 - W330: 0.2.0:. 5300080 0». E0 H9000” 0». 003000000. 0: €00.

03.0.8

—

$3.00

 

156

Proportion of War Parameter (Pr0pwar)

 

i0
1

C
p—a
1

F3
0
l

-.T

-.N

 

2398

 

230.00

<08

53.8

q _ q u q i q

5.0.8 gems 50.0.8 60.0.8 $0.8 08.0.8
030.8 $3.8 $2.8 $3.8 $8.8 $0.8 63.8

3.050 3 - WEBB: 3:00 $380000 0». 90 5:000" 0». $3 0:

00800008.. H.0m 0:0

080.8

‘

Exfoo

 

157

Concentration Parameter (Lag Three)

-.N r

-.w

 

 

 

 

 

$3.8 53.8

u q u q a d d a a

03.0.8 00.0.8 63.8 $0.08 83.8 03.0.8 $00.8
500.8 $3.8 $3.8 32.8 63.8 $0.8 $0.8 $8.8 33.8

4000

Emca mm .. W380: 3:00 93:080. 0». E0 3:000" 0m

00:00:0030: 0: $00.. H.0m H300

158

Concentration Parameter (Lag Four)

 

-.f

-.m

 

 

 

 

83.8 83.8

_ q n — a H — A d

5.88 $8.8 $8.8 $8.8 $3.8 $3.8 $8.8
58.8 58.8 $8.8 $2.8 $3.8 $8.8 $8.8 $8.8 $8.8

<08

Ewan we .. WEBMS 3:9. mmmamam om $6 5839 om

00583330: 0: $8.. bum woe:

BIBLIOGRAPHY

BIBLIOGRAPHY

Ashley, Richard K. 1986. “The Poverty of Neorealism,” in Robert O. Keohane (ed.),
Neorealism and Its Critics. New York: Columbia University Press.

Babst, Dean V. 1964. “Elected Governments: A Force for Peace.” The Wisconsin
Sociologist, 329-14.

Babst, Dean V. 1972. “A Force for Peace.” Industrial Research, 14:55-58.

Barbieri, Katherine. 1996. “Economic Interdependence: A Path to Peace or a Source of
Interstate Conflict?” Journal of Peace Research, 33:29-49.

Benoit, Kenneth. 1996. “Democracies Really Are More Pacific (in General): Reexamining
Regime Type and War Involvement.” Journal of Conflict Resolution, 40:636-657.

Brecher, Michael and Hemda Ben Yehuda. 1985. “System and Crisis in International
Politics.” Review of International Studies, 11:17-36.

Bremer, Stuart A. 1992. “Dangerous Dyads.” Journal of Conflict Resolution, 36:309-341.

Bremer, Stuart A. 1993. “Democracy and Militarized Interstate Conflict, 1816-1965.”
International Interactions, 18 :23 1-249.

Bueno de Mesquita, Bruce and David Lalman. 1992. War and Reason: Domestic and
International Imperatives. New Haven: Yale University Press.

Bueno de Mesquita, Bruce and Randolph M. Siverson. 1995. “War and the Survival of
Political Leaders: A Comparative Study of Regime Types and Political Accountability.”
American Political Science Review, 89:841-853.

Bull, Hedley. 1995 [1954]. The Anarchical Society: A Study of Order in World Politics,
second edition. New York: Columbia University Press.

Buzan, Barry. 1984. “Economic Structure and International Security: The Limits of the
Liberal Case.” International Organization, 38:597-624.

Buzan, Barry and RI. Barry Jones, (eds.) 1981. Change and the Study of International
Relations: The Evaded Dimension. London: Francis Pinter.

Buzan, Barry, Charles Jones, and Richard Little. 1993. The Logic of Anarchy: Neorealism
to Structural Realism. New York: Columbia University Press.

159

160

Chan, Steve. 1984. “Mirror, Mirror on the Wall... Are the Freer Countries more Pacific?"
Journal of Conflict Resolution, 28: 617-648.

Chan, Steve. 1997. “In Search of Democratic Peace: Problems and Promise.” Mershon
International Studies Review, 41 159-91.

Davidson, Russell and James G. MacKinnon. 1993. Estimation and Inference in
Econometrics. Oxford: Oxford University Press.

Dessler, David. 1989. “What’s at Stake in the Agent-Structure Debate?” International
Organization, 43 :44 1 -473.

Deudney, Daniel and G. John Ikenberry. 1991. “The International Sources of Soviet
Change.” International Security, 16:74-1 18.

Dixon, William J. 1993. “Democracy and the Management of lntemational Conflict.”
Journal of Conflict Resolution, 37:42-68.

Dixon, William J. 1994. “Democracy and the Peaceful Settlement of lntemational
Conflict.” American Political Science Review, 88: 14-32.

Doran, Charles F. and Wes Parsons. 1980. “War and the Cycle of Relative Power.”
American Political Science Review, 74(4):947-965.

Doyle, Michael. 1983. “Kant, Liberal Legacies and Foreign Affairs (Part 1).” Philosophy
and Public Aflairs, 12:205-235.

Doyle, Michael. 1983. “Kant, Liberal Legacies and Foreign Affairs (Part 11).” Philosophy
and Public Aflairs, 12:323-53.

Doyle, Michael. 1986. “Liberalism and World Politics.” American Political Science
Review, 80:1151-1170.

East, Maurice and Philip M. Gregg. 1967. “Factors Influencing Cooperation and Conflict
in the lntemational System.” International Studies Quarterly, 11:244-269.

East, Maurice and Charles K. Hermann. 1974. “Do Nation-Types Account for Foreign
Policy Behavior?” in James N. Rosenau (ed.), Comparing Foreign Policies. New York:
Wiley.

Engle, Richard F. 1982. “Autoregressive Conditional Heteroskedasticity with Estimates of
the Variance of United Kingdom Inflation.” Econometrica, 502987-1007.

161

Enterline, Andrew J. 1996. “Driving While Democratizing (DWD).” International
Security, 20:183-196.

Farber, Henry S. and Joanne Gowa. 1995. “Politics and Peace.” International Security,
20:123-146.

Fearon, James. 1994. “Domestic Political Audiences and the Escalation of lntemational
Disputes.” American Political Science Review, 88:577-592.

Fukuyarna, Francis. 1989. “The End of History?” The National Interest, Summerz3-18.

Gaddis, John Lewis. 1986. “The Long Peace: Elements of Stability in the Postwar
lntemational System.” International Security, 10299-142.

Gasiorowski, Mark. 1986. “Economic Interdependence and lntemational Conflict: Some
Cross-National Evidence.” International Studies Quarterly, 30:23-38.

Gasiorowski, Mark and Solomon W. Polachek. 1982. “Conflict and Interdependence.”
Journal of Conflict Resolution, 26:709-729.

Gates, Scott, Torbjom Knutsen, and Jonathon Moses. 1996. “Democracy and Peace: A
More Skeptical View.” Journal of Peace Research, 33:1-10.

Gaubatz, Kurt Taylor. 1996. “Kant, Democracy, and History.” Journal of Democracy,
72136-150.

Gaubatz, Kurt Taylor. 1996. “Democratic States and Commitment in lntemational
Relations.” International Organization, 50:109-139.

Geller, Daniel S. 1992. “Power Transition and Conflict Initiation.” Conflict Management
and Peace Science, 12:1-16.

Geller, Daniel S. 1993. “Power Differentials and War in Rival Dyads.” International
Studies Quarterly, 37:173-194.

Gelpi, Christopher. 1997. “Democratic Diversions: Governmental Structure and the
Extemalization of Domestic Conflict.” Journal of Conflict Resolution, 41 :255-282.

Gilpin, Robert. 1987. The Political Economy of International Relations. Princeton:
Princeton University Press.

Gleditsch, Nils Petter. 1992. “Democracy and Peace.” Journal of Peace Research, 29: 369-
376.

I62

Gleditsch, Nils Petter. 1995. “Geography, Democracy, and Peace.” International
Interactions, 20:297-323.

Gleditsch, Nils Petter and Havard Hegre. 1997. “Peace and Democracy: Three Levels of
Analysis.” Journal of Conflict Resolution, 41:283-310.

Gochman, Charles S. and Zeev Maoz. 1984. “Militarized Interstate Disputes, 1816-1976.”
Journal of Conflict Resolution, 28:585-615.

Goertz, Gary and Paul F. Diehl. 1992. “Toward a Theory of lntemational Norms: Some
Conceptual and Measurement Issues.” Journal of Conflict Resolution, 36:634-664.

Goldstein, Joshua S. 1988. Long Cycles: Prosperity and War in the Modern Age. New
Haven: Yale University Press.

Granato, Jim. 1992. “An Agenda for Econometric Model Building,” in James Stimson, ed.,
Political Analysis. Ann Arbor: University of Michigan Press.

Gurr, Ted Robert, Keith Jaggers and Will H. Moore. 1990. “The Transformation of the
Western State: The Growth of Democracy, Autocracy, and State Power Since 1800.”
Studies in Comparative International Development, 25273-108.

Haas, Michael. 1965. “Societal Approaches to the Study of War.” Journal of Peace
Research, 22307-323.

Haas, Ernst B. 1983. “Words Can Hurt You; or, who Said what to whom About Regimes,”
in Stephen D. Krasner, (ed.), International Regimes. Ithaca: Cornell University Press.

Haggard, Stephan and Beth A. Simmons. 1987. “Theories of lntemational Regimes.”
International Organization, 41 :491-517.

Hamilton, James D. 1994. Time Series Analysis. Princeton: Princeton University Press.
Hendry, David F. 1995. Dynamic Econometrics. Oxford: Oxford University Press.

Hensel, Paul R. 1996. The Evolution of Interstate Rivalry. Ph.D. dissertation, University of
Illinois.

Hermann, Margaret G. and Charles W. Kegley, Jr. 1996. “Ballots, a Barrier against the Use
of Bullets and Bombs: Democratization and Military Intervention.” Journal of Conflict
Resolution, 40:436-460.

Hewitt, J. Joseph and Jonathan Wilkenfeld. 1996. “Democracies in lntemational Crisis.”
International Interactions, 22: 123-142.

163

Houweling, Henk and Jan Siccama. 1988. “Power Transitions as a Cause of War.” Journal
of Conflict Resolution, 31 :87-102.

Houweling, Henk and Jan Siccama. 1991. “Power Transitions and Critical Points as
Predictors of Great Power War.” Journal of Conflict Resolution, 35:642-658.

Huntington, Samuel P. 1991. The Third Wave: Democratization in the Late Twentieth
Century. Norman: University of Oklahoma Press.

Huntley, Wade L. 1996. “Kant's Third Image: Systemic Sources of the Liberal Peace.”
International Studies Quarterly, 40:45-76.

Jaggers, Keith, and Ted R. Gurr. 1995. “Tracking Democracy's Third Wave with the
Polity III Data.” Journal of Peace Research, 32:469-482.

James, Patrick. 1997. “Elaborated Structural Realism and Crises in World Politics.” Paper
presented at the Annual Meeting of the International Studies Association, Toronto.

James, Patrick, Eric Solberg, and Murray Wolfson. 1995. “An Identified Systemic Test of
the Democracy-Peace Nexus.” Paper presented at the Annual Conference of the Peace
Science Society (lntemational), Columbus.

Judge, George G., William E. Griffiths, R. Carter Hill, and Tsoung-Chao Lee. 1980. The
Theory and Practice of Econometrics. New York: Wiley.

Kacowicz, Arie M. 1995. “Explaining Zones of Peace: Democracies as Satisfied Powers?”
Journal of Peace Research, 32:265-276.

Kalman, Rudolf E. 1960. “A New Approach to Linear Filtering and Prediction Problems.”
Journal of Basic Engineering, Transactions ASME. Series D, 82:34-45.

Kant, Immanuel. 1970a. “Idea for a Universal History with a Cosmopolitan Purpose,” in
Hans Reiss (ed.), Kant Political Writings, Cambridge: Cambridge University Press.

Kant, Immanuel. 1970b [1795]. “Perpetual Peace: A Philosophical Sketch,” in Hans Reiss
(ed.), Kant Political Writings, Cambridge: Cambridge University Press.

Kant, Immanuel. 1970c. “The Metaphysics of Morals,” in Hans Reiss (ed.), Kant Political
Writings, Cambridge: Cambridge University Press.

Kegley, Charles W., Jr. and Margaret G. Hermann. 1995. “Military Intervention and the
Democratic Peace.” International Interactions, 21 :1-21.

Keohane, Robert O. 1984. After Hegemony: Cooperation and Discord in the World
Political Economy. Princeton: Princeton University Press.

164

Keohane, Robert O. 1986a. “Realism, Neorealism, and the Study of World Politics,” in
Robert O. Keohane (ed.), Neorealism and Its Critics. New York: Columbia University
Press.

Keohane, Robert O. 1986b. “Theory of World Politics: Structural Realism and Beyond,” in
Robert O. Keohane (ed.), Neorealism and Its Critics. New York: Columbia University
Press.

Keohane, Robert O. and Joseph Nye (eds.). 1972. Transnational Relations and World
Politics. Cambridge: Harvard University Press.

Keohane, Robert O. and Joseph Nye. 1977. Power and Interdependence: World Politics in
Transition. Boston: Little, Brown.

Keohane, Robert O. and Joseph Nye. 1987. “Power and Interdependence Revisited.”
International Organization, 41:725-753.

Kindleberger, Charles P. 1978. Economic Response: Comparative Studies in Trade,
Finance, and Growth. Cambridge: Harvard University Press.

King, Gary, Robert O. Keohane, and Sidney Verba. 1994. Designing Social Inquiry.
Princeton: Princeton University Press.

Krasner, Stephen D. (ed.). 1983. International Regimes. Ithaca: Cornell University Press.

Lake, David A. 1988. Power, Protection, and Free Trade: International Sources of US.
Commercial Strategy, 1887-1939. Ithaca: Cornell University Press.

Lake, David A. 1992. “Powerful Pacifists: Democratic States and War.” American
Political Science Review, 86:24-37.

Layne, Christopher. 1994. “Kant or Cant: The Myth of the Democratic Peace.”
International Security, 19:5-49.

Learner, Edward E. 1983. “Let’s Take the Con out of Econometrics.” American Economic
Review, 73 :3 1-44.

Leamer, Edward E. and Herman Leonard. 1983. “Reporting the Fragility of Regression
Estimates.” Review of Economics and Statistics, 65:306-317.

Lemke, Douglas and William Reed. 1996. “Regime Types and Status Quo Evaluations:
Power Transition Theory and the Democratic Peace.” International Interactions, 22:143-
164.

165

Levy, Jack S. 1988. “Domestic Politics and War.” Journal of Interdisciplinary History,
18:653-73.

Mansfield, Edward D. 1992. “The Concentration of Capabilities and the Onset of War.”
Journal of Conflict Resolution, 3623-24.

Mansfield, Edward D. 1994. Power, Trade, and War. Princeton: Princeton University
Press.

Mansfield, Edward D. and Jack Snyder. 1995. “Democratization and the Danger of War.”
International Security, 30:5-38.

Maoz, Zeev. 1997. Domestic Sources of Global Change. Ann Arbor: University of
Michigan Press.

Maoz, Zeev and Nasrin Abdolali. 1989. “Regime Types and lntemational Conflict, 1816-
1976.” Journal of Conflict Resolution, 33:3-35.

Maoz, Zeev and Bruce Russett. 1992. “Alliance, Contiguity, Wealth, and Political
Stability: Is the Lack of Conflict Among Democracies a Statistical Artifact?” International
Interactions, 17:245-267.

Maoz, Zeev and Bruce Russett. 1993. “Normative and Structural Causes of Democratic
Peace.” American Political Science Review, 87:624-38.

Mearsheimer, John. 1990. “Back to the Future: Instability in Europe After the Cold War.”
International Security, 1525-56.

Midlarsky, Manus I. 1995. “Environmental Influences on Democracy: Aridity, Warfare,
and a Reversal of the Causal Arrow.” Journal of Conflict Resolution, 39:224-262.

Mitchell, Brian R. 1988a. International Historical Statistics: Europe 1750-1988. ?
Mitchell, Brian R. 1988b. International Historical Statistics: The Americas. ?

Modelski, George and William R. Thompson. 1988. Seapower in Global Politics, 1494-
1993. London: Macmillan.

Morgan, T. Clifton and Sally Howard Campbell. 1991. “Domestic Structure, Decisional
Constraints and War.” Journal of Conflict Resolution, 352187-211.

Morgan, T. Clifton and Valerie L. Schwebach. 1992. “Take Two Democracies and Call Me
in the Morning.” International Interactions, 17:305-320.

Morgenthau, Hans J. 1948. Politics Among Nations. New York: Knopf.

166

O’Donnell, Guillermo, Philippe C. Schmitter, and Laurence Whitehead (eds.). 1986.
Transitions from Authoritarian Rule: Prospects for Democracy. Baltimore: The John
Hopkins University Press.

Oneal, John R., Frances H. Oneal, Zeev Maoz, and Bruce M. Russett. 1996. “The Liberal
Peace: Interdependence, Democracy, and lntemational Conflict, 1950-1985.” Journal of
Peace Research, 33 :1 1-28.

Oneal, John R. and Bruce M. Russett. 1997. “The Classical Liberals Were Right:
Democracy, Interdependence, and Conflict, 1950-1985.” International Studies Quarterly,
41:267-294.

Organski, A.F.K. and Jacek Kugler. 1980. The War Ledger. Chicago: University of
Chicago Press.

Pagan, Adrian R. 1990. “Three Econometric Methodologies: A Critical Appraisal,” in
C.W.J. Granger, ed., Modelling Economic Series. Oxford: Oxford University Press.

Polachek, Solomon W. 1980. “Conflict and Trade.” Journal of Conflict Resolution, 24:57-
78.

Pollins, Brian R. 1989a. “Does Trade Still Follow the Flag?” American Political Science
Review, 83:465-480.

Pollins, Brian R. 1989b. “Conflict, Cooperation, and Commerce: The Effects of
lntemational Political on Bilateral Trade Flows.” American Journal of Political Science,
33:737-761.

Puchala, Donald J. and Raymond F. Hopkins. 1982. “International Regimes: Lessons from
Inductive Analysis.” International Organization, 26:245-275.

Raknerud, Arvid and Havard Hegre. 1997. “The Hazard of War: Reassessing the Evidence
of the Democratic Peace.” Journal of Peace Research, 34:forthcoming.

Ray, James Lee. 1993. “Wars Between Democracies: Rare or Nonexistent?” International
Interactions, 3 :25 1-276.

Ray, James Lee. 1995. Democracy and International Conflict: An Evaluation of the
Democratic Peace Proposition. Columbia: University of South Carolina Press.

Ray, James Lee and J. David Singer. 1972. “Measuring the Concentration of Power in the
lntemational System.” Sociological Methods and Research, 1:403-43 7.

167

Reiter, Dan and Allan C. Stam, III. 1997. “Democracy, War Initiation, and Victory.”
Unpublished manuscript.

Raymond, Gregory A. 1994. “Democracies, Disputes, and Third-Party Intermediaries.”
Journal of Conflict Resolution, 38:24-42.

Raymond, Gregory A. 1996. “Demosthenes and Democracies: Regime Type and
Arbitration Outcomes.” International Interactions, 22:1-20.

Rosenau, James N. 1966. “Pre-Theories and Theories of Foreign Policy,” in R. Barry
Farrell (ed.), Approaches to Comparative and International Politics. Evanston:
Northwestern University Press.

Rosenau, James N. 1990. Turbulence in World Politics: A Theory of Change and
Continuity. Hemel Hempstead: Harvester Wheatsheaf.

Rousseau, David L., Christopher Gelpi, Dan Reiter, and Paul K. Huth. 1996. “Assessing
the Dyadic Nature of the Democratic Peace, 1918-88.” American Political Science Review,
90:512-533.

Ruggie, John Gerard. 1975. “International Responses to Technology: Concepts and
Trends.” International Organization, 29:557-584.

Ruggie, John Gerard. 1983. “International Regimes, Transactions, and Change: Embedded
Liberalism in the Postwar Economic Order,” in Stephen D. Krasner, (ed.), International
Regimes. Ithaca: Cornell University Press.

Ruggie, John Gerard. 1986. “Continuity and Transformation in the World Polity: Towards
a Neorealist Synthesis,” in Robert O. Keohane (ed.), Neorealism and Its Critics. New
York: Columbia University Press.

Rummel, R. J. 1968. “The Relationship Between National Attributes and Foreign Conflict
Behavior,” in J. David Singer (ed.), Quantitative International Politics: Insights and
Evidence. New York: Free Press.

Rummel, R. J. 1979. Understanding Conflict and War: Volume 4, War, Power, Peace.
Beverly Hills: Sage.

Rummel, R. J. 1983. “Libertarianism and lntemational Violence.” Journal of Conflict
Resolution, 27:27-71.

Rummel, R. J. 1985. “Libertarian Propositions on Violence Within and Between Nations.”
Journal of Conflict Resolution, 29:419-455.

168

Rummel, R. J. 1995. “Democracies ARE Less Warlike Than Other Regimes.” European
Journal of International Relations, 1:457-479.

Russett, Bruce. 1990. Controlling the Sword: The Democratic Governance of National
Security. Cambridge: Harvard University Press.

Russett, Bruce. 1993. Grasping the Democratic Peace: Principles for a Post-Cold War
World. Princeton: Princeton University Press.

Salmore, Stephen A. and Charles F. Hermann. 1969. “The Effect of Size, Development and
Accountability on Foreign Policy.” Peace Science Society Papers, 14:16-30.

Simon, Michael W. and Erik Gartzke. 1996. “Political System Similarity and the Choice of
Allies.” Journal of Conflict Resolution, 40:617-635.

Singer, J. David. 1961. “The Level of Analysis Problem in lntemational Relations.” World
Politics, 14:77-92.

Singer, J. David, Stuart A. Bremer and John Stuckey. 1972. “Capability Distribution,
Uncertainty, and Major Power War, 1820-1965,” in Bruce M. Russett (ed.), Peace, War,
and Numbers, Beverly Hills: Sage.

Singer, Max and Aaron Wildavsky. 1993. The Real World Order: Zones of Peace/Zones of
Turmoil. Chatham: Chatham House Publishers.

Siverson, Randolph M. and Juliann Emmons, 1991. “Birds of a Feather: Democratic
Political Systems and Alliance Choices in the Twentieth Century.” Journal of Conflict
Resolution, 35:285-306.

Siverson, Randolph M. and Joel King. 1979. “Alliances and the Expansion of War,” in J.
David Singer and Michael Wallace (eds), To Augur Well, Beverly Hills: Sage.

Small, Melvin and J. David Singer. 1976. “The War-Proneness of Democratic Regimes,
1816-1975.” Jerusalem Journal of International Relations, 1:50-69.

Small, Melvin and J. David Singer. 1982. Resort to Arms: International and Civil Wars,
1816-1980. Beverly Hills: Sage.

Spiro, David E. 1994. “The Insignificance of the Liberal Peace.” International Security,
19:50-86.

Stam, Allan C., III. 1996. Win, Lose, or Draw: Domestic Politics and the Crucible of War.
Ann Arbor: University of Michigan Press.

169

Starr, Harvey. 1992. “Why Don’t Democracies Fight One Another—Evaluating the
Theory-Findings Feedback Loop.” Jerusalem Journal of International Relations, 14:41-59.

Stein, Arthur A. 1982. “Coordination and Collaboration: Regimes in an Anarchic World.”
International Organization, 36:299-324.

Strange, Susan. 1982. “Cave! Hic Dragones: A Critique of Regime Analysis.”
International Organization, 36:479-496.

Thompson, William R. 1983a. “Uneven Economic Growth, Systemic Challenges, and
Global Wars.” International Studies Quarterly, 27:341-355.

Thompson, William R. 1983b. “Cycles, Capabilities, and War: An Ecumenical View,” in
William R. Thompson (ed.), Contending Approaches to World System Analysis, Beverly
Hills: Sage.

Thompson, William R. 1996. “Democracy and Peace: Putting the Cart Before the Horse?”
International Organization, 50:141-174.

Thompson, William R. and Richard Tucker. 1997. “A Tale of Two Democratic Peace
Critiques.” Journal of Conflict Resolution, 41:428-454.

Thucydides. c 400 B.C./1951. History of the Peloponnesian War. John H. Finley, Jr.,
transl. New York: Modern Library.

Wallensteen, Peter and Margareta Sollenberg. 1996. “The End of lntemational War?
Armed Conflict, 1989-1995.” Journal of Peace Research, 33:353-370.

Waltz, Kenneth. 1959. Man, the State, and War. New York: Columbia University Press.

Waltz, Kenneth. 1962. “Kant, Liberalism, and War.” American Political Science Review,
56:33 1-340.

Waltz, Kenneth. 1979. Theory of International Politics. Reading: Addison-Wesley.
Waltz, Kenneth. 1986. “Reflections on Theory of International Politics: A Response to My
Critics,” in Robert O. Keohane (ed.), Neorealism and Its Critics. New York: Columbia

University Press.

Weede, Erich. 1984. “Democracy and War Involvement.” Journal of Conflict Resolution,
28:649-664.

Weede, Erich. 1992. “Some Simple Calculations on Democracy and War Involvement.”
Journal of Peace Research, 29:377-383.

170
Wright, Quincy. 1942. A Study of War. Chicago: University of Chicago Press.

Young, Oran. 1980. “International Regimes: Problems of Concept Formation.” World
Politics, 32:331-356.

Young, Oran. 1983. “Regime Dynamics: The Rise and Fall of lntemational Regimes,” in
Stephen D. Krasner, (ed.), International Regimes. Ithaca: Cornell University Press.

Young, Oran. 1991. “Political Leadership and Regime Formation: On the Development of
Institutions in lntemational Society.” International Organization, 45:281-308.

Zinnes, Dina and Jonathan Wilkenfeld. 1971. “An Analysis of Foreign Conflict Behavior
of Nations,” in Wolfram F. Hanreider (ed.), Comparative Foreign Policy. New York:
McKay.