Download as pdf or txt
Download as pdf or txt
You are on page 1of 26

Alliances, Then Democracy: An Examination of the Relationship between Regime Type and

Alliance Formation
Author(s): Douglas M. Gibler and Scott Wolford
Source: The Journal of Conflict Resolution, Vol. 50, No. 1 (Feb., 2006), pp. 129-153
Published by: Sage Publications, Inc.
Stable URL: https://www.jstor.org/stable/27638478
Accessed: 03-12-2019 01:18 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

Sage Publications, Inc. is collaborating with JSTOR to digitize, preserve and extend access to
The Journal of Conflict Resolution

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Alliances, Then Democracy
AN EXAMINATION OF THE RELATIONSHIP BETWEEN
REGIME TYPE AND ALLIANCE FORMATION

DOUGLAS M. GIBLER
Department of Political Science
University of Kentucky
SCOTT WOLFORD
Department of Political Science
Emory University

Previous studies have not established a consistent link between regime type and alliance formation,
despite the relevance of the decision to ally in a number of arguments about regime type and alliance behav
ior. The inconsistency in these findings turns largely on choice of research design and variable definition;
when the dependent variable is alliance formation, democratic dyads are unlikely to ally, but when the
dependent variable is the presence of an alliance tie, democratic dyads are likely to be allied. Under a stan
dard research design, the authors find both claims to be true and propose a test of an explanation for this ten
dency of autocratic states to democratize in alliances. They show that the presence of a defense pact with all
neighboring states reduces the likelihood that a state will be targeted with a territorial militarized dispute,
reduces the level of state militarization, and increases the likelihood of democratic transitions.

Keywords: Alliances; democracy; external threat; cooperation

IVecent studies of the relationship between domestic politics and alliance behavior
tend to focus on questions of the duration (Bennett 1997; Reed 1997) or the reliability
of alliance commitments (Leeds 2003; Gartzke and Gleditsch 2004), but consensus is
equally lacking on the issue of the link between regime type and alliance formation.
Without adequate explanations for and evidence of the processes that lead alliances to
possess particular mixes of regime types, our explanations of events after the forma
tion of the alliance are incomplete. Accordingly, in this study, we (1) resolve an empir
ical debate regarding the propensity for democratic states to ally with one another and
(2) propose and execute a test of an explanation for this phenomenon that has implica
tions for other parts of the literature on alliances and domestic politics. Specifically,

AUTHORS' NOTE: We thank Ashley Leeds, Matt Gabel, and Kirk Randazzo for comments on earlier
drafts and presentations of this work. Replication data are available at http://www.uky.edu/~dgibler and
http://jcr.sagepub.eom/cgi/content/full/50/l/l 29/DC1/.

JOURNAL OF CONFLICT RESOLUTION, Vol. 50 No. 1, February 2006 129-153


DO?: 10.1177/0022002705281360
? 2006 Sage Publications
129

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
130 JOURNAL OF CONFLICT RESOLUTION

we argue that alliances often serve as mechanisms for reducing territorial threat and,
consequently, render transitions to democracy more likely among their members.
We demonstrate that evidence connecting democratic dyads to alliance formation
turns critically on relatively few defense pacts, all formed in the postwar period when
some of the alliance members were not yet democratic. In turn, the treatment of these
cases influences substantially the quantitative evidence linking democratic dyads to
alliance, often producing contradictory results across identical independent variables.
For example, studies demonstrating no link between dyadic democracy and alliance
formation often focus their empirical tests on the first year of alliance formation, while
others demonstrating a positive relationship usually estimate the probability of
observing a democratic, allied dyad during any year of the alliance. However, if allied
states tend to democratize after the formation of the alliances, then a relationship
between dyadic democracy and alliance should manifest even when this relationship
did not exist at the time of formation.
Our results carry important implications for several current debates in international
relations. For example, some theories of cooperation derive from the likelihood for
democracies to prefer other democracies for formal commitments (Leeds 1999).
Cooperation in security is also linked to regional clustering of democracies (Gleditsch
2002) and to the tendency for increases in the strength of the democratic community to
affect the likelihood of democratization in other states (Kadera, Crescenzi, and Shan
non 2003). None of these stories, however, explains why democracy is negatively
associated with alliance formation, while it remains positively associated with being
allied. In addition, answering questions of democracies' reliability as alliance partners
depends critically on whether democracies do, in fact, seek out their own kind as alli
ance partners (Leeds 2003; Gartzke and Gleditsch 2004). If however, states have a ten
dency to become democratic after the formation of the alliance, then the strategic char
acter of the decision to ally may require a new explanation. Accordingly, we offer a
new theory of regime type and alliance commitments that sheds new light on each of
these issues.
Instead of regime type determining the choice of allies, we argue that certain types
of alliances preserve peace by reducing the territorial threat to member states, and as
the presence of territorial threat can hamper transitions to democracy,1 these alliances
may contribute to the democratization of their member states. We make this argument
in four parts. First, we outline extant studies demonstrating a relationship between
regime type and alliance behavior that is inconsistent across changes in research
design. Second, we outline a theory that explains how alliances might increase the
chances for democratization, linking the experience of territorial threat to the internal
development of the state. Our third and fourth steps turn to providing empirical sup
port for the arguments that the relationship between democracy and alliance is present

1. Boix (2003, 228) makes one of the best recent points on this question, demonstrating that periods
of peace associated with "the settling of territorial claims" on the European continent enabled stable eco
nomic growth and diversification, which in turn made possible the establishment of democratic governance
in the Western part of the continent. In a separate paper (Gibier 2005), we note that twenty of twenty-five of
Przeworski and Limongi's ( 1997) state-years that were most likely to be democracies yet remained dictator
ships had territorial claims or were targeted by territorial disputes.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 131

in only a few important cases, and in these cases, member states often democratize
after the alliance forms. We begin with a discussion of the few important cases in the
democracy-alliance relationship, showing that each alliance treaty contained provi
sions for the removal of territorial issues between member states and the resolution of
international borders, as well as collective security in the event of an attack from out
side. Finally, we offer statistical evidence that the presence of a defense pact with all
neighboring states (1) reduces the probability that a state will be targeted by a territo
rial militarized interstate dispute, (2) reduces the level of militarization in a state, and
(3) increases the probability of transitions to democracy.

DEMOCRACY AND ALLIANCES

Siverson and Emmons (1991) first identified the relationship between formal alli
ances and democracy. Separating the Correlates of War alliance data into two samples,
1920 to 1939 and 1946 to 1965, they observed that democracies allied with each other
at a higher rate than with other regime types in both samples, but the bias was espe
cially pronounced during the early years of the cold war. The bias toward jointly demo
cratic alliance dyads held in their tests even after the addition of controls for power
status and contiguity.
Simon and Gartzke (1996) confirm the slight bias toward jointly democratic alli
ance dyads, but they also challenge the results of Siverson and Emmons (1991) as a
product of cold war bipolarity. Using a greatly extended temporal domain, 1815 to
1992, Simon and Gartzke demonstrate that the largest correlation between joint
democracy and alliance formation exists during the cold war years. Indeed, 7 percent
of the entire sample was either Warsaw Pact or North Atlantic Treaty Organization
(NATO) alliance dyads, and the removal of these cases from the data dropped the cor
relation by 49 percent?from .25 to .13. These results and analyses on similarities of
regime score among alliance dyads led Simon and Gartzke to argue that the slight bias
toward alliance in jointly democratic allied dyads is the product of overall distributions
of power in the system and the ideological nature of the cold war alliance system
(Simon and Gartzke 1996, 633), although this argument is never directly tested.
Lai and Reiter (2000) provide some resolution to the debate with the most compre
hensive tests to date of the factors predicting alliance in a dyad. Using the same tempo
ral domain as Simon and Gartzke (1996) but providing the first multivariate test of the
proposition, Lai and Reiter confirm that democratic dyads are not more likely to be
allied but also establish a positive relationship between regime type similarity and alli
ances during the cold war.2 When the dependent variable is limited to defense pacts,
the relationship reverses, and democratic dyads are more likely to ally during the cold
war years. However, the defense pact results may be driven by several large, multilat

2. Leeds et al. (2002) use a different alliance data set (the Alliance Treaty Obligations and Provisions
[ATOP] data set) to show that democratic dyads are more likely to ally in the century and a half prior to World
War II, but Gibier and Sarkees (2004) point out that these results are most likely due to one large multilateral
treaty not included in the Correlates of War data set. Even using the ATOP data set with all joiner states
included, democracies are not likely to ally with each other prior to World War II.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
132 JOURNAL OF CONFLICT RESOLUTION

eral European and North/South American defense pacts as exclusion of these regions
led to statistically insignificant results for the joint democracy variable.
Overall, the literature confirms a link between alliance formation and regime type,
but this relationship is not consistent across time, space, or research design. Indeed,
differences across these studies are large enough to alter the direction of the coeffi
cients in empirical tests, but unfortunately, current theories relating regime type to alli
ances do not account for the variation in results across these studies.
The best theory we have to explain the relationship between regime type and alli
ance behavior ascribes the tendency toward cooperation for similar regime types to be
the product of domestic audience costs. Democracies suffer larger costs from reneging
on public international commitments, as defection is linked with political vacillation
and leads to punishment at the polls. Therefore, democratic leaders tend to make more
credible commitments, making democracies attractive alliance partners for one
another. Conversely, autocratic leaders have greater autonomy at home and suffer
fewer costs from defection. This makes autocratic leaders prone to more commit
ments, but their commitments are likely to be less reliable. Leeds (1999) develops this
argument and finds support with data on cooperation; she also finds that democracies
tend to be the most reliable alliance partners, honoring their alliance commitments in
times of war (Leeds 2003). More recent studies have cast doubt on the empirical find
ing, however, as both Leeds and Gigliotti-Labay (2003) and Gartzke and Gleditsch
(2004) find that democracies are not systematically more likely to honor alliance com
mitments than other regimes. The theories underlying each result, however, depend on
an accurate accounting of whether and why democracies end up in alliances with each
other.
The reliability of democratic partnerships is also crucial for most contractual ver
sions of regime-based arguments. The level of open debate afforded by democratic
institutions increases transparency during international bargaining, and this transpar
ency allows outsiders to gauge the level of commitment held by domestic institutions
for any agreements that are reached with that state. In other words, the "contracting
advantages" of democracies make them more sought after partners and also make it
more likely that their agreements will last (Lipson 2003).
Nevertheless, neither audience costs nor contractual arguments adequately explain
why the expected relationship between regime type and alliance is inconsistent across
regions; nor can the theory explain the negative relationship between joint democracy
and alliance prior to 1945. These anomalies, combined with the inconsistencies across
research design, make us believe the nature of the relationship may be misunderstood.
Consider first the observation that democratic dyads are more likely to be allied
than other dyadic regime types. Unlike previous research, the research design used by
Lai and Reiter (2000) does not identify whether a specific alliance/orras in any given
year. Instead, the dependent variable measures whether a dyad is allied, making it
unimportant whether the alliance was newly formed or in its fifth, tenth, or thirtieth
year. This research design ignores the effects that regime transitions have after states
are already allied. As we argue later, if alliances remove territorial threat in the dyad,
then this type of peace is a factor determining whether contiguous states become dem

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 133

ocratic. The research design would then observe allied dyads during later years of the
alliance even though the alliance was formed with nondemocratic members.
While analyses by Siverson and Emmons (1991) and Simon and Gartzke (1996) do
differentiate between alliance formation and the continuation of alliances in a dyad,
both research designs yield little information on the joint effects of regime type and
contiguity. Siverson and Emmons disaggregate newly formed alliance dyads by both
regime type and contiguity but do not provide a method of assessing the relative
impact of each variable. Simon and Gartzke control for contiguity by eliminating all
contiguous, land border dyads from the data set, but this is obviously inappropriate for
yielding information regarding possible interactions between regime type, contiguity,
and alliance formation.
We argue that the alliance literature thus far points to the conclusion that a large por
tion of alliances, especially those during the cold war, are formed by contiguous states
and often remove territorial threat in the dyad. By effectively stabilizing the interna
tional borders, alliances eliminate one of the major obstacles to democratization in the
state. We outline the rationale behind our argument in the following sections and begin
by showing how the existing alliance literature points toward this conclusion.

ALLIANCE FORMATION, ALLIED DYADS, AND BORDERS

That an alliance can serve as a method of resolving border issues is not a new claim.
Gibier (1996,1997), for example, demonstrates that many alliances serve as territorial
settlement treaties, resolving territorial claims among the states involved in the alli
ances, and these alliances lead to peace much more often than not. Similarly, Powers
(2001) finds that multilateral economic unions often contain provisions for mutual
defense amid assurances of border sanctity; this level of peace allows for trade to
develop among the alliance members. Alliances providing for institutionalized mech
anisms of dispute resolution and border stability, in addition to incentives for trade,
may also increase the likelihood of democratic regime transitions (Pevehouse 2002).
Our argument, however, is somewhat different. Instead, we understand the forma
tion of an alliance between neighboring states as often signaling the acceptance of the
border that divides them, or at least the abeyance of latent claims. In turn, the alliance
functions?directly or indirectly?as a means of removing territorial threat in the
dyad. Territorial settlement treaties are explicit attempts to resolve border issues, but
other alliances may resolve territorial issues as well. When states form coalitions to
coordinate an attack, their territorial claims often become less important, and states
signaling resolve in order to deter attack are much less likely to force claims against
other states in their coalition. In other words, even alliances formed for external secu
rity concerns can have indirect effects on the borders of allied states.3
If external security concerns are combined with coordinated efforts to assure alli
ance members of the sovereignty of their states and the inviolability of their borders,
democratization becomes more likely. As we demonstrate later, the treaty terms of
three large, regional defense pacts carry explicit guarantees of international border
security for all states within the alliance as well as pledges for common defense should

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
134 JOURNAL OF CONFLICT RESOLUTION

these guarantees be breached. Therefore, while these defense pacts may have served as
a signal of resolve against Soviet expansion or as attempts to integrate the states within
the alliance, the pledges for border security among alliance members combined to
eliminate the imminent territorial threat that leads to centralization of the state. Since
these defense pacts often include all neighboring states, the territorial integrity of the
alliance members is reinforced, making their borders stable in the face of all but the
most serious threats to regional and global territorial norms. But why is territorial
threat so inimical to the development of democratic government?
One hundred years ago, Otto Hintze ([1906] 1975) attributed the emergence of
"self-government" and decentralized political authority in states such as England,
Switzerland, and the United States to particular international conditions?defensible
and/or uncontested borders?and the exigencies of military organization associated
with defending territory. Continental states, whose borders were often contested and
difficult to defend, developed highly centralized state structures to mobilize mass
armies for national defense. Insular England and mountainous Switzerland, however,
needed only localized militias for territorial defense. These decentralized organiza
tional structures checked the development of extractive and coercive power at the cen
ter. Peaceful border relations, in effect, create the social conditions for democracy,
demanding less political centralization and less mass mobilization than sensitive or
unstable border relations. The exigencies of war and military organization thus condi
tion the internal organization of the state.4
Conflict behavior is often associated with the organization of the state (Howard
1979; Mann 1988; Rasier and Thompson 1989; Tilly 1990), and in the century since
Hintze, a number of scholars have linked the necessities of war making and the
strength, scope, and size of the state (Titmuss 1959; Freid 1961; Desch 1996). Desch
(1996, 243), for example, asserts that persistent threats to security create a "maximal"
state, large in both size and scope, which dominates and extracts from society to a great
degree. As the level of international threat varies, the necessity of the state to extract
from society should vary as well, leading to changes in modes of state organization.
We extend this logic to suggest that the large standing armies that result from an
increase in extractive and coercive central power stand in the way of decentralization
and, ultimately, democracy.
Thompson (1996) builds on this in his explanation of the democracy-peace rela
tionship. As states abandon expansionist foreign policies, often associated with drives
for regional hegemony, they are able to reduce capability and authority concentration,
opening the way for domestic decentralization and liberalization. Suggesting that
regional stability encourages domestic liberalization by reducing the threat from out

3. Alliances are by no means the only method of resolving territorial issues. Mediation by regional
and international organizations, mediation by other third parties, and even changes in capabilities within the
dyad are all events that can spell the end of a territorial dispute.
4. Vasquez (1993, 141ff) echoes the early arguments of Hintze ([1906] 1975) in his assertions that
the socialized disposition to territoriality impels human collectivities to define their borders with aggressive
displays and to defend them with power politics behaviors. As borders stabilize and as territorial issues
become fewer, the circumstances under which states resort to power politics behaviors should fall accord
ingly (see also Maoz 1989).

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 135

side, he identifies a curvilinear relationship between peace and democracy; the


chances for democratization rise quickly following the abandonment of an expansion
ist foreign policy and the beginnings of a regional zone of peace.
Each of the theories based on the international threat environment coincides well
with growing evidence that democracy obtains only when no domestic group has
enough political power to dominate the state and exclude others from power
(Przeworski 1988; Vanhanen 1990; Olson 1993; Boix 2003). The military presents the
likeliest path to political domination, and, ceteris paribus, a reduction in the political
power of the military should increase the likelihood of a democratic transition. Territo
rial threat in this sense links state-level theories of democracy via group competition to
theories of democracy via regional peace and diffusion (Gleditsch 2002).5
Our explanation of the interaction between international threat and state develop
ment complements stories relating domestic transitions to external conflict. Mansfield
and Snyder (1995, 2002), for example, have found that incomplete democratic transi
tions that stall in the face of a weak institutional environment often lead to war. Domes
tic elites turn to nationalist rhetoric in order to mobilize followers and shore up politi
cal support, touching off nationalist bidding wars and frequently incoherent foreign
policies that result, ultimately, in a failed or stalled transition and the initiation of con
flict. Territorial grievances are plausible lightning rods around which elites might
build support for belligerent policies, and, as Mansfield and Snyder note, Ecuador and
Peru democratized in fits and starts through the 1980s and 1990s in the shadow of a
dispute over their shared border. When transitions can occur in the absence of territo
rial issues, however, then threatened elites have one less "expedient strategy"
(Mansfield and Snyder 2002, 301) around which to galvanize nationalist support and
stall the transition at an incomplete stage.
We find confirming evidence for our argument linking alliances to a reduction of
territorial threat in the growing consensus that alliance ties decrease overall conflict
within the dyad (see, e.g., Bremer 1992; Oneal and Russett 1997; Reed 2000). Given
the strong effect territorial issues have on conflict (Holsti 1991 ; Vasquez 1993; Hensel
2001; Huth 1996; Huth and Allee 2003), it seems reasonable to speculate that the
allied peace may at least partially result from a reduction in territorial threat among
member states. The same interests that provided the rationale for the alliance may also
lead alliance members to forego their claims against the territories of other alliance
members, at least for the duration of the alliance.6

5. Note that this argument reverses the causal connection between democracy and conflict identified
by several authors (Mitchell and Prins 1999; Bueno de Mesquita et al. 2003). Rather than joint democracy
leading to peaceful resolution of disputes, we argue that the processes generating democracy also remove the
hardest issues to resolve (territorial) from the agenda of these states. We develop this argument in much
greater detail in Gibier (2005).
6. The resolution of territorial threats could indirectly be the result of external threat forcing internal
cohesiveness in the dyad or more directly result from provisions in the treaty. Of course, our argument does
not preclude the possibility that alliances may be provocative to states outside the alliance (Vasquez 1987,
1993). We only examine the tendency of defense pacts to remove territorial threat for contiguous states
within the alliance. Also significant is the unique nature of the alliances containing the bulk of democratic
dyads?few in number, the alliances containing democratic dyads are regional pacts, containing internal
security provisions. These are much different from the provocative, capability-aggregating alliances linked
to international war participation (Gibier and Vasquez 1998).

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
136 JOURNAL OF CONFLICT RESOLUTION

We are not the first to argue that alliance ties might be responsible for the diffusion
of democracy through time and space. Kadera, Crescenzi, and Shannon (2003), for
example, find that the presence of a strong democratic international community
increases the likelihood that democratic government will survive. They posit that alli
ance ties, specifically the reliability of democratic alliance partners, may be one of the
causal mechanisms that dampen conflict for these states. This may also help explain
why democracies tend to cluster regionally (Gleditsch, 2002), as democracies develop
within the bounds of regional collective security pacts.
It is not necessary to argue that there is an overt attempt by alliances as institutions,
or even the alliance members individually, to democratize neighboring members.
Rather, the alliance removes the threat to territory, effectively eliminating a major
obstacle to democratic government, and therefore, we should not necessarily see
direct action by alliance institutions such as NATO creating democracy (Reiter 2001
2002). Instead, we should see a gradual increase in the number of democracies, devel
oping in the dyads that have allied.7
This interpretation would obviously explain the defense pact results obtained by
Lai and Reiter (2000), especially how their results were more robust for North Ameri
can and European dyads. Regional defense pacts, or defense pacts formed by contigu
ous states, often alleviate a source of threat to the territory of alliance members. This,
in turn, removes a major obstacle to democratization as land armies and the centraliza
tion of the state become less important. As allied states turn democratic, the likelihood
of observing jointly democratic and allied dyads increases. The first step to confirming
our argument, then, is the direction of the causal arrow?namely, whether contiguous
democracies are forming these alliances or are becoming democratic after the alliance
forms.
Our theory also suggests that democratic dyads should be less likely to form alli
ances, as states suffering no territorial threats would have few reasons to seek alliances
and as democracies would have little reason to control similar regimes via alliance.
While both research designs that focus on alliance formation demonstrate a small bias
favoring democratic alliance formation at the dyadic level, we believe this is due prin
cipally to research design. Simon and Gartzke (1996) exclude the main cases of inter
est to our theory by omitting all contiguous dyads, and a strong negative bias in these
cases could reverse their results, while Siverson and Emmons (1991) employ a much
shorter temporal domain, with the end of their data (in 1965) coming well before most
democratization in alliances occurs. These assertions can be tested empirically. If our
argument is correct, then theories of regime type and alliance preference misinterpret
the direction of the causal arrow?alliance formation influences regime type while
democracy does not affect alliance choice?suggesting that we should observe the
following dyadic relationships:

7. We realize that our argument is probabilistic since Bueno de Mesquita (1981) and Ray ( 1990) note
that some alliances lead to conflict in the dyad. Interestingly for our argument, however, is Bremer's (1992)
study showing that contiguity eliminates the significant relationship between alliances and war, once again
suggesting that contiguity is related to both alliances and conflict.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 137

1. Democratic dyads are unlikely to form alliances.


2. But democratic dyads are likely to be allied.

The rationale for hypotheses 1 and 2 is based on the predicted effects of external
threat on the centralization of the state. We expect that defense pacts signal the willing
ness of alliance members to accept the territorial status quo. This acceptance reduces
territorial threat and the need for maintaining a highly militarized state that often ham
pers the transition from autocracy to democracy. Stated as hypotheses, we expect the
following relationships to obtain for state-level analyses:

3. States in defense pacts with contiguous states are less likely to be targeted by territorial
militarized interstate disputes.
4. States in defense pacts with contiguous states will be less militarized than other states.
5. Autocracies in defense pacts with contiguous states will be more likely to experience
transitions to democracy.

RESEARCH DESIGN

DYADIC-LEVEL TESTS

Since we argue that the inconsistencies of the alliance literature thus far have been
products of differences in research design, we chose to adopt the models and data used
by Lai and Reiter (2000) to test hypotheses 1 and 2 but slightly alter their dependent
variable in two simple ways. First, we updated Lai and Reiter's original dependent
variable, alliance in the dyad-year, by using the new Correlates of War formal alliance
data set (Gibier and Sarkees 2004). We also include a second dependent variable that
captures alliance formation. To code this variable, we include all dyads forming an
alliance during the year the original alliance treaty was signed; we exclude joiner
dyads. NATO provides a good example of our coding rules?the alliance was formed
in 1949, and all dyads part of the alliance in 1949 are coded as alliance formation
dyads for that year; all alliance formation dyads and all subsequent alliance dyad
years from 1950 on, including dyads that join NATO, are coded as allied dyad-years.8
Tests on the first dependent variable confirm the research by Lai and Reiter, providing
evidence for our second hypothesis, while tests using the alliance formation dependent
variable assess the support for our first hypothesis.
For both analyses using alliance ties as the dependent variable, we include a one
year lag that measures whether the dyad was allied in the previous year, expecting this
variable to have a positive effect in both models. As a large number of dyads engaged
in multiple, overlapping regional and security alliances, the formation of an alliance to
cooperate on one issue will likely influence the willingness to cooperate on other
issues. For example, the United States and Great Britain were allied via three different
alliances during the late 1950s, 1960s, and early 1970s; the first alliance was dedicated

8. We also conducted analyses including joiner dyads as alliance formation cases; the inclusion of
these dyads does not alter the results that follow in any substantive way.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
138 JOURNAL OF CONFLICT RESOLUTION

to cooperation in Europe (NATO), the second in Southeast Asia (Southeast Asia


Treaty Organization [SEATO]), and the third in the Middle East (Baghdad Pact).9
We keep all of Lai and Reiter's (2000) independent and control variables for these
analyses. First, the effects of regime type are assessed using measures of joint democ
racy and regime similarity in the dyad. Regime type is based on the 21-point combined
autocracy-democracy scores from Polity IV. Joint democracy is dichotomous and is
coded as 1 for all dyads possessing two states that score 5 or higher on this scale;
regime similarity is the absolute difference of the two regime scores (Lai and Reiter
2000, 213). The second group of independent variables captures cultural similarities
in the dyad. Using the Cultural Composition of Interstate System Members data set
from the Correlates of War project, they code dummy variables for dyads with similar
ethnicity, language, and religion (Lai and Reiter 2000, 214). The third group of inde
pendent variables captures the level of joint threat experienced by the two states of the
dyad. These variables include whether the states were on opposite sides of a milita
rized interstate dispute (MID), whether the states of the dyad had the same enemy in an
MID, and the total number of MIDs experienced by each state in the dyad; each of
these measures was coded using data from the ten years prior to each dyad-year (Lai
and Reiter 2000, 214). Finally, Lai and Reiter (p. 215) specify variables for overall
trade (the lower score of each state's exports and imports within the dyad divided by its
gross domestic product), the presence of at least one major power, distance between
capitals controlling for contiguity, and a learning measure based on Reiter's (1996, 84)
earlier work.10 We keep each of their variables, permitting an easier comparison to
their results.

STATE-LEVEL TESTS

Our other three hypotheses are state-level expectations regarding the effects of alli
ances on regime type. Our principal variable of interest is the presence of a defense
pact with each neighboring state. As we hypothesized earlier, we believe that a defense
pact can stabilize an international border, and this will, in turn, lead to reductions in ter
ritorial threat, reduced levels of militarization, and, eventually, democratization in the
state. We expect an absence of territorial threat to be achieved only when a state has a
defense pact with all its neighbors; we code this dummy variable as present only for
these cases and absent otherwise.
We employ three dependent variables to assess each part of our causal logic. First,
we examine whether borders stabilized via defense pacts remove territorial threat to

9. Almost 19 percent of the data (13,316 of 70,768 dyad-years) had multiple alliances; the multiple
alliances ranged to a high of five in a total of 20 dyad-years. In 1,564 dyad-years, the dyad increased its num
ber of alliances (from zero to one, from one to two, etc.) from the previous dyad-year, and these are coded as
alliance formation dyads.
10. A state is coded as 1 if it had a lesson favoring alliance, -1 if it had a lesson favoring neutrality, and
0 if it had no lesson. A state is coded as having a lesson favoring neutrality if it was neutral during World War I
or II and was not invaded or if it was allied during the war and was invaded. A state is coded as favoring alli
ance if it was allied during a world war and was not invaded or was neutral during the war and was invaded
(Reiter 1996, 84). The dyadic learning variable varies from -2 to 2, according to the cumulative learning
experiences of each state in the dyad.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 139

the dyad, and for these tests, we employ a dichotomous dependent variable that cap
tures whether the state was the target of a territorial MID in a given year.11 Second, we
argue that stable borders reduce the need for militarization of the state, and for these
analyses, we measure militarization as the percentage of total state population serving
in the military. Both the population and the military personnel figures are from the
Correlates of War Composite Index of National Capabilities (CINC) data set. Third
and finally, we examine the effects that stable borders have on the likelihood of a dem
ocratic transition. We use a dichotomous measure of democracy, with states scoring
greater than 5 on the combined Polity IV scale labeled democracies.
We include several control variables in the state-level analyses. First, for all analy
ses, we add a control that captures the total percentage of countries in a Correlates of
War region that are democratic; this measure is derived by dividing the number of
democracies by the total number of states in each region. Regional democracy has
been shown to be one of the best predictors of democratic transition (Gleditsch 2002),
and we believe this clustering process may affect territorial disputes and militarization
as well.
For our analysis of factors affecting the likelihood of democratization, we include
in the post-1945 model the natural logarithms of the gross domestic product (GDP)
and the total population, a dummy variable for major oil-exporting countries, and a
percentage estimate of the total state population that is Muslim. The democratization
literature has demonstrated that these variables often affect regime type, and we use
data from Fearon and Laitin (2003) to operationalize these controls. To test our argu
ment using a longer temporal domain, we replace the GDP and population measures
with the Correlates of War CINC score for each state; this measure is heavily corre
lated with GDP (see Organski and Kugler 1980) and also includes population esti
mates from 1816 to 2000. For the militarization and democratization models, we also
add a dummy variable for states involved in a militarized interstate dispute, as disputes
are likely to increase the number of military personnel. We also include this variable as
a control in the democratic transition model, but the empirical evidence linking dispute
involvement to decreased likelihood of democratic transitions is weak (Reiter 2001
2002).12

SUMMARY STATISTICS: DEMOCRATIC DYADS AND ALLIANCE

As the data in Table 1 demonstrate well, the evidence connecting regime type to
alliance formation is based on an incredibly small number of alliances containing the
bulk of democratic dyads. NATO dominates the relationship, with more than 55 per
cent (3,259 of 5,885) of the jointly democratic allied dyad-years found in this alliance,
and together, NATO, the Organization of American States (OAS) defense pact, and the
1948 defense pact among Western European states account for more than 90 percent

11. We include only militarized interstate disputes (MIDs) initiated by contiguous states, as these are
the type of conflict most likely to constitute a threat to the home territory of the state.
12. Our substantive results in the democratic transition model do not change if this variable is omitted.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
-4.
(Z)

TABLE 1
Jointly Democratic Dyad-Years in Alliance, 1816-1992

Jointly
Name of Alliance Democratic Signatories

Four Type II alliances with jointly democratic dyads 2 of 17 (11.8)


Nine ententes with jointly democratic dyads 13 of 245 (5.3)
Twenty-one defense pacts with jointly democratic dyads 19 of 183 (10.4)
Organization of American States (OAS) 3 of 171 (1.8)
Defense pact: United Kingdom, France, and the BENELUX states 9 of 9 (100)
North Atlantic Treaty Organization (NATO) 45 of 55 (81.8)
91/680 (13.4)

NOTE: Totals for each alliance are listed second; alliance percentages are listed in parentheses.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 141

of the dyad-year cases.13 However, democracies do not seem to choose alliances con
sisting solely of other democracies, as only 1 of the 37 alliances containing at least one
democracy consisted solely of democratic states at formation, and most of the eventual
democratic dyads become democratic many years after the formation of the alliance.
Excluding NATO and the 1948 Western European defense pact, democratic dyads
are greatly outnumbered among their other alliance signatories, representing fewer
than 10 percent of the alliance formation dyads that include them. The OAS, for exam
ple, includes only 3 democratic dyads out of 171 total alliance formation dyads in
1947. However, cross-sectional time-series research designs that focus on the dyad
year would conclude that the OAS included more than 28 percent of all jointly demo
cratic allied dyad-years, as many of its member states democratized after the
formation of the alliance.
NATO and the 1948 defense pact represent interesting cases for our argument.
While most signatories to these alliances had fairly strong democratic traditions, last
ing almost thirty years in many states, both alliances were formed by fledgling
regimes?among continental European states, the average regime in NATO was less
than five years old and less than four years in the 1948 pact. The older states in these
alliances?Britain in both pacts and the United States and Canada in NATO?shared
no land borders with the other alliance signatories. More important, the United States
was capable enough to provide common defense for NATO members, and both alli
ance treaties carried explicit terms accepting the status quo borders and providing for
collective security should borders be violated. As we argue in the last section, both of
these factors greatly reduced the need for large land militaries in the allied states;
indeed, most of Western Europe is now able to legislatively cap their military expendi
tures at a small fraction of their respective gross domestic products.
While democracies are more likely to ally in defense pacts, thirteen alliances were
either neutrality/nonaggression pacts or ententes, and these alliances may provide an
answer to Lai and Reiter's (2000) study demonstrating that democratic dyads are not
likely to be allied prior to 1945. Almost all of these alliances were formed during this
period and contained provisions settling the distribution of far-flung colonies among
noncontiguous major states. As such, these alliances would not provide for a reduction
in threat to the territorial homeland of the alliance signatories, thereby stabilizing bor
ders, and the number of joint democracies therefore remains relatively constant
throughout these alliances. For example, the 1907 agreement signed by Entente Cordi
ale members Britain, France, and Spain was an attempt to limit German colonial
expansion in North Africa but provided no guarantees of homeland border stability.
Similarly, a bilateral agreement signed three years earlier promised French support to
Spain in pressing its disputed claims to Moroccan territory, but this agreement, too,
included no mention of cooperation against territorial threat on the Continent.

13. Ikenberry (2001) argues that many of the post-World War II alliances served as pacts of restraint
initiated by the victors. Following the war, the United States had a large power advantage over all rivals,
enabling a coordination effort with states of similar regime that all suffered direct costs from the war.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
142 JOURNAL OF CONFLICT RESOLUTION

TABLE 2
Probit Analyses of Alliance Formation

Model 1 Model 2 Model 3

Dependent Variable Alliance Formation Alliance Formation Alliance Formation

Data Set All Dyads All Dyads All Dyads


Time Period 1816-1992 1816-1945 1950-1992

Allied in previous year .178 (.033)*** .516 (.044)*** .523 (.050)***


Regime type
Joint democracy -.271 (.043)** -.281 (.086)*** -.568 (.089)***
Polity difference -.013 (.002)*** .005 (.004) -.025 (.003)***
Culture
Joint religion .337 (.027)*** .341 (.037)*** .032 (.052)
Joint language .444 (.034)*** .455 (.041)*** .045 (.070)
Joint ethnicity -.165 (.047)*** -.116 (.055)** .045 (.075)
Threat
Conflict relations .040 (.048) -.066 (.060) .140 (.081)*
Joint enemy .636 (.028)*** .467 (.036)*** .063 (.080)
Amount of threat -.012 (.003)*** .032 (.004)*** -.033 (.006)**
Other controls
Distance -.012 (.000)*** -.011 (.001)*** -.014 (.001)***
Major power .287 (.035)*** .063 (.043) .384 (.074)***
Learning .093 (.019)*** .108 (.032)*** .120 (.036)**
Trade -16.419 (7.624)**
Constant -2.345 (.040)*** -2.648 (.061)*** -2.181 (.069)***
411,033 97,043 301,947
Wald x (12) 3566.120*** 2295.450*** 1217.260***
Pseudo-7? 0.215 0.215 0.207

*p<A0. **p<.05. ***/?<.01.

The formation of large regional pacts such as


a departure in alliance politics, creating the co
type and making allied dyad-years more preva
ously.14 We explore this contention in the next
and the likelihood of observing an alliance in a

14. It is also interesting to note that neither the North


original Organization of American States (OAS) charter ha
its member states. While NATO was "founded on the princ
its charter, specific provisions and ancillary agreements en
signed until the final years of the cold war (Reiter 2001-

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 143

REGIME TYPE, ALLIANCE FORMATION, AND ALLIED DYADS

We first estimate the factors causing alliance formation, using probit analyses clus
tered on each dyad. We include models predicting alliance formation for every dyad
year, 1816 to 1992, and we split the sample into two groups, one for years prior to and
including 1945 and one for 1950 and after, with the latter sample including the level of
trade flow as an independent variable.15 In each model, we include a control identify
ing whether the dyad was already allied in the previous year. The results are listed as
models 1 to 3 in Table 2.
Our principal variable of interest, joint democracy, is negative and significant for
each sample?the presence of joint democracy makes alliance formation less likely in
the dyad, confirming our first hypothesis.16 We arrive at this result after controls for
previously allied dyads and distance, which are both statistically significant across
each sample as previous alliances increase the likelihood of an alliance forming but
distance decreases that likelihood. These variables are probably both proxies for dyads
likely to cooperate in general; in the case of previously allied dyads, this suggests that
dyads cooperating in one alliance, with its attendant issues, will be more likely to form
alliances with additional dyads or encompassing a wider range of issues. As our exam
ple above notes, the United States and Britain cooperated formally in alliances con
cerning Europe (NATO), Southeast Asia (SEATO), and the Middle East (Baghdad
Pact). The only other variable demonstrating consistent results across each sample is
the learning variable, and that increases the likelihood of alliance formation. Interest
ingly, the overall level of trade has a strong, negative effect on the likelihood of alliance
formation, suggesting that alliances do not follow trade, as some have suspected
(Gowa 1994; Lai and Reiter 2000; see also Long 2003 for a discussion of the complex
relationship between alliances and trade).
Overall, our models are quite poor predictors of alliance formation, judged by each
pseudo-7?2 and especially compared to the models found in Lai and Reiter (2000) and
Gibier and Sarkees (2004). Much of this is probably due to the rarity of alliance forma
tion in the dyad and the inherent difficulty in correctly guessing those rare events, but it
is also clear that none of the variables, save for distance and joint democracy, serves as
a consistent, statistically significant predictor across each time period. Even the vari
ous measures of conflict and threat do not do well at explaining alliance formation.

15. The temporal domain is limited to 1950 by the availability of trade flow data. We performed addi
tional analyses, without the trade variable, on data from 1919 to 1992 and from 1946 to 1992, but the sub
stance of our results does not change in either case.
16. Ray (2003) argues strongly against inclusion of control variables that, by definition, are related to
the key explanatory variable. In this case, according to Ray (p. 18), we should not include a variable measur
ing regime-type differences in the same model with joint democracy. We therefore estimated two additional
models (one with each variable in isolation) for both alliance formation and allied dyad-years to check the
robustness of our results. In none of the four cases, however, did the two variables change signs or statistical
significance as compared to the model estimating the variables jointly. We present the results for the full
models here to ease comparisons with the results in Lai and Reiter (2000); results of the other models are
available in the Web appendix supporting this article at www.uky.edu/~dgibler and http://www.yale.edu/
unsy/jcr/jcrdata.htm.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
144 JOURNAL OF CONFLICT RESOLUTION

TABLE 3
Probit Analyses of Allied Dyads

Model 4 Model 5 Model 6

Dependent Variable Allied Allied Allied

Data Set All Dyads All Dyads All Dyads


Time Period 1816-1992 1816-1945 1950-1992

Allied in previous year 4.035 (.023)*** 3.614 (.046)*** 4.397 (.032)***


Regime type
Joint democracy .171 (.032)*** -.567 (.117)*** .105 (.038)***
Polity difference -.010 (.002)*** .002 (.004) -.014 (.002)***
Culture
Joint religion .306 (.021)*** .474 (.046)*** .234 (.026)***
Joint language .372 (.036)*** .425 (.049)*** .440 (.053)***
Joint ethnicity -.036 (.042) -.039 (.066) -.043 (.064)
Threat
Conflict relations -.113 (.040)*** -.187 (.065)*** -.091 (.066)
Joint enemy .108 (.027)*** .143 (.048)*** .101 (.048)**
Amount of threat .024 (.002)*** .057 (.005)*** .011 (.002)***
Other controls
Distance -.012 (.000)*** -.014 (.001)*** -.012 (.001)***
Major power -.010 (.033) .131 (.057)** .022 (.045)
Learning .155 (.017)*** .130 (.035)*** .165 (.019)***
Trade 6.553 (3.110)**
Constant -2.312 (.033)*** -2.698 (.074)*** -2.282 (.044)***
411,033 97,043 301,947
Wald x (12) 46984.200*** 9586.170*** 28150.570***
2
Pseudo-7? 0.880 0.811 0.922

**p<.05. ***/?<.01.

We are much better able to predict what dyads are going to be allied in a given year,
as Table 3 demonstrates. Indeed, the overall pseudo-7?2 for the cold war sample is over
.92, and that is only 10/100ths higher than the lowest R2 model, the sample of dyads
prior to 1945.17
In Table 3, joint democracy has a positive, statistically significant effect on the like
lihood of a dyad being allied in the overall sample of states. However, this relationship
is produced by the cold war dyad-years since the sample of dyads prior to 1945 pro
duces a negative relationship. These results are consistent with our argument that the
large regional alliances formed after World War II contributed to the occurrence and

17. The lagged dependent variable accounts for much of the overall fit of the model. However, we also
ran several models omitting the lagged alliance variable, and these analyses generate pseudo-? s that are still
almost twice the size of the alliance formation models.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 145

maintenance of joint democracy in the dyad. Joint democracies are not likely to form
alliances, but allied dyad-years are likely to be democratic during the cold war years.
Several other variables retain their statistical significance in the allied dyads model.
Political difference is again negative in the full sample and the post-1950 sample but
not statistically significant for dyads prior to 1945. Two of the culture variables exhibit
consistently positive results across all three samples, as do variables measuring the
presence of a joint enemy and increased levels of threat to the dyad. Prior conflict rela
tions have no effect. The controls for previous alliances, state learning, and distance
also maintain their significance?prior alliances, positive previous experiences, and
proximity increase the likelihood of alliance in the dyad.
Interestingly, and contrary to our results for alliance formation, trade flows now
have a strong, positive effect on alliance. This may confirm some of the inferences
drawn by Powers (2001) that alliances can serve as tools alleviating barriers to trade in
the dyad and reduce the likelihood of transnational support for insurgent groups, a
main factor disrupting international trade; this may also be further proof that trade and
commitments often have regional effects (Simmons 2000). But of course both of these
arguments are consistent with our overall contention that certain alliances are agree
ments to stabilize borders, removing possible threats to security, as democracy and
trade grow uninhibited once both domestic and international threats are removed or
stifled.
In sum, we have a clear, consistent picture of alliance formation and the likelihood
of alliance in any given dyad-year. Distance and past relationships are strong predic
tors of both dependent variables; in fact, as confirmed by our analyses of relative
changes in rates of correct prediction, these independent variables are in fact the best
indicators of formation and alliance in each of our models.18 The culture variables also
have a positive and generally consistent effect across our models, but the variables that
proxy conflict relations and threat do not perform well, with signs and statistical sig
nificance changing across temporal domains and dependent variables.19
The results for joint democracy are not consistent across the models either, but we
believe that these inconsistencies are explained well by our theory. Joint democracies
are never likely to form alliances, and joint democracies are also not likely to be allied
prior to 1945. We expected the first result, and the second is consistent with our theory
but does limit its applicability to the large regional alliances of the cold war period.
This positive relationship between joint democracy and alliance after 1950 leaves
open the possibility that alliances have an effect on regime type, and we examine that
relationship in the next section.

18. See Web site replication files at http://www.yale.edu/unsy/jcr/jcrdata.htm.


19. These results are also consistent with the observation that alliances serve multiple purposes, not
always aggregating capabilities among member states (Morrow 1991).

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
146 JOURNAL OF CONFLICT RESOLUTION

TABLE 4
Logit Analyses of the Likelihood of Being Targeted
by a Territorial Militarized Interstate Dispute (MID)

Target of Any Target of Any


Territorial MID, Territorial MID,
1816-2000 1946-2000

Independent variables
Presence of defense pact with all contiguous neighbors -0.571 (0.113)*** -0.601 (0.
Democratic state (> 5 on combined Polity IV scale) -0.566 (0.116)*** 0.103 (0.14
Percentage of Correlates of War (COW) region that
is democratic 0.719 (0.217)*** -0.729 (0.259)***
COW Composite Index of National Capabilities score 2.292 (0.939)* 8.961 (1.1
Constant -2.500 (0.057)*** -2.244 (0.068)***
n 10,958 6,227
Likelihood ratio (LR) -2910.74 -1721.14
LR%2 152.05*** 173.36***
NOTE: Dependent variable: target of any territorial MID. Temporal domain: 1816-2000 and 1946-2000.
Robust standard errors are reported in parentheses next to the coefficients. Dummy variable controls for
Israel; please see text.
*p < .10. ***/?<.01.

DO ALLIANCES CONTRIBUTE TO DEMOCRATIC DEVELOPMENT?

As we have already demonstrated, three alliances account for more than 90 percent
of the democratic dyads in the system prior to 1993. Each alliance was formed in the
years immediately following World War II and contained provisions for the mainte
nance of borders and collective security in the event of attack. The OAS, for example,
contained Article 21, providing for the resolution of borders,20 and Article 28, which
affirmed a common defense.21 The North Atlantic Treaty was signed two years later
and provided European states with similar pledges for collective defense and the
pacific resolution of disputes among members. Article 5 contains the collective secu
rity clause, while Article 1 pledges peaceful dispute resolution among alliance par
ties.22 These clauses also mirror language found in the solely European alliance signed

20. Obviously, the United States and other members of the OAS sometimes targeted the leadership of
alliance member states. However, the type of targeting usually employed (assassinations and surreptitiously
fomenting rebellion) would not necessarily affect our argument as these acts do not constitute direct threats
to core territories, necessitating the building of large land armies capable of capturing the state. Only revi
sionism by bordering states controls this process in threatened regimes.
21. Article 28 of the charter contains the following provisions: "Every act of aggression by a State
against the territorial integrity or the inviolability of the territory or against the sovereignty or political inde
pendence of an American State shall be considered an act of aggression against the other American States."
This collective security agreement was a pledge to maintain the borders settled by Article 21 : "The territory
of a State is inviolable; it may not be the object, even temporarily, of military occupation or of other measures
of force taken by another State, directly or indirectly, on any grounds whatever. No territorial acquisitions or
special advantages obtained either by force or by other means of coercion shall be recognized!'
22. Article 5 of NATO is as follows: "The Parties agree that an armed attack against one or more of
them in Europe or North America shall be considered an attack against them all." Article 1 of NATO pledges

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 147

the prior year; Article 4 of that treaty commits to collective security, while Article 8
calls for the peaceful settlement of disputes among members.23
Given these treaty terms, we have some evidence that the large, regional pacts were
indeed meant to reduce territorial threat among alliance member states, and this is con
sistent with the first part of our logic regarding the process toward democratization. To
determine if the treaty terms had their intended effect, we conducted a logistic regres
sion that assesses whether the presence of a defense pact with all contiguous neighbors
decreases the likelihood of being targeted by a militarized dispute over territory by one
of its contiguous neighbors.24 The results of this analysis are presented in Table 4.
We estimated the same model over two different temporal domains, one model for
the post-World War II period and one for the entire 1816-2000 era.25 International bor
ders stabilized via defense pacts have a statistically significant (p < .001) and negative
effect on the likelihood of being targeted by a territorial MID in both models. Democ
racies are less likely to be the targets of territorial disputes over the entire 1816-2000
time period, but the effect of democracy is not statistically significant in the post-1945
model. Conversely, increases in the level of regional democracy increase the likeli
hood of being targeted by territorial disputes over the entire time period, but this rela
tionship reverses direction for the post-1945 model. Finally, increases in state capabil
ities also increase the likelihood of being targeted by a territorial dispute.26
Our results demonstrate that borders stabilized via defense pacts affect the peace in
an important way. Having defense pacts with all contiguous states decreases the likeli
hood of being targeted by territorial disputes, and this reduces the overall level of
threat to the state. This effect holds even after controlling for regime type and the

parties "to settle any international dispute in which they may be involved by peaceful means in such a manner
that international peace and security and justice are not endangered, and to refrain in their international rela
tions from the threat or use of force."
23. Article 4 of the "Treaty Between Belgium, France, Luxembourg, the Netherlands, and the United
Kingdom of Great Britain and Northern Ireland" stipulates, "If any of the High Contracting Parties should be
the object of an armed attack in Europe, the other High Contracting Parties will, in accordance with the pro
visions of Article 51 of the Charter of the United Nations, afford the Party so attacked all the military and
other aid and assistance in their power." Article 8 specifies procedures for International Court of Justice
jurisdiction over disputes as well as additional measures for mediation among alliance members.
24. We exclude noncontiguous territorial disputes because these are unlikely to generate the same
type of militarization process we identify in our theory.
25. An examination of our data showed that one state is an extreme outlier in the data set; we therefore
added a dummy variable to the model controlling for the presence of Israel. This dummy variable has a
strong, positive influence on the dependent variable, as Israel has been targeted by a large number of territo
rial disputes since its inception in 1947. We provide estimates of the coefficients with this dummy added, but
our results for stable borders do not change if the model is estimated without controlling for Israel. The
democracy variable, however, does switch signs, meaning that not controlling for Israel leads to the conclu
sion that democracies as a whole are likely to be targeted by territorial disputes.
26. One could argue that regional organizations also served as third-party mediators in decreasing
both the seriousness and number of territorial disputes, thereby eliminating the sources of territorial threat
that hamper democratization. The OAS, for example, has explicit provisions that allow the organization to
provide good offices and other mediation roles if member states request assistance. Despite these treaty pro
visions, however, OAS mediation was attempted in only three OAS states (Honduras, El Salvador, and Nica
ragua); two of these three states (Honduras and El Salvador) were threatened by territorial disputes (Huth
and Allee 2003) or claims (Hensel 2001) at the time. This suggests that mediation by the regional organiza
tion is not solely responsible for the elimination of territorial claims and disputes in the Western Hemisphere.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
148 JOURNAL OF CONFLICT RESOLUTION

TABLE 5
Ordinary Least Squares Estimates of Level of Militarization in the State

Ln(Military Ln(Military
Personnel/ Personnel/
Total Population), Total Population),
1816-2000 1946-2000

Independent variables
Presence of defense pact with all contiguous neighbors -0.204 (.018)*** -0.423 (.02
Democratic state (> 5 on combined Polity IV scale) 0.065 (.025)** -0.146 (.030
Percentage of Correlates of War (COW) region that
is democratic 0.419 (.043)*** -0.203 (.053)***
COW Composite Index of National Capabilities score 4.022 (.178)***
Ln(GDP) 0.554 (.012)***
Participating in militarized dispute 0.270 (.019)*** 0.306 (.02
Constant -5.388 (.014)*** -9.324 (.096)***
n 15,112 8,336
R2 0.082 0.290
NOTE: Dependent variable: Ln(military personnel/total population). Temporal do
1946-2000. Robust standard errors are reported in parentheses next to the coefficient
**/?<.05. ***/?<.01.

effects of regional democracy. We believe this may point to one me


"zones of peace" can be created. The next part of our argument tests
our theory?whether stabilized borders also decrease the level of mil
state.
Recall that we measure militarization as the percentage of the total population in
full-time military service; using the Correlates of War CINC measure, we divide the
military personnel measure by the total population and use the natural logarithm of the
ratio. We estimate our stabilized border variable, the presence of democracy, the
regional democracy measure, two different measures of capabilities, and the presence
of an MID as predictors of the militarization ratio. The results of our analyses are
presented in Table 5.
Consistent with our argument regarding territorial threat, the stabilization of bor
ders via defense pacts decreases the level of state militarization in both models. The
presence of democracy and regional levels of democracy only decrease militarization
in the post-1945 model; employing the full data set produces a positive relationship
between democracy, regional democracy, and the ratio of military personnel. Finally,
increases in GDP or the CINC measure also increase state militarization, as does par
ticipation in a militarized dispute.27 The differences between these models are most

27. These results converge well with the literature on military expenditure free riding in NATO. For
example, we know that the small economy states of NATO spent proportionately less on defense spending
compared to the United States, at least until 1966 and probably throughout the 1970s as well (Russett 1970;
Sandier and Forbes 1980; Oneal and Diehl 1994); the smaller states were advantaged because the United
States was providing a larger share of alliance resources?the public good of deterrence against the Soviets
(Olson and Zeckhauser 1966). In economic terms, the United States spent 8.1 percent of its gross domestic

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford/ALLIANCES, THEN DEMOCRACY 149

likely the result of colonial influences?the few democracies prior to 1945 often held
far-flung territories, necessitating an increased number of military personnel to main
tain these interests. Of course, as these militaries were stationed abroad, this type of
militarization would not pose the same type of regime costs as those of a standing
army, forced by external threat.
We believe that these results pose an interesting puzzle for current research on dem
ocratic clustering (Gleditsch 2002; Thompson 1996). If regional democracy does con
trol the success of democratic transitions, this process does not occur due to decreases
in the overall level of threat to the state, where "zones of peace" foster nascent democ
racies. At least this process does not occur once controls are added for borders stabi
lized through alliances.
Thus far, we have provided evidence that several regional defense pacts contained
clauses providing for the stabilization of the borders of their member states, and the
results in Tables 4 and 5 demonstrate that the presence of these defense pacts on each
border often reduces the level of territorial threat to the state. Reduced territorial threat
is important because of its association with a reduction in the overall level of state
militarization, and according to our theory, the military often represents a major
impediment to democratic transitions, representing a "big gun" capable of dominating
the state. The final step in testing our theory, then, is to more systematically evaluate
whether these reductions in territorial threat also led to democratization.
To do this, we estimate two separate hazard models that assess the likelihood of
regime transition in a state-year. We separate the population of state-years to include
only autocratic state-years (those states scoring less than 6 on the combined autocracy
democracy Polity IV scale) and all first-year democratic states; the latter represent
successful democratic transitions for this sample. We estimate both models with con
trols for right censoring?when state-year transitions are still possible but the tempo
ral domain expired. Coefficients with a magnitude less than 1 can be interpreted as
suggesting the variable inhibits democratic transitions.

product (GDP) on defense, while Western Europe was averaging less than 4 percent between the years 1950
and 1986 (Oneal and Diehl 1994, 387). Oneal and Diehl (1994) show that, although Olson and Zeckhauser
(1966) correctly model the correlation between economic capability and defense burden (predicting the
presence of shirkers, for example), NATO military spending as a whole was responsive to increases in Soviet
threat while individual alliance members also reacted to their private security interests.
The responsiveness of NATO-partner military spending to both Soviet threat and the dictates of private
security interests matches well our contentions regarding the role of the military in society. Because the
United States was capable of providing the high military spending necessary to match the Soviet threat, the
European partners were able to maintain nascent democracies without the threat from external foes and
internal institutional struggles, and since the United States was a mature democracy, it suffered little institu
tional harm from increasing its military posture; as Boix (2003) argues, democracy is a self-sustaining equi
librium outcome even in the face of changes in the initial social conditions that made the transition possible.
As private and Soviet threats grew toward the end of the 1960s, the individual members increased their rela
tive defense burdens, but by this time, group competition had been largely institutionalized.
Interestingly, Oneal (1990) notes that the exceptions to the public goods model during the late 1960s
and early 1970s were most likely caused by three countries. Portugal increased its military spending in
response to colonial independence wars, while the rivalry between Greece and Turkey over Cyprus and other
territories influenced much of their spending. In these exceptions, we find added support for our argument as
all three countries were not democratic at the time.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
150 JOURNAL OF CONFLICT RESOLUTION

TABLE 6
Cox Hazard Model Estimates of Regime Transitions

Transition to Transition to
Democracy, Democracy,
1816-2000 1946-2000

Independent variables
Defense pact with all neighbors 2.062*** 1.781*
Militarized interstate dispute involvement 0.964 1.226
Log of gross domestic product 1.027
Log of population 1.289***
Composite Index of National Capabilities score 1.405*
Major oil producer 0.413*
Predominantly Muslim country 0.997
Percentage of region that is democratic 33.210*** 38.196***

Log-likelihood -509.236 -353.815


Likelihood ratio % (7) 109.72*** 84.18***
Number of observations 163 countries, 128 countries,
119 transitions; 87 transitions;
8,385 total observations 3,951 total observations

NOTE: Dependent variable: transition to democracy. Temporal domain: 1816-2000 and 1946-2000.

The results of both hazard models are listed in Table 6, and once again as we
expected, having a defense pact with all neighboring states decreases the time before a
successful transition to democracy occurs. This result is statistically significant (p <
.05) in both models, even after controlling for the strong positive effect of regional
democracy (p < .001). Interestingly, and consistent with the literature on democratiza
tion (Ross 2001), the presence of oil hampers transitions. We believe this also fits with
our "big gun" theory, as control of a major oil industry can provide enough capital to
successfully dominate the state. Capability levels, measured by CINC score or popula
tion estimates, also increase the likelihood of democratic transitions.
These results are obviously supportive of our contention that large, regional
defense pacts strongly influence the domestic institutions of allied states. We have not
demonstrated, nor do we argue, that all defense pacts lead to democratization. Instead,
we show that a large number of states are democratizing within multilateral regional
pacts, and we suggest the plausible hypothesis that these two events are connected.
Current cooperation theory cannot explain why joint democracies do not form alli
ances but are likely to be allied, and we have presented an alternative theory that
explains this relationship. Our theory consistently explains why stable borders
decrease territorial threat and the level of militarization of the state while also connect
ing these processes to observed increases in the likelihood of democratic transitions.
Of course, these results are tentative. We suffer from the same small-TV bias for
which we criticize the current alliance literature?only a few cases of alliances are sta
bilizing state borders. Nevertheless, the consistent results across each link in our

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford/ALLIANCES, THEN DEMOCRACY 151

causal logic provide empirical support connecting territorial threat to regime transi
tions and alliances as a mechanism of democratic diffusion and border stability.

CONCLUSION

In this article, we find that democratic dyads are unlikely to form alliances but that
they are likely to be allied, and we argue that these results are linked to the removal of
territorial threat by defense pacts. This implies that alliances may be uniquely able to
shed light on the observations that democracies tend to cluster regionally (Gleditsch
2002) and that the strength of the democratic community is correlated with transitions
to democracy, as their international nature makes borders likely candidates for the dif
fusion of both certain types of conflict and regime type. If alliances function as diffu
sion agents for democracy, then the association between neighboring democracies and
democratization may be a by-product of stabilizing borders, ensured by mutual
defense agreements. Our results also contribute to the debate on alliance reliability by
casting a different light on the question of whether regimes differ in their utility for
signing and participating in alliance commitments (Gartzke and Gleditsch 2004); our
results suggest that, in many cases, the choice to ally with a democracy may not be stra
tegic in the sense that democracies are chosen or avoided as alliance partners based on
their expected behavior, as some democratic allies join the alliance as non
democracies, causing these alliances to become democratic over time.
Alliance researchers should also consider our argument that the alliance-democracy
relationship is driven statistically by three large regional defense pacts?NATO, OAS,
and a European defense pact?all formed in the years following World War II. Theo
ries about the reliability or duration of democratic alliances, or any theories that sug
gest uniquely democratic alliance behaviors, must all infer their results from an
incredibly small number of cases, as most alliances are formed by nondemocratic
states and nearly every alliance contains a majority of nondemocratic dyads.

REFERENCES

Bennett, D. Scott. 1997. Testing alternative models of alliance duration, 1816-1984. American Journal of
Political Science 41 (3): 846-78.
Boix, Carles. 2003. Democracy and redistribution. New York: Cambridge University Press.
Bremer, Stuart. 1992. Dangerous dyads: Conditions affecting the likelihood of interstate war, 1816-1965.
Journal of Conflict Resolution 36 (2): 309-41.
Bueno de Mesquita, Bruce. 1981. The war trap. New Haven, CT: Yale University Press.
Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Siverson, and James D. Morrow. 2003. The logic of
political survival. Cambridge, MA: MIT Press.
Desch, Michael. 1996. War and strong states, peace and weak states? International Organization 50:237-68.
Fearon, James D., and David D. Laitin. 2003. Ethnicity, insurgency, and civil war. American Political Sci
ence Review 97 (1): 75-90.
Freid, Morton H. 1961. Warfare, military organization, and the evolution of society. Anthropologia 3:134-47.
Gartzke, Erik, and Kristian Skrede Gleditsch. 2004. Why democracies may actually be less reliable allies.
American Journal of Political Science 48 (4): 775-95.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
152 JOURNAL OF CONFLICT RESOLUTION

Gibier, Douglas M. 1996. Alliances that never balance: The territorial settlement treaty. Conflict Manage
ment and Peace Science 16 (1): 15-91.
-. 1997. Control the issues, control the conflict: The effects of alliances that settle territorial issues on
interstate rivalries. International Interactions 22 (4): 341-68.
-. 2005. Bordering on peace. Mimeo, University of Kentucky.
Gibier, Douglas M., and Meredith Reid Sarkees. 2004. Measuring alliances: The Correlates of War formal
interstate alliance dataset, 1816-2000. Journal of Peace Research 41 (2): 211-22.
Gibier, Douglas M., and John A. Vasquez. 1998. Uncovering the dangerous alliances, 1495-1980. Interna
tional Studies Quarterly 42 (4): 785-807.
Gleditsch, Kristan Skrede. 2002. All international politics is local. Ann Arbor: University of Michigan
Press.
Gowa, Joanne. 1994.Allies, adversaries, and international trade. Princeton, NJ: Princeton University Press.
Hensel, Paul. 2001. Contentious issues and world politics: The management of territorial claims in the
Americas, 1816-1992. International Studies Quarterly 45 (1): 81-109.
Hintze, Otto. [1906] 1975. Military organization and state organization. In The historical essays of Otto
Hintze, edited by F. Gilbert. New York: Oxford University Press.
Holsti, Kalevi J. 1991. Peace and war: Armed conflicts and international order, 1648-1989. Cambridge,
UK: Cambridge University Press.
Howard, Michael. 1979. War and the nation-state. Daedalus 108 (4): 101-10.
Huth, Paul K. 1996. Standing your ground. Ann Arbor: University of Michigan Press.
Huth, Paul K., and Todd Allee. 2003. The democratic peace and territorial conflict in the twentieth century.
New York: Cambridge University Press.
Ikenberry, G. J. 2001. After victory: Institutions, strategic restraint, and the rebuilding of order after major
wars. Princeton, NJ: Princeton University Press.
Kadera, Kelly M., Mark J. C. Crescenzi, and Megan L. Shannon. 2003. Democratic survival, peace, and war
in the international system. American Journal of Political Science 47 (2): 234-47.
Lai, Brian, and Dan Reiter. 2000. Democracy, political similarity, and international alliances, 1816-1992.
Journal of Conflict Resolution 44 (2): 203-27.
Leeds, Brett Ashley. 1999. Domestic political institutions, credible commitments, and international cooper
ation. American Journal of Political Science 43:979-1002.
-. 2003. Alliance reliability in times of war: Explaining state decisions to violate treaties. Interna
tional Organization 57 (4): 801-29.
Leeds, Brett Ashley, and Jennifer Gigliotti-Labay. 2003. You can count on me? Democracy and alliance reli
ability. Paper presented at the annual meeting of the American Political Science Association, September,
Philadelphia.
Leeds, Brett Ashley, Jeffrey M. Ritter, Sara McLaughlin Mitchell, and Andrew G. Long. 2002. Alliance
treaty obligations and provisions, 1815-1944. International Interactions 28 (3): 237-60.
Lipson, Charles. 2003. Reliable partners. Princeton, NJ: Princeton University Press.
Long, Andrew G. 2003. Defense pacts and international trade. Journal of Peace Research 40:537-52.
Mann, Michael. 1988. States, war, and capitalism: Studies in political sociology. Cambridge, UK: Basil
Blackwell.
Mansfield, Edward D., and Jack Snyder. 1995. Democratization and the danger of war. International Secu
rity 20 (1): 5-3*.
-. 2002. Democratic transitions, institutional strength, and war. International Organization 56 (2):
297-337.
Maoz, Zeev. 1989. Joining the club of nations: Political development and international conflict, 1816-1976.
International Studies Quarterly 33 (2): 199-231.
Mitchell, Sara M., and Brandon Pries. 1999. Beyond territorial contiguity: An examination of the issues
underlying democratic interstate disputes. International Studies Quarterly 43 (I): 169-33.
Morrow, James D. 1991. Alliances and asymmetry : An alternative to the capability aggregation model of
alliances, American Journal of Political Science 35:904-33.
Olson, Mancur. 1993. Democracy, dictatorship, and development. American Political Science Review
87:567-76.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms
Gibier, Wolford / ALLIANCES, THEN DEMOCRACY 153

Olson, Mancur, and Richard Zeckhauser. 1966. An economic theory of alliances. Review of Economics and
Statistics 48:266-79.
Oneal, John R. 1990. The theory of collective action and burden sharing in NATO. International Organiza
tion 44:379-402.
Oneal, John R., and Paul F. Diehl. 1994. The theory of collective action and NATO defense burdens: New
empirical tests. Political Research Quarterly 48 (June): 373-96.
Oneal, John R., and Bruce Russett. 1997. The classical liberals were right: Democracy, interdependence, and
conflict, 1950-1985. International Studies Quarterly 4:267-94.
Organski, A. F. K., and Jacek Kugler. 1980. The war ledger. Chicago: University of Chicago Press.
Pevehouse, Jon C. 2002. Democracy from the outside in? International organizations and democratization.
International Organization 56:515-49.
Powers, Kathy L. 2001. International institutions, trade and conflict: African regional trade agreements from
1950-1992. Ph.D. diss., Ohio State University.
Przeworski, Adam. 1988. Democracy as a contingent outcome of conflict. In Constitutionalism and democ
racy, edited by J. Elster and R. Slagstad. New York: Cambridge University Press.
Przeworski, Adam, and Fernando Limongi. 1997. Modernization: Theories and facts. World Politics 49 (Jan
uary): 155-83.
Rasier, Karen A., and William Thompson. 1989. War and state-making: The shaping of global powers.
Boston: Unwin and Hyman.
Ray, James Lee. 1990. Friends as foes: International conflict and wars between formal allies. In Prisoners of
war, edited by Charles S. Gochman and Alan Ned Sabrosky, 73-92. Lexington, MA: Lexington Books.
-. 2003. Explaining interstate conflict and war: What should be controlled for? Conflict Management
and Peace Science 20 (Fall): 1-31.
Reed, William. 1997. Alliance duration and democracy: An extension and cross-validation of "democratic
states and commitment in international relations." American Journal of Political Science 41 (3): 1072
78.
-. 2000. A unified statistical model of conflict onset and escalation. American Journal of Political Sci
ence 44 (\): M-93.
Reiter, Dan. 1996. Crucible of beliefs: Learning, alliances, and world wars. Ithaca, NY: Cornell University
Press.
-. 2001-2002. NATO and democracy. International Security 26 (Winter): 230-5.
Ross, Michael. 2001. Does oil hinder democracy? World Politics 53 (April): 325-61.
Russett, Bruce M. 1970. What price vigilance? New Haven, CT: Yale University Press.
Sandier, Todd, and John F. Forbes. 1980. Burden sharing, strategy, and the design of NATO. Economic
Inquiry 18:425-47.
Simmons, Beth A. 2000. International law and state behavior: Commitment and compliance in international
monetary affairs. American Political Science Review 94 (4): 819-35.
Simon, Michael W., and Erik Gartzke. 1996. Political system similarity and the choice of allies: Do democ
racies flock together or do opposites attract? Journal of Conflict Resolution 40 (4): 617-35.
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 (2): 285-306.
Thompson, William. 1996. Democracy and peace: Putting the cart before the horse? International Organiza
tion 50(1): 1996.
Tilly, Charles. 1990. Coercion, capital, and european states, A.D. 990-1990. Cambridge, UK: Basil
Blackwell.
Titmuss, Richard. 1959. Essays on "the welfare state." New Haven, CT: Yale University Press.
Vanhanen, Tatu. 1990. The process of democratization: A comparative study of 147 states, 1980-88. New
York: Crane Russak.
Vasquez, John A. 1987. The steps to war: Toward a scientific explanation of Correlates of War findings.
World Politics 40:108-45.
-. 1993. The war puzzle. Cambridge, UK: Cambridge University Press.

This content downloaded from 129.186.253.194 on Tue, 03 Dec 2019 01:18:16 UTC
All use subject to https://about.jstor.org/terms

You might also like