OUP user menu

Intergovernmental Organizations (IGOs) and Interstate Conflict: Parsing Out IGO Effects for Alternative Dimensions of Conflict in Postcommunist Space

Elizabeth Fausett, Thomas J. Volgy
DOI: http://dx.doi.org/10.1111/j.1468-2478.2009.00578.x 79-101 First published online: 1 March 2010


While the relationship between intergovernmental organizations (IGOs) and conflict has captured the attention of international relations scholars for decades, the empirical results of this research agenda have presented contradictory conclusions regarding the pacifying effect normally attributed to IGOs. We address these contradictions by refocusing primarily on potential IGO effects on low-severity conflicts. We examine new states in the postcommunist space spanning Europe and Central Asia as a useful research site to explore these relationships in the post-Cold War era. We argue that especially in the case of newly emerging states, where there is little institutional memory and long-term experience in foreign affairs, IGOs expose differential policy preferences between members, and such information should be associated with the likelihood of increased low levels of conflict. We find a strong association between shared IGO membership and low severity conflict, a significant relationship between low and high severity conflict, and differences between IGO membership effects on low versus high severity conflict, consistent with our theoretical argument.

On January 1, 2006, Gazprom, the Russian state-owned oil company, substantially curtailed the amount of gas it supplied to Ukraine. The cutback came after Russia requested and Ukraine refused to pay $230 per 1,000 cubic meters of gas, a striking increase from the prior fee of $50. The change in resource policy toward Ukraine followed the election of a pro-Western Ukrainian president, and left little doubt that the clash over energy represented both an economic and a political conflict between Russia and its former territory (BBC News 2006). The issue between the two parties coincided with growing conflicts between Russia, several of the newly formed states of the former Soviet Union (for example, the Baltic States), and those of its former allies that had migrated into the EU (for example, Poland). In addition, the Ukrainian-Russian conflict led to increasing concerns among the more established EU states about growing EU dependence on Russia's natural resources and increasing Russian belligerence toward the “near abroad,” the EU and the United States.

Events such as the Gazprom conflict constitute substantial issues of concern in international politics, and with important consequences for numerous states, both directly and indirectly affected by the conflict. Yet, much of the systematic, large-N analyses of conflict have focused only on the most severe cases, such as militarized interstate disputes (MIDs), wars, and crises. Cases such as the one noted above, despite their importance, would not register in any of those databases, and are typically not accounted for by most conflict models. Concern with war and military interactions among states has dominated a large portion of international relations research, and because of their destructive potential, rightly so. Yet, MIDs and wars represent an extremely small part of the range of conflicts appearing between states. As Figure 1 illustrates, between 1990 and 2000, there were 59 episodes of militarized disputes (MIDs) between states that were either allies of the former Soviet Union or emerged as new republics from a disintegrated USSR (Ghosn and Palmer 2003). At the same time, Reuters captured more than 5,226 occurrences of conflictual behavior between these same states during the same 10-year time span.1


Count Comparisons of Militarized Interstate Disputes and the IDEA Events within the Former Soviet Sphere of Influence, 1990–2000

We assume that the study of conflicts between states that encompass a broader range of behaviors than the most conflictual of events is a worthy focus of scholarly inquiry, and we are far from the first to make this claim (for recent examples, see Pevehouse 2004; Dixon 2007; Hensel, Mitchell, Sowers, and Thyne 2008). Our work is particularly influenced by an earlier effort (Kinsella and Russett 2002) that created a conflict model analyzing MIDs within a broader context of less severe conflicts.

While conflicts that do not contain coercive components represent a range of concerns between states that may be worthy of scholarly focus for their substantive implications alone, we suggest another rationale for focusing on lower as well as higher levels of conflicts. It is quite plausible that the antecedents of many severe conflicts in international politics lie in lower level disputes and conflicts between states that, when unchecked, can escalate to much higher levels of severity. Thus, studying the dynamics associated with lower level conflicts may help us to understand better one path to conflicts of high severity and security threats for states.

Numerous studies have examined the pacifying effects of intergovernmental organization (IGO) membership on high levels of interstate conflict, albeit with mixed results.2 This effort seeks to contribute to that literature in two ways. First, while much of the research has explored the effects of IGOs on high severity conflict, we focus primarily on IGO effects on stimulating low severity conflict, which in turn may increase higher levels of hostility between states. Second, we examine this relationship for newly emerging states in a region experiencing the turbulence of post-Cold War international politics.

Why Would IGOs Stimulate Conflict between Members?

As much of the IGO literature has noted, IGOs can be very effective in mediating among conflicting parties, reducing uncertainty by conveying information, assisting in problem solving, helping to expand states’ conceptions of self interest, helping to integrate states into regional and global norms (Russett, Oneal, and Davis 1998:444–445), providing organizational mechanisms through which states can pursue their objectives more efficiently (Katzenstein, Keohane, and Krasner 1998), and socializing states into the modern international system (Thomas, Meyer, Ramirez, and Boli 1987; Finnemore 1996; Schimmelfennig 2005).3 Integration into the constellation of IGOs providing such functions should have positive effects on conflict amelioration between states, including newly emerging ones.

Yet, we argue that membership in IGOs can be a double-edged sword and especially so by inducing low severity conflicts as a function of joint IGO membership between states. We begin with the assumption that conflicts between two states (A and B) do not occur in vacuum, but stem from a series of factors that become manifest in divergent policy preferences of mutual salience. Differences between A and B's policy preferences4 may stem from numerous causes, including domestic politics and foreign policy interests and the goals of each. In the Gazprom case, the conflict between Ukraine and Russia in part stemmed from Russian reaction to changing Ukrainian domestic political dynamics and their potential consequences for Ukrainian foreign policy preferences toward both Russia and the West.

Divergent policy preferences alone, depending on their salience, may not escalate to low severity conflict. Some dyads have divergent policy preferences but as they either seldom interact or appear relevant to each other, divergence in preferences is unimportant, and this is one reason why large N empirical studies focus on “politically relevant” dyads, where divergent preferences would likely matter. Contiguity appears important for political relevance as most states pay more attention to their neighbors compared to others whose activities outside of the neighborhood may carry fewer consequences.5 Just as important, the salience of divergent preferences between actors is in part a function of prior experiences and institutional memory embedded in states’ foreign policy infrastructures (Crescenzi 2007). Institutional memory and historical patterns of foreign policy experience act as filters or buffers for state interactions, creating some path dependency through preexisting reputations and histories institutionalized between A and B.

Divergent policy preferences, even if between politically relevant dyads, may not lead automatically to conflictual behavior between the two parties. Factors exogenous to the dyad may dampen or highlight policy differences. The level of turbulence in the neighborhood may be one such factor. Where security and economic relations are stable and pacific, such as in the EU core, we would expect far less amplification of policy differences than in the more turbulent space of the Middle East, or in the postcommunist space of emerging new states and considerable uncertainty over territory, resources, and ethnic and national identity. Of course what is turbulence in the neighborhood and how a state adapts to it is also likely to be a function of past experiences with previous disturbances embedded in institutional memory (assuming that such institutional memory exists and is available to policymakers).

The literature on intergovernmental organizations points to comembership in IGOs between A and B as having a significant effect on their conflict behavior, by dampening the conflict flowing from strongly divergent policy preferences, or by ameliorating the linkage between divergent policy preferences and the expression of conflict, or indirectly by minimizing the divergence in their policy preferences. However, virtually all of this literature focuses on the conflict inhibiting effects of IGOs regarding high severity conflicts. For example, the number of organizational memberships within a dyad is associated with a decreased likelihood of militarized interstate disputes (Russett et al. 1998) while preferential trade agreements (PTAs) are likely to depress high severity conflicts between members with heightened trade (Mansfield and Pevehouse 2000). Meanwhile recent literature is persuasive in arguing that strong institutions, those with design features that contain extensive legal mechanisms, bureaucratic organizations, relative independence, and/or decisional mechanisms requiring substantial participation by state policymakers, seem to have stronger negative effects on major conflicts between states than institutions designed with more fragile organizational mechanisms (McCall Smith 2000; Boehmer et al. 2004; Gartzke, Nordstrom, Boehmer, and Hewitt 2005; Haftel 2007). In fact, some recent scholarship has argued that failing to parse out the effects of strong institutions leads to findings that may either wash out the effects of organizational membership, or worse still, indicate positive relationships between organizational membership and conflict, by watering down the effects of “strong” IGOs in models of conflict (for example, see Gartzke, Li, and Boehmer 2001; Kinsella and Russett 2002).

We argue that generally IGO comembership may have four (non-mutually exclusive) effects on conflicts between A and B. First, an IGO may provide a substitute arena (for example, the WTO) for pursuing disputes that arise out of policy differences and goals (Dixon 2007). To the extent that an IGO is designed to address and contain such disputes, and offers an alternative for direct conflicts between the two parties outside of the organizational setting, it can ameliorate conflicts and perhaps even resolve differences amicably. Potentially, such substitution effects are available for both high and low severity conflicts, but likely occurs only when policy divergence between A and B is clear and such divergence is not easily addressed outside of the institutional setting.6

Second, various IGOs with strong design features are able to provide mechanisms to mediate conflicts between members, whether or not those conflicts arise inside or outside of the organizational setting (Powers and Goertz 2006). This role is played once conflicts have typically reached a fairly high level of severity, and are less likely to be of relevance to low severity conflicts.

Third, IGO membership may yield a socialization function, especially for newer states, socializing them into the norms of the region or the global community of states (Thomas et al. 1987; Russett et al. 1998:444–445). Presumably, this effect, and again depending on the type of organization and how it is designed, is not automatic, takes effect over a considerable period of time, and while it may apply to broad norms of conduct, it may not eliminate a large range of policy differences between A and B and their escalation to at least low levels of conflict.

Finally, and most important for our concerns, IGOs should have an impact on their members by providing information relevant to the pursuit of their foreign policy objectives in general and more specifically about the divergence in policy preferences between states (along with the salience of those policy preferences). By participating in IGOs, members can generate more information from two different sources. One is through reporting by the bureaucracy about matters falling under the scope and jurisdiction of the organization. Second, and perhaps just as important, through routinized interactions, A and B are exposed to substantial amounts of information about each others’ policy preferences and the salience of those preferences. Such increased opportunities for more information about each other through greater interaction can clarify policy similarities or differences, and can amplify policy differences, making the dyad politically more relevant. While increased information is often treated in the literature as potentially ameliorating conflict between members, it is just as feasible that information received through the constellation of IGOs creates familiarity that may breed discontent and highlight differences of interests, policies, and values. While greater information gained about capabilities and intentions between A and B under conditions of high conflict may help to contain conflicts of high severity, it is equally likely that additional information about policy preferences may create greater friction at lower levels of conflict. The negative effects of such information about capabilities and intentions may be particularly the case when institutional memory filters are not present.

The process we describe would not be unusual for students of legislative politics in democratic political systems where two routes exist through which interactions between legislators can generate greater conflict: interest articulation and interest aggregation.7 The “norms of bounded competition” (Dixon 1994) and institutional arrangements (Bueno de Mesquita 2003) in democratic political systems create dispute resolution mechanisms that allow substantial conflicts and deep policy cleavages to be resolved without the use of force. However, their legislative institutions neither eliminate competition nor conflicts within the political system; on the contrary, they encourage competition and representations of the rich diversity of intra-societal policy preferences within political institutions (at least at the legislative level), including issues that are not easily amenable to compromise.

Within both democratic domestic political systems and within intergovernmental organizations clear processes exist to encourage interest articulation—the representation of broad and substantial differences in interests, values, and policy preferences—a process through which the clarification and amplification of policy differences creates some conflict between members. The existence of interest articulation in intergovernmental organizations is true by definition: there is no reason for the existence of these institutions if they cannot reflect the interests of their members.8 The process of interest articulation can highlight similarities in policies preferences in IGOs, but it is just as likely to identify differences between two states, to make the A-B dyad more politically relevant, and to exacerbate conflict within the dyad.9 Furthermore, the strength of interest articulation by A or B may serve to further amplify initial differences.10

It is also likely that in many IGOs (as actors do in domestic, democratic institutions) members pursue some version of interest aggregation: searching for partners, allies, and supporters for their side, by trying to “capture” (Krebs 1999) the organization or at least membership support for their policy preferences. Under conditions of divergent policy preferences between A and B, B's search for public support for itself and/or for condemnation of A's position is likely to amplify differences further and increase the potential for conflict within the dyad. This may have been the case between Greece and Turkey through their joint involvement in NATO (Krebs 1999).11

Thus, as IGOs provide for their members forums for state interaction, states, by articulating and seeking support for their interests, convey information about the content and salience of their policy preferences and are able to learn about the policy preferences of other members. The generation of such information is likely to create opportunities for increased, albeit low severity conflicts between states as divergent preferences are exposed. These effects may be amplified especially for newer states that lack the institutional memory and history of prior relationships that could filter the information being received through IGOs. While it may be the case that proximity increases the probability of “at least occasional confrontation and conflict” (Waltz 1970:205–222) for all states, this may be particularly true for states with limited previous experience with each other.

Alternatively, it is plausible that information received through interactions within IGOs could reveal not only divergent preferences but also bargaining space for cooperation and compromise, providing opportunities for conflict avoidance even when divergent preferences are exposed.12 While this argument is plausible as well, we think it is less likely for states with little or no experience in international affairs. Nevertheless, the empirical test we offer below can address both arguments, by restricting the research design to newly emerging states, and testing for information effects by observing whether or not a positive versus a negative relationship is found for low severity conflicts.

The literature on the linkage between high severity conflict and IGO membership has increasingly narrowed the scope of effects to those organizations with strong institutional design characteristics. We suggest that this strategy is inappropriate when focusing on IGO effects on low severity conflicts. States can and do join all types of IGOs, including those with weak design characteristics. Weaker organizations may not be strong enough to ameliorate high severity conflicts between states, but they may not be neutral in the development of low severity conflicts between states. We suggest that, ceteris paribus, to the extent that weakly structured IGOs function to expose states to differences of preferences, interests, and values (and/or to amplify existing differences) they are likely to increase at least lower levels of conflicts between members.

Thus, we expect that both strongly designed13 and weakly designed organizational structures—regardless of the extent to which they may vary in ameliorating high levels of conflicts between states—should be similar in affording members opportunities to find additional information to highlight differences in policy preferences, objectives, and styles of engagement, all of which may lead to at least low levels of conflict with each other as a function of their mutual participation in these organizations. If in fact most of these organizations allow a relatively broad range of interest articulation and some aggregation by members, then the likelihood of finding differences (or reinforcing and amplifying perceptions of differences) in preferences, objectives, and styles—as a function of joint participation and interaction—should be one outcome of organizational membership and yield at least higher frequencies of low levels of conflict within dyads. For states relatively new to the international arena, and bereft of much institutionalized memory about potential friends and enemies, this effect should be particularly salient and should function as well to stimulate higher levels of conflicts between these states.

The explanation we create for IGO comembership producing increased amounts of low severity conflict for dyads is silent on whether or not such IGO comembership may act as well to reduce (or increase) high severity conflict for the same dyads. Our silence is due to two factors: first, our theory is not designed to address the direct impact of IGOs on high severity conflict within dyads. Second, we are interested in dyads where institutional memory has yet to be developed in emerging states, and there is little in the literature addressing the high severity conflict ameliorating functions of IGOs for these states. Nevertheless, we compare below conflict models for low and high severity conflicts for two reasons. First, we do so in order to assess whether or not the production of low severity conflicts creates consequences for higher severity conflicts. Second, our theoretical sketch clearly suggests that IGO effects may work differently for low severity conflicts than they do for conflicts of high severity. We demonstrate both by comparing low and high severity models of conflict.

Why Focus on Postcommunist Space?

We focus our analysis on a region we call postcommunist space. It consists of three sets of actors: countries that were closely allied with the Soviet Union in Eastern Europe during the Cold War and were virtually recreated afterwards, states that emerged from the dissolution of the Soviet Union following its collapse, and those states that emerged from the former Yugoslavia, a state that was neither controlled by nor closely allied with Moscow yet was critically restructured following the end of the Cold War. Excluding a handful of microstates, the states arising in postcommunist space represent virtually all the new states created after the end of the Cold War (and within the time frame of our analysis).14

These states are of substantive, theoretical, and methodological interest for a number of reasons. First, and with the possible exception of Russia, they emerged on the international scene lacking recent (or any) experience with foreign affairs, and needing practical knowledge and information regarding interstate affairs salient for their well being. Second, all of these states were in the midst of restructuring both their domestic and foreign policies and their political systems, either through fragile democratic transformations or through equally difficult autocratic political consolidation. Third, these states are in a region of considerable turbulence and uncertainty as Soviet Union based (and enforced) institutions disappeared simultaneously with the appearance of a host of new issues including appropriate boundaries, rights of citizenship, ethnic divisions, and conflicting claims over ownership of resources. This combination of tumultuous geopolitical space, newly emerging states, and transitioning democracies can easily create a combustible mix for interstate conflict (for example, Mansfield and Snyder 1995; Bennett 2006).

From a research design perspective, the geographical space of postcommunist states provides a rich research site for our inquiry, creating a combination of not only a turbulent environment, but also one in which we are able to observe the effects of IGO participation relatively free of foreign policy institutional filters. One nearly impossible factor to control in studies of conflict revolves around the effects of institutional memory (for example, Crescenzi 2007) in a state's foreign policy apparatus filtering its foreign policy experience with other actors. Seldom do states come to participate in IGOs with a relatively clean slate of little knowledge regarding politically relevant actors in the neighborhood. By focusing on new and newly reemerging states in postcommunist space, we are able to approximate states with little institutional experience in their new, regional environment,15 and therefore we can gauge more directly the potential effects of IGO membership on low severity conflicts.16

The extent to which such institutional foreign policy memory is absent to filter relations between postcommunist states is difficult to assess systematically, but it does appear to be the case. We recognize that this is not true for Russia, which inherits the foreign policy apparatus of the former Soviet Union. Nevertheless, the foreign policy apparatus utilized by almost all of the other states was created typically by either ex-communist elites with little or no experience in foreign policy (for example, in Kazakhstan, Uzbekistan, Kyrgyzstan, Belarus, Ukraine), or where new governments sought to expunge the communist legacy (for example, Poland, Czech Republic, the new Baltic States). Few states besides Russia salvaged any institutional memory of foreign policy relationships,17 and especially ones relevant for postcommunist relations.18

While controlling for the possible path dependency toward dyadic conflict by removing institutional foreign policy memory and extensive previous experience for our sample of states, we are nevertheless creating a research site that is far from being pacific. Apart from Russian aspirations to establish order in the region, it is a site of competing economic and political interests toward the region by major powers from outside (for example, the United States, the EU, Japan, China). New states within the postcommunist space may quickly come to establish divergent preferences in political and economic policies under pressure from both outside forces and an emerging Russia. Differences in policy preferences, proclivities toward the West versus the Commonwealth of Independent States (CIS), and festering issues ranging from material resource ownership to issues of identity combine to guarantee that the potential effects of IGO membership are competing against other—substantial factors associated with conflict. Therefore, using this research site provides a combustible mix of turbulent space, rapid movement into a variety of IGOs, and a short period when foreign policy bureaucracies lack institutional memory that would otherwise condition responses to uncertainty. If the organizational structures through which these states interact have an independent effect on their conflict behavior, we should find evidence of it using this research site.19

A Conflict Model for States in Postcommunist Space

Based on our discussion, we offer two predictions regarding the relationship between membership in IGOs and the conflict activities between states in postcommunist space. Focusing on dyads, our primary hypothesis posits a positive and significant relationship between the frequency of joint IGO membership and lower levels of conflict. As these states have a limited history of interaction and virtually no institutional memory of prior relationships, we do not expect that low severity conflicts will be easily contained, and therefore hypothesize as well a significant relationship between less severe and more severe levels of conflict.

Note that we are not making a prediction between IGO membership and its effects on high severity conflict. While we employ models of low and high severity conflict (below) that include both “weak” and “strong” IGOs, we do so for conflicts of high severity in order to a) control for variables associated with the Kantian peace, and b) to demonstrate empirically that IGO membership appears to operate differently for high severity conflicts than for conflicts of low severity for the states in our sample.

In their pioneering work, Kinsella and Russett (2002) argue that a narrow focus on MIDs creates a selection effect and only through a broader modeling of both MIDs and other conflicts can researchers uncover the conflict ameliorating effects of the Kantian peace and its separate dimensions of IGOs, democratic regimes, and interdependence. They create a model based on opportunity (power ratio, proximity, and major power involvement in the dyad) and willingness (democracy, dependence, economic openness, IGO and alliance memberships) to engage in conflict, and apply the model to dyads for the 1951 to 1992 time frame. We borrow liberally20 from their effort in creating our conflict models, but we alter a number of the variables to better suit the conditions associated with states newly emerging on the regional and international stage. Most critically—as the propositions above indicate—and unlike their work, we expect that for newly emerging states both the effects of IGOs on low severity conflicts and the effects of low severity conflicts on high severity conflicts should be of salience, and in both cases associated with increased conflict.

Conflict Variables

We measure the frequency of conflict within dyads through the use of the IDEA events database. The database is generated by a computerized contextual analysis, relying on verb-recognition software, of Reuter's newsfeeds spanning the period from 1990 to 2004.21 The use of events data and the IDEA database, while not without flaws, nevertheless allows researchers to address simultaneously temporal and cross-section research designs, to begin to tackle potential causal mechanisms, and to distinguish between different levels of conflict severity.22

Conflict events are drawn from the IDEA project, aggregated to the non-directed dyadic year. First, we identify all conflict events; second, we create two separate dimensions of conflict representing those events of low severity and those of high severity. These divisions are accomplished by applying the Goldstein (1991) cooperation/conflict scale to the data, similar to Kinsella and Russett's (2002:1055) use of events data available for the Cold War era. Designed for easy interpretation, this scale extends from −10 (most conflictual) to +8.3 (most cooperative) and attempts to capture an intuitive understanding of conflict at the state level. An excerpt of the divided sample of conflictual behavior, as per the Goldstein scale, is illustrated in Appendix 1. A contextual analysis of the scale as well as a frequency mapping identified −6.8 as the most natural cut-point for the transition between low and high severity conflicts with high severity conflicts on the scale ranging from −10 to −6.8 and low severity conflicts spanning the range of −0.1 to −6.7.

These procedures generate two versions of the dependent variable. Both variables are measured in terms of their frequency—rather than their intensity—within each dyad, for any given year. The count rather than the intensity option is chosen in order to provide comparability with Kinsella and Russett's (2002) findings on dyads in dispute.23

Comembership in Intergovernmental Organizations

The primary independent variable is shared IGO membership within dyads. We use the COW IGO conceptualization24 and database (Pevehouse, Nordstrom, and Warnke 2003). We use the FIGO database (Volgy, Fausett, Grant, and Rodgers 2008) and overlay it on COW IGO to differentiate between “strong” versus “weak” organizations. As it is reasonable to assume that IGO effects are not instantaneous, we lag the dyadic membership data one year. In addition, as shared memberships are not evenly distributed and are also likely to suffer from diminishing returns, the variable is log-transformed.

Other Conflict Considerations

In order to test our predictions, we create a broad context to account for variation in conflict across dyads by integrating the joint IGO membership variable into three models of conflict. These include an aggregate model (Model 1), along with models of low (Model 2) and high severity conflict (Model 3).

The low severity conflict model is built with the following considerations in mind. We start with the notion that low severity conflicts emanate from differences in policy preferences within the dyad. The greater the divergence in policy preferences, the more likely that conflicts will occur between two states. Given our theoretical framework, and especially in the absence of institutionalized foreign policy memory, divergent policy preferences constitute a critical threshold for assessing the independent effects of IGO membership on these states.

We develop two measures of policy preferences. One is a generic policy affinity measure, based on similarities within dyads using UN General Assembly roll call votes. We are not alone in either using such voting behavior as a policy preference measure (see Tomlin 1985; Gartzke 1998, 2000, 2006; Voeten 2000; Lemke 2002), or noting its drawbacks.25 It is clear that state policy preferences and state behavior are not identical, although we concur with Gartzke that the resulting “distortions will be least intense where the value in making choices is most modest” (Gartzke 2006:2), as is the case in UNGA voting behavior.

Our measure of policy preferences provides comparisons between the voting behaviors of dyad members. Drawn from UNGA roll call votes, individual states’ voting scores are calculated by averaging each state's vote over the number of resolutions for the session.26 The absolute value of the distance between individual state scores is the dyad policy affinity score. Lower numbers represent smaller distances between policy preferences and thus higher policy affinity between states, while larger numbers correspond to greater disagreements in policy affinity within the dyad. As policy preferences become more divergent (the policy affinity score becomes larger), the likelihood of conflict should increase.

While we use voting behavior as a general measure of preferences, we are concerned that this measure may not be sufficiently sensitive for the most salient policy concerns of newly emerging states in postcommunist space. Much of the UNGA agenda, while ultimately relevant for these states, may focus on global issues of less immediate concern than those within the neighborhood, and may be less useful for measuring policy preferences in the early years of emergence. Our research site allows a second, more specific preference measurement: alliance similarity. Alliance portfolios have been used extensively in previous studies but appear to be relatively insensitive to short term fluctuations in policy preferences (except for the most dramatic types). However, for our research site, alliance commitments appear to work somewhat differently. As nearly all the states in our region (with the exception of Russia) have emerging, untested, and typically small militaries during these early years, we assume that alliance commitments are primarily signals of security fears and security policy concerns between states that are developing their foreign policy orientations. Therefore, we create a security concern variable and code it as “1” if the dyad pair shares at least one political-military alliance in that dyadic year, as reported by Leeds, Ritter, Mitchell, and Long's (2002) ATOP version 3.0 data.27

We believe that this treatment of alliance treaties within the postcommunist space is more appropriate as a measure of security concern rather than a primary manifestation of security coordination. A closer examination of a number of these treaties exemplifies the point. Shortly before their internationally recognized split, the Czech Republic and Slovakia signed a treaty in November of 1992, stipulating defense pact and consultation pact components, and specific guidelines to deal with internal disagreements (Leeds et al. 2002: article V). Both Romania and Hungary (non-aggression pact, September 1996) and Poland and Ukraine (non-aggression pact, consultation pact, neutrality pact, May 1992) signed similar documents. Each of these dyads faced border anxiety in a geopolitically fluid time frame and concerns over the treatment of minority populations within neighboring territories, and the treaties included commitments to noninterference in internal state affairs.28 The explicit attention within these treaties to problems facing new or newly reemerging states in a transitory phase underlines potential and/or existing policy differences and concerns between dyad members.29 While we expect that treaties would have a role in mediating conflicts of high severity, the expression of security concerns embedded in them should predict as well to increased low severity conflict within the dyad.30

We are not sure that the democratic peace applies to our sample of transitioning states, some of which are reaching higher levels of institutionalized democracy gradually, some more quickly, and others still are in turbulent transition, but we agnostically include level of democracy as a possible constraint on low levels of conflict. We use Polity IV scores (Marshall and Jaggers 2006), ranging from 0 (least democratic) to 10 (most democratic). As the benefits of mutual democracy should be constrained by the least democratic state within the dyad (Dixon 1994), we use the “weakest link” score for the dyad.

Contiguity has been consistently identified as a critical determinant of conflict relationships, and we expect that physical distance will reduce the relevance of actors to each other, even when they have opposing policy preferences (for example, Bremer 1992). We create a measure of distance, calculated as the distance (per 10 miles) between capital cities while we account for contiguous states by reverting the measure to zero when states share a direct border. The data are obtained through EUGene (Bennett and Stam 2000) and were originally part of the COW project (Small and Singer 1982).

We add relative capabilities to the model as a control variable, testing the notion that imbalanced strength may inhibit expressions of even low levels of conflict, although we anticipate that especially in early stages of new state development, relative capabilities may be difficult to calculate by states unless the differences are enormous. We calculate relative capabilities using the dyadic ratio of each state's CINC scores (the states’ average share of the total international population, urban population, military expenditures, military personnel, iron and steel production, and energy consumption) as developed by Singer, Bremer, and Stuckey (1972) and provided by EUGene.

Kinsella and Russett integrate into their model both economic interdependence and openness as conflict ameliorating factors. These considerations are difficult to apply to the states in our sample for both substantive and methodological reasons. Economic relations are just developing within these dyads and reliable information on trade relations between them is scarce.31 However, more is known about the economic capabilities inherited from the Cold War era and we expect that policymakers have better information on their economies than on the value of their military capabilities and power potential. Economic capabilities in general and wealth specifically may be significant, either dampening conflict as dyads with wealthier economies have sufficient resources to pursue a variety of strategies, including conflict avoidance, or conversely, they have sufficient resources to engage in conflict. We measure national wealth using data from the Penn World Tables (Heston, Summers, and Aten 2002; real GDP per capita, constant prices, chain series, scaled per hundred). In keeping with a “weakest link” principle (Dixon 1994), we use the lowest GDP-PC of the dyad as an indicator of relative dyadic wealth.

Finally, we add a dummy variable for Russian presence in the dyad (“1” if Russia is one of the dyad members; “0” if otherwise). To the extent that Russian policymakers have sought to reorder postcommunist space (for example, Willerton and Beznosov 2007), we expect Russian involvement in the dyad to be a considerable source of turmoil for new states. In addition, as the major power in the region, Russian presence within a dyad is expected to correlate highly with conflicts of all types (Kinsella and Russett 2002:1061).

Our model for high severity conflicts is nearly identical to the one for low severity, with one addition and one changed expectation. We add the low severity conflict variable to test our second prediction that low severity conflict is salient for higher severity. The change in expectation/direction (as noted above) is in regard to security preferences: we expect that similarity in alliance commitments at high levels of conflict within the dyad should predict to reduced conflict as these treaties contain provisions and mechanisms for mediating high levels of conflict (potential violence) between signatories.

The states constituting the region (along with their birth/rebirth years) are noted in Appendix 2. Descriptive statistics on each of the variables are presented in Table 1. We include all possible dyads in postcommunist space as they come into existence starting with the end of the Cold War.

View this table:

Descriptive Statistics

Shared IGO membership2,52413812.70026.6888
Shared IGO membership (logged)2,52403.6382.38830.5975
Low severity conflict2,5240911.17325.0187
High severity conflict2,5240350.09570.9307
Security concern2,524010.67630.4680
Policy affinity2,52400.39130.07400.0715
Russian presence2,524010.08500.2789
Level of democracy (weakest link)2,5240102.99263.2793
National wealth (weakest link)2,5249.8569136.659841.106521.5523
Relative capabilities2,5240.003610.36060.2798

Results and Discussion

Following Kinsella and Russett (2002), for purposes of comparison we generate three models. Model 1 is the aggregate for all conflicts while Models 2 and 3 disaggregate the results by level of conflict severity. In all three models of conflict we include a one year lag of the dependent variable, and we do so for both methodological and substantive reasons.32 Because of the over-dispersion occurring throughout the population of conflictual behavior, the negative binomial is well suited for event data.33Table 2 displays the results of the analysis for all three models.

View this table:

Joint IGO Membership on Conflict among Dyads, 1990–2000

All conflict [1], Coef. (SE)Low severity conflict [2], Coef. (SE)High severity conflict [3], Coef. (SE)
Shared IGO membership (lagged)0.2619** (0.1284)0.4085** (0.1307)0.1735 (0.1637)
All conflict (lagged)0.0839*** (0.0221)
High severity conflict (lagged)0.379*** (0.0946)
Low severity conflict (lagged)0.067** (0.0254)
High severity conflict0.2028* (0.1079)
Lower severity conflict0.028** (0.0112)
Security concern0.3098 (0.2249)0.4581** (0.2091)−0.7064* (0.4034)
Policy affinity−1.876* (1.0474)−1.4547 (1.0118)−1.3166 (1.9422)
Russian presence2.5528*** (0.2972)2.5423*** (0.2742)2.3051*** (0.4506)
Level of democracy−0.0799** (0.0247)−0.0779** (0.0247)0.0074 (0.0523)
Distance−0.0181*** (0.0017)−0.0175*** (0.0017)−0.016*** (0.0026)
National wealth0.0043 (0.0039)0.0065 (0.004)−0.0157* (0.0086)
Relative capabilities0.1024 (0.373)0.1662 (0.3658)0.1619 (0.8295)
Constant−0.8073* (0.4495)−1.5829*** (0.4066)−1.8881* (1.0256)
Alpha2.1188 (.2883)1.9167 (.2842)3.6950 (1.1940)
Log Psuedo-Likelihood−1,840.6954−1,755.9783−415.20415
Prob > χ20.0000.0000.000
  • (Notes. Coefficients are unstandardized, standard errors are robust, clustered on dyads and indicated in parenthesis. *p > .10, **p > .05, ***p > .001.)

Focusing on the primary concern regarding the potential effects of IGO comembership on conflict, the aggregate model shows that the relationship is significant and positive, appearing to indicate that dyads with higher amounts of IGO comembership experience increased amounts of conflict. When conflict is disaggregated into low (Model 2) and high severity (Model 3), increased levels of joint membership in IGOs continue to be associated with a statistically significantly larger probability of low severity conflict behavior within dyads, consistent with our primary hypothesis.34 The contending explanation that IGOs would increase bargaining space, and thus lead to conflict avoidance, does not appear to be borne out, suggesting that information effects are largely exacerbating low severity conflict among states with limited institutional memory and experience. As negative binomial model coefficients are not linearly related to the dependent variable and their interpretation aside from direction is difficult, we graph the impact of IGO comembership on conflict while holding all other variables constant at their mean. Figure 2 illustrates the expected rate of low severity conflictual events concurrently with the number of shared IGO memberships in the dyad; the graph indicates both a strong initial upswing and a steady rise associated with increased IGO interaction.


Impact of IGO Memberships on the Rate of Low Severity Conflict among Dyads

Controlling for the range of potential considerations associated with low severity conflict, a one standard deviation increase in IGO comembership corresponds to a 3.7% increase in low severity conflict within the dyad when all other variables are held at their means. A different way of assessing the effect of IGO membership is to hold all other independent variables at their median values, a more real-life scenario than mean values. Doing so, we find that a one standard deviation increase in IGO comemberships (from 12 to 19) corresponds to an increase in low level conflict by 4.1%. Thus, we find some evidence in the data supporting the hypothesis that increases in the opportunity for interaction and information-sharing through IGOs within a dyad is associated with increased friction, at least at a low severity level.

In order to have a yardstick of relevance, we compare in Figure 3 the relative impact of IGO comembership among states with the relative impact of the level of democracy on the overall incidence of low severity conflict events. The graph is calculated with all other variables held at their means. While democracy's impact is fairly moderate for Model 2, IGO membership has a stronger effect, particularly for values at one standard deviation below the mean and greater.35


Impact of IGO Memberships and Democracy on Low Severity Conflict

Model 3 illustrates what we had expected: joint IGO membership appears to act differently on high severity conflict than on low severity conflict for this sample of dyads: the relationship between the IGO variable and high severity conflict is not significant.36 We note two plausible reasons. First, IGO membership may have immediate effects for low severity conflicts by instantaneously exposing states to differences in foreign policy orientations and interests, while many of the IGO effects purported to be ameliorating high severity conflict require substantial participation by members over a longer period of time than covered in the time frame of this study. Second, it is plausible as well that by aggregating all IGOs into our analysis, we are watering down the effects of stronger IGOs with complex design characteristics that are most likely to ameliorate severe conflicts between states. In order to test this notion, we recalculated COW IGO membership by restricting the data to membership in organizations that meet 11 criteria for institutional autonomy and centralization, using the procedures developed in a previous effort (Volgy et al. 2008). Using this restricted population of formal intergovernmental organizations (FIGOs), we substituted FIGO comembership in the high severity model.37 Nevertheless, the results failed to show a significant impact on high severity conflict.38

Regarding our second hypothesis, we find a significant relationship between higher and lower severity conflict. The impact of low severity conflict is noted in Model 3 and illustrated in Figure 4. The incidence of low severity conflict is associated with a statistically significant rise in high severity conflict, even though we control for high severity conflict in the previous year. The size of the effect, however, requires some interpretation. At first glance, the impact of low severity conflict on high severity appears to be quite small: as low severity conflict increases by one standard deviation from its mean, and all other variables are held at their means, there is an associated 0.23% increase in the expected frequency of high severity conflict among dyads in our sample. However, this apparently small impact needs to be considered in the context of high severity conflicts being low probability events, along with an assessment of conditions under which high severity conflict emerge. In our model, dyads experience the highest levels of conflict when they are not allied and contiguous, and when Russia is included in the dyad.39 Within this profile and holding all other variables at their mean values, we find that an increase from 0 low severity conflict to 1 is accompanied by an expected increase of 4.79% in the count of high severity conflict events. This magnitude illustrates that even under conditions when high severity conflicts are already well predicted by other considerations (including the lag of high severity conflict), low level conflict continues to contribute to high severity conflicts.


Impact of Low Severity Conflict on High Severity Conflict among Dyads

Turning to the control variables in our models, we note that the expected relationship between our first measure of policy divergence and conflict are not borne out. Although in the correct direction, the relationships with the dependent variable are statistically insignificant in both disaggregated models and barely significant in Model 1. We suspect that the lack of significant findings here are because of a combination of methodological and substantive reasons elaborated above.40

The measure of security consideration through mutual alliance commitments—as a more salient indicator of policy concerns in the neighborhood—does appear to work as we had suggested earlier. The aggregate model in Table 2 shows an insignificant, albeit positive, relationship between the security measure and conflict in general, masking two different dynamics shown when we consider low and high severity separately. As we had predicted, dyads with formal treaty commitments within them are associated with significant levels of low severity conflict; at the same time, these treaty obligations are negatively related to high severity conflicts,41 corresponding to a 5.7% increase in low severity (and a 1.2% decrease in high severity) conflicts within dyads when all other variables are held at their means.

Clearly, the presence of Russia has a very strong effect on conflicts within dyads. The expected rate of conflict for dyads with Russian presence is 10 and 12.7 times greater for high and low levels of conflict, respectively, holding all else constant at their means. For high conflictual events, this accounts for about an 11% increase when all other variables are held at their means. For low severity conflict however, this effect more than doubles. Concurrent with both our expectations and recent findings (Willerton and Beznosov 2007), it is not surprising that Russia, as a major power located in the region, would hold such great influence on the likelihood of conflict.

Consistent with expectations, distance is significant: the likelihood of conflict, at high or lower severity levels, recedes as the distance between states increases. Further, as expected, power ratios are not, once Russia is treated as a separate variable. Wealth appears to matter, but primarily in inhibiting conflict at the more severe level.

Overall, both low levels and high severity conflicts among dyads in postcommunist space seem to be heavily influenced by geographic proximity, involvement with Russia, and alliance security ties. Nevertheless, opportunities for increased interaction and access to further information about policy preferences through IGO comembership continue to be associated with higher frequencies of low severity conflicts. In turn, and despite a number of controls introduced in the analysis, the relevance of low severity conflict for conflicts of high severity remains salient in the calculus involving the evolution of conflicts in this region.


More than a decade after their abrupt (re)entrance into the international arena, the Central and Eastern European states and their Asian counterparts represent a conglomeration of young (and fairly young) states thrust into the middle (literally and figuratively) of established powerful states and long-standing competition in a period of dynamic transformation in the international system. This effort has sought to address two related puzzles with respect to these states. The first is in regard to the possible differential effects of IGO membership on lower versus higher levels of conflict between them. The second and related concern is about the impact of low levels of conflict on conflicts of higher severity for such newly emergent states.

Our theoretical sketch led us to a research design focusing on states exhibiting a set of unusual characteristics: newly emergent, with new foreign policy establishments that have little institutional memory relevant to presently salient foreign policy relationships, and acting in a context of dramatic change with a major power (Russia) in the neighborhood seeking to reestablish some semblance of control over the region. Using this type of site for our research allowed us to look at possible opportunity effects for processing information through IGOs without typical institutional filters (ongoing foreign policy bureaucracies) in a dangerous and turbulent neighborhood where policymakers need to respond to rapidly changing circumstances.

We suspect that our findings are not trivial. Recent research has argued that previous findings of insignificant or positive relationships between IGO membership and certain forms of conflict are because of a failure to control for design differences across organizations, but consideration of IGOs with strong design features should continue to exhibit the conflict-ameliorating functions of intergovernmental organizations (for example, Gartzke et al. 2005). The corollary to this argument is that researchers should filter out weaker IGOs in their analysis. However, our research suggests that this approach would be a mistake. While institutions with strong design characteristics may ameliorate high levels of conflict between states that are comembers of those organizations, it appears likely that for some states comembership in both strong and weakly designed IGOs create opportunities for greater low severity conflict, which in turn may increase as well high severity conflict between those states.

The choice of research site however raises concern regarding the extent to which these findings are generalizable to other newly emergent states, or to newly emerged states in more stable neighborhoods, or during more stable global conditions.42 Neither is it clear that our findings are necessarily applicable to states that are not new. It is plausible that states with established foreign policy machineries and substantial histories of experience with their neighbors are less likely to react in a similar manner to the opportunity of witnessing differences in goals, interests, policies, or styles exhibited through shared IGOs. We suspect that at one extreme—for major powers—additional information and interaction with other states through IGOs may generate insubstantial amounts of low level conflicts. For other states, however, this remains an empirical question worth pursuing further, including the secondary linkage between lower versus higher levels of conflict severity.

Expanding the sample to a larger number of states, organizations, and a longer time frame will allow further investigation as well regarding the types of organizations that are most likely to produce increased opportunities for low levels of conflict within politically relevant dyads. Do members of these dyads pay equal amounts of attention to each other in global, inter-regional, regional, or sub-regional organizations? We assume that the agendas of regional and sub-regional organizations are most relevant for these states where the principal foreign policy challenges exist in the immediate neighborhood. However, the effects of globalization are widespread and information generated from inter-regional and global organization, even with what appears as a less immediately salient agenda, may contribute to generating low levels of conflict for politically relevant dyads. Likewise, a longer time frame will allow some determination about whether or not these effects are more likely under stable versus unstable global conditions (Cold War versus post-Cold War turbulence), including and perhaps especially, the salience of low levels of conflict for conflicts of higher severity.

The idea that low severity conflict may impact significantly on high severity conflict has consequences as well for our understanding of conflict dynamics. Much of the empirical literature, either by data gathering choices or by theoretical assumptions regarding “noise,” or “real” conflict, seems to have ignored the conditions under which lower severity conflicts arise, and the potential consequences they may have for higher severity conflicts. Our findings suggest that—at least for these states—low levels of conflict may represent more than just background noise and they may in fact influence more severe conflict relationships between states. This argument is not inconsistent with Kant's original perspective on perpetual peace.43

We found in our analysis no significant negative relationship between IGO comembership and high severity conflict. This was so even when we sought to reduce comembership to formal intergovernmental organizations that contain strong design characteristics associated with bureaucratic centralization and some level of autonomy and independence. It is plausible that this was so because our filters did not eliminate sufficient numbers of IGOs that only minimally meet the criterion of strongly designed organizations. It is more likely that this outcome is a function of the research site: these dyads are in the midst of considerable turbulence and Russian involvement in the region is so extensive that it may overcome the strength of institutions with conflict ameliorating functions.44

These results suggest as well a third possibility worthy of future research: when it comes to high severity conflicts, the payoffs from IGO involvement (routinized meetings of top officials, increased information, organizational conflict mediation processes, socialization dynamics, etc.) may require more time spent inside of organizations than the relative short time frame used in this study. Anecdotal information suggests the same: involvement in the EU by newer states has created more conflict between members (for example, Polish objections to liberal social rules) and early failure to live up to promises (Hungarian and Romanian backsliding on Roma rights) in the short-run, although these conflicts may be alleviated as longevity in the EU increases. Most research linking dyadic membership in IGOs with high levels of conflict fails to control for the duration of immersion in these organizations by members of the dyad. Our research implies that this may be a fruitful arena of further pursuit. We plan to do so for this research site as more data become available in order to observe dyadic interactions and characteristics beyond 2000. We are particularly interested in uncovering the extent to which further immersion into the EU by some and into the constellation of CIS states by others have an impact on the relationships we have found in the earlier period.


Appendix 1: Conflict Division

View this table:

Appendix 2: Postcommunist Space States (Birth/Rebirth Year)

View this table:


  • Authors' notes: Earlier versions of this article were presented at the 18th Peace Science Society Annual Conference, November 2–4, 2007 and the 4th Journeys in World Politics Workshop, October 9–11, 2008. The authors would like to acknowledge Gary Goertz, William Dixon, Sara McLaughlin Mitchell, Andrei Tsygankov, and three anonymous reviewers for their insightful comments on previous drafts. Replication data and command files are available on the Authors' Web site (http://www.u.arizona.edu/∼elfauset/research.php).

  • 1 Strictly speaking MIDs and conflict events do not represent equivalent units as MIDs episodes are “united historical cases” of aggregated, interrelated conflictual incidents for dyads where only the highest level of hostility reached is recorded (Jones, Bremer, and Singer 1996:168; Pevehouse 2004:253–254). Nevertheless, the numbers of actual dyadic conflicts dwarf the number of conflict episodes represented by MIDs.

  • 2 The most recent reviews of the empirical literature linking IGO membership and conflict amelioration include Boehmer, Gartzke, and Nordstrom (2004); Chan (2005), and Pevehouse and Russett (2006), while Rapkin and Thompson (2009:13) underscore the range of contradictory findings.

  • 3 We are not claiming that policymakers in these states make explicit choices to join IGOs because they necessarily understand (understood) these benefits. Some had little choice, either because they were coerced to join or because they could not conduct effective foreign and economic policies without joining. Others may have joined to lock in domestic policy preferences (for example, Schimmelfennig 2005; Mansfield and Pevehouse 2006). Others still perhaps made strategic choices about joining various intergovernmental organizations. Our more modest claim here is that regardless of why they joined, we should expect to find significant impacts on their foreign policy interactions with other states.

  • 4 For an excellent discussion of policy preferences and their salience, see Bueno de Mesquita (2003).

  • 5 This is of course less true when a member of the dyad is a major global power, capable of reaching into the neighborhood.

  • 6 We assume that considerations of state sovereignty would prevail when states can easily address policy differences outside the institutional setting.

  • 7 We are aware that the analogy between democratic legislative systems and state membership in IGOs is not isomorphic. States participate voluntarily while sub-systemic actors in democracies are constitutionally constrained from leaving these structures, forcing the latter actors to engage in more compromise and conflict resolution. Yet, this difference only underscores the greater willingness and desire by states to articulate their interests in IGO settings, even if they have come together to have IGOs facilitate their own objectives.

  • 8 Except for the rare condition when potential members are coerced to join an organization by a dominant power and the organization exists for the sole purpose of facilitating coordination between the dominant power and the other members, a condition that may have existed during the Cold War for the Warsaw Pact and COMECON.

  • 9 Unless of course the organization is composed of homogeneous units with highly similar policy preferences. This could be the case with very narrow functional organizations designed for coordination problems between highly homogeneous states, but this condition does not appear to apply to postcommunist space.

  • 10 This argument is applicable across various kinds of institutions and mandates, be they functional, social, economic, etc. While it is likely that states often come together in organizations because of generalized overlapping and mutual interests and particular goals, this does not preclude substantial differences in policy preference regarding either outcome or enactment.

  • 11 The Greek-Turkish case of NATO impact on policy preferences through NATO interaction is salient as that case represents a rivalry and a situation of finely tuned institutional memory. Yet, the IGO effect according to Krebs (1999) appears to have persisted nonetheless.

  • 12 We are grateful for the suggestion made by an anonymous reviewer of this manuscript.

  • 13 By “strongly designed” or strong, we are referring to IGOs that require routinized meetings of high level officials representing their states, a substantial bureaucracy, and ongoing funding for the organization—factors corresponding with some amount of organizational centralization and autonomy. Below, we provide one operational definition (formal intergovernmental organizations) of strong IGOs.

  • 14 By focusing on states in postcommunist space, we are excluding a number of states that came into existence after the end of the Cold War. However, with two exceptions (Namibia and Eritrea), during the period under examination, all the excluded ones are microstates with different foreign policy dynamics and virtually no resources with which to impact their external relations. Excluding microstates, states in postcommunist space account for 91% of all new states for the time period under examination.

  • 15 We assume that this would be less likely when Russia is part of the dyad. This should be true for a number of reasons including historical experience with the Kremlin, and Russian movement immediately after the dissolution of the Soviet Union to influence post-Soviet space. We account for this phenomenon in our model (below).

  • 16 We are separating here the effects of long-term institutional “memory” from immediate previous experience, which we include in our conflict model (below) by lagging the dependent variable.

  • 17 One notable exception is Georgia, during Eduard Shevardnaze's leadership of that state; the break-up of Yugoslavia, for example, left its foreign policy machinery in Serbia, cleansed of actors from the other parts of the multi-ethnic republic, and presumably not in very good shape.

  • 18 We recognize that many of the political elites in the new republics of the old USSR had substantial and close ties with Russian elites, but these relationships did not extend to foreign affairs. Elites in the restructured states of Eastern Europe may have had far less in common with their Russian counterparts, and in fact many of them came to office with very strong, negative perceptions of Russians and other neighbors; yet even such negative images likely required some reinforcement based on foreign policy interactions inside and outside of organizational structures.

  • 19 Ideally, we would have broadened our research design to compare states in postcommunist space with newly emerging states following decolonization in Africa, another region of turbulence, and populated by new states with no institutional memory. Unfortunately, the available databases (COBDAB and WEIS) underreport extensively low severity conflicts for Africa during that period, making comparison virtually impossible for low severity conflict effects.

  • 20 Kinsella and Russett restrict their analysis to politically relevant dyads; we do not. As our theory suggests, the very definition of politically relevant dyads in this restricted geographical space may be partially dictated by the effects of IGO involvement.

  • 21 Because of restrictions imposed by the availability of IGO membership data, this time period is reduced to the 11-year span of 1990 to 2000.

  • 22 For an explanation of event data and techniques associated with machine-coded event data, as well as an extensive validity test, see Gerner, Schrodt, Francisco, and Weddle (1994). For further information on the IDEAs data set in particular, see Bond, Bond, Oh, Jenkins, and Taylor (2003).

  • 23 Kinsella and Russett (2002:1055) count the existence of a conflict if it reaches a severity level of −2.2 or higher on the Goldstein scale, which corresponds to what Kinsella and Russett deemed as strong verbal hostility or higher. We find no theoretical reason for this cut-off level for conflict, and so we include all low level conflicts. We think this is especially appropriate for new states experiencing their first interactions in the region without the benefit of institutional memory or previous foreign policy experience.

  • 24 An IGO is defined as such if it is “(1) is a formal entity, (2) has [three or more] [sovereign] states as members, and (3) possesses a permanent secretariat or other indication of institutionalization such as headquarters and/or permanent staff” (Pevehouse, Nordstrom, and Warnke 2005:9–10).

  • 25 For a review of several preference measures, including UNGA votes, see Lemke (2002). For an expanded justification for the viability of UNGA votes as preferences see Volgy, Frazier, and Ingersoll (2003).

  • 26 UNGA votes from contested resolutions are recoded along a Likert scale where a “no” vote is recorded as the lowest value, a “yes” is the highest, and “abstain” or “absent” provide the middle point. We draw our data from the Inter-university Consortium for Political and Social Research (ICPSR) project on United Nations General Assembly roll call voting.

  • 27 We code defense and non-aggression pacts and we note the possibility that alliances and IGO comemberships could represent a case of double counting. However, we find this highly unlikely in our case. Very few of the IGOs represented here are of a political-military nature while alliance treaties after 1989 are overwhelmingly bilateral (Leeds and Anac 2005). Furthermore, virtually all multilateral alliances in this region are either not included in the IGO population (the CSTO and the Shanghai Five) or came into existence after our analysis concludes (SCO).

  • 28 Specific treaty data for these examples are drawn from Case files 4845, 4370, 4225 (Leeds et al. 2002).

  • 29 While the language within these treaties undoubtedly conforms to some form of convention, we believe that we can accept the provisions within these treaties as representative of actual concerns among treaty members.

  • 30 Thus, this indicator appears to function at two levels. In general, it signals security concerns between two states; however, to the extent that it also indicates the creation of conflict management mechanisms when needed, we expect that during episodes of high conflict severity, the indicator also reflects some institutional capacity for managing conflicts.

  • 31 Traditionally, trade (as a function of trade dependence between the dyads) would be included in the model, but because of the lack of data (reducing the population size by more than 86%), this variable is excluded. Results using this variable are available from the authors; its inclusion produces results similar to those reported below.

  • 32 We substitute the lagged dependent variable in lieu of “peace years” used in a number of conflict models, as we are observing newly emerging states with limited experience in international politics. We assume as well that immediate experience in terms of the previous year's conflict within a dyad is likely to be salient for a foreign policy apparatus.

  • 33 This over-dispersion is in large part because of the under-occurrence of conflict within the system. Ordinarily, one might resort to a zero-inflated binomial, however as our analysis seeks to understand the role IGOs play in the reduction or promotion of conflict rather than a pure theorization of the causes of conflict itself, the negative binomial model appears to be a more appropriate choice. The significance of the alpha parameter demonstrates this.

  • 34 We argue that both “weak” and “strong” IGOs should contribute to low severity conflicts; disaggregating IGO comembership into the two types, yields similar results for each (results are available from the authors).

  • 35 As the distribution of the IGO variable is skewed to the right, values outside of the empirical range are dropped and the lines are connected. For all graphical points represented, other than the minimum, the IGO variable has a greater impact than democracy.

  • 36 Kinsella and Russett (2002) find a similar result in their analysis of MIDs and high severity conflict using a different sample of states and a different time frame (1951 through 1992).

  • 37 These models are not shown here but are available from the authors on request.

  • 38 A methodological explanation is plausible as well: the relationship between low severity conflict and IGO membership suggest the plausible presence of multicollinearity in the model of high severity conflict, diminishing the impact of IGO membership. Testing this possibility we found the IGO variable to be insufficiently correlated with any of the other independent variables to create overly high standard errors.

  • 39 For example, this is characteristic of both Georgian-Russian and Azerbaijan-Russian relations in 1992.

  • 40 We noted earlier caveats regarding the use of UNGA voting methodology for assessing policy preferences of newly emerging states in post-Soviet space, as well as substantive issues raised by this measure. Additionally, we note that we also controlled for whether or not each state in the dyad eventually sought membership in the EU or the CIS; however the resulting dummy variables were insignificant in the models.

  • 41 These findings appear to parallel those of Kinsella and Russett (2002:1059) as well; they suggest that alliance commitments appear to signal “salience” between states in a dyad, allowing for both conflict management but disputes as well.

  • 42 As noted earlier, one appropriate comparison would have been with newly emerging states following decolonization in Sub-Saharan Africa, but is prohibited by data availability and reliability issues involving low severity conflict events in that region.

  • 43 Kant's treatise on the possibility of perpetual peace emphasizes the tripod of democracy, interdependence, and IGOs. Yet, Kant hinted at the potential negative effects of interdependence, suggesting that increased “intercourse” among states not uniformly democratic and not governed by a federation will increase conflictual relations (Kant [1795], 1957:142). This point is similar to our argument regarding low severity IGO effects (the authors thank Otto Hieronymi for this perspective).

  • 44 The role of Russia in postcommunist space is consistent with large N empirical studies of conflict that have created a series of regional “dummy” variables and found that they impact significantly on general models of conflict, suggesting substantial regional variation that is likely based on different characteristics associated with regions, and perhaps with the role of major powers in those regions.


View Abstract