Do Intergovernmental Organizations Promote Peace?
Charles Boehmer
Erik Gartzke
Timothy Nordstrom *
I. Introduction
Traditional paradigms in world politics offer starkly contrasting interpretations of the role of intergovernmental organizations (IGOs). Liberal institutional theory argues that IGOs foster nonviolent conflict resolution and constrain the advent of disputes. Functionalists conceive of IGOs as capable of transforming state preferences and promoting pacific global relations. Students of the democratic peace have now added IGOs to the prescriptive liberal cocktail.1 Together, the three pillars of the "Kantian tripod"—democracy, cross-border trade, and international organizations—appear to diminish the likelihood of militarized contests in some large-sample quantitative studies. Realists, by contrast, have long argued that IGOs reflect, rather than effect, world politics. Quantitative support for a link between IGOs and peace also appears attenuated, with IGOs increasing the probability of conflict when violations of certain statistical assumptions are addressed.2
Conventional accounts of the role of IGOs in influencing war and peace seem to us to be incomplete and overly categorical in their praise as well as in their criticism. We develop a more nuanced alternative to [End Page 1] the contrasting visions of classical arguments. Bargaining theory explains war as the result of both material and informational conditions. States often disagree and power is omnipresent, but conflict among nations results more often in diplomatic bargains than in violent contests. Diplomacy fails and warfare occasionally ensues when states differ in their beliefs about power and interest. Third parties, such as IGOs, can most effectively foster peace by credibly informing competitors or by intervening in ways that carefully prevent participants from converting new strategic conditions into different, more extractive demands. As such, our explanation constitutes a "middle path" in which IGO effectiveness is a variable rather than a parameter. The theory offers a number of empirical implications—only some of which are explored here—even as it suggests two improvements in research design. First, IGOs are not all created equal. If IGOs vary in their institutional structure, mandate, and member cohesion, then it follows that they can also vary in their impact on interstate disputes. We identify IGO attributes that are likely to make some of them more effective at fostering agreements among states. Second, states are not all equal, either. The sometimes positive statistical relationship between IGO membership and militarized conflict can be accounted for if the countries that are most active internationally are also those most likely to belong to intergovernmental organizations. We introduce a control variable for state engagement in world politics that accounts for an observed tendency of IGOs to be associated with disputes.
The article proceeds as follows. We briefly review relevant studies of intergovernmental organizations. We then outline an argument for why states fight, why they join IGOs, and how IGOs can influence international conflict. Next, we explain our research design, introduce variables to measure the institutionalization and contentiousness of IGOs, and report our findings. The tests presented are a necessary first step in analyzing the theory. Because of space limitations and the need to maintain comparability with previous studies, some of the most intriguing implications of our argument await further study. While individual results can be interpreted as favoring realist or liberal arguments, collectively our findings challenge classical accounts. IGOs are not broadly effective in the way they should be if international organizations alter preferences or form a web of constraining commitments. Neither are IGOs universally ineffective, as suggested by critics. The bargaining approach suggests that only those IGOs with certain attributes are likely to promote peace. We show that institutionalized IGOs reduce the risk of militarized disputes but that other IGOs have little direct influence on conflict. We also show that cohesion among IGO members increases IGO effectiveness. IGOs with a security mandate are more effective at promoting peace [End Page 2] than are economic organizations. Finally, controlling for engagement in the international system helps to explain previous anomalous findings. We conclude with a discussion of the implications of the study.
II. Literature: Linkages Between IGOs and Peace
Research on interstate conflict and international institutions broadly speaking or on intergovernmental organizations in particular can be summarized as follows. First, it is a matter of debate how and when IGOs matter. Second, there is limited systematic examination of IGO efficacy even while existing evidence appears to be contradictory. Third, controversy and anecdotal evidence suggest that IGOs have a nonuniform effect on state behavior; they may both increase and decrease conflict. Fourth, states select into IGOs in ways that could correlate with the impact of IGOs on disputes.
Realists, liberals, and others have long debated the utility of IGOs.3 Most agree that IGOs matter in some form or in certain contexts, but there is no consensus as to the extent of IGO influence and how best to treat IGOs analytically as an aspect of world politics. By holding that IGOs matter, participants in the debate usually mean that they should be capable of altering state behavior. On one side in the debate, constructivists, functionalists, and liberal institutionalists contend that IGOs are (or can be) a central component of world order. On the other side, many realists argue that IGOs are only marginally influential in world politics and that IGOs typically reflect status quo power relations.
Constructivists, functionalists, and institutionalists argue that global politics is increasingly organized around regimes and institutions that foster cooperation by providing information and organizational structure, promoting norms and common belief systems, and reducing transaction costs.4 Realpolitik should become less important with the [End Page 3] multiple channels of interaction and the growing salience of nonstate actors, leading to a decline in militarized violence.
Neoliberal institutionalists accept much of the functionalist/Kantian prescription for international politics, but they also accept the realist tenets of systemic anarchy, the importance of power, and the preeminence of states' interests.5 Collective security and cooperation are nevertheless feasible where states share common or complementary interests and where states value absolute over relative gains.6 IGOs help foster cooperation by encouraging reciprocity through regularizing interactions.7 Indeed, the institutionalist perspective may be construed as the strongest advocate for IGOs, given the lack of alternatives once one accepts anarchy as endemic to international affairs.
While conceptual work by functionalists, liberals, and realists is extensive, the empirical literature on institutions is decidedly sparse. It is unclear whether IGOs are as effective (or ineffective) as partisans claim. Indeed, the limited scope of empirical research may have retarded theoretical development. Theory stands to advance most quickly when confronted with contrasting evidence. The literature suggests that IGO heterogeneity and factors such as regime type and major power status influence IGO functionality. Nonetheless, current analysis fails to address these claims fully, and although several studies call for assessments of institutional efficacy, much remains to be done.8 [End Page 4]
Most existing empirical research involves qualitative case studies.9 Recently, however, democratic peace researchers have become interested in the role of IGOs and have provided the strongest extant evidence to date that IGOs alter state behavior.10 Oneal and Russett argue in a series of articles that dyads that share more IGO memberships are less likely to experience disputes. This result appears sensitive to the choice of sample and econometric technique. IGOs either increase conflict or have no effect when assessing all dyads or when adding nonlinear controls for temporal dependence. Other studies find little or no indication that IGOs bring peace. Jacobson, Reisinger, and Mathers and Domke offer regression analysis of IGOs and interstate conflict, but neither reports significant results for intergovernmental organizations.11 Oneal and Russett and Gartzke, Li, and Boehmer produce results suggesting that IGOs increase interstate conflict among members.12
At least three factors appear responsible for these disparate findings. First, IGOs are institutionally heterogeneous. In the three decades since Singer and Wallace first highlighted the issue, no study has investigated heterogeneity in a systematic fashion.13 NATO and the EU exemplify institutionalized IGOs with an ability to alter state behavior. Other organizations with more moderate or limited institutional structures, such as the Southern Cone Common Market (MERCOSUR) or the Economic Community of West African States (ECOWAS), hold promise. IGOs such as the South Asian Association for Regional Cooperation (SAARC) or the Economic Community of the Great Lakes Countries (CEPGL) have little ability to influence member behavior. Attempts to treat organizations with different institutional structure and scope homogeneously distorts the apparent impact of the minority of IGOs with a genuine ability to influence state foreign policies. [End Page 5]
Second, states with extensive interests abroad are more likely to act internationally and are more likely to be members of international organizations. Even IGOs that are successful in reducing tensions between rivals may appear unsuccessful if at least part of the impetus for membership in IGOs follows from diplomatic need. States that share in many IGO memberships are more likely to interact—cooperatively, competitively, and occasionally violently—than are other states.
Realists will not be surprised by findings suggesting that IGOs fail to function as liberals would expect. According to Mearsheimer, IGOs have no international impact independent of that of powerful states.14 Major powers use IGOs to mold the global system and maintain or enhance their own power. "For realists, the causes of war and peace are mainly a function of the balance of power, and institutions largely mirror the distribution of power in the system."15 Moreover, concerns over relative gains and cheating block substantial cooperation even while institutions fail to alter system structure. Not all realists dismiss international institutions, however. Schweller and Preiss contend that traditional realists have always understood that IGOs can play a role in altering state behavior.16 Modified structural realists argue that institutions act as mechanisms through which powerful states exert control, impose rules, bind other states, and collude to preserve the status quo. In short, IGOs act as intervening variables between power and world politics.
A third challenge to traditional explanations involves the lack of an explicit theory of contests. For IGOs to influence dispute behavior, they must impinge on the causal processes that lead states to fight. An evaluation of the utility of IGOs as a means for promoting peace necessarily involves linking the capabilities and actions of IGOs with the decision calculus of states in conflict. Fearon argues that variables such as the distribution of power, threat, or interests—while key to determining the content of negotiated bargains (that is, who gets what)—tell us relatively little about why states occasionally resort to overt force.17 The symmetry inherent in dispute behavior requires that the "causes" of a contest must be remedied for the contest to end. If, for example, states fight because of an imbalance of power, then war must lead to a balance of power in order for peace to be restored. Uncertainty about the balance [End Page 6] of power, rather than the balance itself, can accommodate the need for a symmetric explanation for war. States fight not because of material or motivational factors per se but because they have private knowledge of the actual status of these factors. Since any claims about how IGOs influence dispute behavior must be embedded in some theory of why disputes occur, we adopt the bargaining approach here.
III. A Theory of IGO Influence on Peace and International Conflict
The causal processes linking IGOs and interstate conflict behavior are complex and indirect. The effect of intergovernmental organizations on disputes depends on the genesis and structure of IGOs, the origins of interstate contests, and the interaction of the two. Traditional approaches offer very different interpretations of IGOs, in part because they emphasize different elements in this complex causal process. We develop a middle path, based on bargaining theory, between the contrasting claims of realists and liberals. Our approach uses simple formal models to illustrate the likely effectiveness of different methods of third-party intervention. (See Appendix 1.) IGOs can promote peace, but success depends on attributes present in only the most cohesive and institutionalized organizations.
After summarizing the bargaining argument, we look within the basic framework for ways that can best diminish the risk of contests.18 Rather than develop complex three-actor models that incorporate additional aspects of strategic behavior, we believe that it is most appropriate at this point to identify which actions a third party can take that are most likely to influence competition between two states. Our goal is to identify how IGOs can best influence conflict behavior, not whether (and when) they will seek to do so.19 We argue that IGOs will have the greatest impact on dispute behavior in a limited number of ways related to mandate, member cohesion, and institutional structure. The resulting theory produces a number of empirical implications, some of which [End Page 7] cannot be evaluated here. Nevertheless, our tests support the theory while challenging existing alternatives.20 We hope to extend analysis of the theory in subsequent research.
The Bargaining Approach to War
Theories of war commonly conflate the motives for conflict with the choice of method for conflict resolution. Costly contests involve at least two elements. First, there exists a zero-sum competition for excludable goods.21 States differ over issues or territory that each cannot simultaneously control. Second, states choose a method of settlement, a choice that is non-zero sum. Fighting deprives victors of benefits and increases burdens on the vanquished, implying that states are collectively better off selecting methods that minimize the transaction costs of bargaining. Since war is costly and since almost all contests end in some settlement, fighting is rational only if states are unable to arrive at the settlements that eventually terminate contests prior to hostilities. A theory of war is thus an account of why states sometimes fail to achieve ex ante the bargains that obtain ex post.
Fearon identifies three conditions that can precipitate costly contests: issue indivisibilities, commitment problems, and information asymmetries.22 States can experience bargaining failures and war if the stakes in dispute are not readily divisible (mutually acceptable bargains are unavailable, either because apportioning the stakes is impossible or because prior action by one of the states has made compromise costly). Fearon discounts indivisibilities as a precipitant because side payments are generally available and appear to be practiced. For example, U.S. negotiators resolved a deadlock in talks ending the Spanish-American War by offering Spain $20 million in reparations.23 We follow Fearon [End Page 8] in arguing that indivisibility is a relatively infrequent cause of interstate war.24
Commitment problems occur when changes in the balance of power, in war costs, or in interests provide disincentives for a rising state to honor existing agreements. A declining state can fight because its strategic position is better in the present than in the future and because its opponent cannot credibly commit under anarchy to honor a contract. As with indivisibilities, contests that result from commitment problems must resolve the commitment problem in order to reconcile the motives for fighting: otherwise there is resort to involve military solutions. This implies that commitment problems lead to particularly intractable contests or knife-edged bargains in which neither party expects its opponent to gain significant advantage in the future.25 States may be able to use IGOs to improve the credibility of commitments, but IGO influence is again limited by capabilities and member cohesion.
Fearon follows Blainey in arguing that uncertainty about relative power, war costs, or interests represents the most common explanation for war.26 States possess private information about variables likely to influence a contest (capabilities, resolve, and so on). While an actor might reveal private information, competition makes revelation problematic. Only by fighting or similar actions can states distinguish resolved or capable opponents from those seeking to bluff. That some states are willing and able to fight does not explain the need for contests unless it is impossible to distinguish these states from others that are less capable or resolved. If states can agree about the content of eventual bargains, then opponents should mutually prefer the anticipated bargain to fighting. In short, disagreements about the nature of the eventual bargains explain the need for contests. [End Page 9]
Fearon's bargaining approach to war leads us to focus on the informational impact of IGOs. Uncertainty and incentives to compete can precipitate costly interstate warfare.27 IGOs can alleviate the motives for contests by credibly communicating information about strategic variables that otherwise remain the private domain of particular states. IGOs can also seek to reduce dispute behavior by sanctioning or intervention, though the bargaining approach suggests a different causal logic than that offered by traditional explanations. Below we examine ways in which third parties can reduce the likelihood or duration of international contests. We then tailor these generic insights to the characteristic features of intergovernmental organizations. First, however, we review ways that third parties are likely to be ineffective in limiting disputes.
Why Third Parties Often Fail to Influence Dispute Behavior
Students of international organization often argue that third parties (such as IGOs) influence states' interest in or ability to pursue conflict by altering the costs or benefits associated with contests. Deterrence attempts to reduce the chances of costly contests by making such contests more expensive or by reducing the odds of victory. Appeasement seeks to alter the probability of contests by increasing the net benefit of settlements made in lieu of fighting. Such arguments are plausible to the extent one believes that states are restricted to fixed and exogenously determined demands. Yet we normally think of anarchy and sovereignty as allowing states wide latitude in the formulation of agreements. Efforts to alter strategic conditions can change either the probability of a contest or the bargains states make in lieu of fighting. Bargaining theory emphasizes that altering cost/benefit parameters facing competitors will often fail to reduce the risk of war significantly.28 [End Page 10]
The strategic nature of contests means that any effort that changes costs or benefits for one actor in competition does so in a roughly reciprocal fashion for competitors: if war becomes more expensive or the net benefits of fighting diminish for one state, then war is relatively cheaper or more appealing for opponents.29 If competitors can convert advantage into better bargains, then simply altering the costs or benefits of fighting for one actor should have relatively little effect on the probability of a dispute. If IGOs intervene, changing the balance of power or interests, but allow competitors to bargain, we should often see little or no reduction in conflict behavior.
Proposition 1. Actions by IGOs that alter the bargaining power between two competitors, when common knowledge, encourage revised demands rather than just prompting a change in decisions to fight.
If competitors are informed about changes in strategic variables and are free to act, then the effects of changes should often be subsumed in bargaining, regardless of whether the change is endogenous to a contest or imposed by outside actors. Even where imposing costs succeeds in compelling peace (as in "corner solutions," where one party prefers any available offer to fighting), the approach is relatively inefficient, since much of the third party's effort is diverted by competitors into obtaining a better bargain. Attempting to deter or appease all parties might appear to be a remedy. However, just as in a chicken game, increasing the cost of fighting (or the benefits of peace) for all competitors only alters the odds of a contest if conflict is prohibitively costly regardless of what opponents demand.30 States may occasionally be uninformed about changes in the balance of power, threat, or interests, or they may be unable to bargain effectively. However, competitors clearly have incentives to be aware of and to act on new or better information. We argue that third parties can promote peace most often and most economically by influencing the informational conditions surrounding contests.
How Third Parties Can Best Promote Peace
IGOs can potentially affect dispute behavior in a variety of ways. First, they can act as mediators and information brokers.31 To the degree that [End Page 11] contests result from private information about capability and resolve, revelation of this information reduces the incentives to fight. Second, third parties can directly intervene in contests but constrain participants from converting advantages into more extractive demands where necessary. While changing the value of strategic variables may or may not diminish dispute behavior, third parties can promote peace by restraining competitors from acting on new conditions or by concealing information about new conditions from states that stand to benefit. Finally, third parties can reduce uncertainty by creating minor contests that have the effect of credibly communicating resolve among relevant parties (signaling). Talk is cheap in international relations.32 If threats of force face costly consequences (sanctions, diplomatic censure), then leaders' words can be more informative, allowing observers to draw more accurate inferences.
Informational Arbitrage
Third parties can act as information arbitrageurs, reducing the risk of war by revealing private information about strategic variables. The bargaining approach suggests that war often results from uncertainty and incentives to bluff. If a third party informs one state of the status of an opponent's military capabilities, for example, then the state can make concessions when necessary.
Proposition 2. IGOs promote peace by revealing private information about competitor states.
Nevertheless, there are barriers to effective third-party promotion of peace through information arbitrage. To function as an arbitrageur, a third party must have information that is not available to at least one of the competing states. In practical terms, the best way to achieve this is to have access to sophisticated administrative and intelligence-gathering capabilities. This implies that IGOs with extensive institutional structures or support from member countries (particularly major powers) will be more effective at promoting peace. Further, even by acting only as information arbitrageurs, third parties stand to disadvantage some parties and so will be seen as strategic participants.33 [End Page 12]
Intervention with Secrecy or Constraint
A second method by which third parties can promote peace is by altering the payoffs for one side in a conflict while keeping the new payoffs secret from other competitors or constraining opponents from using the new conditions to advantage themselves in bargaining. A third party, such as an IGO, can assist one side in a conflict or, alternatively, punish one or both parties. Unfortunately, as noted above, when a third party intervenes, the beneficiary of new strategic conditions has an incentive to seek to extract additional concessions from its opponent. If instead, the beneficiary does not know of the new strategic conditions or if the beneficiary is restrained from demanding greater concessions from its opponent, then intervention is likely to have a greater effect on peace.
Proposition 3. IGOs promote peace by intervening secretly or by constraining states that stand to benefit from IGO intervention.
Keeping changes secret is arguably the least taxing method in terms of resources for third parties. All that is necessary to promote peace is that the beneficiary be uninformed of new conditions. However, secrecy may also be the least effective method, especially for IGOs. Once new strategic conditions are revealed, failure ensues.34 Members of IGOs may have divergent objectives. States with partisan interests have incentives to divulge IGO secrets to fellow partisans in a conflict. Thus, IGOs that possess or develop a consensus among members are more likely to be successful in promoting peace, while IGOs housing divergent interests will more often face problems using this method.
Proposition 4. Preference homogeneity among IGO members increases the effectiveness of efforts to promote peace through intervention and secrecy.
Third parties can also change incentives for one side in a contest while constraining opponents from changing their demands. If an IGO punishes a state for fighting or provides the state with a subsidy for agreeing to a bargain, while its opponent is forced to make its regular offer, [End Page 13] then competitors will more often avoid disputes. The approach may be more difficult to implement than informational arbitrage or signaling, since it requires third parties to take two simultaneous actions (intervene and constrain). If the IGO intervenes to appease or deter one state, it must also restrain the counterpart. The institutional strength of the organization, as well as the cohesion and capabilities of members, will be critical in implementing intervention and constraint.35
Costly Signaling
The third option available to third parties in promoting peace is to confront states contemplating warfare with an additional layer of intermediate challenges. The bargaining approach implies that states fight because force represents one of the only available practical means for generating differential costs and thus for distinguishing between resolved competitors and those seeking to bluff.36 One way of reducing the frequency of disputes, then, is to create other methods of demonstrating resolve. Third parties can impose sanctions, embargo products, or take other actions that make it more costly for competitors to compete. Again, it is not the deterrent effect of these efforts that reduces the frequency of disputes.37 States can simply change their demands in response to new opportunities or constraints. Instead, the peace-producing effect of sanctions is to remove some of the uncertainty that surrounds competition. States that endure sanctions or other third-party punishment inform opponents by distinguishing themselves from other, less-resolved competitors.
Proposition 5. IGOs promote peace by facilitating costly signaling among states through the imposition of sanctions, embargoes, or other nonmilitarized punishments.
Costly signaling can be implemented even when the ability of the IGO to constrain competitors is partial or incomplete. As long as at least some competitors find sanctions burdensome, resulting differences in behavior allow observers to learn something new about relative resolve [End Page 14] or capabilities. Competitors that interact in the presence of IGO sanctions or other punishments will more often arrive at bargains prior to the onset of violence. The potential to act in ways that allow states to signal implies that IGOs need not have sufficient capabilities to impose their will on states in order to be effective. Still, IGOs that are primarily interested in security and that are willing to sacrifice economic and social interests in the pursuit of peace stand to be most effective. Third parties must also possess a mandate to punish states for political acts. IGOs with economic or social mandates may be disinclined or unable to punish states for purely political behavior. Thus, promoting peace through signaling is probably most effectively achieved through security-oriented IGOs.
Proposition 6. IGOs with a mandate as security organizations are more effective at promoting peace through costly signaling than are those with nonsecurity objectives and interests.
Bargaining failures can be reduced if competitors are confronted with "tests" prior to escalation to militarized acts. Holding punishments in reserve may be problematic, since it invites misperception on the part of both competing states and the third party. Concern about appearing ineffective may be misplaced. Even if punishments fail to deter, the resulting reduction in uncertainty about relative resolve among competitors allows for ex ante settlements. Punishments directly inhibit contests only if they are so large that the target finds any offer preferable to fighting. It seems unlikely in most contexts that states contemplating costly violent force will be cowed by the punishments available to most IGOs. Even more troubling, delay in implementation can create added uncertainty, increasing the likelihood of a contest. The ability of the third party to act decisively and with sufficient sanctioning power to influence events and inform competitors depends once again on mandate, institutional structure, member cohesion, and the presence of major powers.
Evaluating Igos
Testing bargaining theories is challenging because of the difficulty in observing information. We thus need to take an indirect route, seeking implications that are uniquely indicative of the theory. The need to adopt existing research designs also constrains our efforts. Our findings suggest, nonetheless, that the theory possesses considerable face validity. While individual results can be interpreted as supporting competing conventional perspectives, other findings contradict each classical view. We believe that our theory best explains the particular mixture of IGO strengths and weakness that we observe. [End Page 15]
We present four hypotheses designed to differentiate between our argument and the conventional alternatives. First, we have argued that disparate findings in the literature about the effect of IGOs on war and peace can be explained by variability in state activity. States that are active abroad share both more IGO memberships and a higher likelihood of involvement in militarized disputes.
Hypothesis 1. After controlling for the level of international activity, dyads with more IGO memberships should appear about as likely to experience militarized disputes as other dyads.
Second, the effectiveness of IGOs at promoting peace should vary with institutional structure. Institutionalized IGOs possess at least some of the resources needed for informational arbitrage, intervention and constraint (or secrecy), or signaling. Realists imply that the notion of IGOs with strong institutions is an oxymoron. Liberals and functionalists argue that merely constructing IGOs may be sufficient to promote peace. All treat IGOs as more or less uniform in their impact on conflict.
Hypothesis 2. Dyads in which states share membership in a large number of IGOs with greater institutional structure are less likely to experience militarized disputes.
Third, IGOs vary in the level of contentiousness among their members. IGOs require consensus to be effective. Dissatisfied members may seek to delay or obstruct interventions, or they may reveal secrets to protégés. Even more serious, IGOs that are undecided about intervention increase uncertainty, potentially leading competitors to fight when a contest might otherwise have been avoided. Traditional approaches fail to emphasize the impact of internal political struggles for IGO effectiveness. Indeed, functionalists see IGOs as ensuring that differences among states gradually "wither away."
Hypothesis 3. Dyads in which states share membership in IGOs with tighter member cohesion are less likely to be associated with militarized disputes.
Finally, IGOs differ in their organizational mandate. IGOs with a security mandate are more likely to influence disputes. This contrasts with the realist assertion that IGOs are hollow shells, but it also differs from functionalist and liberal claims that IGOs broadly transform interstate relations.
Hypothesis 4. Dyads in which states participate in more IGOs with a security mandate have a lower probability of experiencing a militarized dispute. [End Page 16]
IV. Research Design
This study uses pooled time-series cross-national data with a unit of analysis of nondirected dyad-years. We focus primarily on the all-dyads sample, though results using politically relevant dyads are generally the same.38 We base our analysis on Oneal and Russett, though data on IGO attributes limit the temporal domain to 1950-91.39 We first replicate Oneal and Russett's findings and then introduce additional variables measuring the attributes of IGOs. We employ both the BTSCS technique detailed by Beck, Katz, and Tucker and General Estimating Equations to correct for temporal dependence in the dependent variable.40 The former method uses a "peace year" variable and cubic splines to account for autocorrelation in militarized disputes. Due to limited space and because the results are comparable, we report only the logit/BTSCS analyses.
We use a new schema to code the IGO data, details of which are provided in Appendix 2. IGOs vary in their degree of institutionalization. At one end of the institutionalization continuum are organizations that contain only a nominal organizational structure. The heads of states or various ministers meet annually or biannually under the auspices of the organization. But agreements seldom require the organization to implement policy. The member states themselves are responsible for coordination and cooperation, which often leads to defections from agreements in the absence of further institutional support. The goals of these IGOs thus often go unfulfilled. Somewhere in the middle of the continuum are organizations with formal structures guided by procedures and rules regarding control of the agenda and voting. The members of some IGOs go further, making decisions that are binding on all members. Many of these organizations also possess bureaucracies that aid implementation and that produce other tangible benefits for members. Last, the most highly institutionalized organizations possess organs or mechanisms of mediation, arbitration, or adjudication aimed at conflict resolution and the enforcement of organizational decisions; or they may possess other benefits such as economic aid that can be withheld from states to influence decision making. These IGOs are also capable of uncovering [End Page 17] private strategic information, especially in cases where states forgo organizational benefits under sanctions, thus signaling resolve. Institutionalized IGOs should be most capable of affecting international conflict. IGOs are coded on a three-point scale of institutionalization.
- Minimal organizations contain plenary meetings, committees, and possibly a secretariat without an extensive bureaucracy beyond research, planning, and information gathering.
- Structured organizations contain structures of assembly, executive (nonceremonial), and/or bureaucracy to implement policy, as well as formal procedures and rules.
- Interventionist organizations contain mechanisms for mediation, arbitration and adjudication, and/or other means to coerce state decisions (such as withholding loans or aid), as well as means to enforce organizational decisions and norms.
Dependent Variable
The dependent variable comes from the Militarized Interstate Dispute data set produced by the Correlates of War Project.41 A MID occurs when a state threatens, displays, or uses military force against another state. We use Maoz's dyadic version (DYMID1.0), which also corrects for some errors in the basic MID data.42 MIDOnset only equals one for the onset year of a MID.43 Onset and duration are conceptually distinct processes to which IGOs might contribute in very different ways. Using Onset also addresses the generic problem of the nonindependence of subsequent dispute years. Of even greater importance here, the tendency for conflict and IGO membership to be serially autocorrelated could produce a misleading positive correlation between the variables.
Independent Variables
We analyze four different IGO membership variables. The first measure counts the number of intergovernmental organizations in which both states of a dyad are members in a given year. We begin by replicating Oneal and Russett's aggregate joint IGO variable.44 The replication variable [End Page 18] ranges in value from 1 to a maximum of 93. Data were collected from multiple sources. Data from 1950 to 1963 were taken from the Correlates of War Intergovernmental Organizations data set.45 Data for 1964 to 1992 come from the Yearbook of International Organizations,46 the Political Handbook of the World,47 and from information provided by individual organizations. We next subdivide IGO memberships according to the three institutionalization levels outlined above, to provide three different variables, one for each category in our typology.
IGO Mandate reflects the type of role framers of the IGO anticipated when the organization was founded, as formalized in the founding documents of the organization (security, economic, or other). In the analysis section, we use IGO Mandate to identify the sample of security IGOs.
Other attributes strengthen or undermine the ability of IGOs to foster interstate cooperation. Organizational efficacy is presumably affected by the degree of cohesion among members, irrespective of mandate or institutionalization. We measure the level of agreement among IGO members using data on the similarity of state voting patterns in the United Nations.48 The similarity of state preferences is measured for all members of a given IGO and ranked in quartiles of contentiousness. Values are then aggregated and averaged across all IGOs within a given dyad, so that the resulting variables can be used in our analysis. Floor Member Contention measures the average contentiousness among members of IGOs in which states of a given dyad share membership. We expect that IGOs that suffer from more contentious memberships are less able to act decisively to prevent or extinguish interstate violence. Similarly, Major Power Contention measures the average divergence of preferences among major power members of IGOs in which the states in a dyad share membership. Finally, we combine the two variables above into a third variable, Difference in Floor-Major [End Page 19] Power Contention, which reflects the degree of contentiousness between the average position of all states in the IGO relative to the mean position of major power members of the IGO.49
Our next explanatory variable controls for the magnitude of state interactions with the international system. There is likely to be a relationship between participating in IGOs and involvement in military contests. States active in numerous IGOs potentially have a greater number of international interests to defend. States that interact extensively also have more opportunities to engage in conflict. States with a large number of diplomatic relations are differentially likely to join IGOs and are more likely to experience conflict. Diplomatic Missions Low is based on Oneal and Russett's "weak link" assumption and equals the value for the state in a given dyad that maintains the fewest foreign diplomatic missions. We anticipate that the higher the value of Diplomatic Missions Low, the more likely it is that a MID will occur. Diplomatic Missions Low controls for international engagement.50
Oneal and Russett Control Variables
To maintain comparability with other studies, the remaining variables are all from Oneal and Russett.51 The two additional "legs" of the Kantian tripod for peace (along with IGOs) are joint democracy and trade interdependence. Both Democracy Low and Trade Dependence Low use the "weak link" assumption. We measure regime type by subtracting each state's autocracy score from its democracy score, as coded in Polity III,52 for an index from -10 (most autocratic) to +10 (most democratic). Democracy Low then equals the lower of the two state regime scores, while Trade Dependence Low equals the lower of the sum of a state's exports plus imports with its dyadic partner divided by its IGO. Again, the higher the lower value in a dyad, the less likely is a MID.
Oneal and Russett identify realist variables that potentially predict conflict. Allies equals one when the states of a dyad are allied, and zero otherwise. Similarly, Major Power Dyad equals one if at least one dyad member is a major power. Major powers by definition have greater opportunity to engage in MIDs. Of course, deterrence may work as well. Capability Ratio equals the ratio of the stronger dyadic state's capability [End Page 20] to the weaker state. We take the natural logarithm of the variable since power arguably has diminishing returns to scale. The greater the relative difference in power in a dyad, the less likely states are to fight. Lastly, distance has proven a reliable predictor of conflict. States are much less likely to fight with geographically distant counterparts. Two variables are included to control for geographic space. First, Contiguity is a dummy variable equal to one when dyad members are contiguous by land or by water within 150 miles (zero otherwise). Second, Distance is the natural log distance between capitals or major ports of dyad members.
Temporal Dependence
One of the primary functions of IGOs is to offer structures that encourage ongoing, amicable interstate interaction. Thus, time becomes a major factor in the analysis of IGOs and peace. States that have coexisted amicably for many years are more likely to remain at peace. IGOs may help to regularize interactions between states, or they may simply reflect a cooperative status quo. Russett, Oneal, and Davis fail to control for temporal dependence.53 Oneal and Russett and Gartzke, Li, and Boehmer find that, after controlling for temporal dependence with the Beck, Katz, and Tucker procedure, IGOs appear to increase the dispute propensity of member states.54 Militarized disputes (MIDs) and other indicators of interstate violence appear to be associated temporally. The probability of a dispute at time t may be statistically related to the duration of peace since the last conflict. Past decisions to cooperate coincide with a lack of motives for disputes and so IGOs appear to reduce disputes when in fact they simply coincide with prior peace.
V. Results
To assess the effects of IGOs on the conflict propensity of states, we first replicate analysis by Oneal and Russett.55 In Table 1 we use logistic regression to estimate MID Involvement, the Oneal and Russett dependent variable, with the Beck, Katz, and Tucker controls for serial [End Page 21] autocorrelation.56 Logistic regression constitutes a more demanding test than GEE, as the latter produces results friendlier to our theory and to that of Oneal and Russett. As reported by Gartzke, Li, and Boehmer and Oneal and Russett, introducing the BTSCS temporal controls implies that the effect of IGOs on disputes is positive and significant.57 Table 1 includes two regressions each for politically relevant dyads and for the larger all-dyads sample. Regression models 1-1 and 1-3 replicate Oneal and Russett, while models 1-2 and 1-4 introduce the control for international activity, Diplomatic Missions, suggested by hypothesis 1. Our contention that the association between conflict and IGO status can be explained by states' international activity is largely substantiated. In model 1-2, adding Diplomatic Missions makes the IGO variable statistically insignificant. In model 1-4, using all dyads, the effect is similar, though not as strong. Joint IGO Membership continues to be positively associated with disputes, though only at the marginal 10 percent level. While it is useful to control activity, more needs to be done to adequately assess the impact of IGOs on war and peace.
We can now begin to relax the assumption that all IGOs are created equal. Estimates in Tables 2 and 4 are again obtained using logistic regression, but with Onset as the dependent variable. The first column of Tables 2 and 4 reports estimates of a model using an aggregate variable including all joint IGO memberships in a dyad. The three remaining columns offer models with IGO membership variables disaggregated by level of institutional structure, beginning with minimally institutionalized organizations in the second column and moving to interventionist organizations in the right-hand-most column. Tables 3 and 5 detail substantive effects for the estimated coefficients.
Model 2-1 shows that there is no statistically significant effect of IGOs in aggregate on war and peace. Once we use a more appropriate dependent variable (Onset) and, as suggested by hypothesis 3, add variables measuring contentiousness among IGO members to the Oneal and Russett model, realist skepticism would appear to be justified. We are not yet finished modifying the Oneal and Russett model, however. Hypothesis 2 asserts that the level of IGO institutionalization is a critical factor. Aggregating IGO data produces one common coefficient estimate for all types of IGOs, while the "true" model arguably contains several coefficients for different IGO categories.
Breaking down the joint IGO variable by level of institutionalization reveals a different, more nuanced picture. The average IGO may be irrelevant [End Page 22]
Click for larger view |
Table 1
Igos, Time Dependence, International Interactions and Militarized Disputes Involvement (1950-91) |
Click for larger view |
Table 2
Igo Institutionalization, Contentiousness, and the Onset of Militarized Disputes (1950-91) |
to peace, but some IGOs significantly inhibit conflict. The coefficient for membership in minimal IGOs is negative but insignificant; the least institutionalized organizations do not discourage disputes. By contrast, joint membership in structured organizations, as reported in model 2-3, significantly reduces the probability of a dispute (a 12 percent decrease, as reported in Table 3). The overall number of organizations shared by states in a dyad is less important for interstate peace than the organization type. We thus find some support for our argument that IGOs require institutional structure in order to contribute to peace. Finally, interventionist organizations have no effect on the onset of MIDs. While this variable is negative and statistically significant in models estimated using GEE, we suspect that the category still contains unobserved heterogeneity. In particular, while both financial lending institutions and collective security organizations figure prominently as interventionist institutions, only the latter are likely to take active, coercive measures to affect the security behavior of states. Hypothesis 4 suggests that we should again disaggregate the data, differentiating between IGOs on the basis of their mandate. Before proceeding, however, we briefly review the other findings from Tables 2 and 3.
Click for larger view |
Table 3
Igo Institutionalization, Contention, and Mid Onset |