Do Intergovernmental Organizations Promote Peace?: by Charles Boehmer, Erik Gartzke, and Timothy Nordstrom

v57.1.001.boehmer.
001-038
6/24/05
9:52 AM
Page 1
DO INTERGOVERNMENTAL
ORGANIZATIONS
PROMOTE PEACE?
By CHARLES BOEHMER, ERIK GARTZKE,
and TIMOTHY NORDSTROM*
I. INTRODUCTION
RADITIONAL 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 tripoddemocracy, cross-border trade, and international organizationsappear 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
* We thank David Bearce, Quan Li, John Oneal, Matt Ruppert, Bruce Russett, Holger Schmidt,
and Erik Voeten for helpful comments. Erik Gartzke thanks Ned Lebow and the Mershon Center,
the Ohio State University, where he was a postdoctoral fellow in 20002001. An earlier draft of the
paper was presented at the annual meeting of the Peace Science Society (International), Tucson, Ariz.,
November 13, 2002. Data for the project are available from the authors.
1
See, for example, Bruce Russett, John R. Oneal, and David Davis, The Third Leg of the Kantian Tripod for Peace: International Organizations and Militarized Disputes, 19501985, International
Organization 52, no. 3 (1998); Bruce Russett and John R. Oneal, Triangulating Peace: Democracy, Interdependence, and International Organizations (New York: W. W. Norton, 2001).
2
See John R. Oneal and Bruce Russett, The Kantian Peace: The Pacific Benefits of Democracy,
Interdependence, and International Organizations, 18851992, World Politics 52 (October 1999);
Erik Gartzke, Quan Li, and Charles Boehmer, Investing in the Peace: Economic Interdependence
and International Conflict, International Organization 55, no. 2 (2001).
World Politics 57 (October 2004), 138
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 2
WORLD POLITICS
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 implicationsonly some of which are explored hereeven 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
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 3
INTERGOVERNMENTAL ORGANIZATIONS
&
PEACE
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
3
John Mearsheimer, The False Promise of International Institutions, International Security 19,
no. 3 (1995); Randall L. Schweller and David Preiss, A Tale of Two Realisms: Expanding the Institutions Debate, Mershon International Studies Review 41, no. 1 (1997). And for reviews of and perspectives on the literature, see Lisa L. Martin and Beth A. Simmons, Theories and Empirical Studies of
International Organizations, International Organization 52, no. 4 (1998).
4
See, for example, Ernst B. Haas, Beyond the Nation-State: Functionalism and International Organization (Palo Alto, Calif.: Stanford University Press, 1964); Robert O. Keohane and Joseph S. Nye, Power
and Interdependence (New York: Harper Collins, 1989); Oran R. Young, International Regimes: Toward a New Theory of Institutions, World Politics 39 (October 1986); idem, The Effectiveness of International Institutions: Hard Cases and Critical Variables, in James M. Rosenau and Ernst-Otto
Czempiel, eds., Governance without Government (Cambridge: Cambridge University Press, 1992); John
Gerard Ruggie, Continuity and Transformation in the World Polity: Toward a Neorealist Synthesis,
World Politics 35 ( January 1983); idem, The False Promise of Realism, International Security 20, no.
1 (1995); James M. Rosenau, Citizenship in a Changing Global Order, in Rosenau and Czempiel,
Governance without Government; and Alexander Wendt, Constructing International Politics, International Security 20, no. 1 (1995).
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 4
WORLD POLITICS
rise of 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
5
Robert O. Keohane, After Hegemony: Cooperation and Discord in the World Economy (Princeton:
Princeton University Press, 1984); Robert Axelrod, The Evolution of Cooperation (New York: Basic
Books, 1984); Robert Axelrod and Robert O. Keohane, Achieving Cooperation under Anarchy:
Strategies and Institutions, World Politics 38 (October 1985); Robert Powell, Absolute and Relative
Gains in International Relations Theory, American Political Science Review 85, no. 4 (1991); Duncan
Snidal, Relative Gains and the Pattern of International Cooperation, American Political Science Review 85, no. 3 (1991); Robert O. Keohane and Lisa Martin, The Promise of Institutionalist Theory,
International Security 20, no. 1 (1995); Charles A. Kupchan and Clifford A. Kupchan, The Promise of
Collective Security, International Security 20, no. 1 (1995); Duncan Snidal and Kenneth Abbott,
Why States Act through Formal International Organizations, Journal of Conflict Resolution 42, no. 1
(1998); Martin and Simmons (fn. 3).
6
Snidal (fn. 5) points out that contrasting predictions of absolute and relative gains depend on the
number of actors involved. For an informative debate, see comments by Joseph M. Grieco, Robert
Powell, and Duncan Snidal, The Relative-Gains Problem for International Cooperation, American
Political Science Review 87, no. 3 (1993). See also James S. Mosher, Relative Gains Concerns When
the Number of States in the International System Increases, Journal of Conflict Resolution 47, no. 5
(2003); and David L. Rousseau, Motivations for Choice: The Salience of Relative Gains in International Relations, Journal of Conflict Resolution 46, no. 3 (2002).
7
Subsequent research suggests that the shadow of the future can actually inhibit cooperation. See
Curtis S. Signorino, Simulating International Cooperation under Anarchy: The Effects of Symmetric and Asymmetric Noise, Journal of Conflict Resolution 40, no. 1 (1996); James D. Fearon, Bargaining, Enforcement, and International Cooperation, International Organization 52, no. 2 (1998); J.
Samuel Barkin, Time Horizons and Multilateral Enforcement in International Cooperation, International Studies Quarterly 48, no. 2 (2004).
8
J. David Singer and Michael Wallace, Intergovernmental Organization and the Preservation of
Peace, 18161964: Some Bivariate Relationships, International Organization 24, no. 3 (1970);
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 5
&
PEACE
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.
Michael Wallace and J. David Singer, Intergovernmental Organization in the Global System,
18151964: A Quantitative Description, International Organization 24, no. 2 (1970); William
Domke, War and the Changing Global System (New Haven: Yale University Press, 1988); Young (fn. 4,
1992); Martin and Simmons (fn. 3).
9
Martin and Simmons (fn. 3) provide a review of the extensive case study literature.
10
John R. Oneal, Bruce Russett, and Michael L. Berbaum, Causes of Peace: Democracy, Interdependence, and International Organizations, International Studies Quarterly 47, no. 3 (2003); Oneal
and Russett (fn. 2); Russett, Oneal, and Davis (fn. 1).
11
Harold Jacobson and Domke (fn. 8) provide typologies of IGOs; Jacobson, Networks of Interdependence: International Organizations and the Global Political System (New York: Knopf, 1984). However, Harold Jacobson, William M. Reisinger, and Todd Mathers use an inappropriate dependent
variable, whereas Domke relies on disaggregated probit estimates for each year; see Jacobson,
Reisinger, and Mathers, National Entanglements in Governmental Organizations, American Political Science Review 80, no. 1 (1986).
12
Oneal and Russett (fn. 2); and Gartzke, Li, and Boehmer (fn. 2).
13
Singer and Wallace (fn. 8).
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 6
WORLD POLITICS
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 interactcooperatively, competitively, and occasionally violentlythan 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 interestswhile 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 bal14
Mearsheimer (fn. 3).

Ibid., 13.
16
Schweller and Preiss (fn. 3).
17
James D. Fearon, Rationalist Explanations for War, International Organization 49, no. 3 (1995).
See also James D. Morrow, Capabilities, Uncertainty, and Resolve: A Limited Information Model of
Crisis Bargaining, American Journal of Political Science 33, no. 2 (1989).
15
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 7
&
PEACE
ance 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
18
For reviews of the rationalist literature on bargaining and war, see Dan Reiter, Exploring the
Bargaining Model of War, Perspectives on Politics 1, no. 1 (2003); and Robert Powell, Bargaining Theory and International Conflict, Annual Review of Political Science 5 (2002). For a critique, see Jonathan
Kirshner, Rationalist Explanations for War? Security Studies 10, no. 1 (2000).
19
The notion that IGOs are selective in their mediation efforts is contrasted by normative and institutional incentives to intervene. Unlike peacekeeping, say, where the investment is high, organizing
talks, pressuring delegates, brokering deals, informing parties, and promoting sanctions are all relatively low-cost actions. These activities are often explicitly encouraged by IGO charters, members, and
interested publics. Indeed, the proliferation of security IGOs, the bulk of which are unequipped to impose peace, would appear peculiar if they were not meant to conduct such activities.
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 8
WORLD POLITICS
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 nonzero 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
20
Formal models serve, among other things, to tie tests of implications of an argument to untested
elements of a theory. See Rebecca B. Morton, Methods and Models: A Guide to the Empirical Analysis of
Formal Models in Political Science (Cambridge: Cambridge University Press, 1999).
21
States can have no incentive to dispute benefits that cannot be denied to the loser or third parties. If players value relative gains, however, war can result from nonrivalrous goods.
22
Fearons list is not exhaustive. Risk acceptance can motivate contests, as can a consumption value
for war. There is also a broader set of commitment problems that is assumed away by Fearons modeling framework. See, for example, Michelle R. Garfinkel and Stergios Skaperdas, Conflict without
Misperception or Incomplete Information: How the Future Matters, Journal of Conflict Resolution 44,
no. 6 (2000); Branislav Slantchev, The Power to Hurt: Costly Conflict with Completely Informed
States, American Political Science Review 97, no. 1 (2003). Still, Fearons characterization of war is coherent, widely addressed in the literature, and tractable.
23
David F. Trask, The War with Spain in 1898 (New York: Macmillan, 1981), 44566. The likelihood of an agreement was greatly assisted by a proposal emanating from Senator Frye that the United
States should offer financial compensation in exchange for the Philippines; Joseph Smith, The SpanishAmerican War: Conflict in the Caribbean (London: Longman, 1994), 199200.
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 9
&
PEACE
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.
24
States often have incentives to seek to generate indivisibilities. Commitment increases bargaining power by making it harder to compromise. Third parties could allow states to extract themselves
from commitments. Blaming an IGO for having to renege on a promise may free a leader from domestic audience costs, for example. It is also possible, however, that IGOs increase opportunities for generating indivisibilities. IGOs allow leaders to formalize agreements that are later binding (or at least
impinging) on domestic politics. We do not explore these issues here.
25
Commitment problems and indivisibilities may be more prevalent in intrastate conflict, where
settlements often include a loss of autonomy for one of the competing parties or where any settlement
implies recognition, which is part of the stakes in the contest. See Barbara F. Walter, Explaining the
Intractability of Territorial Conflict, International Studies Review 5, no. 4 (2003); Monica Duffy Toft,
The Geography of Ethnic Violence: Identity, Interests, and the Indivisibility of Territory (Princeton: Princeton University Press, 2003); James D. Fearon and David Laitin, Explaining Interethnic Cooperation,
American Political Science Review 90, no. 4 (1996); idem, Ethnicity, Insurgency, and Civil War, American Political Science Review 97, no. 1 (2003).
26
For a succinct and highly intuitive account of the bargaining argument, see also James D. Morrow, How Could Trade Affect Conflict? Journal of Peace Research 36, no. 4 (1999).
v57.1.001.boehmer.001-038
10
6/24/05
9:52 AM
Page 10
WORLD POLITICS
Fearons 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
27
Erik Gartzke, War Is in the Error Term, International Organization 53, no. 3 (1999). The
asymmetric information argument is not theoretically sufficient to explain war. A sufficient account of
the causes of war requires that researchers access the private information of competing states.
28
Fearon (fn. 17) provides a formal proof. We use Fearon as the basis for our theory. Robert Powell, In the Shadow of Power: States and Strategies in International Politics (Princeton: Princeton University Press, 1999); idem, The Inefficient Use of Power: Costly Conflict with Complete Information
American Political Science Review 98, no. 2 (2004). Garfinkel and Skaperdas (fn. 22); and Slantchev (fn.
22) offer alternative bargaining frameworks in which commitment problems are a more important
cause of war and contests can occur under full information. Which characterization of bargaining is
most general empirically remains a subject of debate. Note, however, that alternative conceptions hinge
on assumptions about the sequence of offers in bargaining. Since states often negotiate about how they
will negotiate (structure is endogenous), and since a satisfactory theory of this aspect of bargaining has
yet to be introduced to international relations, we adopt the simplest model of bargaining as a place of
departure. The theory we supply is shown here to better explain observed behavior than conventional
liberal or realist alternatives.
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 11
&
PEACE
11
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 partys 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
29
The security dilemma makes clear the zero-sum nature of dynamics in the balance of power.
However, realists have failed to extend the insight to diplomatic bargaining. Factors that alter the likely
outcomes of contests should alter the demands of diplomats in roughly the same fashion.
30
Changing the payoffs in the chicken game alters play only if changes are ordinal. The effect is
stronger in a bargaining game, where competitors set payoffs endogenously through their offers.
31
A review of the traditional literature on the effects of third-party mediation can be found in
Lilach Gilady and Bruce Russett, Peace-Making and Conflict Resolution, in Walter Carlesnaes,
Thomas Risse, and Beth Simmons, eds., Handbook of International Relations (London: Sage, 2002).
v57.1.001.boehmer.001-038
12
6/24/05
9:52 AM
Page 12
WORLD POLITICS
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
opponents 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
32
For a discussion of credible diplomacy, see Anne Sartori, The Might of the Pen: A Reputational
Theory of Communication in International Disputes, International Organization 55, no. 3 (2002).
33
IGOs must also convince states of the veracity of information. Paradoxically, a third-party intent
only on averting a contest may be less effective than a biased actor with preferences over the outcome of
bargaining. Randall Calvert demonstrates that information from biased political sources can be informative; Calvert, The Value of Biased Information: A Rational Choice Model of Political Advice, Journal
of Politics 47, no. 2 (1985). Andrew Kydd extends this logic to the context of third-party mediation of
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 13
&
PEACE
13
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,
interstate conflicts; Kydd, Which Side Are You On? Bias, Credibility, and Mediation, American Journal of Political Science 47, no. 4 (2003). In a subsequent paper, Kydd points out that reputation can be
used to make unbiased mediators more credible, though the balance between costs and credibility is
critical. See Andrew Kydd, The Honest Broker: Mediation and Mistrust (Manuscript, Harvard University, Cambridge, July 2004). Similarly, Robert W. Rauchhaus relaxes the assumption that mediators prefer peace; Rauchhaus, Asymmetric Information, Mediation and Conflict Management
(Manuscript, University of California, Santa Barbara, 2003).
34
Secret threats lack the consequences often viewed as critical to successful deterrence. See Robert
Powell, Nuclear Deterrence Theory: The Search for Credibility (Cambridge: Cambridge University Press,
1990). Uncertainty about third-party actions could increase the onset of disputes.
v57.1.001.boehmer.001-038
6/24/05
14
9:52 AM
Page 14
WORLD POLITICS
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 peaceproducing effect of sanctions is to remove some of the uncertainty that
surrounds competition. States that endure sanctions or other thirdparty 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
35
Neutral third parties, while arguably less effective at information arbitrage, are likely to be more
effective at intervention and constraint. Biased IGOs may fail to constrain their favored faction, leading
to more extractive demands and a higher risk of war.
36
Actors can autonomously reveal resolve through costly actions, but burning money also weakens bargaining power. We should see such actions only when the reduction in uncertainty leads to payoffs exceeding the cost in resources and the loss of leverage. See David Austen-Smith and Jeffrey S.
Banks, Cheap Talk and Burning Money (Manuscript, Northwestern University, Evanston, Ill., and
the California Institute of Technology, Pasadena, Calif., 1998).
37
Gartzke, Li, and Boehmer (fn. 2).
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 15
&
PEACE
15
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.
v57.1.001.boehmer.001-038
16
6/24/05
9:52 AM
Page 16
WORLD POLITICS
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
with greater institutional structure are less likely to experience militarized
disputes.
IGOs
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 protgs. 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.
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 17
&
PEACE
17
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 195091.39 We first replicate
Oneal and Russetts 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 un38
All results mentioned but not published in the study are available from the authors.
Oneal and Russett (fn. 2). We collected data on institutionalization for 297 IGOs, coding 198 as
minimal, 52 as structured, and 47 as interventionist. Details for defunct IGOs are often unavailable.
40
Nathaniel Beck, Jonathan N. Katz, and Richard Tucker, Taking Time Seriously: Time-SeriesCross-Section Analysis with a Binary Dependent Variable, American Journal of Political Science 42, no.
4 (1998).
39
v57.1.001.boehmer.001-038
18
6/24/05
9:52 AM
Page 18
WORLD POLITICS
covering 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.
1. Minimal organizations contain plenary meetings, committees, and possibly a secretariat without an extensive bureaucracy beyond research, planning,
and information gathering.
2. Structured organizations contain structures of assembly, executive (nonceremonial), and/or bureaucracy to implement policy, as well as formal procedures
and rules.
3. 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 Maozs dyadic version (DYMID1.0), which also corrects for
some errors in the basic MID data.42 MID Onset 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 Russetts aggregate joint IGO variable.44 The replication vari41
Daniel Jones, Stuart Bremer, and J. David Singer, Militarized Interstate Disputes, 18161992:
Rationale, Coding Rules, and Empirical Patterns, Conflict Management and Peace Science 15, no. 2
(1996); Charles S. Gochman and Zeev Maoz, Militarized Interstate Disputes, 18161976: Procedure,
Patterns, and Insights, Journal of Conflict Resolution 28, no. 4 (1984).
42
Zeev Maoz, Dyadic Militarized Interstate Disputes (DYMID 1.1) Dataset, Version 1.1 (Codebook, Department of Political Science, Tel Aviv University, 1999).
43
For a useful discussion of the debate over onset and ongoing MIDs, see D. Scott Bennett and
Allan Stam, Research Design and Estimator Choices in the Analysis of Interstate Dyads: When Decisions Matter, Journal of Conflict Resolution 44, no. 5 (2000).
44
Oneal and Russett (fn. 2). Replication is approximate given idiosyncrasies in data and coding.
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 19
&
PEACE
19
able 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 Floor45
Wallace and Singer (fn. 8). COW data are reported in five-year intervals. We interpolate missing
values, as do Oneal and Russett (fn. 2).
46
Union of International Associations, Yearbook of International Organizations (Munich: K.G. Saur,
various years).
47
Arthur Banks and Thomas C. Muller, Political Handbook of the World (Binghamton, N.Y.: CSA
Publications, various years).
48
Affinity provides annual measures of the similarity of dyadic voting positions in the General Assembly (194696). Data construction is discussed elsewhere. See Erik Gartzke, Kant We All Just Get
Along? Motive, Opportunity, and the Origins of the Democratic Peace, American Journal of Political
Science 42, no. 1 (1998); idem, Preferences and the Democratic Peace, International Studies Quarterly
44, no. 2 (2000). We use the United States as the reference country to convert dyadic values to
monadic values. We also examined a variable based on nominate coding of state ideal points used in
Erik Voeten, Clashes in the Assembly, International Organization 54, no. 2 (2000). Designation of
ideal points involves additional assumptions about state utilities that are not immune to controversy
and that yield nonintuitive results in our analysis.
v57.1.001.boehmer.001-038
20
6/24/05
9:52 AM
Page 20
WORLD POLITICS
Major 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 Russetts 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
states 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 states 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 states capability
49
We also created an alternative measure of IGO member contentiousness by inverting and summing quartiles. The resulting variables correlate highly with the IGO institutionalization variables.
50
In future research we plan to explore selection effects in IGOs using a different unit of analysis.
51
Oneal and Russett (fn. 2).
52
Keith Jaggers and Ted R. Gurr, Transitions to Democracy: Tracking Democracys Third Wave
with the Polity III Data, Journal of Peace Research 32, no. 4 (1995).
v57.1.001.boehmer.001-038
6/24/05
9:52 AM
Page 21
&
PEACE
21
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
53
Russett, Oneal, and Davis (fn. 1).

Ibid; Gartzke, Li, and Boehmer (fn. 2).
Oneal and Russett (fn. 2). We use Oneal and Russetts data and statistical model to establish a
baseline for comparison. The changes in data and variable construction mentioned previously are reported in subsequent tables. We also replicated all regressions using the ReLogit procedure, but found
no substantive differences in results. See Gary King and Langche Zeng, Explaining Rare Events in
International Relations, International Organization 55, no. 3 (2001); idem, Logistic Regression in
Rare Events Data, Political Analysis 98 (2001).
54
55
v57.1.001.boehmer.001-038
22
6/24/05
9:52 AM
Page 22
WORLD POLITICS
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 ir56
57
Beck, Katz, and Tucker (fn. 40).

Oneal and Russett (fn. 2); Gartzke, Li, and Boehmer (fn. 2).
0.003**
0.009***
8.852*
0.040***
0.154*
0.134***
0.051**
0.155***
0.008***
0.016***
0.016***
0.053
0.408***
0.005
0.009
0.040
33.610
0.110
0.232
0.588
0.052
0.017
0.175
0.282
0.133
0.005
1.516
0.006
0.002***
0.012***
23.848
0.057*
0.182
0.217**
0.072
0.239
0.015***
0.028***
0.030***
0.057
0.641**
n = 33337
n = 20708
Log Likelihood = 4267.3 Log Likelihood = 2234.2
0.008
0.055
18.253
0.245
0.301
0.757
0.125
0.589
0.106
0.164
0.074
0.033
1.462
0.004***
0.009***
10.603**
0.040***
0.157**
0.153***
0.049***
0.135***
0.008***
0.015***
0.014***
0.082
0.406***
n = 149404
Log Likelihood = 5732.4
0.016
0.063
31.073
0.191
0.369
1.586
0.361
1.711
0.100
0.152
0.051
0.134
1.617
0.006*
0.002***
0.012***
22.617*
0.055
0.200
0.195***
0.080***
0.198***
0.013***
0.025***
0.023***
0.082
0.701**
n = 110769
Log Likelihood = 3111.1
0.015
0.010
0.058
43.245
0.037
0.304
1.582
0.333
1.214
0.170
0.266
0.090
0.127
1.671
All Dyads
Coefficient
S.E.
Model 1-4
Membership in
All IGOs
9:52 AM
*p < .10, **p < .05, ***p < .01
Model 1-3
Membership in
All IGOs
6/24/05
Model 1-2
Membership in
All IGOs
Politically Relevant Dyads Politically Relevant Dyads

All Dyads
Coefficient
S.E.
Coefficient
S.E.
Coefficient
S.E.
Model 1-1
Membership in
All IGOs
TABLE 1
DEPENDENCE, INTERNATIONAL INTERACTIONS AND MILITARIZED DISPUTES INVOLVEMENT (195091)
Joint IGO Membership

Diplomatic Missions, Low
Democracy, Low
Interdependence, Low
Capability Ratio
Allies
Geographic Contiguity
Distance between Capitals
Major Power Dyad
Peace Years
Spline 1
Spline 2
Spline 3
Constant
Variable
IGOS, TIME
v57.1.001.boehmer.001-038
Page 23
TABLE 2
0.0078
0.0536
0.2051
0.0857
0.0086
0.0325
10.3706
0.0418
0.065
1.9381
0.2965
0.9416
0.3580
0.0012
0.0003
0.0001
2.2606
Joint IGO Membership

Joint Membership, Minimal
Joint Membership, Structured
Joint Membership, Interventionist
Floor Member Contention
Major Power Contention
Dif FloorMP Contention
Democracy, Low
Capability Ratio
Allies
Major Power Dyad
Peace Years
Spline 1
Spline 2
Spline 3
Constant
n = 109965
Log likelihood = 2327.6
0.0117
0.0652
0.0877**
0.0875
0.0023***
0.0115***
14.7170
0.0479
0.1920
0.1869***
0.0711***
0.1882***
0.0347***
0.0003***
0.0002
0.0000**
0.7447***
S.E.
Coefficient
n = 109965
0.0162
0.0649
0.0866**
0.0887
0.0023***
0.0118***
14.0504
0.0476
0.1868
0.1880***
0.0695***
0.1873***
0.0347***
0.0003***
0.0002
0.0000**
0.7468***
S.E.
Model 2-4
Membership in
Interventionist IGOs
0.0429**
0.0094
0.0672
0.0602
0.0865**
0.2022
0.0804
0.0738
0.0024***
0.0081
0.0121**
0.0352
13.0465
12.9022
0.0473
0.0405
0.1833
0.098
0.1892***
1.9281
0.0738*** 0.2906
0.1889***
0.9357
0.0345*** 0.3577
0.0003*** 0.0012
0.0002
0.0003
0.0000**
0.0001
0.7815*** 2.3040
S.E.
n = 109965
0.0854
0.0508
0.2080
0.0854
0.0098
0.0253
5.4724
0.0372
0.040
1.9390
0.3175
0.9299
0.3578
0.0012
0.0003
0.0001
2.1044
Coefficient
Model 2-3
Membership in
Structured IGOs
9:53 AM
*p < .10, **p < .05, ***p < .01

a
Contention variables based on the Affinity data.
Coefficient
0.0069
0.0045
0.0661
0.0615
0.0861**
0.2128
0.0879
0.0596
0.0023***
0.0083
0.0117
0.0353
14.1792
12.6852
0.0480
0.0393
0.1917
0.094
0.1885***
1.9279
0.0714*** 0.2922
0.1892***
0.9417
0.0347*** 0.3577
0.0003*** 0.0012
0.0002
0.0003
0.0000**
0.0001
0.7551*** 2.3941
S.E.
Model 2-2
Membership in
Minimal IGOs
ONSET OF MILITARIZED DISPUTES (195091)
6/24/05
n = 109965
Coefficient
Variable a
Model 2-1
Membership in
All IGOs
IGO INSTITUTIONALIZATION, CONTENTIOUSNESS, AND THE
v57.1.001.boehmer.001-038
Page 24
v57.1.001.boehmer.001-038
6/24/05
9:53 AM
Page 25
&
PEACE
25
TABLE 3
IGO INSTITUTIONALIZATION, CONTENTION, AND MID
Based on Model 2-3

Baseline
Joint Membership, Structured
Democracy, Low
Capability Ratio
Allies
Major Power Dyad
ONSET
Probability a
% Change from Baseline
0.0018
0.0015
0.0018
0.002
0.0019
0.0022
0.0015
0.0017
0.0017
0.0017
0.0121
0.0013
0.0044
12.12
4.08
14.89
8.02
22.97
15.41
2.61
5.45
3.92
587.82
24.83
152.71
a
prob after 1 standard deviation increase, or max value if binary, in given x variable holding others
at their means or mins.
relevant 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.
Results for the IGO contentiousness variables are consistent across
the four models in Table 2. As suggested by hypothesis 3, organizations
in which members share a substantially common agenda are less likely
to get bogged down in intramural argument and more likely to act.
S.E.
Coefficient
Coefficient
n = 109965
0.1244
0.0677
0.0889**
0.0833
0.0023***
0.0119***
13.9332
0.0473
0.1725
0.1889***
0.0727***
0.1894***
0.0346***
0.0003***
0.0002
0.0000**
0.7633***
S.E.
n = 109965
0.0881
0.1065
0.0645
0.0500
0.0879**
0.2224
0.0866
0.0594
0.0023***
0.0083
0.0115*** 0.0348
13.7941
12.5133
0.0472
0.0375
0.1863
0.0461
0.1850***
1.9267
0.0670*** 0.2986
0.1849***
0.9265
0.0346*** 0.3575
0.0003*** 0.0012
0.0002
0.0003
0.0000**
0.0001
0.7208*** 2.3331
S.E.
Model 4-4
0.1335***
0.0694
0.0859*
0.0853**
0.0024***
0.0116***
12.8378
0.0479
0.1843
0.1838***
0.0692***
0.1878***
0.0343***
0.0003***
0.0002
0.0000**
0.7313***
S.E.
n = 109965
0.5055
0.0151
0.1414
0.1814
0.0096
0.0339
5.3972
0.0355
0.0046
1.9187
0.2954
1.0096
0.3481
0.0011
0.0003
0.0001
2.1192
Coefficient
Membership in
Interventionist IGOs
9:53 AM
n = 109965
0.13557 0.0635**
0.1388
0.0272
0.0690
0.0727
0.2016
0.0868**
0.2152
0.1122
0.0870
0.0325
0.0085
0.0023***
0.0084
0.0327
0.0118*** 0.0375
8.6261 13.8807
14.9055
0.0379
0.0477
0.0364
0.0483
0.1904
0.1598
1.9383
0.1894***
1.9087
0.2998
0.0747*** 0.2932
0.9343
0.1870***
0.9459
0.3552
0.0343*** 0.3576
0.0012
0.0003*** 0.0012
0.0003
0.0002
0.0003
0.0001
0.0000**
0.0001
2.2372
0.7789*** 2.4007
Coefficient
Model 4-3
Membership in
Structured IGOs
6/24/05
* p < .10, **p < .05, ***p < .01

a
Contention variables based on the Affinity data.
Joint Security IGO Membership

Joint Security Mem, Minimal
Joint Security Mem, Structured
Joint Security Mem, Interventionist
Democracy, Low
Capability Ratio
Allies
Major Power Dyad
Peace Years
Spline 1
Spline 2
Spline 3
Constant
Variable a
Model 4-2
Membership in
Minimal IGOs
Model 4-1
Membership in
All IGOs
TABLE 4
INSTITUTIONALIZATION, CONTENTIOUSNESS, OF SECURITY IGOS AND THE ONSET OF MILITARIZED DISPUTES (195091)
v57.1.001.boehmer.001-038
Page 26
v57.1.001.boehmer.001-038
6/24/05
9:53 AM
Page 27
&
PEACE
27
TABLE 5
SECURITY IGO INSTITUTIONALIZATION, CONTENTION, AND MID ONSET
Based on Model 3-4
Baseline
Joint Security Mem, Interventionist
Democracy, Low
Capability Ratio
Allies
Major Power Dyad
Probability a
% Change from Baseline
0.0017
0.0013
0.0017
0.0019
0.0020
0.0021
0.0014
0.0017
0.0016
0.0017
0.0115
0.0013
0.0047
22.0
1.2
9.9
17.8
22.5
20.1
2.6
5.2
0.5
574.4
23.3
173.6
a
prob after 1 standard deviation increase, or max value if binary, in given x variable, holding others
at the means or mins.
While contention among organizational members is consistently associated with a higher probability of conflict, the effect is not significant.
By contrast, disagreement between major power members significantly
increases the likelihood of militarized disputes. As reported in all four
models of Table 2, dyads that share IGO memberships with a high degree of contentiousness among major power members are at higher risk
for involvement in a dispute (roughly a 15 percent increase, as reported
in Table 3). The effect is much weaker for contention between major
powers and floor members of IGOs. Thus, our findings offer some support for the argument that major power cooperation bears on IGO efficacy. The results also show, however, that IGOs have an impact
independent of major powers; IGOs are more than simply hollow passthroughs for major power influence.
Concerning the control variables, note also that while major powers
and states that maintain extensive diplomatic ties globally are more
likely to become involved in a MID, the effects of alliances and relative
power appear minimal. The impact of relative power appears insignificant (literally), as suggested by proposition 1. As predicted by advocates
of the Kantian peace, democracy appears to reduce MID onset by about
15 percent, although the substantive effect of trade dependence is minimal. As we would also expect, the risk of a MID appears to increase dramatically among contiguous states and in dyads that include at least
one major power. Diplomatic Missions is significant and positive in all
v57.1.001.boehmer.001-038
28
6/24/05
9:53 AM
Page 28
WORLD POLITICS
models, regardless of which estimation technique, which dependent

variable, and which set of dyads are examined. States that are more involved in the international system tend to be more conflict prone.
These are also the states most likely to join large numbers of IGOs in
order to advance or safeguard their global interests. Failing to control
for variability in international engagement confounds efforts to identify the effect of IGOs on peace.
While the results in Table 2 show that structured IGOs reduce the
risk of interstate conflict, they also fail to demonstrate that interventionist organizations play a similar role. Hypothesis 4 suggests a course
of action. The interventionist category contains a large proportion of financial IGOs and other organizations that contain no explicit security
mission. For this reason, we test our models again, this time further disaggregating IGOs to identify those with mandates addressing international security. Table 4 reports logit estimates for security IGOs, while
Table 5 illustrates the substantive impact of relevant variables from
Table 4. It appears that security organizations in general reduce militarized disputes, as reported in model 4-1, but subsequent regressions reveal that the level of institutionalization matters greatly. In fact,
minimally institutionalized security organizations (model 4-2) marginally contribute to interstate conflict. As with our other findings, there is
no indication that the mere presence of organizations encourages stability or discourages conflict escalation. Even for structured security
IGOs (model 4-3) that possess codified rules and procedures facilitating
ordinary interaction, attempts at conflict resolution may fail. Only interventionist security IGOs significantly reduce interstate conflict. Table
5 reports that membership in this type of IGO lowers the probability of
a MID by 22 percent. Without foundational motives or explicit institutional mechanisms, IGOs lack the ability to intervene in crises in a manner likely to foster dispute resolution. Conversely, IGOs with a security
mandate and sufficient institutional structure can promote peace. This
contingent nature of IGO effectiveness appears to be lacking in traditional interpretations but is consistent with the argument presented
here. The contribution of IGOs to peace depends not so much on forging common bonds or on converting international anarchy as it does on
creating conditions that encourage effective interstate bargaining.
VI. CONCLUSIONS
The objective of this study has been to determine whether intergovernmental organizations contribute to interstate peace. Previous research
v57.1.001.boehmer.001-038
6/24/05
9:53 AM
Page 29
&
PEACE
29
produced disparate findings, including the troubling possibility that

IGOs might be associated with increased conflict. With several improvements in theory and research design, we find reason for cautious
optimism. Although both liberal and realist expectations for IGOs are
not without basis, our results do not seem to be fully consistent with either classical view. Rather, the contingent nature of IGO effectiveness
best supports our alternative theoretical perspectivethat IGOs can variously reduce or exacerbate conflict or have no effect, depending on the
level of institutionalization, member cohesiveness, and organizational
mandate. IGOs can constrain or inform if they possess sophisticated institutional structures capable of managing information or of intervening in a manner that shifts the balance of power or interests without
allowing demands to shift as well. IGOs that possess a mandate to engage in security diplomacy are also more effective. The role of IGOs as a
venue for deliberation does not in itself appear to be a direct contributor to interstate peace. Finally, IGOs can create conflict when they add
to international uncertainty.
This article demonstrates that not all IGOs contribute to interstate
peace. In particular, IGOs that lack extensive institutional structure pose
no barrier to interstate conflict. Minimally institutionalized IGOs lack
the ability to collect and disseminate information not commonly available to belligerents. The effect is most pronounced in security-oriented
IGOs where the prospects for peace increase only for those organizations that contain extensive institutional structure. We also show that
IGOs confronted with high levels of contentiousness, particularly
among major powers, are less effective at ameliorating interstate conflict. Disparate claims and findings about the effect of IGOs on peace
can in large part be attributed to the tendency to treat all IGOs as equivalent. Having now only begun to relax rigid assumptions of IGO uniformity, we find encouraging results.58
Returning to the traditional debate, our study suggests that whether
IGOs matter depends on what it is that IGOs are intended to accomplish. While intuitive, discussions of IGO effectiveness have tended to
ignore the possibility that some IGOs make a difference, whereas others
do not. The presence of IGOs in and of themselves does relatively little
to influence international conflict. IGOs may extend common norms
and enhance global prosperity, but these objectives differ from a direct
contribution to peace. IGOs are treated as a panacea in liberal and constructivist accounts. We believe that we have demonstrated that the
58
Aggregating memberships at the level of the dyad is far from ideal. We use the dyad as the unit
of analysis to facilitate comparison of our results with other quantitative studies of IGOs.
v57.1.001.boehmer.001-038
30
6/24/05
9:53 AM
Page 30
WORLD POLITICS
mere presence of IGOs in the absence of extensive institutional structure

and a mandate has little significance for whether states fight. The ability of IGOs to impinge on warfare is limited by the causes of war and by
the resources of the IGO.
Conversely, realist criticism of IGOs, while offering a useful counterpoint, is equally categorical and flawed. Realists correctly point out that
the countries that are most active in IGOs are also most active internationally. We have shown that the positive correlation found in previous
studies between IGOs and conflict can be explained by the relationship
between international engagement and IGO membership. Realists err,
however, in claiming that IGOs are simply a proxy for power. Even after
taking into account the contribution of major powers, the balance of
capabilities, alliance ties, and international involvement, institutionalized and security-oriented interventionist IGOs reveal themselves to
have an independent impact on the advent of disputes. As so often
happens, claims of all or none must give way to more nuanced and contingent arguments in the light of empirical evidence. Do intergovernmental organizations promote peace? It depends.
APPENDIX 1: FORMAL THEORETICAL ANALYSIS
OF THE PROPOSITIONS
This appendix provides formal arguments supporting the six propositions in the text. Many patterns of interaction are possible between autonomous actors. Ideally, a model of bargaining would incorporate a
framework in which the structure of interaction among players is determined endogenously. States in the real world negotiate over how
they will negotiate. Unfortunately, mimicking such a framework formally is extremely complex. Further, somewhat arbitrary assumptions
about structure are still required to model preliminary negotiations,
which in turn determine subsequent bargaining patterns. A second best
option would be to mirror the patterns of bargaining we observe empirically. There remains considerable debate over what assumptions best
characterize interstate negotiations. Several authors argue that iterated
games are more realistic than single-shot static games.59 We see iterated
games as advantageous primarily when addressing issues of timing or
duration, such as the insight that states can fight initially to signal resolve for subsequent bargaining.60 Factors that preempt or terminate
59
Powell (fn. 28, 2004); idem, Bargaining and Learning while Fighting, American Journal of Political Science 48, no. 2 (2004); Slantchev (fn. 22).
60
R. Harrison Wagner, Bargaining and War, American Journal of Political Science 44, no. 3 (2000).
v57.1.001.boehmer.001-038
6/24/05
9:53 AM
Page 31
&
PEACE
31
contests can be illustrated using static models. Indeed, states themselves

often devise bargaining structures similar to single shot ultimatum type
games, particularly when time is precious or when war waits in the
wings. While in theory states can bargain indefinitely, in practice they
frequently seek to confine negotiations temporally and spatially. The
Dayton Accords, for example, occurred over a specified three-week period in November 1995. At least one key participant credits success at
Dayton to a finite game structure; last-minute bargaining that resulted
in an agreement required that there be some terminus to bargaining.61
At some point, failure to accept an opponents demands implies that
bargaining will not continue. The target of a challenge must provide a
final answer, and the challenger must decide whether to accept a compromise, fight, or back down. Static models represent this moment of crisis in dynamic games and in real-world negotiations. The chief virtue of
non-sequential games, however, is practical. They are considerably easier
to solve than are dynamic games and they avoid thorny questions about
temporal discounting. In the present application, where we seek to address multiple extensions of the basic model, simplicity is an important
virtue. We use an approach similar to Fearon because it is simple and
logical and it creates a foundation for more elaborate explanations.
THE BASIC MODEL
To begin with, imagine that two states (A and B) compete over some
issue or good (k, where 1 k > 0). States keep any division of the stakes
but forfeit contest costs (ci , where ci > 0 and where i [A,B]). To keep
the game simple, imagine that A makes a take-it-or-leave-it offer (d,
where k d 0) to B. If B accepts the offer, then it receives d, while A
receives (kd). If instead B rejects the offer, then a contest ensues. Suppose that the victor gets k (minus costs) while the loser obtains zero
(again minus costs). Assume also that p is the probability that A wins
and (1p) is the probability that B wins (where 1 p 0). Thus, fighting has an expected value of [p k + (1p) 0 cA ] = [p k cA ] for A and
[p 0 + (1 p) k cB ] = [( 1 p) kcB] for B. We list below utility functions for each player, where f is Bs dichotomous fight decision.62
UA = (1f ) (k d) + f [p k + (1p) 0 cA ]
UB = (1f ) d + f [p 0 + (1p) k cB]
61
(1)
(2)
Richard Holbrooke, To End a War (New York: Random House, 1998).

Game-theoretic models often include the status quo as an outcome. We treat the status quo as a
special case of the equilibrium in which the bargain in the game equals the bargain ex ante.
62
v57.1.001.boehmer.001-038
6/24/05
9:53 AM
32
Page 32
WORLD POLITICS
Obviously, whether A and B fight depends on d. If d < [(1p) kcB ],

then B prefers to reject As offer. If d [(1p) k cB], then B accepts
As offer and peace, or at least an absence of war, obtains. In the jargon
of bargaining theory, [(1p) kcB ] is Bs reservation price. Similarly, A
is willing to make an offer d such that (kd) [p k cA ] d [(1p)
k + cA ]. This is As reservation price, since A cannot willingly make an
offer d > [(1p) k + cA ]. The region between [(1p) k + cA ] and [(1p)
k cB ] is the bargaining space of size [(1p) k + cA ][(1p) k cB ] =
cA + cB. For risk-neutral or risk-averse actors with positive costs for
fighting, the bargaining space is a positive interval. There always exists
some bargain that is mutually preferred to fighting.
What does As offer look like? Payoffs for A are strictly decreasing in
d, [(UA )/(d)] < 0, up to the point where B prefers to fight rather than
accept d. Thus, A optimizes by just meeting Bs reservation price (d* =
[(1p) k cB ]). Since B must be indifferent over equivalent outcomes,
B is said to weakly prefer As offer to fighting.63 This basic bargaining
game never leads to a military contest. The sole equilibrium in the fullinformation game is for A to offer d* and for B to accept.
Suppose instead that we assume that the cost of fighting remains
public information for A (cA ) but that B is privately informed about its
costs (cB cB 0, where cB equals some arbitrarily high boundary for cB).
Introducing private information requires use of the Perfect Bayesian
Equilibrium (PBE) solution concept rather than Nash subgame perfection above. If A is uncertain about what B prefers, then As best offer is
one that maximizes As utility function, subject to As beliefs about Bs
type at each information set. As beliefs are characterized by the probability distribution over types of B. To keep things simple, we assume
that Bs costs are distributed uniformly over the interval (cB~ U [0, cB]).
A must estimate the probability that B will fight as the portion of types
that will reject a given offer d, Prob( f = 1) = [(cB. 0)/(cB0)], where cB = (1
p) k d is the value of cB or the type player B that is just indifferent
between fighting and accepting a given offer d. Substituting As estimate
of f into equation (1) and solving for [(UA )/(d)], gives [(2 (d k (1p))
+ cA cB )/(cB )]. Setting [(UA )/(d)] equal to 0 and solving for d as an optimal offer yields the following (where k d* 0):
d* =
1
[2k (1 p) + cA cB ]
2
(3)
63
Having A sweeten its offer by making an arbitrarily small concession leads B to strictly prefer d
to fighting. However, this clutters the model while yielding no new substantive conclusions.
v57.1.001.boehmer.001-038
6/24/05
9:53 AM
Page 33
&
PEACE
33
Setting Bs reservation price equal to equation 3 and solving for cB , we

get cB = [1/2](cBcA ). If cB < cB , then B rejects As offer and A and B fight.
Otherwise, if cB cB , B accepts A s bargain. We now have an equilibrium solution for a game that involves both war or peace. The next section uses this basic model to derive the ten propositions offered in the
body of the text.
SKETCHING PROOFS OF THE PROPOSITIONS
Proposition 1. Actions by IGOs that alter the bargaining power between two
competitors, when common knowledge, encourage revised demands rather than
just a change in decisions to fight.
Suppose that some third party I raises Bs war costs or lowers Bs probability of victory (1 p). We examine each possibility in turn. We can
treat this added cost (h, where h > 0) as a dead weight loss for fighting.
Where the distribution of types B was originally bounded [cB ,0], it is
now bounded [cB + h,h]. Similarly, Bs reservation price becomes [(1p)
k d h]. If A again offers equation 3 to the distribution of types B,
then cB+h = [1/2](cB cA. 2h) < cB Prob( f = 1 | h > 0) < Prob( f = 1 | h
= 0). In this conception of third-party intervention, the probability that
B accepts A s demand is strictly increasing in I s influence on Bs war
costs through the impact of h.
Yet if A is strategic, it no longer prefers its original demand. We can
recalculate a solution to the modified game based on adding h. Prob( fh
= 1) = [(cBh h)/(cB + h h)], where cBh = (1p) k d h. Substituting into
the modified equation (1) and solving for [(UA )/(d)] yields [(2(dh
k (1p)) + cA cB )/(cB )]. Setting this equal to 0 and solving for d, we
get d* h = [1/2][2k (1p) + cA cB2h]. Solving for UB( f = 1) = UB( f = 0) reveals that cBh = cB = [1/2]( cB cA). Unless the change is very large (where
boundary conditions limit the offer), the effect of h tends to be felt in
terms of the bargains without reducing the likelihood of a contest. The
same types of B reject d*h as reject d*. I s intervention does not increase
the chances for peace.64
Suppose instead that I alters the probability of victory for one side or
the other. Equation 3 is a function of relative capabilities (the size of p),
but the probability of fighting depends only on the relationship between cB and cB . By construction, interventions that change the probability of victory have little impact on whether bargaining fails. Since As
64
Actors may be limited in their ability to optimize by the bounded nature of parameter values or
by constraints imposed by third parties (assuming opponents are restrained from altering demands).
v57.1.001.boehmer.001-038
34
6/24/05
9:53 AM
Page 34
WORLD POLITICS
probability of victory, p, is a parameter, we can substitute some other

value p such that p > p or p < p without changing the results in any
substantive way. Imagine that I attempts to influence the probability of
victory for both parties. Trivially, the zero-sumness of relative power
means that such an attempt is futile. Increasing p to p > p means that
(1p) > (1p) and that p p = (1p) (1 p).
Finally, suppose that an effort is made to appease B. Imagine that I
provides B with an inducement in the form of a side payment (m, where
m > 0) to accept As offer. If A makes its original offer, the probability
that B rejects d will of course go down, but A again has incentives to
alter its demand. It is easy to show that inducements (m) not to fight
and punishments (h) for fighting are equivalent in expected utility
terms. We omit the proof here to save space.
Proposition 2. IGOs promote peace by revealing private information about
competitor states.
Imagine that both B and I learn the value of cB , but not A. If I can reveal
this information, then A can make an offer that B will accept, leading to
peace. The argument holds even if we relax the assumption that I is fully
informed. Suppose that I can only provide A with an approximation of
Bs costs, say along some interval [cBlow,cBhigh], where (0 cBlow < cBhigh cB ).
As offer to B is then d* = [1/2][2k (1p) + cA cBhigh ], and the probability
of fighting,
Prob( f = 1 | [cBlow
,c ~ ]) = [(cB cBlow)/(cBhigh cBlow)]
~
~ Bhigh
[(cBhigh. cBlow )/(cB 0)] = [([1/2](cBhigh.cA) cBlow)/(cB)]. Since Prob( f = 1) = [(cB
0)/(cB 0)] = [([1/2](cB cA))/(cB )] and since, by definition, cB cBhigh, it
follows that Prob( f = 1 | [cBlow,cBhigh ]) Prob( f = 1).
Proposition 3. IGOs promote peace by intervening secretly or by constraining states that stand to benefit from IGO intervention.
We can return to the setup for proposition 1. Suppose that I imposes h

on B, but conceals this from A. State A offers d* from equation 3, which
is now accepted by more types of state B, [cB + h,h] . cB+h < cB Prob( f
= 1 | h > 0) < Prob( f = 1 | h = 0), as in proposition 1.
Alternatively, if h cannot be concealed from A, I can potentially force
A to retain the default offer (d*). Suppose that I imposes a punishment,
w = (d*d* h), on A, where > 0 scales the magnitude of the punishment for an offer less than d*. Solving for As optimal offer produces d*h.
= [1/2][2k (1 p) 2h + cA. cB (1 )]. If 1 wants to exactly cancel the
effect of punishment of B on As offer, then it should impose sanction
= [2h/(cB )]. Given that [(UB)/(f )] = [1/2][cB (1 ) cA.2cB ], the prob-
v57.1.001.boehmer.001-038
6/24/05
9:53 AM
Page 35
&
35
PEACE
ability of a contest is strictly decreasing in . I may choose any value of

to reduce the probability of fighting, depending again on I s institutionalization or help from major powers.65
Proposition 4. Preference homogeneity among IGO members increases the
effectiveness of efforts to promote peace through intervention and secrecy.
Suppose that n members of an

equation 4:
IGO
have preferences as specified by
Ui = (1 f )[vi + (k d)] + f [p (. k) + (1p) 0]
(4)
Let i [1,n] represent each member of I. The lowercase vi, (v > 0)

reflects is payoff for achieving peace and , (0 [(vi )/(cA )]), is is
degree of bias.66 All other terms are as previously defined. For simplicity, assume that the first n 1 members of I are neutral ( = 0), while
member n favors A ( = 1). Neutral members of I strictly prefer concealing h from A, but a single partisan member can prefer revealing h if
vi < cA + h. In fact, any member such that > [(vi )/(cA + h)] prefers to reveal h and allow A to obtain greater concessions from B, at the risk of a
higher probability of a dispute.
Proposition 5. IGOs promote peace by facilitating costly signaling among
states through the imposition of sanctions, embargoes, or other nonmilitarized
punishments.
Imposing a punishment on B for fighting shifts the distribution of benefits toward A without necessarily reducing the probability of a contest.
An alternative is for I to impose h prior to bargaining. Without committing to particular substantive details, suppose that B can incur h
early in the game. This ex ante sanction on B requires two conditions.
First, some types of B, those with low war costs, will suffer h anyway.
Second, precisely because the first condition is not true for high war
cost types B, embracing h prior to bargaining can prove informative to
A. In preparing its offer, A can identify two different groups of players
B, one set with low war costs and one set with high war costs. A can tailor its offers to each group, reducing the range of types that reject both
offers. Equations 5 and 6 below detail players utility functions.
65
This result depends on the inability of B to make a counteroffer. In an iterated game, B could use
the vulnerability of A to extract additional concessions, much as A does to B in proposition 1.
66
Using 1 as the upper bound would both allow that the third party is more eager for a contest than
A (not likely) and eliminate the possibility that I values the stakes more than A.
v57.1.001.boehmer.001-038
36
6/24/05
9:53 AM
Page 36
WORLD POLITICS
UA = s [(1fs=1) (k ds=1) + fs=1 (p k cA)]

+(1s)[(1fs=0) (k ds=0) + fs=0 (p k cA)]
UB = s [(1fs=1) ds=1 + f ((1 p) k cB)h]
+(1s)[(1fs=0) ds=0 + f ((1p) kcB h)]
(5)
(6)
Define cBCRIT as the type of player B such that UBs=1 = UBs=0. Let cBs=0 and
cBs=1 be the type of player B indifferent between fighting and not fighting in each subgame. cBs=0 = cBCRIT, since types B that will fight in the s =
0 subgame at least weakly prefer the s = 1 subgame. Solving as previously for As optimal offer in each subgame yields d*s=1= [1/3][3k (1 p)
* = [1/3][3k (1 p) + c 2c h]. The probability
+ 2cA. cB + h] and d s=0
A.
B.
of a contest in the s = 1 subgame is [(2cA cB + h)/(cA2cB + 2h)]. Prob(s
= 1) = [(2cB. c.A2h)/(3cB )], so that the probability of a contest given
signaling, Prob( f = 1 | s = 1), equals Prob(s = 1) Prob( fs=1 = 1) = [(cB.
2cA.h)/(3cB )] Prob( f = 1) = [(cB cA)/(2cB )].
Proposition 6. IGOs with a mandate as security organizations are more effective at promoting peace through costly signaling than those with non-security
objectives and interests.
Up to this point we have assumed that it is costless for I to impose

sanctions. Imagine that the price for I of imposing h on B is h, where
0 1. Equation 7 below is I s utility function:
UI = v [(1 fv=1 i + fv=1 0)]+(1 v)[(1 fv=0)(i + h) + fv=0 0]
(7)
I suffers an opportunity cost for sanctioning (v is I s dichotomous sanction decision). Substituting the probability of fighting, Prob( f = 1),
from proposition 1, and the probability of fighting with signaling,
Prob( f = 1 | s = 1), from proposition 5 for I s estimates of the probability of a contest in the v = 0 and v = 1 subgames, respectively, and taking
the partial with respect to the sanction decision, we get [(UI )/(v)] =
[(i (cA + cB + 2h)3h (cA + cB ))/(6 cB )]. Setting [(UI )/(v)] = 0 and
solving, we are able to define = [(i (cA + cB + 2h))/(3h (cA + cB))] such
that UIv=0 = UIv=1. Values of > UIv=0 > UIv=1 and vice versa. The
likelihood that I sanctions is thus a function of the relative size of i and
h. Security organizations (i > h) more willingly sanction, pursuing
peace at the cost of prosperity, for example. But non-security organizations may have mixed objectives (i < h), considering prosperity as
part of their mandate as much or more than political conflict.
v57.1.001.boehmer.001-038
6/24/05
9:53 AM
Page 37
APPENDIX
&
PEACE
37
2: IGO INSTITUTIONAL CODING CRITERIA
We score the institutionalization of IGOs in three categories. A joint

IGO membership variable is created for each category, as well as an indicator that combines the second and third categories.
MINIMAL
This category is composed of two IGO types. First, some organizations
exist almost exclusively on paper. There is little evidence of any apparatus or organization. The English Commonwealth (prior to 1965) or the
Group of Ten fall into this category. These organizations often give
themselves a name and then the heads of state of member states or
other ministers meet on a regular basis, whether annually, biannually, or
otherwise. For example, the Central African Customs and Economic
Union seems to hold meetings only of heads of states. These organizations lack any bureaucratic, executive, or judicial organs that possess any
formalized power.
Second, some IGOs contain minimal organizational structure, often
committees or councils where ministers or other representatives of
member states meet and discuss issues. There may be a few rules on
procedures, such as voting, that are nonbinding or do not compel states
to take action. The ability to enforce agreements and the implementation of decisions is optional for member states. The most centralized
feature in these organizations is often a secretariat or similar executive
organ that generally possesses few autonomous powers or functions beyond the logistics of organizational meetings and communication.
There are many examples of this type of organization, among them, the
Arctic Council and the Central American Common Market.
STRUCTURED
Almost every organization appears on paper to possess several deliberative and administrative organs, but most often the linkages between
bodies are unguided by codified procedures and little work is done beyond committees. At the intermediate institutionalization level, however, member states relinquish minimal amounts of their sovereignty to
support IGO projects and missions. Organizational decisions are made
by formal voting, and/or other codified procedures guide member interactions. Organizational operations (appointments and elections to
positions) are also well specified. Additionally, a bureaucracy often exists to carry out decisionssome of which are binding on member
v57.1.001.boehmer.001-038
38
6/24/05
9:53 AM
Page 38
WORLD POLITICS
statesand manage programs that produce concrete benefits. In cases

where there exists no extensive bureaucracy, other bodies or organs assume concrete administrative powers over organizational programs
without directly relying on state decisions. For example, the secretariat
of the Central European Initiative makes some binding decisions. Of
course, all member states have the ability to ignore IGO dictates, but
most IGOs never attempt to directly influence state behavior in this
manner. Intermediate-level structural IGOs are scored as a two.
INTERVENTIONIST
These IGOs possess clear mechanisms for coercing or influencing state
behavior. This can be realized in several ways. First, these IGOs include
organizations with overt security or conflict-resolution statements in
their official missions that allow them to inject themselves into interstate conflicts through established structures and mechanisms of mediation, arbitration, and adjudication. These organizations typically also
possess many of the institutional characteristics regarding extensive
codification of rules and procedures between and within the executive,
legislative, and judicial structures. Any organization that possesses a judiciary structure is included in this category. The European Union is
one such example. Another example is the League of Arab States
(more commonly known as the Arab League). Second, some of these
IGOs have the ability to influence state behavior by directly manipulating the opportunity costs associated with interstate conflict. IGOs may
sanction states by withholding economic benefits in the form of grants,
loans, credits, or access to other resources. The World Bank and IMF are
high-profile examples. All IGOs in this category are scored a three.

Do Intergovernmental Organizations Promote Peace?: by Charles Boehmer, Erik Gartzke, and Timothy Nordstrom

Uploaded by

Copyright:

Available Formats

Do Intergovernmental Organizations Promote Peace?: by Charles Boehmer, Erik Gartzke, and Timothy Nordstrom

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Do Intergovernmental Organizations Promote Peace?: by Charles Boehmer, Erik Gartzke, and Timothy Nordstrom

Uploaded by

Copyright:

Available Formats

v57.1.001.boehmer.

World Politics 57 (October 2004), 138

the contrasting visions of classical arguments. Bargaining theory explains

than are economic organizations. Finally, controlling for engagement in

rise of multiple channels of interaction and the growing salience of

Most existing empirical research involves qualitative case studies.9

Mearsheimer (fn. 3).

cannot be evaluated here. Nevertheless, our tests support the theory

in arguing that indivisibility is a relatively infrequent cause of interstate

If competitors are informed about changes in strategic variables and are

contests result from private information about capability and resolve,

Third parties can act as information arbitrageurs, reducing the risk of

Nevertheless, there are barriers to effective third-party promotion of

INTERVENTION WITH SECRECY OR CONSTRAINT

Keeping changes secret is arguably the least taxing method in terms of

Bargaining failures can be reduced if competitors are confronted with

We present four hypotheses designed to differentiate between our

Second, the effectiveness of IGOs at promoting peace should vary with

Third, IGOs vary in the level of contentiousness among their members.

IV. RESEARCH DESIGN

covering private strategic information, especially in cases where states

able ranges in value from 1 to a maximum of 93. Data were collected

Major Power Contention, which reflects the degree of contentiousness

Russett, Oneal, and Davis (fn. 1).

autocorrelation.56 Logistic regression constitutes a more demanding

Beck, Katz, and Tucker (fn. 40).

*p < .10, **p < .05, ***p < .01

Politically Relevant Dyads Politically Relevant Dyads

Joint IGO Membership

Joint IGO Membership

*p < .10, **p < .05, ***p < .01

ONSET OF MILITARIZED DISPUTES (195091)

IGO INSTITUTIONALIZATION, CONTENTIOUSNESS, AND THE

Based on Model 2-3

% Change from Baseline

* p < .10, **p < .05, ***p < .01

Joint Security IGO Membership

% Change from Baseline

models, regardless of which estimation technique, which dependent

produced disparate findings, including the troubling possibility that

mere presence of IGOs in the absence of extensive institutional structure

contests can be illustrated using static models. Indeed, states themselves

Richard Holbrooke, To End a War (New York: Random House, 1998).

Obviously, whether A and B fight depends on d. If d < [(1p) kcB ],

Setting Bs reservation price equal to equation 3 and solving for cB , we

probability of victory, p, is a parameter, we can substitute some other

We can return to the setup for proposition 1. Suppose that I imposes h

ability of a contest is strictly decreasing in . I may choose any value of

Suppose that n members of an

have preferences as specified by

Ui = (1 f )[vi + (k d)] + f [p (. k) + (1p) 0]

Let i [1,n] represent each member of I. The lowercase vi, (v > 0)

UA = s [(1fs=1) (k ds=1) + fs=1 (p k cA)]

Up to this point we have assumed that it is costless for I to impose

2: IGO INSTITUTIONAL CODING CRITERIA

We score the institutionalization of IGOs in three categories. A joint

statesand manage programs that produce concrete benefits. In cases

You might also like

p < .10, p < .05, p < .01

p < .10, p < .05, p < .01

* p < .10, p < .05, *p < .01