Trustees of Princeton University

The Kantian Peace: The Pacific Benefits of Democracy, Interdependence, and International Organizations, 1885-1992 Author(s): John R. Oneal and Bruce Russett Reviewed work(s): Source: World Politics, Vol. 52, No. 1 (Oct., 1999), pp. 1-37 Published by: Cambridge University Press Stable URL: http://www.jstor.org/stable/25054099 . Accessed: 12/10/2012 05:03
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

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

Cambridge University Press and Trustees of Princeton University are collaborating with JSTOR to digitize, preserve and extend access to World Politics.

http://www.jstor.org

THE KANTIAN
The Pacific Benefits Interdependence, Organizations,

PEACE
of Democracy,

and International 1885-1992

By JOHN R. ONEAL andBRUCE RUSSETT*

JUST

that "re years ago Immanuel Kant suggested a "commercial of international constitutions," trade, publican spirit" and a federation of interdependent would the basis republics provide

over

two hundred

even starker in the nuclear era than for perpetual peace. The alternative, sort: "a vast grave where in 1795, would be peace of a different all the horrors of violence and those responsible for them would be buried."1 Kant declared, we have a duty to work for peaceful in Consequently, ternational relations. Though he emphasized the absolute character of he was no idealist; rather, he believed that natu imperative, on to act in ways ral processes based self-interest individuals impelled a was also that would and Kant eventually produce lasting just peace. in the anarchic realistic. He acknowledged inter that war was inherent this moral act pru cautioned that nations must system and therefore was established. of dently until the federation interdependent republics But he also knew that the mechanisms of power politics produce only not temporary respite from conflict, lasting solutions. national the past half century much of the world has been at peace. Un causes and its is the funda that trajectory, phenomenon, derstanding mental for international relations scholars seek to today. We challenge Over show that Kant s realistic guidance for this task. Most useful theory provides now agree that the con political can be traced in demo part to the so-called no states have established democratic fought scientists statement of liberal

temporary peacefulness cratic peace, wherein

* We thank the Carnegie Corporation of New York, the Ford Foundation, and the National Sci ence Foundation for financial support; Zeev Maoz for comments; and Jennifer Beam, Margit Buss Brian Radigan, and Jacob Sullivan for data collection and mann, Soo Yeon Kim, Yury Omelchenko, management. 1Kant, Perpetual Peace: A Philosophical ed. Hans Reiss (Cam Sketch, inKants Political Writings, Lutz-Bach and Matthias Press, 1970), 105. See also James Bohman bridge: Cambridge University mann, eds., Perpetual Peace: Essays onKants Cosmopolitan Zafe?/(Cambridge: MIT Press, 1997).

WorldPolitics 52 (October 1999), 1-37

2 international wars with

WORLD one

POLITICS and the use or threat of force

another

among them, even at low levels, has been rare.2 This view is incomplete, it fails to recognize the pacific benefits of the other lib however, because eral elements of Kant s program for peace. Moreover, the term hides the theoretical the about this sep controversy processes vigorous underlying arate peace?over is really whether democracy to the degree which the empirical phenomenon concerns are These theoretical and empirical relations democracies among peaceful during simply a even existed linked. its cause in other and over eras.

If, for example, the cold war era were

of their shared security interests vis-?-vis the op consequence a in alliance then their would system posing bipolar world, peacefulness same con be spuriously related to the character of their regimes. The to the clusion would result if the democratic peace could be attributed power of the United hegemonic allies or to East-West differences differences peace terns in to suppress conflict its among to in preferences unrelated underlying not then expect to find a separate One would states in other different pat evincing periods States to show these questions by reporting that peaceful relations among the twentieth the century4 Extending the effect of the changing on interstate relations.5 address

regimes.3 among democratic of interstate relations. We

1885-1992, analyses covering democracies existed throughout historical character domain also allows of the international

us to assess system

2 in the social science literature, war is defined as a conflict between two By convention recognized of the international system that results in at least one thousand battle deaths. The sovereign members on militarized most international disputes (MIDs), compiled by Stuart Bremer and his complete data are available at The democracy data we employ colleagues, http://pss.la.psu.edu/MID_DATA.HTM. were Third Wave with the compiled by Keith Jaggers and Ted Robert Gurr, "Tracking Democracy's Peace Research 32, no. 4 (1995), available at http://isere.colorado.edu Polity III Data,"Journalof* Both data sets are produced from the demo /pub/datasets/polity3/politymay96.data. independently cratic peace research program, and the initial codings, from the 1980s, precede it. Reviews of the Peace: Problems and Promise," Mershon Inter program include Steve Chan, "In Search of Democratic national Studies Review 41, no. 1 (1997); James Lee Ray, "Does Democracy Cause Peace?" Annual Review Peace to Kantian of Political Science 1 (1997); and Bruce Russett and Harvey Starr, "From Democratic in the International System," in and Conflict Peace: Democracy Manus Midlarsky, ed., Handbook ofMichigan Studies, 2d ed. (Ann Arbor: University Press, forthcoming). 3 Politics Henry F?rber and Joanne Gowa, "Common Interests or Common Polities?" Journal of 2 (1997); Gowa, Ballots and Bullets: The Elusive Democratic Peace (Princeton: Princeton University ofWar 57, no.

Press, 1999); Douglas Lemke andWilliam Reed, "Regime Types and Status Quo Evaluations," International Interactions 22, no. 2 (1996); Erik Gartzke, "KantWe All Just Get Along? Opportunity, Willingness and the Origins of the Democratic Peace,"American JournalofPoliticalScience 42, no. 1 (1998). 4 The MIDs data (fn. 2) are unavailable after 1992, and data on dyadic trade are sparse and unreliable before 1885. In any event the further back one goes into the nineteenth century, the rarer are instances of democracy, intergovernmental and high levels of economic interdependence. The MIDs organizations, data include only disputes between recognized states and not, for example, extrasystemic (i.e., colonial) in support of a recognized government. interventions actions, covert operations, or domestic military 5 We will not here offer a new theory on why democracy produces peaceful relations. A recent state ment et al., "An Institutional is Bruce Bueno de Mesquita of the Democratic Peace," Explanation American Political Science Review 93, no. 4 (1999).

KANTIAN In

PEACE our analysis be of economically

we the Kantian perspective, keeping with expand the democratic the influence peace, incorporating yond in international trade and joint memberships important The classical liberals

organizations. ex of the eighteenth and nineteenth centuries as well as to control of government pected popular interdependence states into have important benefits. Commercial relations draw pacific a web of mutual self-interest that constrains them from using force against to what contribute and democracy interdependence in addi the "liberal peace." Kant emphasized, the of international law and Our benefits tion, organization. previous war era indicate at least, of the cold those that, years during analyses did reduce the trade and networks of intergovernmental organizations another. Thus called we have of militarized interstate We the benefits earlier of democracy.6 and later years. these effects were on top of disputes; in show here that they also operated one

number

Our Objectives

and Method

are often considered anti the liberal and realist perspectives Although we our s in Kant conduct with thetical, analysis keeping philosophical account tests of the Kantian realist peace while important taking into as as Kant that both influences. We matter, believe, did, perspectives
"The Classical Liberals Were Right: Democracy, John R. Oneal and Bruce Russett, Interdepen International Studies Quarterly 40, no. 2 (1997); Russett, Oneal, and dence, and Conflict, 1950-1985," and Militarized David R. Davis, "The Third Leg of the Kantian Tripod: International Organizations 1950-85," International 52, no. 3 (1998); Oneal and Russett, "Assessing the Disputes, Organization Liberal Peace with Alternative Specifications: Trade Still Reduces Conflict," fournal of Peace Research 36, no. 4 (1999). Here we extend this line of research in three ways: (1) providing a conceptual syn thesis of Kantian and realist theories that treats conflict as inherent but subject to important con the temporal domain for trade and IGOs into the nineteenth straints; (2) extending century; and (3) and Kantian theories about systemic influ assessing realist theories regarding the role of the hegemon on the inter ences in away that addresses, among others, constructivist and evolutionary perspectives a national system. Note that the Kantian influences may be mutually reinforcing in dynamic system of "Kant's Third Image: Systemic Sources of the Liberal feedback loops, as suggested byWade Hundey, Peace," International Studies Quarterly 40, no. 4 (1996); and Russett, "ANeo-Kantian Perspective: De mocracy, Emanuel Interdependence, Adler and Michael and International Barnett, in in Building Security Communities," Organizations in Comparative Perspective (New York eds., Security Communities 6

Press, 1998). Cambridge University We and others have begun to address some of these links, such as greater trade between democra that trade is diminished between conflicting states, the effect of democracy, trade, cies, the possibility in international organizations, and the effect of conflict on de and peace in increasing membership mocracy. On the first, see Harry Bliss and Russett, "Democratic Trading Partners: The Liberal Con Politics 58, no. 4 (1998), and James Morrow, Randolph Siverson, and Tessa Tabares, nection,"fournal of "The Political Determinants of International Trade: The Major Powers, 1907-90," American Political Science Review 92, no. 3 (1998); on the second, see Soo Yeon Kim, "Ties That Bind: The Role of Trade in International Conflict Processes" (Ph.D. diss., Yale University, 1998); on the third, see Russett, Oneal, of and Davis (this fn.); and on the last, see Oneal and Russett, "Why An Identified Systemic Model Peace Nexus' Does Not Persuade," Defence and Peace Economics 11, no. 2 (2000). the Democratic

4 both consider conflict

WORLD and

POLITICS of violence to be inherent in an

the threat

states. The Hobbesian of sovereign of this un element to realist is it is but also central derstanding theory, deeply embedded in the liberal tradition. Kant accepted Hobbes's of a state of description anarchic world that a balance of power could prevent and believed among nations realists acknowledge, war; but history has shown all too clearly, as most that this "peace" is tenuous. Kant, however, was convinced that a gen war republics that rested more and on the three Kantian

uine, positive peace could be developed within a "federation" of liberal

interdependence, international supports?democracy, on law and organizations?than not aworld is Kant envisioned by remain

power politics. The pacific federation state but a federation whose members confederation^ sees democratic tional rooted linked or collective

governance, to law as the means the security dilemma by which supersede in the of the international system. For states not much anarchy

linked only by sovereign, that is, Liberalism, security arrangements. economic interna and interdependence,

tion, Kantian

remains. In addi the threat of violence by these ties, however, must states states liberal fear those illiberal that remain outside the confederation.7 we

that the international system is anarchic begin by assuming and power is important. Yet despite the inherent possibility of violence, or even at all times states do not where realist principles fight all others are and dis dominate. constrained alliances, Rather, they by power, tance. States must be concerned with the balance of power and the co Thus in alliances. Many states, expressed are irrelevant to these calculations: in general, the farther moreover, two states are, the fewer are the issues over which to apart fight and the one to the another. therefore realists are less threat they pose Ultimately to states that have the and incentive concerned only with opportunity a in constraints because these conflict.8 engage provide Accordingly to assess the additional of the Kantian which baseline impact against incidence of national interests influences, model. To constraints, we them incorporate the realist variables we hypothesizing against incentives that other as central features will of our theoretical less fre impor relations; add measures for the three Kantian use force

(1) democracies

quently, especially tant trade creates

democracies; (2) economically of peaceful for the maintenance

7 Michael W. Doyle, Ways ofWar and Peace (New York W. W. Norton, 1997), chap. 8; David Lake, " American Political Science Review States andWar "Powerful Pacifists: Democratic 86, no. 4 (1992). 8 of Conflict and Peace" (Manuscript, Yale and Theories "Inherency, Contingency, Birger Heldt, and Harvey Starr, Inquiry, Logic, and International Politics (Colum 1998); Benjamin Most University, of South Carolina Press, 1989), chap. 2. bia: University

KANTIAN (3) international organizations a in peace moting variety of ways. tem is far from realist and

PEACE constrain decision by pro international sys states, we expect both explicitly makers

Since

the modern

a of democratic pacific federation and Kantian factors to affect interstate level have

relations. We

consider how realist and liberal influences at both the dyadic and the
systemic addressing ing Kantian Evidence membership accepted jected fect of more of the international altered the functioning system, the role of the leading state and the influence of the chang variables over time. for the pacific benefits in intergovernmental of economic and interdependence (iGOs) is less widely and it has been sub have assessed the ef

organizations is that for the democratic peace, to less extensive critical scrutiny. We alone than IGOs on conflict expectations than those at the dyadic the regarding level of

theoretical diverse

that democracies those who

impact No one hypothesizes concerning democracy. are more are other to likely fight each other than poli conflict over the division its intellectual of the gains from

analysis. Moreover, of trade and IGOs are

ties; but the liberal view of the pacific effects of trade is contradicted by
expect trade and school and and descen by the dependency predecessors and small states.9 As for IGOs, dants, who expect conflict between large a states form or view be that international organi might join plausible to zations often without with their manage?albeit success?disputes an realists regard the UN being adversaries, commonly, example. More as to the at the institutions irrelevant international security issues nearly lations.10 Even role theorists among those who the hold that trade or IGOs play a positive choice advanced vary. Rational economic interests that are reflected the use impor about

heart of high politics, with no effect independent of existing power re

in promoting peace, reasons

actors' complementary political emphasize in maintaining interstate relations?interests peaceful

in the decisions of national leaders. Fearful of the domestic political

avoid of losing the benefits of trade, policymakers consequences in of force against states with which economically they engage tant trade. But one can also devise constructivist explanations how the communication of shared associated identity.11 with trade builds sentiments

cross-national

9 and Conflict," Mershon International Stud A useful review is Susan McMillan, "Interdependence iesReview 41, no. 1 (1997). 10 "The False Promise of International International Institutions," John Mearsheimer, Security 19 (Winter 1994-95). 11 Emanuel Adler Adler and Barnett and Michael Barnett, in Theoretical in "Security Communities Perspective," Social Theory of International Politics (New York Cam evidence that trading contacts expand elites' views of

bridge University

(fn. 6); Alexander Wendt, Press, 1999). For microlevel

6 Even

WORLD

POLITICS

realists acknowledge that international institutions like NATO to preserve peace among their members the de help by supplementing terrent effect of sheer Liberals the power. military potential emphasize of institutions information for communicating and facilitating bargain see institutions as instruments for expand constructivists ing,12 while conceptions ing people's IGOs vary Because widely of these of identity, and self-interest. relatedness, in their functions and capabilities, any or all be correct in a particular instance.13 As with

may explanations we do not to re the consequences of democratic institutions, attempt solve these theoretical debates here. Instead we seek to offer an empir on interstate ical assessment of the effect of the Kantian influences In expanding the historical domain of the Kantian one en

relations. peace,

counters hurdles (and opportunities) that arise from marked changes in

the nature and of political the importance of international trade, regimes, stan the measured the role of international As organizations. by

dard data on political regimes, Polity III,14 the average level of democ in a racy in the international system has risen since the early 1800s, the mean level of eco pattern that is sporadic and wavelike.15 Similarly, as measured nomic ratio the of bilateral trade to by interdependence gross quent those domestic I but rose again in subse fell after World War product in the number clear is the growth of IGOs, though years. Most of a and sustenance associated with the creation truly global largely emerge and only after World War II. These trends for the

economy

1885-1992 period are shown in Figure l.16Higher

should, interdependence, membership flict for the pairs of countries affected; but we number of democracies trade grows, increases, there will well. The be influences important systemic effect of the Kantian influences time as well as cross-nationally. IGO

levels of democracy,

of course, reduce con also expect that as the and IGOs pairs proliferate, of states as be

on other should, we

hypothesize,

apparent

over

see Daniel Lerner, "French Business Leaders Look at EDC," Public Opinion Quarterly and Bruce Russett, Community and Contention: Britain and America in the Twentieth MIT Press, 1963), (Cambridge: chap. 9. Century 12 and Lisa Martin, Robert O. Keohane "The Promise of Institutionalist Theory," International Se no. 1 (1995); Lisa Martin Studies of Interna "Theories and Empirical and Beth Simmons, curity 20, tional Institutions," International Organization 52, no. 4 (1998). 13 For a review of some relevant hypotheses and findings, see Russett, Oneal, and Davis (fn. 6). 14 Jaggers and Gurr (fn. 2). 15 in the Late Twentieth Century (Norman: Samuel P. Huntington, The Third Wave: Democratization their self-interest, 24, no. 1 (1956); of Oklahoma Press, 1991). University 16 For graphing purposes the scale for bilateral trade/GDP has been increased by two orders of mag nitude and that for IGOmembership has been reduced by one order of magnitude.

KANTIAN
-AVGDEM.AVGDEP-AVGIGO |

PEACE

Year

1 Figure Kantian Variables

cross-sectional time-series statistical method?pooled Our regres sion analysis of data regarding pairs of states (dyads) observed annually at hand. It considers in states' variance ?is well suited to the purposes across in each year and in in militarized involvement dyads disputes we can determine the time. Consequently relations through dyadic across thousands as a function of of differences of conflict likelihood states in dyadic relations or in the in and of of annually pairs changes over a than one of more ternational system from year to year period in the Kantian variables hundred years. By measuring change through we can to from their their effects time, begin systemic disentangle strictly dyadic influences. over time in the average level of democracy, interdepen Changes of the norms elements dence, and IGO involvement capture important con of the international for instance, and institutions system. Wendt, tends norms that world a Lockean politics has system wherein the right to anarchy slowly evolved from Hobbesian dilemma is ameliorated the security by states to exist; these, in turn of sovereign

recognizing

8 effectively elimination: limit

WORLD POLITICS
states the use of force.17 Thus, whereas twenty-two internationally its formal are no longer to subject states were

forcibly occupied or absorbed during the first half of the twentieth cen
tury, no state has lost World War II.18 The within which since conquest sovereignty through of a Kantian of states, emergence subsystem use of force is con the unprovoked have illegitimate, may

recognized

tributed directly to this evolutionary development and affected the probability that force will be used primarily by states that are not par
ticularly democratic, nizations. If democracies then the latter will of weakening tocracies, system lations spheres growth for interdependent, are more have likely than to be concerned inwar, whether or involved in international to win orga

are autocracies about with

their wars, implications or other au

themselves

the security democracies

as the number in the international of democracies especially re most great powers are democratic, their peaceful grows.19 If should reduce the incentive for war for all states across their of influence. among increases and stimulates economic If globalization states will have to be states, nonliberal interdependent

concerned lest they be punished by global markets and trading states

international violence that disrupts trade and invest instigating even little mutual trade may find it pru ment; antagonistic dyads with norms dent to avoid conflict.20 If international and institutions for

17 see Wendt (fn. (fn. 11). On some systemic effects of a high proportion of democracies, Hundey Three Levels of Analysis," four and Havard Hegre, "Peace and Democracy: 6); Nils Petter Gleditsch nal of Conflict Resolution 41, no. 2 (1997); Sara McLaughlin Scott Gates, and Havard Hegre, Mitchell, no. 6 (1999); and Lars in Democracy-War "Evolution Dynamics," fournal of Conflict Resolution 43, Erik Cederman, the Democratic Peace as a Collective "Back to Kant: Reinterpreting Learning Pro at Los of California cess" (Manuscript, Political Science Department, University Angeles, December 18 see Bruce Russett, J. David For dates of independence, Small, "National Po Singer, and Melvin no. litical Units in the Twentieth Century: A Standardized List," American Political Science Review 62, 3 (1968). Germany and Japan temporarily lost sovereignty afterWorld War II, but soon regained it as two states). Kuwait was in 1990-91; but a large, diverse coalition of briefly occupied (Germany states under the aegis of the United Nations in order to protect the sovereignty forced Iraq to withdraw if one regards its unification of established states. South Vietnam is an exception to this generalization in 1976 as the result of external conquest rather than of an internationalized with North Vietnam civil war. Whereas state extinction as a consequence of international war has become rare, the ideology of ethnic self-determination has led to the breakup of many states and empires. 19 A counterhypothesis would be that as democracies become more numerous and more confident in to pursue coercive relationships their individual and collective strength, they may become emboldened with those autocracies that remain. For evidence that democracies do win most of their wars, see Bruce Bueno "War and the Fate of Regimes: A Com de Mesquita, Siverson, and Gary Woller, Randolph no. 3 (1992); Lake (fn. 7); and Allan C. Stam parative Analysis," American Political Science Review 86, ofMichigan (Ann Arbor: University Press, 1996). III, Win Lose orDraw 20 Thomas L. Friedman, TheLexus and the Olive Tree (New York: Farrar, Straus, and Giroux, 1999); of Production and International and Stephen G. Brooks, "The Globalization Security" (Ph.D. diss., forthcoming).

1998).

Yale University,

KANTIAN resolving grow, disputes or international

PEACE

states may be to use impelled to regional organizations help settle their disputes costs rather than accept the political, military, and economic imposed as the liberal for in force. Thus by community penalties using military creases in the Kantian at the system level may have beneficial influences effects on the behavior of dyads that are not particularly democratic, or involved in international economically interdependent, organizations. is not an ecological This do not make inferences about fallacy.21 We at the from conflict In information about conflict level. dyadic systemic all our analyses we address the incidence of militarized among disputes states. We of the of purely dyadic char consequences pairs investigate acteristics for dyadic behavior, but we also do consider the effects of in the international system. To capture the effects evolutionary changes scores of of such systemic changes, we use the annual mean democracy, as a bilateral trade of domestic (GDP), and gross proportion product

even nonliberal

joint memberships

in IGOsgraphed in Figure 1.They effectively gauge

norms and institutions the pervasiveness of changes in international success the and document of liberal states in the com example of the of the also consider the influence among nations. We petition leading on the interstate relations. We this aspect of state, hegemon, investigate measures of the relative power of the hege leading realist theories with states' satisfaction mon, of its own security. with the status quo, and the hegemon's sense

Historical
As

Domain,

Key Variables,

and Sources

of Data

our spans the years 1885 to 1992, it enables us to examine analysis the effects of democracy, economic and international interdependence, over a war and for a few years before the cold organizations long period after. Realists often contrast the dynamics of bipolar and multipolar over is there their for in systems, consequences though disagreement terstate relations. By Waltz's criteria, the international world as is neither system was mul

tipolar for the centuries preceding 1945 but bipolar during the cold
war.22 And the current, post-Soviet is best understood polar but perhaps
21

unipolar,

nor multi bipolar at least as measured by

Identified byW. S. Robinson, and the Behavior of Individuals, "Ameri "Ecological Correlations Sociological Review 15, no. 3 (1950). On how some inferences can be made, see Gary King, A Solu tion to theEcological Inference Problem (Princeton: Princeton University Press, 1997). 22 Kenneth Waltz that defines polarity and says that it is the power of the units (states) themselves see not the number or power of the alliances Waltz, they lead; Theory of International Politics (Reading, Mass.: Addison-Wesley, 1979), 98-99. Thus the formation of two opposing alliance systems prior to on the World War I did not change the structure of the multipolar system. Waltz's systemic emphasis can

10

WORLD

POLITICS

the relative power of the United States (in its capabilities if not in its will to control or shape events). These theoretically based distinctions
us to consider the effects of the Kantian require ent international structures, though evaluation necessarily remains War tentative. variables within differ war era of the post-cold

We
mentary, 1946-49.

omit from our analyses all but the first year of bothWorld War I
trade data for those years are frag II, because bilateral are for the immediate and years, 1919-20 postwar they con all but the first year of the world wars, which Omitting as between our democracies results and autocracies evidence or between two biases but of the demo

andWorld

sisted of conflicts autocracies, cratic peace,

against finding assurance that our results are not de it also provides of our variables termined by these dramatic but atypical events.23 Most we concentrate on in Here and data are discussed previous publications. new. is what

Dependent Variable: Involvement Militarized Disputes War We use the Correlates of

We or code or both states

in

(COW)data on interstate disputes (MIDs).

one in which in a dispute a demonstration of force, the other. The variable DISPUTE Some researchers a urge

each year that a dyad was involved to use force, made threatened force

actually used military 1 if a dispute was equals increases

against and 0 if not. ongoing that the dyad will eases some

that only the initial year of a dispute be noted since a dispute in one
the chances experience in dispute but raises oth

year

problems subsequent years.24 This procedure as all our theories ers. If leaders are rational, assume, to escalate, whether their reevaluate positions, quently maintain of wars, the existing the prolonging

agree with Blainey: strategy. We or of wars and the prolonging shortening

they will fre or deescalate, "The beginning of pe

1945 and all previous years of modern of nuclear weapons would also imply a break between definition would the end of the bipolar cold war system is more problematic. Waltz's history. Dating Dixon and argue for a break at the end of 1991, when the Soviet Union was dissolved. But William a decisive shift in the pattern of Soviet-American conflict in 1988; see Dixon Stephen Gaarder show of Soviet-American and Gaarder, "Presidential Succession and the Cold War: An Analysis Relations, 194$-1992,n Journal ofPolitics 54, no. 1 (1992). 23 F?rber and Gowa (fn. 3) express this concern. 24 of Interstate War," the Likelihood Stuart A. Bremer, "Dangerous Dyads: Conditions Affecting Journal of Conflict Resolution 36, no. 1 (1993); Katherine Barbieri, "International Trade and Conflict: at the annual meeting of the International Studies As The Debatable Relationship" (Paper presented Minn., Beck, Jonathan Katz, and Richard Tucker, sociation, Minneapolis, February 1998); Nathaniel Analysis," American Journal of 'Political "Taking Time Seriously in Binary Time-Series-Cross-Section in fn. 49 below. Science 42, no. 4 (1998). See, however, our comment effects

KANTIAN

PEACE

11

same riods of peace all share the same causal framework. The explana same are in framework and the factors vital each tory understanding war and we state in the of 166 sequel investigated peace."25 Moreover,

multiyear disputes during the post-World War

more than half the course

II era and found that

a over in the level of force involved change employed arose as the first was or that a new of the dispute dispute in Thus we report analyses of states' involvement concluding. disputes rather than of just their onset; but as in earlier studies of the cold war

era,26 we reestimated key analyses using only the first year of disputes without material differences from those reported below. finding

Dyadic
We not

Independent

Variables

variables by one year to ensure that they were lag all independent a some to be affected For by dispute explained. explanatory this precaution is clearly important; variables for example, conflict may constrain limit trade just as trade may conflict. A similar reciprocal can be and for international organizations relationship imagined as many states formed conflict, among IGOs?though hardly all?are For that maintain relations. other variables such considerations peaceful are prone to conflict, countries proximate Geographically not but the frequency of their disputes does affect their proximity. To be we the variables. all This however, consistent, lag independent not put to rest all about the does direction of precaution questions is a reasonable step at this time.27 All causality, but it in the appendix. listed by their acronyms the variables are are irrelevant.

DEMOCRACY We a summary measure of the III data to compute Polity on score character of the every country's regimes, subtracting political on the score its scale. The scale from autocracy democracy resulting to +10 for variable (DEM.) ranges from -10 for an extreme autocracy use the a can result from the states. Because democratic dispute a function is primarily of conflict of a single state, the likelihood state in less constrained of the degree of constraint the by experienced link in the chain of peaceful each dyad. As that state is the weak the most actions
25 Geoffrey Blainey, The Causes ofWar, 3d ed. (New York: Free Press, 1988). 26 Oneal and Russett (fn. 6,1999). 27 Kim (fn. 6), using a simultaneous equation model, finds that the effect of trade on conflict ismuch (fn. 6) construct a model for predicting stronger than the reciprocal one. Russett, Oneal, and Davis IGO is an effect, but it that includes, among other factors, the absence of conflict. There membership isweaker than the influence of IGOs on conflict.

12

WORLD POLITICS

state in a relations,28 we expect that the less democratic dyad is (DEML) more the stronger determinant of interstate violence. the Conversely, more that it democratic constrained will be from engaging in state, the a more we In and the the dispute dyad. previous analyses peaceful that the difference between states' political found, as Kant had expected, also affects the likelihood of conflict. Democratic-autocratic regimes were the most in the cold war era; two autocracies dyads conflict-prone were were less likely to fight, and two democracies the most peaceful. We reconsider these findings below.

The Polity III regime scores exhibit some problems of comparability over time. Until 1918 about 40 percent of British males (dispropor
tionately working class) were disfranchised by residence requirements;

female suffragewas granted partially in 1918 and fully only in 1928.29 In theUnited States women obtained the vote only in 1920, and blacks
were excluded until the 1960s. Swiss women achieved systematically are reflected in 1971. Some of these the franchise in the changes only in the international in rising levels of democracy Polity data and hence 6 to 7 on the de For from the United goes system. Kingdom example,

mocracy scale in 1880, to 8 in 1902, and jumps to 10 only in 1922. But Switzerland is coded at 10 from 1848, as is the United States from
on of these restrictions 1871. The consequences political participation not for foreign be trivial. In the United contemporary policy may women are more averse use of to for the States, significantly example, are men force than and vote in part on this basis.30 Thus the military in earlier of women from the franchise could exclusion have periods states reduced the tendency of even the most "democratic" profoundly to avoid conflict.
28 of International Conflict," American William J. Dixon, "Democracy and the Peaceful Settlement Political Science Review 88, no. 1 (1994). 29 Trevor Wilson, The Myriad Faces ofWar: Britain and the Great War, 1914-1918 (Cambridge, En The English Parliament (Harmondsworth: gland: Polity Press, 1986), 660-61; Kenneth MacKenzie, 1950), 106. Penguin, 30 and Jonathan Nagler, "Explaining the Gender Gap Carole Kennedy Chaney, R. Michael Alvarez, inU.S. Presidential Elections," Political Research Quarterly 51, no. 2 (1998). To take such changes into uses an account, Zeev Maoz adjusted threshold of democracy for all countries that shifts upward in 1870 (for general male suffrage) and 1920 (female suffrage); see Maoz, Domestic Sources of Global ofMichigan Press, 1996), 54. Our use of unadjusted democracy scores Change (Ann Arbor: University I. Kristian Gleditsch and thus leans against our hypothesis of democratic peace before World War minus Autocracy Michael Ward note that our continuous measure, Democracy score, has the virtues of is unstable over time; being symmetric and transitive; but the relative importance of its components see Gleditsch inModern "Double Take: A Re-examination and Autocracy andWard, of Democracy no. 3 (1997). For the this aggregated mea Polities," Journal of period 1880-1969 Conflict Resolution 41, con sure is for executive recruitment; subsequently largely influenced by the degree of competition the relatively stable earlier period straints on the executive are the main determinant. Fortunately war years we add here. As no covers all the analysis of the democratic peace afterWorld War pre-cold II has yet addressed the 1969 break, we too leave that for later investigation.

KANTIAN

PEACE

13

ECONOMICINTERDEPENDENCE
For most of the post-World War ian variable is straightforward, Fund reports statistics regarding to influence II era the measurement because bilateral the International trade. Since of this Kant Monetary is expected we di important, trade

dyadic relations only if it is economically sum vide the of a country s exports and imports with its partner by its in the standard references for the years after 1950.31 GDP, as reported we As with the influence of democratic institutions, expect the likeli

hood of a dispute to be primarily a function of the freedom of the less

mea trade-to-GDP is, the bilateral on trade with sure of the state its dyadic economically a tests We also for effect of asym report partner positive (DEPENDL). on conflict, as metric theorists. by dependency dependence proposed we move back to the years before World War na When II, however, constrained that tional 1920-38 economic data become of Nations more problematic. the League in current the years During data on bilat contemporary state to use force, less dependent

nor for the period before 1914. BeforeWorld War I the annual editions
of The Statesman's Yearbook offer the closest approximations, but these rates for data are less standardized, the appropriate exchange converting unit are less certain, and more data are the data to a common missing.34

compiled rates.32While the ac along with exchange are the and data of these less than in curacy comparability undoubtedly are no institu the later IMF reports, they are the best available. There tional compilations of trade data for the years of the two world wars, eral trade values,

31 International Monetary IMF, 1993; Fund, Direction of Trade (ICPSR 7623) (Washington, D.C.: for Political and Social Research). Consortium distributed by Ann Arbor, Mich.: Inter-University Robert Summers et al., The Penn World Table (Mark 5.6a) (Cambridge, Mass.: National Bureau of Economic Research, 1995). Due to missing data for trade and/or GDP, the great majority of dyads in are omitted, as are most Arab-Israeli volved in the Korean and Vietnam Wars dyads. Since most of no trade, our those are conflicting democratic-autocratic analysis is likely to be biased dyads with against the liberal hypotheses. Because these conflicts spanned several years, excluding these cases mit in the time series, as does omitting all but the first year of igates the problem of temporal dependence states the world wars. Also omitted are roughly 2,500 communist dyad-years: non-IMF members. These traded among themselves but did not report it to the IMF and generally had little conflict. Had we been able to include them, the post-1950 sample would have been increased by only about 2 percent. 32 International Trade Statistics (Geneva: League of Nations, annual volumes). League of Nations, 33 Martin Epstein, ed., The Statesman's Yearbook, 1913 (London: Macmillan, 1913), and earlier an nual editions by other editors. 34 re took several steps to minimize We trade data in this period. We used information missing one state s exports to another to infer its we between known partner's imports; interpolated garding values of trade and used the average value of a dyads trade to extrapolate; and we assumed, for those states for which we had data, that there was no trade between any two if neither reported any exports or we have trade data for 61 percent of the and imports with the other. As a result dyads 1885-1913 conducted several tests to see if these methods might have biased our results. First we 1920-38. We zero values of trade, and then we and extrapolations. Analyses dropped all interpolations dropped all with the remaining "reaT data, 1885-1940, revealed little change in the results. We also determined

14 Because of these difficulties Yearbook

WORLD POLITICS
we collected alternative estimates for bilat

eral trade in the 1885-1949 period, compared them with the data from
The Statesman's data from our the and the League of Nations, and adjusted as principal appropriate.35 on mea numerator Information of the dependence dyadic trade, the for the pre-1950 era, however. To cal sure, is only half of the problem sources the economic domestic of trade we importance No comprehensive products. provides estimates need estimates dollars dollars collection of GDP data of nations' ex

culate gross

ists, but Maddison a two-step procedure of countries. number GDP estimates where they were on

in constant in current Maddison

for fifty-six for a large dollar

countries in all regions of theworld for 1870-1992.36 We used these in

to estimate we the GDPs First, states' total regressed annual energy the year, and various s constant

located,

the region consumption, terms. Annual interactive

War energy consumption, collected by the Correlates of

(cow) project,

as is a good correlate of the size of national economies, Morgenstern, than 93 percent noted twenty-five Knorr, and Heiss years ago.37 More GDPs was inMaddison's Based on the coeffi of the variance explained. the constant dollar GDPs in this analysis, we were able to estimate these con for a large number of other countries. Second, we converted stant dollar estimates to current dollars, Maddison's U.S. dollar using GDP deflator. cients

MEMBERSHIPS JOINTIGO
The influence of international variable, last Kantian on interstate the conflict, organizations IGOs is assessed by the number in which of both

that the sample of dyads for which we have trade data is unlikely to be biased. To do this, we created a variable (MISSING) that equaled 1 if DEPENDLwas missing and 0 otherwise and then changed all miss zero. We then estimated equation 1 below with the variable MISSING added. ing values of DEPENDL to It was not statistically significant, indicating that the incidence of disputes among the dyads for which trade (or GDP) data are missing does not differ from that for the dyads for which data are available. 35 for each region of the world and for the United These include volumes by Brian R. Mitchell of Commerce, Press, various years); U.S. Department University Kingdom (Cambridge: Cambridge Historical Statistics of the United States: Colonial Times to 1970 (New York: Basic Books, 1976); and rates come from Katherine Barbieri's data posted at http://pss.la.psu.edu/TRD_DATA.htm. Exchange U.S. Federal Reserve Bank sources, The Statesman's Yearbook, and Global Financial Data Company, www:globalfindata.com. 36 for Economic the World Economy, 1820-1992 (Paris: Organization Angus Maddison, Monitoring "A Long and Development, 1995). His U.S. dollar GDP deflator is found inMaddison, Cooperation on Institute of Economic of Groningen, Run Perspective Research, University Saving" (Manuscript, October 1991). 37 Klaus Knorr, and Klaus P. Heiss, Long Term Projections of Power: Political, Oskar Morgenstern, Economic, andMilitary Forecasting (Cambridge, Mass.: Ballinger, 1973); and also John R. Oneal, "Mea Balance of Power, International Interactions 15, no. 2 (1989). Base of the East-West suring theMaterial

KANTIAN PEACE
states in a dyad share membership, as reported by the Yearbook

15
of Inter

national Organizations?* Simply counting joint memberships

from tions. down 0 to over measure 130 for some dyads in recent of the importance and effectiveness organizations that are weak

(ranging

an ideal years) is far from of international organiza and strong, regional and

It includes

global, functional and multipurpose. Ideally the total should be broken

and some organizations but this is hard to given special weight, do as a practical matter and there is little theory to guide the attempt. For now we use the simple count of joint in intergovern memberships IGO. mental is this variable labeled organizations; CAPABILITY RATIO The on states' use of first of the realist constraints force is rel military a ative power, specifically the balance of power within idea dyad. The that an equal balance of power may deter conflict has deep historical of capabilities is more roots, as does the idea that a preponderance as to which side suggests, however, that it is preponderance that deters military action.39 Our index of rel ative power of the ratio of the (CAPRATIO) is the natural logarithm to state's index of that the weaker member stronger military capability make these calculations forces.40 using the COW data on pop and military

to preserve the peace by likely uncertainty reducing would win a contest of arms. Recent empirical work

in each dyad. We ulation, industry, ALLIANCE Allies because political are

thought generally common they share and economic

to

fight

each

other

less than other often for this

states

security interests as well. We

interests.

They control

share other influence it equals 0

using a variable (ALLIES)that equals 1 if the members of a dyad were

linked by amutual otherwise.41 defense treaty, neutrality pact, or entente;

38 We extended the data from the sources in Russett, Oneal, and Davis (fn. 6 ). 39 Bremer (fn. 24); Jacek Kugler and Douglas Lemke, eds., Parity and War: Evaluations and Exten re sions of the War Ledger (Ann Arbor: University of Michigan Press, 1996). Waltz (fn. 22), 117-23, views the balance of power literature and states his own version. 40 Data are from J. David Singer andMelvin Data Small, National Military (Ann Arbor: Capabilities of the data ofWar Project, 1995); the date of final modification of Michigan, Correlates University was December 28,1994. 41 We updated J. David Singer, Alliances, 1816-1984 Corre of Michigan, (Ann Arbor: University Treaties and Al lates ofWar Project, 1995), with material from N. J. Rengger, with John Campbell, liances of the World, 6th ed. (New York: Stockton, 1995).

16

WORLD

POLITICS

CONTIGUITY AND DISTANCE exists when for interstate violence The potential a can reach the other with effective military dyad pecially likely natural most as one goes farther back reasons to have the most at least one member force. For most of

states

the ability to do so is determined foremost by geographic proximity, es

are in history. Furthermore, neighbors to territorial boundaries, fight?over of cross-border ethnic groups, and so reduces to the capability fight and is extremely this finding strong terms in our regression analy DISTANCE is the natural

on. Thus

the grievances resources, the constraint of distance research. we include

of the incentives

to do so as well; two different

in previous

this effect as fully as possible. the great circle distance in miles between the capitals of of logarithm the major ports for the largest countries); the two states (or between

Accordingly, ses to capture

using the logarithm acknowledges a declining marginal effect. Addi

that equals 1 if two states or other (via colonies indirectly contiguous depen directly or a are land boundary It equals 0 if they share dencies). separated by nature of colo of the widespread less than 150 miles of water. Because = are not correlated nial empires, these two measures (r2 0.21), highly in constraining II. The effect of distance up toWorld War especially those with is less for the great powers: the land, sea, conflict, however, or (in the last to deliver substantial air forces or half-century) capability COW has these destructive The identified power globally. major project tionally are not include NONCONTIG, or on the basis of a consensus of historians. To give full consider powers ation to realists' concerns, we add a third variable, MINORPWRS, coded 1 and 0 for those that include at if a dyad is composed of minor powers our view that conflict be consistent is least one great power. with (To we reverse the and but subject to constraints, endemic terminology two variables from those in research of the last reports previous coding where used CONTIG and MAJOR. This has no effect on our statistical Note that other than to reverse the sign of the coefficients. analyses, some one or two also include major powers.) contiguous dyads states In most of the analyses below, we include all possible pairs of COW member is data information for which available, using regarding we system ship in the international limit our study to the politically ous states and dyads containing to believe These that such a restriction are much more dyads to generate these cases. Thus we do not as relevant dyads, identified contigu at least one continue We major power. makes good theoretical sense, however. to engage in military disputes. Polit we a measure

likely

KANTIAN

PEACE

17

ically relevant dyads constitute just 22 percent of all the dyads forwhich
we account for 87 percent of all the dis data; nevertheless they are In other words, the politically relevant dyads putes. twenty-four a militarized to times more than are those we likely experience dispute have deemed are to be "irrelevant." dyads liance ported contagion commitments. below to be some And among disputes of into conflicts effects being drawn We of the include all dyads in most sure we are not causes of the ignoring the consequences these through other al re

have

analyses these other the non

but we disputes,42 relevant pairs.

also explore

of including

Systemic
kantian

Independent
systemic and

Variables
relative dyadic measures

To clarify the influence of the international system on the likelihood of

create three conflict, we dyadic three realist variables, the latter effect on interstate relations. The forward means derivations of DEM, DEPEND, system-level Kantian variables and to capture the hegemon's designed are variables three Kantian straight our measures: we the of basic simply computed and IGO for each year. These are the measures

(omitting the years of the world wars) graphed in Figure 1. In the

as AVGDEM, AVGDEPEND, are identified and below, analyses they AVGIGO. We the that the greater these systemic measures, hypothesize more normative and institutional will reflect the the global system and the rule democracy, interdependence, assess to the standing of each dyad in each year of law. It is also possible three relative to our three annual Kantian averages. Thus we calculated = / the standard relative dyadic measures: RELDEML (DEML AVGDEM) = / the of DEM; deviation AVGDEPEND) RELDEPENDL (DEPENDL constraints associated with were identify the dyads that in involved and democratic, intergovernmental interdependent, at each in time. by the standard organizations point By dividing standard deviation of IGO.These measures we can

standard deviation of DEPEND; and RELIGO= (IGO AVGIGO) / the

most

coefficients. these estimated compare directly measures in a single equation allows relative and systemic Combining us to compare variables values of the Kantian the effect of changing to versus the relative of time dyads cross-sectionally standing through deviations,
42 As recommended Reed, byWilliam of the Peace sented at the annual meeting tober 1998). "The Relevance Science Society of Politically Relevant Dyads"(Paper pre [International], New Brunswick, N.J., Oc

18 the annual means. We

WORLD expect the

POLITICS systemic and relative variables to

make independent contributions to the frequency of dyadic disputes.

REALIST SYSTEMIC MEASURES Hegemony. prominent international most We also relations. state create theories three variables systemic the hegemon's regarding associated influence with on

realist

powerful constrain weaker

to deter adversaries from using and the ability a way to its interests. A in detrimental crude but military measure state is its share of all reasonable of the power of the leading in each year. As before, we use COW data the major powers' capabilities to make this calculation. force in all cases. is not obvious of the hegemon Through it is not clear whether much of recent history any state was truly hege It is generally in the thirty years before World War monic.44 agreed that Identification I the United Kingdom was closer than relative hegemonic, although United the interwar era the United States States was declining. During economic and than the had greater military clearly strength potential was United but its actual military power only about equal. Kingdom; Moreover, volvement its geographic in the Central position European and isolationist system. policy limited we its in accept Consequently, its power any other country to both Germany to being and the

the peace might sphere of influence

that the theory postulates Hegemonic-stability in the system, has the ability to the hegemon, states from to violence.43 This power to resorting keep as the hegemon's be manifested dominance within

Organski and Kugler's judgment that Britain was the hegemon in the
II years, if any state interwar period as well.45 In the post-World War can be said to have been it is the United States. Hence we hegemonic, of the hegemon's power in the first sixty years analyzed

use the proportion of capabilities held by the United Kingdom as the

measure and

has that of the U.S. after 1945. Our systemic indicator (HEGPOWER) reasonable face validity, declining from 33 percent in 1885 to 14 per cent in 1913, and dropping under 11 percent by 1938. America's hege mony is manifest immediately following World War II, when it controlled 52 percent of the major powers' capabilities. This declined
43 Robert Gilpin, War and Change inWorld Politics (Cambridge: Cambridge University Press, 1981). 44 "The Mysterious of American Bruce Russett, Decline or, IsMark Twain Really Hegemony, Dead?" International Organization 32, no. 2 (1985). 45 A. F. K. Organski of Chicago Press, 1980). and Jacek Kugler, The War Ledger (Chicago: University see David Sacko, "Measures of On measurement, (Paper presented at the annual meeting Hegemony" of the Peace Science Society [International], New Brunswick, N.J., October 1998).

KANTIAN

PEACE

19

to 26 percent by the early 1980s but rose to 29 percent with the collapse of the Soviet Union.
with the status quo. The power-transition Satisfaction theory origi not advanced consists of propositions by Organski nally only about the influence of an imbalance of power but also about the role constraining status states' with the satisfaction quo. States rising in power played by tional effort Lemke system it dominates. to subsume the democratic contend hegemon and Reed extend this rationale in an

will challenge a hegemon only if they are dissatisfied with the interna

ory.46 They because the used benefits cratic

the peace within power-transition that democracies have fought less historically has been democratic since the end of the that

Napoleonic Wars. First Britain and then theUnited States, it is argued,

an international to construct its power system to itself and its democratic allies. Thus mostly with the status quo state and reinforced provided democracies'

created by the most demo powerful accounts its of alliances for the system by we assess democratic and Like Lemke this Reed, peace among dyads. a measure view by computing of each state's satisfaction with the status its portfolio between of alliances and quo based on the correspondence as indicated measure that of the hegemon, the tau-b of statistical as by sociation. Then we multiply the scores of the two states in a dyad to create a measure of joint satisfaction indi (SATISFIED).47 This measure satisfaction

cates the degree towhich each dyad is content with the distribution of
benefits state. the leadership of the dominant tensions. Both hegemonic-stability Hegemonic theory and power transition hold that the international will be more peace system theory to its the hegemon ful when is strong relative rivals. The principal achieved under

concerns for its may also affect the system by transmitting hegemon own to other states. International tensions the security involving hege monic for its allies, its rivals, its power are likely to have consequences rivals' allies, and even neutral states. "When elephants as the gets trampled," adage goes. that when small animals metaphor, It is also possible, ones will fight, big fight, the grass to extend the be drawn in.

see an op states may intervene in conflicts because Large they ongoing or to achieve ten avoid losses. Either way, international portunity gains
46 Lemke and Reed (fn. 3). 47 We added 1 to each state s tau-b score to make it positive. The tau-b index of the similarity of al liance portfolios was introduced by Bruce Bueno de Mesquita, "Measuring Systemic Polarity," Journal no. 2 (1975). It was as a of Conflict Resolution 19, adapted dyadic measure of satisfaction byWoosang Science 35 (1991), and and Great Power War,"American Kim, "Alliance Transitions JournalofPolitical subsequently used by Lemke and Reed (fn. 3).

20

WORLD

POLITICS

sions may be contagious. of To assess this view, we created a measure sense of its own its the hegemon's defense security, calculating spending as a share of its GDP (hegdef).48 We that the global system hypothesize is commit will experience more numerous when the hegemon disputes more of its resources to the the In such times, pre ting military. hegemon assess the greater threats to its interests. To scope of sumably perceives we consider whether rises mosdy in disputes for involvement contagion, or states. its the hegemon for the and for unallied allies, itself, hegemon

Results
We evaluate the Kantian First we is primarily peace, 1885-1992, logistic regres employing assess the effects of democracy, interdependence,

and IGOsusing a simple dyadic specification. In this view the likelihood

eco state less constrained determined by the to which the We also consider the nomically politically. degree polit of the other member of a dyad affect ical and economic characteristics of conflict or

sion analysis.

the likelihood of amilitarized

temic and cross-sectional consider

dispute. Next we disentangle the sys

of the Kantian of trends dyadic in the variables underlying and involve interdependence, we state in in variables on

influences the effects

conflict. We and each ment

dyads degree of democracy, in IGOs relative to these annual realist tenets

central vestigate the international We examine

averages. Finally systemic the role of the leading regarding in militarized interstate for a total

system. the involvement of states Because noted

nearly 6,000 pairs observations. 150,000 conditions wars

observed

gins with disputes in the year 1886 that are explained by reference to
the two world earlier, we do not consider or after the first year of conflict the immediate postwar years; that in 1885. As otherwise equations We adjust significance clustering the in our re the coefficients indicated, we estimate the general (GEE) using estimating equation estimate for first-order and (ARl) autoregression errors that take into ac using robust standard of our data by dyads. Thus we respond to the con

annually, of the lagged variables

of disputes of almost be

the analysis

is,we exclude disputes for 1915-20 and 1940-46.

Unless gression method. statistical count

validity

48 of the COW index of militarily is a component relevant capabilities. On the Military expenditure see John R. Oneal and of our measure, "The Effect of Alliance Mem Hugh Carter Whatley, on National Defense Burdens, 1953-88," International Interactions 22, no. 2 (1996). Changes bership in this index for the hegemon's military burden correlate highly with changes in the average military burden for all the major powers.

KANTIAN PEACE
cerns their doubts cally raised by Beck, recommended about

21

on GEE rather than on Katz, and Tucker. We rely solution for because of dependence temporal its appropriateness, given the strong, theoreti especially

relation since a between trade and the time elapsed specified our We last reestimated have, however, dyad's dispute.49 key equations as a check on our our Because using their method findings. hypotheses are directional and we have corrected in the as for these violations sumptions statistical underlying significance. regression analysis, we report one-tailed tests of

Evaluating
Weak-Link

the Kantian
Specification

Peace Using

the

Our first test is the simplest.We

a function

expect the likelihood of conflict to be

to which state of the degree the less constrained primarily to use is each of several dimensions is free force. This the along military a state is more to weak-link that this likely precipitate assumption or on in break the peace: the less the political economic constraints that we state's use of force, the greater the likelihood of violence. Consequently score and the lower bilateral include the lower democracy trade-to number

GDP ratio. The

in international of joint memberships organi a zations is it the Kantian measure; inherently dyadic completes a measure in the regression include of the specification. We equation an indicator of whether of balance and the members of a power dyadic the two are minor

are allied. We also control for the distance dyad separating or not are and whether both states, whether contiguous, they Our first then takes the form: equation powers.50 DISPUTE= DEML + DEPENDL + IGO + ALLIES + CAPRATIO

+ NONCONTIG+ DISTANCE+MINORPWRS (1) The results of estimating equation 1, found in the first column of Table 1, provide strong support for the pacifying influence of democ
49 On GEE, see Peter J. Diggle, Kung-Yee Liang, and Scott L. Zeger, Analysis of Longitudinal Data in StataCorp, Stata Statistical (Oxford: Clarendon Press, 1994). We used the computing algorithms 1997). For Beck, Katz, and Tucker's Software, Release 5.0 (College Station, Tex.: Stata Corporation, see fn. 24.We methods, express our doubts that the effects of the theoretical variables and of time are as Beck, Katz, and Tucker's method GEE allows separable, requires, in Oneal and Russett (fn. 6,1999). in the time series but gives the theoretical variables primacy in accounting for for temporal dependence interstate disputes. Beck, Katz, and Tucker introduce the PEACEYRSvariables into the estimation process as Duration De coequals of the theoretical variables. See also D. Scott Bennett, "Parametric Methods, Data Revisited,"'American 43, no. 1 (1999). pendence, and Time-Varying Journalof"PoliticalScience 50 are found in Oneal and Russett Our recent specifications (1997); and Russett, Oneal, and Davis draw on Barbieri (fn. 24). (fn. 6). The controls, from Oneal and Russett (fn. 6,1999),

Models

of the

Kantian
1.1886-1992

Peace,

1886-1992:

1 Table Predicting

Involvement

2.1886-1992 Peaceyears Correction

Variable Lower democracy

Simplest,All Dyads
-0.0658**

-0.0628***

(deml) Trade/GDP (dependl)

International organizations

(0.0106)
-57.8650**

(0.0093)
-31.0726**

(15.4901)
-0.0010

(10.6036) 0.0160* (0.0042)

-0.1913***

(IGO) Capability ratio (capratio)

Alliances

(0.0379)
-0.2337**

(0.0502)
-0.2511

(0.0401)
-0.3691**

(alliances)
Noncontiguity

(0.1659)
-2.0038**

(0.1574)
-1.5864***

(noncontig)
Log distance

(0.1836)
-0.4647**

(0.1532)
-0.3615***

(distance)
Only minor powers

(0.0571)
-1.8392**'

(0.0498)
-1.7208***

(minorpwrs)
Constant

(0.1706)
-1.9349**

(0.1351)
-1.6174***

Chi2 P of Chi2 Log Likelihood

Pseudo N R2

(0.4731) 1354.80 0.0000

(0.4060) 1920.45 0.0000 -5732.4260 0.284 149,404

one-tailed test but wrong sign

149,373
*p<.05; **p<.01; ***p<.001, one-tailed tests; #p<.001,

33

KANTIAN PEACE

23

state in a the less democratic racy and trade: the more democratic dyad more is is and the likeli the the trade, greater economically important are measures hood of peace. The lower democracy and dependence

both significant at the .001 level.The number of joint memberships

does IGOs, however, in this specification not have a statistically significant < is a consequence This .40). (p effect of two on conflict things:

in
the

inclusion of all possible dyads in the analysis (not just those thought to
in the number of interna be politically and the rapid growth relevant) over time. The tional organizations realist variables perform generally as at the the is only indicator of alliance expected, though significant .07 level: (1) a preponderance of power rather than a balance deters as are those whose states are prone to conflict; (2) contiguous fight, are homelands and (3) major powers are in geographically proximate; are in disputes more volved than are smaller states. All these variables a at the .001 level.51 onset (or first year of the only) significant Using as the variable produced nearly identical results. dispute dependent

Column
using ries. The most not Beck

2 of Table
coefficients

1 shows the results of estimating equation 1

in the time se for temporal dependence are and significance levels similar. The usually IGO. Its coefficient is now involves the variable four times its standard error.52

et al.'s correction

notable

only positive

exception but nearly

To see if the pacific benefits of the Kantian variables are limited to

the cold war era, we first reestimated column equation 1 for just the early years,

1886-1914
Comparing analysis smaller
51 To

and 1921-39 using GEE.The results appear in column 3.

them with 1 shows much

the same pattern as the < score for all years. Both the lower democracy .001) and the (p < ratio (p bilateral trade-to-GDP .004) are highly significant.

test the robustness of these results, we estimated separate regressions for each theoretically in status. The signs and teresting variable with just the controls for distance, contiguity, and major-power one were consistent with those in the multivariate exception. Joint regressions, with significance levels < .001) reduced conflict in the restricted IGOmemberships also reesti analysis. We significantly (p because this variable has mated equation 1 after dropping the measure of economic interdependence the most missing values. The pacific benefits of democracy remained strong (p < .001). Joint member was associated with a reduction in conflict (p < .02) when DEPENDL was significandy ship in IGOs, too, states share memberships in international organizations. omitted. Not surprisingly, interdependent 52 < .001). We suppress coefficients for the four spline segments to save space. All are significant (p IGOs are the only ones not ro In this equation, and others presented subsequently, the coefficients for As our results suggest, joint for adjusting for temporal dependence. bust to the different methods in IGOs is most correlated of the three Kantian variables with the years of peace since a membership our work on IGOs. GEE We also esti preference for preceded dyad's last dispute. Our methodological < .001) or fixed effects logistic regression. Greater democracy mated equation 1 using conditional (p < .05) continued to be associated with aswas the ex and interdependence relations, peaceful dyadic (p in IGOs and a greater capability ratio increased the prospects istence of an alliance. Joint membership on at least one are for dyads that experienced the observations results based of conflict. These 20,289 zero. cases were dropped because the dependent variable always equaled dispute; 129,092

24 Democracy benefits Gowa53 other?an and

WORLD POLITICS
interdependence had strong peace-inducing effects

during the multipolar period after 1885 and before the cold war. The
are strongest in the interwar of democracy years, but, as toWorld War I democra also reports, by the decade leading important shift that is obscured by using the years

cies had become less likely to engage inmilitarized disputes with each 1886-1914 as the period of analysis. In light of this evidence, the ab
peace era?becomes in the nineteenth its presence century?not answer to be the anomoly The explained. in that of democratic inclusiveness politics of the international coefficients system.

sence of democratic in the cold war may lie more than

in the lower in characteristics The

century period similar before

Our measure of joint memberships

1885-1939. 1940 other for the early years is even weaker

in IGOs is insignificant for the

1 are in equation reasonably effect of alliances period. The

and the entire < all years are considered. .19) than when (p an indicator 1 for the We also estimated after equation creating war years and terms with 1989-92 interactive each post-cold forming variables. changed have been The results indicate that the influence of has not in this short span of time and the benefits but IGOs are more important relevant states, politically are not dyads importantly influences we have modeled.

of the three Kantian democracy of interdependence constraints

on the threat

reduced, or use of force.

In the past we limited our analyses in the belief that the relations of most influenced

to the other

and economic by the political states affects our results, we To see how including all possible pairs of states and those 1 reestimate the equation contiguous using just pairs of that contain at least one major power?the politically relevant dyads?

for all years, 1885-1992. This excludes dyads that in the great majority
of cases had no reasonable cause the states were of all three to conflict, to engage in armed conflict be opportunity too far apart and had few issues over which to fight. of the Kantian peace. in international For the dyads most does organizations

The
benefits prone

last column of Table 1 provides strong support for the pacific

elements joint membership

reduce the likelihood of conflict (p < .04).The benefits of democracy

< .001) and interdependence (p sults for the extended period,

Russett,

and Davis, Oneal, are more and ered; they significant

53 Gowa 54 Oneal, 1816-1976

< re .002) remain apparent. These (p are consistent in with those 1886-1992, were where the 1950-85 consid years only statistically.54
(fn. 3), 409, analyze lower-level MIDs for the likelihood of conflict only after 1919.

(fn. 3), 98-100. and Davis Russett, (fn. 6). F?rber and Gowa and find that democracy affects significantly

KANTIAN

PEACE

25

Our tests with all possible pairs understate the pacific benefits of IGOsbecause most of these dyads do not have significant political military relations. The probability that a nonrelevant dyad will become involved in a dispute is only 1/18 that of amajor-power pair; it is 1/44
that of a contiguous and involve dyad. Democracy, interdependence, ment states from in IGOs constrain is no realis but if there force; using tic possibility in conflict, of then the absence of two states engaging these constraints a will not increase the incidence of violence. With all obscures the hypothe dyads included large number of false negatives in 1 The sized relationship. variables interesting equation theoretically as are state of relations, in such the irrelevant they simply explaining numerous ir for example. Burma and Ecuador, are, between Including re shown with relevant dyads can bias the results, as we have recently to trade.55 gard With fects pute the easiest way to show the substantive logistic regression, is to estimate the probability of a militarized of the variables for various illustrative ef dis

same is often used in dyads. The procedure For studies. report the effect example, epidemiological epidemiologists on an that individual will contract of various risk factors the probability are some of their cancer. As in our variables analyses, independent lung not to intervention and for (for example, age, heredity, gender; subject us distance to some others are amenable and contiguity), while degree for us al and of "policy" control (for example, diet, exercise, smoking; inference and IGOs). By statistical liances, democracy, interdependence, an event can in of estimate the the reduction probability they, and we, a amount. given a baseline to make For this, we calculated against which probability because these states assumed the dyad is contiguous, comparisons. We

occurring if any one risk factor for a typical individualwere different by

we find evidence of democratic peace by 1900. Earlier than However, using interactive terms for years, states were not democratic by contemporary standards. As democracy that even the most democratic states may and their antagonisms with authoritarian the common interests of democracies developed, in F?rber and Gowa's analyses is have become more substantial. Support for the benefits of democracy to exclude consideration of all years of the world wars. Due to possible si weakened by their decision not control for alliances. Since alliances show little impact in our analy multaneity problems, they do ses, this may not matter. For results for trade that agree with ours, see Christopher Way, "Manchester and Empirical Evaluation of Commercial Liberalism" Revisited: A Theoretical (Ph.D. diss., Stanford 1997). For results that differ from ours, see Barbieri (fn. 24); and idem, "Economic Inter University, to Peace or a Source of Interstate Conflict?" Journal of Peace Research 33, no. 1 dependence: A Path of interde (1996). Our analyses to date indicate that this is primarily due to our different measures to GDP. The results reported in Oneal and not weight trade by its contribution pendence: Barbieri does are robust to several alternative show that the pacific benefits of trade, 1950-92, Russett (fn. 6,1999) samples, and estimation procedures. specifications, 55 Oneal and Russett (fn. 6,1999).

26 are

WORLD POLITICS

we set each continuous variable prone to conflict. Then particularly at its mean value for the contiguous that the lower de dyads, except more score was made to its median is which value, rep pendence equal resentative. We that the pair of states is not allied and does postulated not include a

that this "typical"dyad would be involved in amilitarized dispute using the coefficients reported in columns 1 and 4 in Table 1. Next we
changed standard allied. The crease these first various two columns in the annual increase of Table 2 give the percentage in a risk of a dyad being involved dispute 1 is based on the coefficients conditions. Column or de under esti the theoretically interesting to the continuous deviation variables measures in succession or by adding the dyad by making a

major

power. We

then

estimated

the annual

probability

mated using all dyads, and column 2 is produced with the coefficients
at the results for just the politically relevant subset of cases.56 Looking is in column it and that dra 1, apparent democracy interdependence

matically reduce the likelihood of conflict. Compared with the typical dyad, the risk that the more democratic dyad will become engaged in a
dispute is reduced by 36 percent. If the dyad is more A autocratic, the

danger of conflict is increased by 56 percent. A higher dyadic trade-to

GDP ratio cuts the incidence of conflict are used

of joint memberships
conflict if all pairs

in IGOshas little effect on a dyad's likelihood of

in the estimation process. When of common mem

by 49 percent.

larger number

of states

analysis is limited to the politically relevant dyads, however, the benefit

of joint memberships in IGOs is clear. If the number

berships is fifty-three rather than thirty-two, the likelihood of conflict is reduced by 13 percent. And when the analysis is limited to politically
relevant pairs, are somewhat The their effects ables. Consider derance fourfold lowers stantially the effects less than when of democracy all dyads and economic are considered. variables is confirmed if interdependence

substantive are

importance compared with the second

of the Kantian the results column

again of power is a standard increase

of changing the realist vari of Table 2. If a state's prepon that reduces deviation the higher, alliance is sub a stan

probability of a dispute by 31 percent, but that resultwould require a

the state. An in the capabilities of the stronger 24 of interstate violence percent. This by or with the dyad is more democratic less than when incidence

dard deviation higher level of bilateral trade.

This baseline probability is .031 among all dyads and .055 for the politically relevant pairs.

KANTIAN

PEACE

27

percentage

of change

Militarized

Table 2 in rlsk for annual for Contiguous Dispute _(1886-1992)_

involvement Dyads3

in a

Based 2. 1. Equation 1 Equation

on 1 3. Equation 2

(AllDyads)
DEML increased decreased by 1 std. dev. -36 by 1 std. dev. DEML 1 std. dev. increased by DEPENDL IGO increased 1 std. dev. by CAPRATIO increased by 1 std. dev.

(Politically Relevant Dyads)

-33

(AllDyads)

+56
-49 -2 -27 -22

+48
-33 -13 -31 -24 -33 -22 -30 -36 -18 -26 -33 +3

ALLIESequals 1
RELDEML increased 1 std. dev. by RELDEPENDL RELIGO increased by 1 std. dev. 1 std. dev. increased by AVGDEML 1 std. dev. increased by AVGDEPENDL AVGIGO increased 1 std. dev. by 1 In each case other variables are held at baseline by increased 1 std. dev.

values.

We

have

argued be

that the characteristics affected are more

of the less constrained

state

largely account for the likelihood of dyadic conflict, but the potential
for violence dyadic may member.57 significantly Democracies as Kant to nature of the other by the at than autocracies peaceful

the national (ormonadic) level aswell as dyadically; but in our previous

found,

and autocracies that democracies expected, one of because the dis another prone political fight particularly tance analysts think that asymmetric interdepen separating them. Other we considered dence may lead to conflict.58 To evaluate these hypotheses are score and trade-to-GDP of the higher ratio, democracy as interactive to 1 and both these variables individually equation adding score or trade-to-GDP terms with the lower ratio.59 democracy the influence
57 Maoz and James Lee Ray, "New Tests of the (fn. 30); Oneal and Russett (fn. 6, 1997); Oneal Political Research Quar for Economic Peace Controlling Democratic 1950-1985," Interdependence, 50, no. 4 (1997). terly 58 and Joseph S. Nye, Power and Interdependence: World Politics in Transition Robert O. Keohane International Stud of Interdependence," (Boston: Little Brown, 1997); John A. Kroll, "The Complexity of theWorld ies Quarterly 37 (September 1993); Immanuel Wallerstein, "The Rise and Future Demise no. 4 (1974); Barbieri (fhn. 24 and 54). System," Comparative Studies in Society andHistory 16, Capitalist 59 is thought to depend on the value of another If the effect of one variable (DEML, DEPENDl) terms and the test should include their interactive DEPENDL* (DEML* DEMh (DEMH, DEPENDh),

research we

28 The

WORLD

POLITICS

in a table but available from the authors, results, not reported that the conflict-prone character of mixed de indicated pairs?one one to II and limited the War mocracy autocracy?was post-World era. animosities be the institutional and ideological Plausibly special tween democrats were dyads and communists, the most had solidified by the cold war, account

for that. In the multipolar period, 1885-1939, dyads consisting of two

democracies and mixed after about 1900. Autocratic peaceful pairs found no evidence similar rates of conflict. We benefits whatever the

that asymmetric interdependence raised the likelihood of amilitarized

dispute. Increasing trade had significant

relative size of the states involved.We did find a declining marginal utility for high levels of economic interdependence.60 Disentangling Influences the Systemic and Cross-National of the Kantian Measures

pacific

Estimating equation 1 indicates that the likelihood of a dispute among

score and the lower is a function all dyads of the lower democracy in in ratio in a dyad but not of states'joint trade-to-GDP memberships that the failure of the IGO vari ternational organizations. We suggested as able to perform expected of irrelevant pairs of states results partly from including large numbers that have no significant relations political

and lack a realistic possibility of becoming engaged in a dispute. By

to contrast, contiguous dyads and those containing limiting the analysis a the of benefits international organizations. major power highlights of joint IGO membership increases We also noted that our measure the contribution of inter rather steadily over time. This may obscure national parisons of inflation. organizations across time to interstate peaceful as with less meaningful, relations nominal by making GDPs in com periods

The

influence of IGO membership

between

can be reconsidered by distin

in international the frequency of states' participation guishing time and the dyads rela organizations through standing of individual at each in time. tive to this annual measure We each point decompose Kantian variable?the lower democracy score, the lower trade-to-GDP

of Multiplicative See Robert J. Friedrich, "In Defense Terms in Regression Equa Multiple DEPENDH). tions," American Journal of Political Science 26, no. 4 (1982). 60 as a inwhich we modeled the effect of interdependence hyperbola suggest that the ben Analyses See Mark Gasiorowski efits of trade increase rapidly and then approach a limit asymptotically. and in the Era of"Detente," Journal of Conflict Resolution "East-West Trade Linkages Solomon Polachek, 26, no. 4 (1982).

KANTIAN PEACE

29

a mea ratio, and the number of joint IGOmemberships?into systemic the value of level of states' sure, score, average democracy interdepen in IGOs (Figure 1), and a cross-sectional dence, or joint membership measure that ranks dyads relative to this annual average. The annual av of the number of joint IGO memberships (AVGIGO), for example, erage time the of international captures organizations, prominence through while the degree of involvement of individual dyads relative to this av

erage (RELIGO)identifies those states that are more

through To distinguish fluences, AVGDEPEND for IGO. Our we the network of IGOs in each year. the systemic and cross-sectional between

(or less) linked

Kantian in

substitute

in equation 1 AVGDEM and for RELDEML DEML, and RELDEPENDL for DEPENDL, and AVGIGO and RELIGO second equation becomes:

+ RELIGO + AVGDEM + DISPUTE = RELDEML RELDEPENDL + AVGDEPEND + AVGIGO + ALLIED + CAPRATIO

+ NONCONTIG+ DISTANCE+MINORPWRS (2) Column

using

1 of Table 3 reports the results of estimating equation 2

variables a negative

all pairs of states. All the relative and systemic Kantian in IGOs have except the annual average of states'involvement rized

sign, indicating that increasing values reduce the likelihood of amilita

As ex all but AVGIGO are very significant statistically. dispute; the three relative measures in the last section, we standardized plained to indi of their estimated coefficients. These permit direct comparison cate that the conflict-reduc trade has greatest important economically

ing benefits,
international

followed
organizations.

by democracy
Two

and joint memberships

systemic vari

in

of the three Kantian democracies

ables also affect the incidence of dyadic disputes: the likelihood of con
flict drops when more important .001 level.61 The there are more with both economically; of the other variables influences in the system and trade is at the variables significant in the equation are

61 To in trend in the likelihood of a dispute over the period 1885-1992. There is a mild downward sure that the systemic Kantian variables were not simply collinear with this secular trend toward de we included in each of the equations reported inTable 3 an indicator of time, creasing rates of disputes, which equals the year minus 1884. The coefficients of the Kantian variables changed very little, and the score and trade-to-GDP ratio remained significant at the .001 level; the measure of average democracy at the .05 level in these tests. If equation 2 is estimated for just the time was never significant in becomes of the average level of interdependence 1885-1939 statistically period, the coefficient at the outset ofWorld War I was higher than itwas significant, primarily because the level of trade at the .001 level. during the interwar years; the average level of democracy remained significant

Table Models Predicting (Dyadic of the Kantian Involvement and Systemic

1886-1992: Peace, inMilitarized Disputes All Influences, Dyads)

2. Systemic Kantian, Heg. Power, 3. Heg. Systemic Kantian, Burden Defense -O.4102**

1. Variable Relative lower democ.

Only Kantian

SystemicVariables
-0.3688***

Satisfaction
-0.3576***

(reldemJ
Relative trade/GDP

(0.0680)
-0.7270***

(0.0677)
-0.7045**

(0.0703)
-0.5149**

(reldependl)
Relative IGO

(0.2333)
-0.1304**

(0.2412)
-0.1060*

(0.2132)
-0.1602**

(religo)
Average democracy

(0.0500)
-0.2383***

(0.0512)
-0.2485***

(0.0502)
-0.2702**

(avgdem)
Average dependence IGOs

(0.0412)
-292.4397***

(0.0412)
-260.3094***

(0.0423)
-355.5549**

(avgdepend)
Average

(36.4178) 0.0043 (0.0109)

-0.2897***

(48.7066) 0.0102 (0.0115)

-0.2787***

(39.7875)
-0.0440**

(avgigo) Capability ratio (capratio) Alliances (allies)

Noncontiguity

(0.0136)
-0.3125**

0.0518
-0.2554

(0.0521)
-0.2186

(0.0135)
-0.3330*

(0.1625)
-2.0080***

(0.1665)
-2.0423***

(0.1636)
-1.9225**

(noncontig) Log distance (distance)

Only minor powers

(0.1803)
-0.4915***

(0.1828)
-0.4637***

(0.1802)
-0.5202**'

(0.0567)
-2.0230***

(0.0597)
-2.0073***

(0.0569)
-2.0694**

(minorpwrs)
Hegemonic power

(0.1893)

(0.1941)
-1.5339

(0.1911)

(hegpower)
Joint satisfaction

(0.9502)
-0.0893

(satisfied)
Heg. defense burden

(0.1057)
17.9704**

(hegdef)
Constant -0.7345 -0.7113

(1.9906)
-0.3735

Chi2 P of Chi2 N
*p

(0.4850) 1559.82 0.0000 149,372

tests

(0.5075) 1530.24 0.0000 147,963

(0.4975) 1529.38 0.0000 149,372

< .05; < .01; < .001, one-tailed **p ***p

KANTIAN relatively

PEACE

31

of a power reduces the likelihood unchanged. Preponderant as do distance, an alliance, or the absence of a in dispute, major power as the the dyad. Using the onset of a dispute variable pro dependent duced nearly identical results. The First, results they show of estimating that dyads equation relatively in time tend more 2 are important for three reasons. or involved in international to be more

ganizations the Kantian

peaceful, supporting IGOs. Second, the results indicate hypothesis regarding the that the statistical of ratio and trade-to-GDP democracy significance

at any point

as well as cross-sectional in equation 1 is the consequence of temporal assurance is valuable variation. This of the robustness of the pacific now have of these Kantian benefits influences. We explicit justification for believing that that states can modify their circumstances by policies the increase democracy, of the and, given significance interdependence, IGO measure, in international relative participation organizations. are it view the that there of in Third, consequences supports systemic creasing democracy eral dyads. The estimated substantive and trade for all pairs of states, not just for the lib coefficients for equation 2 allow us to compare the measures. We of the relative and cross-sectional

importance of conflict for various calculate the probabilities hypothetical we assume as before In the baseline that the risk, dyads. calculating at its mean set each continuous is and variable (or me contiguous dyad again dian for the trade ratio) for this subset of cases. We make it does not include a major power. We allied and assume 1 of Table the dyad un estimate the one at a fi

annual probability that this representative dyad would be involved in a

in column the coefficients using dispute time we change each continuous variable 3. Then a standard by deviation;

nally we make the dyad allied. Column 3 of Table 2 gives the annual probabilities of a dyad being
in a dispute under these conditions. The effects of the cross to their po sectional Kantian rank dyads according variables, which are to the annual sition relative averages, again substantial. systemic a score the risk of conflict For dyads with relative democracy higher increase is 30 percent below the baseline rate; a standard-deviation a 36 means in relative dependence lower probability of con percent involved

flict; and when states' participation in IGOs is higher the likelihood of

conflict Kantian these is reduced variables effects with substantive of the significance by 18 percent. The for interstate relations again emerges by comparing the realist variables. those that result from changing

32

WORLD

POLITICS

A higher capability ratio means lowering the danger of violence by a third, andwhen two states are allied the probability of conflict is lower The risk of a dispute drops by 26 percent if the systemic average of the
a standard score increases to deviation (from -0.47 democracy by it falls 33 percent if the systemic +1.26); average of the trade-to-GDP to .006). There ratio rises by a standard deviation 30 is (about percent no if the in for states' average systemic effectively participation change IGOs grows. Thus, effects throughout two of the Kantian the international have powerful or institu normative system. By in the number of liberal states constrains the systemic variables by 22 percent. Note also the effects of the Kantian systemic variables.

an increase tional means, use of force even are not democratic or by dyads that interdependent.62 to those states that The effect of IGOs is limited, however, participate more of these international forums relative to other pairs. jointly in

Assessing
In our evaluate

the Hegemon's
we claim

Influence

on Dyadic

Conflict

last analyses a central

power-preponderance that conflict becomes clines relative

the role of the hegemon. We first investigate of hegemonic and of the theory stability Both of these realist theories theory.63 predict as the power of the state de leading likely rivals. At the same time, we also address for the democratic sense of its own expenditures In a final test

more

to its principal

the argument that it has been the power of the (democratic) hegemon
to reward we consider by cated its allies whether ratio the that accounts peace. the hegemon's of its military insecurity, to its gross domestic of conflict globally. as indi

a is associated with danger product, heightened assess the We of the hegemon's relative power and states' importance two terms to 2: with the status quo by adding satisfaction equation the of held the HEGPOWER, by proportion major powers' capabilities

Britain (through 1939) and the U.S.

measure member's of joint

our (after 1945); and SATISFIED,

of each dyadic based on the similarity satisfaction, to that of the of allies leading power. It is appropri portfolio the level of conflict in the international system, then

ate theoretically to include both in the same equation. If the hegemon

is able to regulate
62 To score and trade ratio were insure that the effects of the annual averages of the democracy truly or were not to and those that democratic confined interdependent, we systemic only dyads relatively added three interactive terms (AVGDEM*RELDEML,AVGDEPEND*RELDEPENDL, and AVGIGO*REUGO) to are not confined to equation 2. The results indicated that the effects of the systemic Kantian variables just those dyads that rank high relative to the annual averages. 63 A. F. K. Organski, World Politics (New York Knopf, 1968); George Modelski, ed., Exploring Long (fn. 39); K. Edward (fn. 43); Kugler and Lemke Cycles (Boulder, Colo.: Lynne Rienner, 1987); Gilpin 1815-1939: An Empirical Test of Gilpins Model and Major Power War, Spiezio, "British Hegemony International Studies Quarterly 34, no. 2 (1990). of Hegemonic Governance,"

KANTIAN

PEACE

33

its influence should be greatest with those states with which

closely leading in that porters. allied. At state the same time the advantages

it ismost
it

for a state of aligning

self closely with the hegemon should be greatest when the power of the
is relatively situation should large vis-?-vis able be most its principal rivals; the hegemon to confer benefits its sup upon

Column 2 of Table 3 suggests that the strength of the leading state

of hegemonic rivals does matter. The measure < seems to reduce .06). Strong hegemony power is nearly significant (p in the international This effect how violence stems, system. apparent a to the of weakened the from ever, (Britain) prevent inability hegemon relative to its principal outbreak wars. of systemwide the coefficient of our measure in the table, In an analysis not reported of hegmonic reversed power signs when was then the first year of each of the world wars was dropped: hegemony

< positively related to the incidence of disputes in the system (p .003). Apparently the pacific benefits of hegemonic strength do not apply
relations. By contrast, we found during normal periods of international no evidence in these analyses that states' satisfaction with the status quo accounts for the democratic in of joint satisfaction peace. The measure

column 2 ofTable 3 is far from statistical significance, while the signifi

cance of relative we Finally as indicated our measure consider and systemic whether is little democracy changed.64 sense of its own the hegemon's of GDP it devotes to military security, expendi

by the proportion

tures (HEGDEF),is related to the likelihood of dyadic conflict.We

add

defense burden to equation 2. The results of the hegemon's are in column 3. As seen there, the defense burden of this test reported hood of dyadic disputes. There feels endangered. are

of the leading state is positively associated (p < .001) with the likeli
when consequences wide-ranging con is of Nor the heightened the hegemon danger flict limited to the world wars, as with hegemonic power, or significant or its allies. In a separate not in for the analysis reported only hegemon the table, we confirmed that other states, too, experience more disputes com has increased the proportion of its resources when the hegemon to the variables mitted Our and relative Kantian systemic military. of states' Even the systemic measure important. now at the is in international organizations significant participation in part upon the .001 level. The effectiveness of IGOs may depend nonetheless remain
64 in evaluating the role played by states' satisfaction with the We tested alternative specifications measures of satis status quo. We assumption, adding the smaller of the tau-b adopted the weak-link two dissatisfied states might also be peaceful; but these faction to equation 2, and investigated whether terms were not statistically significant.

34 major pendent powers not

WORLD POLITICS
a need to and presumably feeling develop, means for protecting their and promoting use, inde interests.

military

A Kantian
Our analyses correct: for the years

System? Past and Future

1885-1992 economic reduce indicate that Kant was and substan inter

tially democracy, in international organizations

interdependence, the incidence of militarized

involvement

state disputes. The pacific benefits of the Kantian influences, especially

to the cold war era but ex of democracy and trade, were not confined tend both forward from that era and back many decades. Moreover, score of the less are substantial. When these benefits the democracy democratic hood all dyads state of conflict a standard deviation, in a dyad is higher by is more than one-third below the baseline a higher bilateral trade-to-GDP the likeli rate among means ratio that

in the system;

the risk of conflict is lower by half. The pacific benefits of democracy in

the twentieth century as Kant believed, In addition, and the suffrage was extended. matured, states may learn from the success and failure of their policies. The benefits in intergovernmental of joint membership organiza tions are more modest for the but nevertheless significant politically relevant major states and at least one dyads containing dyads?contiguous For these the of power. dyads, probability particularly dangerous by a standard of inter deviation. The pacific benefits are also in the trend this when variable apparent more states involved by a stan all dyads, pairs of to the annual embroiled average systemic in interstate violence. are 18 are clear, and the from the nineteenth century change an is consistent with view: democratic institutions evolutionary

a dispute drops by 13 percent when the number of joint memberships

in IGOs is greater national organizations is eliminated: among in IGOs relative dard deviation percent less likely to become

of the Kantian the influences averages systemic By distinguishing from the standings of each dyad relative to the annual means, we also showed benefits of democracy and trade over time as well as cross-sec on con of The effects the influences systemic Kantian tionally. dyadic flict there those when cies are The international important. are more democracies and when not democratic those or is more peaceful when system trade is greater. All dyads?even less dispute-prone effect constraining

of norms

systemic and institutions trade

interdependent?become increase. The Kantian variables

and when

there are more democra that emerge when states holds for many is economically important

KANTIAN PEACE

35

even for those that participate to only a limited degree in the Kantian
subsystem.65 in IGOs rose steadily, Over the period 1885-1992 states' participation a or but there is little evidence of trend toward increased democracy over the span of time. A long interdependence complete two aspects of toward greater interdependence be masked may by our data. First, the over time. Less and more sample changes developed are states before World War I. peripheral probably underrepresented economic trend

Only with the establishment of the IMFand UN

mation on states' wealth and dyadic level of bilateral trade become

agencies does infor

reasonably complete. the average be overstated Thus, may interdependence in the early years. Second, in the late 1950s and the decolonization 1960s created dozens of new states that were less democratic and less into the global economy than the states already in the sys integrated the average scores for democracy and interdependence. tem, lowering as noted, the we use overstate And of that the dem codings democracy ocratic character of states in much of the nineteenth before suf century was extended frage Both democracy World War II. The especially late 1970s, to women and and those without do show interdependence number of democracies after the cold war jump after has grown steadily since the ended. Trade grew in rapidly property. a marked

the 1970s. Since 1987 these phenomena have been followed by a pre
cipitous

of interstate wars, despite the entry of drop in the number new states into the Our results for the many early post-cold system.66 war years cover but indicate that the beneficial ef 1989-92, only they fects of democracy, and IGOs continued past the end interdependence, our of the cold war. Moreover, of the 1885-1992 sug analyses period gest that the relative peace of the past decade owes less to the systemic effects As chiefly of power and hegemony for the realist influences, distance, power effect than some to influences. growing Kantian of the dyadic characteristics? and minor Kantian at the power status?also have level not also

preponderance, for alliances Realist is. The

reduce the likelihood of disputes. This

lack of a robust abolished occurred does not power when politics.

is not surprising, though the

influences systemic

variables

make a difference in the incidence of dyadic conflict. Both world wars

the hegemonic Britain, as work always hypothesized. state, was weak. Yet more normal During hegemony periods of

65 them. See the references in fnn. 17 and 19 and the textual discussion accompanying 66 Third World War (Lanham, Md.: Rowman, Littlefield, 1999). Monty G. Marshall,

36 international hegemon was relations, powerful

WORLD POLITICS
there were than when more militarized itwas weak; the disputes when and when the hegemon

felt threatened (as evidenced by higher military spending relative to its gross domestic product), the likelihood of disputes rose throughout the
system. Democracies autocracies common likely fought that shared two world their wars some side by side, along with interests. Was the democracies' It is more freedom

alignment that shared an

purely interests

strategic a result of interests? strategic in democracy and economic

role. By contrast, dis alliances had no systematic played important to the cold war. For the influence era, prior pute-inhibiting post-1945 a is it when effect of alliances strains belief evident, strong reasonably to to attribute interests. that effect primarily Of course the strategic as understood cold war was about national substantially by security realists. But it was also about a clash The and economic systems. political and peoples of the free-market physical ties were their The security and national fundamentally dominant governments, felt not were one only threatened of two different classes, that their but also

democracies independence allied with

that their prosperity and especially their political and economic liber
at stake. Hence of life.67 they another to preserve way war era is full of affirmations post-cold built on and prosperity freedom, democracy, Some may be just rhetoric, but sophisticated understand common

the importance of markets. interdependent

about

actors economic global in the role that interdependence their In plays prosperity. 1999 NATO fought a war against Serbia in the name of democracy and a dictatorial not in human that did government rights Europe, against a we shall see whether constitute time threat. In will peace strategic hold democracies and interdependent states, but to call the among * a a now of extinct period in world democratic peace poli byproduct its tics"68 sounds very like a premature of death. report we are toward a synthesis of Kantian and progressing Analytically, Kant argued realist influences and of dyadic and systemic perspectives. a more to that three naturally tendencies operate occurring produce so democ world. Individuals desire to be free and prosperous, peaceful leads to the growth of international racy and trade will expand, which to facilitate these processes. law and organizations Peace, therefore,
67 we have tried to show that the democratic By controlling for states' interests, fact of the cold war; see Oneal and Russett, "Is the Liberal Peace Just an Artifact ests? Assessing Recent Critiques," International Interactions 25, no. 3 (1999). 68 Gowa (fn. 3), 114. peace is not an arti of Cold War Inter

KANTIAN PEACE
does not transformation of humanity upon a moral depend even devils are self-interested can and calculate.69 For Kant, the Enlightenment, this was evidence of an ordered universe or the outcome and individuals certain: would reason would not not always as

37
as long a child of and, per

haps, of providential design. Yet he did not think that the process was
mechanical and states always their enlightened interests. Human agents must that of war, and change behavior. including our world, The current unipolar character?inevitably transitory?of an with no other state close to the power of the United States, provides to build a peace based not on force but also opportunity only military on Kantian not last forever. does Hegemony principles. Consequently, should be extended and deepened, the "cosmopolitan law" democracy of commerce and international law and for human respect expanded, Kant would say this is a moral rights institutionalized. imperative. act prevail, with conformity learn from experience, in

_Appendix:
ALLIES: AVGDEM: 1 if dyad members linked

Variables_

average democracy AVGDEPEND: average dyadic AVGIGO: average number of dyadic score

or entente treaty, neutrality pact, by defense score for all states in a year trade to GDP ratio for all states in a year shared IGO memberships

CAPRATIO: logarithm of ratio of higher to lower power capability in a dyad

democracy DEMH: higher lower democracy DEML: DEPENDH: DEPENDL: DISTANCE: HEGDEF: score in a dyad in a dyad ratio trade-to-GDP ratio

higher dyadic lower dyadic trade-to-GDP

in a dyad in a dyad

DISPUTE:involvement in dyadic dispute

or in miles of dyadic distance between ports logarithm major capitals to its GDP ratio of leading state's military spending HEGPOWER: of all major powers state's proportion of the capabilities leading IGO: number of international shared by a dyad organization memberships a MINORPWRS: not include 1 if major power dyad does

NONCONTIG: 1 if dyad is not contiguous by land border or less than 150 miles of water
RELDEML: DEML of DEM AVGDEM/standard deviation AVGDEPEND/standard deviation RELDEPENDL: DEPENDL RELIGO: IGO AVGIGO/standard of IGO deviation SATISFIED: leading tau-b measure state of similarity of dyad members' of DEPENDL to that of the

alliance

portfolios

69Kant(fn.l),112.