John R. Oneal
Research in political science on the causes of war was dominated until recently by realists.* After the Second World War, scholars reacted to the idealism of the interwar years by asserting the primacy of power. National capabilities, they maintained, shape the behavior of states and determine the outcomes of their interactions as each pursues its national interests in the absence of world government. The onset of the Cold War confirmed the relevance of the realist critique; but E. H. Carr, Hans Morgenthau, and others of the early postwar years, had little to offer beyond a conceptual framework and maxims for the conduct of foreign affairs. Later, Kenneth Waltz sought to inject greater theoretical rigor.
War cannot be explained by reductionist theories, he maintained. National leaders must respond to the constraints imposed by the global distribution of power, so the use of force is not associated with particular forms of government or economic systems. War is rooted in the international system. Even the actions of a country with purely defensive motives may have violent conse-quences because others may be unsure its intentions are peaceful. The solution is a self-organizing balance of power, with networks of alliances to preserve the independence of states while minimizing armed conflict. Waltz, like his prede-cessors, relied on historical examples to illustrate his arguments. Thus, twenty years ago, international relations research was predominantly realist in approach, oriented to the international system, and classical in its scholarship.
Today, liberal rather than realist theories dominate research on the causes of war in political science. Social scientific studies show that democracies are
* The author would like to thank Margit Bussmann, Karl DeRouen, Michael Mousseau, and Bruce Russett for their helpful comments.
unlikely tofight one another; and economic interdependence, too, increases the prospects for peace. Importantly, democracy and trade have effects at least as great as those of power and alliances, and they are more amenable to manipulation. These conclusions are based on quantitative studies of many pairs of states through time. By considering virtually all countries for more than a century, these analyses have great statistical power.
In the next section I review the pre-eminent position of realism twenty years ago, and discuss the reasons for its decline. Then I provide evidence that liberalism has dominated the discourse over the past ten years. To indicate the contributions democracy and interdependence make in reducing the risk of war, I report the results of new analyses of over 12,000 pairs of states over the period 1885–2001, using the same statistical techniques employed by medical epidemiologists. These tests are conducted using a liberal–realist model (LRM) of armed interstate conflict that incorporates key elements from both schools of thought. The contribution economic development makes is also discussed.
Key elements of realism also receive support in these analyses, but realism does not provide a path to world peace. The best hope is continued liberal reforms—the expansion and deepening of democracy and capitalism and the inclusion of more countries in the international economic system.
Fortunately, globalization has advanced rapidly in recent decades, and the prospects for its continued expansion, despite the travails of the Great Reces-sion, are good.
Realist theories of the causes of war
After the First World War, idealists presented moralistic arguments that inspired efforts to abolish war by international agreement. This culminated in the League of Nations and the Kellogg–Briand Pact. The stark evidence of failure presented by the Second World War led to the realist reaction, notably in the publication of the second edition of E. H. Carr’s The Twenty Year Crisis, and thefirst of many editions of Hans Morgenthau’s Power Among Nations, the most widely used international relations text in American colleges.1
Central to Morgenthau’s analysis was the inevitability of the quest for power by states wishing to remain independent. This makes a balance of power necessary if states are to preserve their sovereignty and minimize the risk of war, an argument common in republican political theory since the time
1Edward Hallett Carr, The Twenty Years’ Crisis, 1919–1939, 2nd edition (London: Macmillan, 1946); Hans J. Morgenthau, Power among Nations: The Struggle for Power and Peace (New York: Knopf, 1948).
of Rome.2But Morgenthau did not offer a testable theory of world politics, and the“balance of power” had many meanings. Realism was a conceptual frame-work that oriented study or a paradigm within which advice on foreign affairs could be offered. It was a wake-up call easily summarized: in an anarchic world—in the world as it is, rather than how we would like it to be—a state must be concerned with power and what it can do to develop it. Its power must be adequate to protect its sovereignty and territorial integrity and to promote its national interests. Its policies can succeed only if backed by power.
The power of potential adversaries is mortally dangerous, and states must be able to mobilize, unilaterally or in combination, power equal or superior to that of its rivals, because there is no world government and international law is an empty phrase.“States may fail to recognize these truths, or may not be able to meet these requirements successfully, but prudent men will recognize the validity of this analysis of international reality and try to conform to the requirements which it poses.”3
Waltz sought to inject greater rigor into realism. His Theory of International Politics, published in 1979, set the standard against which theoretical works would be judged for a generation.4Waltz, like Morgenthau, emphasized the importance of the balance of power; but he focused on the structure of the international system, not diplomacy and foreign policy. In Waltz’s neo-realism, what matters are the number of great powers and the distribution of militarily relevant capabilities. A bipolar system is best because it is simplest, allowing the rivals to focus their attention. The danger of war is also reduced by uncertainty about who would win a contest of arms. Neither state can be confident of victory so, especially with all that is at stake, each will act cautiously. The superpowers should also prevent adventurism by small states, fearing that they will be drawn into a costly war. Waltz supported his theoret-ical claims in traditional, humanist fashion, by erudite argument and telling examples; but his historical references were illustrative, not systematically generated or subject to statistical testing.
The importance of realism and Waltz’s standing are indicated in a 1990 survey of thefield widely used in graduate schools. In Dougherty and Pfaltz-graff’s reader, 104 of the 575 pages are devoted to chapters on “Power and Realist Theory” and “Systemic Theories of Politics and International Rela-tions.” Waltz and J. David Singer, whose leadership in the Correlates of War project is discussed below, are the most frequently cited researchers.5There is no chapter on liberal theory; indeed, no entry in the index for liberalism or the
2Daniel H. Deudney, Bounding Power: Republican Security Theory from the Polis to the Global Village (Princeton, NJ: Princeton University Press, 2007).
3Inis L. Claude, Power and International Relations (New York: Random House, 1962), 36.
4Kenneth Waltz, Theory of International Politics (Reading, MA: Addison-Wesley, 1979).
5James Rosenau is the most cited scholar, but most are to a volume he edited.
democratic peace. Only isolated references are made to Adam Smith and other liberals,first as the utopian foil for Carr, and then dismissively in the intro-duction to a chapter on“Imperialism and Economic Causes of International Conflict.” Montesquieu is given one page in a section on “The Origins of Modern Pacifism.” Norman Angell’s view that war in the industrial age is an unprofitable anachronism is briefly discussed in the same section. Dougherty and Pfaltzgraff note only in passing that there has never been a war between two democracies.6
Realism’s dominant position twenty years ago is confirmed in a recent review of the quantitative literature published from 1970 to 2000.7 From 1985 to 1989, 48 per cent of data-based articles were realist in orientation.
Except for 1970–4, when 49 per cent were, this was the highest percentage for anyfive-year period over the thirty years examined. Only 16 per cent of the articles, 1985–9, were liberal in theoretical focus, versus 8 per cent in 1970–4.
Despite realism’s success, there was no agreement among realists on key points. Doubts about the advantages of bipolarity had been expressed early on by Karl Deutsch and Singer, who argued that war should be less frequent in multipolar systems than in bipolar ones. Then, cross-cutting cleavages are more likely, and the consequences of uneven development, anticipated by Lenin, can be offset by shifting alliances.8Nor was there a consensus regarding the consequences of an equal balance of power for bilateral relations. Waltz believed balanced bipolar systems are most peaceful; but Kenneth Organski, echoing Thucydides and Hobbes, argued that peace, if not justice, is most likely when one state holds a preponderance of power.9Then, the expected winner is evident, so war is unlikely. Organski, like Waltz, was primarily interested, however, in the systemic implications of the bilateral balance between the leading state and potential challengers. Because a preponderance of power discourages conflict, a unipolar system is best, he believed. A strong leading state will substitute partially for world government.10 For Waltz, unbalanced power simply meant unchecked aggression.
The consequences for world peace of the polarity of the system—whether one, two, three, or more great powers, is best—and the effect of concentrated
6 James E. Dougherty and Robert L. Pfaltzgraff, Jr., Contending Theories of International Relations:
A Comprehensive Survey, 3rd edition (New York: Harper and Row, 1990), 5, 197, 223, and 356.
7 Thomas C. Walker and Jeffrey S. Morton,“Re-Assessing the ‘Power of Power Politics’ Thesis: Is Realism Still Dominant?” International Studies Review, 7 ( June 2005), 341–56.
8 Karl W. Deutsch and J. David Singer,“Multipolar Power Systems and International Stability,”
World Politics, 16 (April 1964), 390–406.
9 A. F. K. Organski, World Politics, 2nd edition (New York: Knopf, 1968); also, Robert Gilpin, War and Change in World Politics (Cambridge: Cambridge University Press, 1981).
10This is the essence of hegemonic-stability theory. Robert O. Keohane, After Hegemony:
Cooperation and Discord in the World Political Economy (Princeton, NJ: Princeton University Press, 1984).
versus diffuse distributions of national capabilities among the leading states could not be settled logically or by selective references to history. Debates were lively from the 1960s through the 1980s; but disagreement persisted.
Indeed, realists remain divided on fundamentals. In a book that “ranks with, and in many respects supersedes, the works of Morgenthau and Waltz,” John Mearsheimer has again argued for the greater peacefulness of bipolar systems.11He addresses fundamental questions: Why do great powers want power, and how much do they want? What is power, and by what strategies do states try to acquire it? He expresses his hope that he can provide convincing answers; but realism, he notes, is a rich tradition with a long history, and disagreements are common, so“there is no consensus among realists on the answers to any of them.”12 The situation has changed little since 1981, when Stanley Hoffmann concluded that“we are all realists now, but there are not two realists who agree either in their analysis of what is, or on what ought to be, or on how to get from here to there.”13
The behavioral revolution in the study of war and peace
Inconclusive debates among realists encouraged the“behavioral revolution”
in political science in the 1960s. J. David Singer, and others of the Correlates of War (COW) project, collected information on wars among the major powers after 1815. In keeping with the then dominant approach, they focused on the international system, emphasizing the structural conditions within which great powers act. They measured national capabilities along three dimensions:
demographic (total and urban population), industrial (energy consumption and iron or steel production), and military (total expenditures and the number of armed forces personnel). Singer et al. correlated measures of the concentration of power with the incidence of war among the major powers, but their results were disappointing.14There was no consistent relationship between the dis-tribution of capabilities and either the frequency or severity of war. Refined analyses in the 1980s confirmed that structural features of the international
11John J. Mearsheimer, The Tragedy of Great Power Politics (New York: W. W. Norton, 2001). The judgment is Samuel Huntington’s, who is quoted on the cover.
12Mearsheimer, The Tragedy of Great Power Politics, 13.
13Stanley Hoffmann, Duties Beyond Borders: On the Limits and Possibilities of Ethical International Politics (Syracuse, NY: Syracuse University Press, 1981), 659. Realism does not meet the standards for a progressive scientific research program. John Vasquez, The Power of Power Politics: From Classical Realism to Neotraditionalism, 2nd edition (Cambridge: Cambridge University Press, 1998). Contrast Fred Chernoff,“The Study of Democratic Peace and Progress in International Relations,” International Studies Review, 6 (March 2004), 49–78.
14J. David Singer, Stuart Bremer, and John Stuckey,“Capability Distribution, Uncertainty, and Major Power War, 1820–1965,” in Bruce Russett (ed.), Peace, War, and Numbers (Beverly Hills, CA:
Sage, 1972).
system—the distribution of power or polarity of the system—are not good predictors of military conflict.15
If realists failed to provide convincing answers twenty years ago regarding the causes of war, liberals did no better. Several quantitative studies sought to determine whether democracies have been more peaceful than non-democratic states. In contrast to Singer et al., these investigators used the individual country, not the international system, as the unit of analysis, examining the behavior of many countries over time. Liberals hypothesized that several mechanisms might operate to make democracies more peaceful.
There might be institutional constraints involving the political accountability of democratic leaders, limits on executive power, competitive political parties, and free media. Democracies might also be more peaceful because non-violent norms of conflict resolution characteristic of their domestic politics are exter-nalized. The results of these studies, too, were inconclusive; but most failed to find a strong association between democracy and more peaceful foreign policies.16
The ascendance of liberal theory through statistical studies of pairs of states
Recently, there has been rapid progress in research on the causes of war by analyzing pairs of states through time. In this quantitative approach, the unit of analysis is the state of relations between two countries in a given year (a “dyad-year”): two states are either involved in a military conflict or not.
Such dyadic analyses allow researchers to address the questions of greatest interest to scholars and policymakers alike: which states are prone tofight one another and when? Considering pairs of states eliminates some serious meth-odological problems. It avoids the ecological fallacy that plagued systemic studies and, unlike research with individual states, it can easily accommodate relational variables like trade, alliances, or the balance of power.
Early works by Solomon Polachek and Bueno de Mesquita were path-break-ing, but Stuart Bremer’s research in the early 1990s—using a liberal–realist model of dyadic interstate conflict—sparked most new studies.17 Bremer incorporated, in a single statistical model, elements from the two major
15Patrick James,“Structural Realism and the Causes of War,” Mershon International Studies Review, 39 (October 1995), 181–208.
16Stephen L. Quackenbush and Michael Rudy,“Evaluating the Monadic Democratic Peace,”
Conflict Management and Peace Science, 26 ( July 2009), 268–85.
17Solomon W. Polachek,“Conflict and Trade,” Journal of Conflict Resolution, 24 (March 1980), 55–78; Bruce Bueno de Mesquita, The War Trap (New Haven, CT: Yale University Press, 1981);
Stuart Bremer,“Dangerous Dyads,” Journal of Conflict Resolution, 36 ( June 1992), 309–41.
schools of international relations amenable to social scientific investigation:
the liberal, in which the character of states’ political regimes, their economic relations, and involvement in international organizations are thought to influence international behavior; and the realist, with its emphasis on the absolute and relative power of nations, alliances, and geographic considerations.
The analyses below confirm that both theoretical traditions help explain whofights whom. Two democracies are unlikely to be involved in a militar-ized dispute, and economic interdependence dramatically improves the pro-spects for peace; but the bilateral balance of power and the ability of states to project military force at a distance also matter. Surprisingly, however, the liberal factors are at least as influential as power; they are much more important than alliances. Most importantly, the liberal variables are amenable to manipulation. All states can become democratic and integrated into the international economy. All cannot enjoy favorable balances of power or become allies. It is this potential for constructive social engineering that gives liberalism its special appeal.
The ascendance of liberalism is easily documented. Consider again Walker and Morton’s survey of the literature. From 1970 through 1974, only 8 per cent of the articles were evaluations of liberal theory; by 1995–2000, 39 per cent were—nearly twice as many as were categorized as realist in orientation (22 per cent). Presidential addresses to the American Political Science Association in 2002, and the International Studies Association in 2008, further emphasize the importance of liberalism. Walker and Morton also document the growing application of the scientific method. In the earliest period, 1970–4, only thirty-seven data-based articles were published in the journals they surveyed.
That grew to sixty-four in 1985–9, and 155 in 1995–2000. About half of all articles in political science journals now include statistical analyses.18
The acceptance of the dyadic liberal–realist model in particular is clearly indicated by its widespread use and the frequency with which this research is cited. Thomson Reuters identified the scholars most frequently cited, 1996–2006, on the subject of armed interstate conflict.19 Nine of the first ten have used the LRM extensively. Thus, political science research has moved from realist to liberal theory, and support for the liberals’ political
18Walker and Morton, “Re-Assessing the ‘Power of Power Politics’ Thesis.” Robert Jervis,
“Theories of War in an Era of Leading Power Peace,” American Political Science Review, 96 (March 2002), 1–14; Nils Petter Gleditsch, “The Liberal Moment Fifteen Years On,” International Studies Quarterly, 52 (December 2008), 691–712. Gary King, Michael Tomz, and Jason Wittenberg,
“Making the Most of Statistical Analyses: Improving Interpretation and Presentation,” American Journal of Political Science, 44 (April 2000), 347–61.
19Thomson Reuters,“Special Topics: Armed Conflict” (2006) < http://esi-topics.com/armed-conflict/authors/b1a.html>.
and economic prescriptions comes not from humanistic studies of the inter-national system, but statistical analyses of pairs of states through time.
Analyzing the onset of fatal militarized disputes, 1885–2001, using the liberal–realist model
In this section, I analyze the behavior of thousands of pairs of states from 1885 to 2001. The liberal–realist model is designed to explain (or predict) the state of relations (armed conflict or peace) between two countries in a year. Ana-lyses with the LRM provide estimates of the probability of a militarized dispute between the United States and the Soviet Union during the Cold War or the US and Russia today, for example, or any other pair of states in particular years.
The LRM can also be used to predict the likelihood of interstate violence for hypothetical cases of theoretical interest. The pooled time series of over
The LRM can also be used to predict the likelihood of interstate violence for hypothetical cases of theoretical interest. The pooled time series of over