DEMOCRACY AND CONFLICT BETWEEN STATES

The realist school of thought in international relations, which greatly influenced both scholarship and policy-making during the cold war, maintains that state behavior is primarily driven by the balance of power among rivals in the international system (Morgenthau 1948; Waltz 1979; Mearsheimer 2001).

This realist position came under increasing scrutiny beginning in the 1980s. Doyle (1983, 1986), for example, compiled a list of liberal societies and interstate wars during the nineteenth and twentieth centuries and found that no two democracies had engaged in a full-scale war. He concluded that “liberal states have created a separate peace, as Kant argued they would, and have also discovered liberal reasons for aggression, as he feared they might” (Doyle 1986: 1151). Rummel (1983, 1985) came to a similar conclusion after subjecting the proposition to somewhat more systematic testing. Path-breaking work by Doyle and Rummel triggered an avalanche of empirical and theoretical investigations into what is now referred to as “the democratic peace.” According to Levy (1988), the democratic peace is the closest thing to an empirical law found in the study of international relations.

Some of the earliest research examined the characteristics of democratic governments and societies that shape the state's general foreign policy orientation and behavior.

Thus, scholars have attempted to explain the democratic peace using a series of related arguments, which identify causal mechanisms operating at different levels of analysis. Ithas been difficult to distinguish the relative explanatory power of these competing arguments because data sets constructed to test arguments at one level of analysis are often not appropriate for testing arguments at other levels. However, progress in both the­oretical development and empirical analysis has reduced this problem in recent years (e.g. Bennett and Stam 2000,2004; Rousseau 2005). The following review of democratic peace theory and research is organized according to the main causal mechanisms identified in the academic literature.

Democratic norms and conflict resolution

Many explanations of the democratic peace emphasize the socialization of political leaders within their domestic political environments (Dixon 1993, 1994; Maoz and Russett 1993; Russett 1993; Huth and Allee 2002). This argument has two parts. First, democratic political elites have risen to positions of leadership within a political system that emphasizes compromise and non­violence. Conflicts of interest in democracies are usually resolved through negotiation and log-rolling. Losing a political battle does not result in the loss of political rights or exclusion from future political competition.

Second, the argument assumes that domestic political norms are externalized by decision makers when they become embroiled in international disputes. Presidents and prime ministers approach conflicts of interest in the international environment in much the same way they approach conflicts in the domestic environment, and with conflict-resolution skills honed by their domestic political experiences. Compared to their counterparts in authoritarian regimes, democratic leaders are more likely to seek negotiation, mediation, or arbitration (Dixon 1994; Raymond 1994). Their approach to international conflict resolution reduces the likelihood that an international dispute will escalate into a militarized crisis and war.

The strong version of the norms argument holds that democratic leaders externalize peaceful practices of conflict resolution in their interactions with all types of regimes. In contrast to this monadic claim, those who emphasize the dyadic nature of the democratic peace argue that although all decision makers are inclined to externalize domestic practices of dispute resolution when dealing with interstate conflicts, this externalization is conditional for democratic decision makers. Democratic leaders exter­nalize their domestic norms only if they expect similar behavior from their foreign counterparts. Because democratic decision makers expect that choices by other demo­cratic leaders are also shaped by norms of peaceful conflict resolution, there is little risk in an attempting to resolve their conflict in accordance with these shared norms.

A related argument highlights the importance of identity formation. Some have suggested that peace between democracies is a function of a common social identity (Risse-Kappen 1995; Hopf 1998, 2002; Kahl 1998/99). Social identities are bundles of shared values, beliefs, attitudes, norms, and roles that are used to draw boundaries between in-group and out-groups. Members of one's own group are viewed as less threatening than members of other groups. If democratic polities use democratic values and norms to define the in-group, the actions and capabilities of other democracies are then viewed as less threatening. Their shared identity will reduce the likelihood that either party will resort to violence to resolve a political dispute. Although realists discount the importance of ideational factors in world politics, liberals and constructivists have long maintained that a shared sense of identity partly accounts for lower levels of international conflict. While liberals tend to focus on a shared liberal identity, constructivists believe that many types of shared identity may reduce interstate conflict. Risse-Kappen (1995), for example, argues that a shared sense of identity among democratic states, and not simply their concern with the balance of power, explains decision making within the North Atlantic Treaty Organization. Laboratory experiments have also demonstrated that shared cultural beliefs and experiences can decrease intersubjective threat perceptions (Mintz and Geva 1993; Rousseau 2006).

Explanations for both the monadic and dyadic versions of the democratic peace imply that as the number of democracies in the international system increase, the number of interstate wars will fall.

Democratic institutions and restraint

Another class of explanations for the demo­cratic peace highlights the institutions of democratic governance, broadly defined, and the domestic political costs of using force (e.g. Morgan and Campbell 1991; Morgan and Schwebach 1992). Decisions to use military force are choices made by leaders based largely on calculations of political costs and benefits. Foreign policy decisions can have costly domestic political repercussions. The expenditure of resources and loss of human life often mobilize opposition groups or fracture ruling coalitions (Mueller 1973; Cotton 1987; Bueno de Mesquita and Siverson 1995).

The nature of political institutions can have a bearing on the credibility of signals sent during an international crisis. Fearon (1994, 1995) argues that in a world of complete information, decision makers can determine each side's expected value for war—the ends sought and the likelihood of achieving them while suffering the costs of armed combat. In such a world, war would be rare; if each side's aims were known, along with the price each was willing to pay in blood and treasure, it would always be possible to strike a bargain acceptable to all without actually having to suffer the costs of war. Unfortunately, we live in a world of incomplete information. A government's willingness to use force is usually private information, and leaders may have an incentive to exaggerate or otherwise misrepresent their resolve in order to strike a better bargain. In this context of incomplete information, signals of resolve are more credible when leaders are likely to pay higher domestic audience costs for bluffing, and democracies are political systems in which audience costs are highest. Moreover, the openness of political debate in democracies provides information to foreign opponents. When the political opposition in a democracy lines up behind the executive during an international confrontation, this is a powerful signal that because the party in power will pay high political costs for backing down, the executive is probably not bluffing (Schultz 1998, 1999, 2001).

Another institutional argument derives from a game theoretic model of political survival developed by Bueno de Mesquita et al. (1999, 2003). Public policies, both domestic and foreign, yield a mix of public and private goods. Public goods, of course, are available to the entire society, whereas private goods can be allocated as leaders see fit. Political systems vary in terms of the proportion of society involved in the selection of political leaders (the selectorate), and the proportion of the selectorate whose support is required to maintain one's position of power or, in the case of a challenger, to unseat the current leader (the winning coalition). Democratic states have large selectorates and large winning coalitions. Autocratic states may have large selectorates, too, when elections have high voter turnouts but are nevertheless rigged, but they always have small winning coalitions. Political survival in a democracy therefore requires that goods, public and private, be distributed among a larger winning coalition than is the case in an autocracy. Public goods serve that purpose well in a democracy because they go to all, while the value of private goods diminishes due to the larger number of recipients. In an autocracy, however, private goods are relatively more important to the leader's political survival because they are distributed among a smaller constituency.

Bueno de Mesquita and his colleagues argue that successful public policies generate the public goods that democratic leaders need in order to stay in office. Autocratic leaders prefer that their policies succeed, but the consequences of policy failure for political survival are not dire as long as the leader has access to resources that can be distributed as private goods to a small winning coalition. What are the implications for the democratic peace? With higher political costs of policy failure, democratic leaders avoid international contests unless they are confident of victory. And once they become involved in a crisis or war, democrats try harder to win. Contests with other democratic states of similar capabilities are to be avoided for this reason, but democratic states are not so disinclined to avoid confronting autocratic states, whose leaders have less to lose politically by backing down. The model does suggest, however, that stronger democracies also face fewer disincentives when confronting weaker democracies; the stronger state is likely to succeed no mat­ter how hard the weaker one tries, and policy success is what counts for political survival.

Democratic political institutions often influence foreign policy decision making in particular ways regardless of the regime type of the opponent. But given a set of institutional constraints, the leaders of democratic states may well behave differently depending on whether or not their opponents are believed to be similarly constrained. For example, Bueno de Mesquita and Lalman (1992) propose a three-part institutional explanation for the dyadic democratic peace. First, the international system is assumed to consist of hawkish states (leaders are uncompromising and predisposed to use force to resolve disputes) and dovish states (leaders are prone to compromise and use strategies of reciprocation). At the same time, there is some uncertainty surrounding which strategy will be adopted by any particular state. Second, domestic institutional structures reduce (but do not eliminate) this uncertainty by signaling a state's most likely strategy. Due to the poten­tial domestic costs of using force, decision makers believe that democracies are more likely than nondemocracies to adopt dovish strategies. Third, when a democracy confronts another democracy, each expects a negotiated outcome and the exercise of restraint when contemplating the use of force. But when an autocracy confronts a democracy, a hawkish leader expects to encounter a dovish one and is likely to exploit the situation. In such a situation, the dove feels compelled to adopt the aggressive strategy of the hawk and may initiate conflict in order to preempt an expected attack. The logic of the argument is dyadic: democratic states pursue strategies involving compromise and nonviolence only when dealing with other democratic states.

Another institutional explanation for the dyadic character of the democratic peace focuses on the difficulty of mobilizing popular support for the use of force. According to Maoz and Russett (1993), the inclusiveness of democratic regimes hinders their ability to rapidly mobilize societal groups in support of military action. Authoritarian regimes, with constituencies spanning a much narrower range of the political spectrum, can more quickly reach the necessary consensus on the use of force. When a dispute emerges between two democratic states, the slow process of mobilization in both states creates opportunities for the resolution of the conflict through noncoercive means. However, when a conflict arises between a democratic state and an authoritarian state, rapid mobilization by the latter forces democratic leaders to find ways to work around normal political processes. That is, the emergency situation encourages the democratic state to adopt the tactics of its nondemocratic opponent.

Critics of the democratic peace thesis

In addition to the studies discussed above, there is considerable additional social sci­entific research that supports one or more of the propositions contained in democratic peace theory, especially as concerns dyadic peace (see, e.g., Maoz and Abdolali 1989; Bremer 1993; Rousseau et al. 1996; Rasler and Thompson 2001; Russett and Oneal 2001; Dixon and Senese 2002; Peceny et al. 2002). But despite the impressive body of evidence, the academic literature includes many studies that aim to refute the democratic peace proposition in part or in whole. Some of the most noteworthy research focuses on the purported risks to peace presented by democratizing states.

While acknowledging that mature democracies rarely fight each other, Mansfield and Snyder (1995,2002a, 2002b) have argued that the process of democratic reform may actually increase the probability of war. Their empirical findings, based on both qualitative and quantitative analyses, have been read not only as partly refuting democratic peace theory, but also as calling into question the wisdom of efforts to promote democracy in other countries, a cornerstone of Western foreign policy following the end of the Cold War. If the condition of being democratic decreases the probability of violent conflict, how could the process of becoming more democratic have the opposite effect? Mansfield and Snyder propose that transitional regimes experiencing high level of political mobilization together with weak institution controls are often tempted to incite external conflict. The intense political competition ushered in by the disintegration of the previous authoritarian government leads elites and would-be leaders to identify issues that can be used to build broad popular coalitions.

The issues that tend to be exploited by elites, according to Mansfield and Snyder, are those that can become the basis for a “belligerent nationalist coalition.” The old elite, including the military establishment, often seek to define themselves as the guardians of the nationalist cause, reminding the populace of the dangers they collectively face. Newly emerging interest groups are also inclined to seize on such issues as group leaders feel compelled to assert their nation­alist credentials as a means of unifying the fragmented interests bubbling to the surface in an unstable political environment. While the masses may not be particularly war-prone at the start of this process, sustained appeals to nationalism from across the political spectrum can quickly create a belligerent popular mood. The intensification of this mood can trigger “blow-back,” a situation in which the leadership feels compelled to behave aggressively having become trapped by their own demagoguery. In their initial research on wars during the nineteenth and twentieth centuries, Mansfield and Snyder (1995, 8) found “that democratizing states—those that have recently undergone regime change in a democratic direction—are much more war-prone than states that have undergone no regime change, and are somewhat more war-prone than those that have undergone a change in an autocratic direction.”

Analyses by Mansfield and Snyder have been challenged on a number of methodological grounds (Enterline 1996, 1998; Weede 1996; Wolf 1996; Thompson and Tucker 1997), and other empirical investigations into the war-inducing effects of democratization report evidence at odds with theirs (Gleditsch and Ward 2000; Russett and Oneal 2001; Oneal, Russett, and Berbaum 2003; Bennett and Stam 2004; Rousseau 2005). Mansfield and Snyder have responded to these critiques in various ways, but their most recent work has included a closer examination of the position of transitioning states along the autocracy­democracy spectrum and its implication for becoming involved in militarized disputes or wars. They have generally concluded that the probability of conflict is increased when autocratic states make incomplete democratic transitions, but not when they make complete transitions. Nor do partially democratic regimes appear dangerous when they are undergoing transitions to full democracy (Mansfield and Snyder 2005).

In challenging some of the core assumptions and arguments of realist theory, it is not surprising that the democratic peace research program has itself come under attack from various angles. Gowa (1999), for instance, argues that the democratic peace is spurious, that it is a function not of democratic governance and conflict resolution but security considerations within the Western alliance in its opposition to the Soviet bloc after World War II. Democratic peace researchers have responded to this and other criticisms on realist grounds by controlling for a number of factors that feature in realist explanations of war and peace, including geography, alliance, the balance of military capabilities, and nuclear armament. They have also controlled for other liberal factors, like wealth, trade, and participation in international organizations. Efforts to model the liberal determinants of peace alongside realist ones have demonstrated that evidence for the democratic peace is quite robust (e.g. Russett and Oneal 2001; Kinsella and Russett 2002; Kim and Rousseau 2005; Rousseau 2005).

Other critics point to the difficulties of measuring democracy, an exceedingly complex social scientific construct. These difficulties led early exploratory research to opt for dichotomous measures of regime type (e.g. Doyle 1986), but the majority of later studies have employed the democracy and autocracy scales developed by the Polity Project (see Marshall and Jaggers 2002). Although indices constructed from the Polity scales have a number of important strengths, they are rough measures and are often insensitive to small but important changes in domestic power configurations. Democratic peace researchers have typically responded to such objections by modifying their measures and retesting their propositions. Rousseau (2005), for example, finds strong support for both the monadic and dyadic democratic peace propositions using new measures and data for institutional constraint.

A more fundamental objection to the measurement of democracy in the democratic peace literature is leveled by Oren (1995, 2003). Polities can be measured along numerous dimensions, and social scientific practice at any given moment reflects the identity evaluation and threat perceptions prevailing during that historical period. Oren argues that in the case of social science research in the United States, the features of democracy considered most important are those that the American political system shares with the political systems of friendly states, while those features it shares with its enemies tend to be downplayed. Oren's claim, therefore, is that the democratic peace is an artifact of a built-in bias of the social scientific community. This is a provocative critique to be sure, though one that has so far not prompted much reaction among democratic peace researchers.

The theoretical underpinnings of the democratic peace project have also been scrutinized. There is a lack of agreement within the research community regarding the exact causal mechanisms responsible for the empirical regularities that are routinely observed. While we know that democracies do not fight other democracies, we are not sure which of the many causal mechanisms is behind the pattern, or which have the most influence in different contexts. Others have examined the logic of both the normative and institutional explanations discussed above, arguing that those explanations imply even more pacific behavior on the part of demo­cratic states than what democratic peace researchers are able to show (Rosato 2003). To some extent, these critiques replay earlier objections to the challenges posed to realist theory. What they have not done, however, is undermine the democratic peace proposition, the research program supporting it, or the implication that the spread of democratic forms of governance enhance the prospects for the resolution of conflicts between states (Ray 2003; Chernoff 2004; Kinsella 2005).