Online Journal of Peace and Conflict Resolution 2.1

PEACE THROUGH COERCION:

THE EFFECT OF FOREIGN MILITARY INTERVENTION

ON DEMOCRATIZATION AND WAR

By David Reilly

Introduction

The relationship between democratization and war has received great scrutiny in recent years as the set of nation-states in the world becomes increasingly democratic. The democratization of the globe has, at least in part, been attributed to external actors in the form of promotion, diffusion, coercion, or encouragement (see, for example, Huntington 1991, Modelski and Perry 1991, Starr 1991). Whitehead (1996) finds that there are only three instances of independent invention of democracy: Sweden, Switzerland, and the United Kingdom. Other cases are broadly identified as occurring due to international change resulting from decolonization and the aftermath of major wars. Given the extensive democratization literature that connects democratic development with the protection of human rights, economic development, social stability, and possibly even systemic peace, some have suggested that there is justification for the active promotion of democracy world-wide (Franck 1992, Lynn-Jones 1998). Many leaders and policymakers have joined this movement in support of democratizing the globe (Boutros-Ghali 1996, Christopher 1995, Clinton 1992, Talbott 1996).

There are a number of untested and uncertain outcomes that could result from such a crusade. The effect of military intervention on the likelihood of a democratizing state engaging in war is one such issue. Although recent research has begun to test the relevance of democratization to participation in conflict, this monadic-level behavior involves a number of complexities that have yet to be addressed. Given that regime transition is often imposed or advanced through military intervention, external force should be included in the calculation.

This research attempts to gauge the effect of military intervention on the probability of a transitioning state engaging in interstate war. By including the variable of military intervention in the analysis, it is anticipated that a clearer indication of the dynamics involving regime transformation and conflict will result. This paper is structured in the following manner: Section I addresses the current literature involving democracy, democratization, and peace. It includes a summary of the findings on this topic, as well as a discussion of the apparent contradictions in these studies. In addition, relevant findings from the analysis of revolution and war are identified. Section II provides the justification for including military intervention in the analysis of democratization and war. Included in this section is an account of the limitations of research to date on the connection between democratization and external forces. Section III involves an attempt to analyze the role of intervention in the democratization-war link. The data used and research design employed are described. The final section of the paper takes stock of the findings presented within, offers explanations for them, and identifies some of the limitations of this analysis. The results indicate that autocratization decreases the probability of war, regardless of whether intervention is involved. Intervention, in and of itself, increases the probability of war, although when transition results from military intervention war becomes less likely. Finally, and most importantly, democratization increases the probability of war when the state acts independently. However, when military intervention plays a part in the process of democratic transition, the likelihood of war decreases. Although the analysis indicates that democratization spurred by military intervention is less likely to result in war than democratic transition absent external action, this finding is tentative and requires additional consideration.

I. Democracy, Democratization, and Peace

One of the strongest empirical regularities in international relations research is the "democratic peace"; the idea that democracies do not engage in war with other democracies. Although there has been debate on the causal mechanisms of this relationship, as well as on its limitations, a consensus is emerging on the absence of war in democratic dyads. Beginning with the theories of Immanuel Kant, there has been an expectation that democracies or republics will be more peaceful than other states in their relations with other democracies because of institutional and normative characteristics that are unique to these regimes. In more recent studies, the explanations posited include structural accounts of executive incentives (see Mintz and Geva 1993) and constraints (Morgan and Campbell 1991, Morgan and Schwebach 1992) and the normative view (see Doyle 1986, Maoz and Russett 1993, Dixon 1994). Democratic institutions and their related incentives and constraints are asserted as the causal mechanisms within the structural explanation. The normative view cites shared democratic norms, values, and practices as the source of democratic peace. Most of these studies have argued that democracy is a sufficient condition for peace: there is a threshold of democracy which, when present for both members of a dyad, precludes war. Critics have argued that neither structural nor normative accounts have provided a conclusive, satisfactory explanation. In spite of this potential shortcoming, it has been commented that the relationship between pairs of democratic countries and peace is "the closest thing we have to a law in international politics" (Levy 1988, 653).

One of the most interesting facets of the research on democratic peace is the apparently contradictory evidence surrounding the dyadic and monadic levels. More specifically, the supporting research on the dyadic aspect of the democratic peace does not appear to extend to the monadic level. Studies tend to indicate that while democracies do not fight other democracies, they are as war prone as autocracies (Chan 1984; Dixon 1993, 1994; Morgan and Campbell 1991; Morgan and Schwebach 1992; Russett and Maoz 1993; Small and Singer 1982). Some research has disputed this issue, however (Hewitt and Wilkenfeld 1996; Ray 1995a, 1995b; Rummel 1983, 1995). When the unit of analysis has been the monadic level of individual states, the results have been more contentious. Furthermore, explanations for this apparent contradiction are limited and unsupported.

Recent studies have examined the relationship between democratization and war as a means of clarifying the uncertainties of the aforementioned debate. By examining the tendencies of democratizing states, it is expected that the monadic aspect of the democratic peace will be revealed. In other words, an examination of states in the process of becoming democratic should offer insight into how the behavior of these states changes in conjunction with political and social modifications.

The relationship between regime change and war, however, is a complicated one. Independent of direction of change (e.g., democratization, autocratization), there is a debate surrounding the dynamics of revolution and war that comes to contradictory conclusions. Realist theories that focus on the balance of power seem indeterminate with regard to the relationship between transition and war because the effects of changing power relations are ambiguous. Competing hypotheses are equally plausible within the realist framework: 1) the increased power of an emerging revolutionary regime could lead that state to aggressive action, whereas decreased or ambiguous power structures have the potential to invite aggressive neighbors to attack; or 2) increased state power could provide security sufficient to alleviate the need to expand and the declining power of a revolutionary state reduces the threat for adjoining states (See Walt 1996, 4; Gilpin 1989; Levy 1987).

There are studies that find a positive relationship between revolution and war. Steven Walt (1996) finds that revolutions augment security competition between states and, accordingly, increase the likelihood of war as perceptions of the balance of threats are altered. As revolution unfolds, the emerging regime and neighboring states must respond to the shifting environment. The result is often conflict as new threats and opportunities for the improvement of a state's security status are perceived. As Walt suggests, revolutions result in new regimes that are more vulnerable and are often in conflict with the allies of the previous regime, and the fear of spreading revolution leads neighboring states to take an aggressive stance against the new regime.

There are studies that also discover the opposite result. Patrick Conge suggests that just as revolution can increase the likelihood of war under particular circumstances, in other instances it is likely to generate peace. Whether or not a revolutionary regime has the "organizational capacity ... to extract resources and convert them into military strength and ... the power of ideas to transcend national boundaries and undermine the ability of opposing regimes to accommodate one another" are critical aspects for understanding the effect of revolution on the likelihood of war (Conge 1996, 6). Depending upon the resources available to a transitioning regime, as well as the organizational structure that would make use of these resources, a regime may recognize its own vulnerability and attempt to preempt conflict with accommodations to neighbors. By presenting a more benign face to potential adversaries, a transitioning regime may reduce the threat of conflict.

Like the research on revolution and war, expectations about the effects of democratization on conflict are divided. Hypotheses about the behavior of democratizing states can be broadly grouped in two categories. The first would expect that the behavior of transitioning states would be more prone to conflict and is based on different, but interrelated, explanations. The internal account expects the more conflicting behavior of democratizing states to stem from domestic instability. The tenuous position of a fledgling regime may lead to the externalization of aggression as a means of generating nationalistic support or "rally 'round the flag" responses from the public or as a diversionary tactic from critical opposition. The external explanation expects that surrounding states will be more aggressive when they observe regime disorder in a neighboring state. The process of democratization involves the breakdown of the regime (Brinton 1965) as well as the subsequent polarization of internal factions (Rustow 1970), both features that increase a state's vulnerability. As groups compete for power and control of the regime internally, the state is more susceptible to attack and, therefore, more likely to be involved in conflict.

There is empirical support for this hypothesis. Mansfield and Snyder (1995, 1996) find that regime change, regardless of type, increases the probability of interstate conflict at the national level.⁽¹⁾ The transition phase of states, whether democratic or autocratic, is likely to be characterized by internal fragmentation manifested in interstate disputes. Their analysis offers the suggestion that emerging democracies may be more prone to war involvement, particularly if the process of democratization is rapid and destabilizing. Ward and Gleditsch (1998) find evidence to support this last aspect; the direction and magnitude of regime change are strongly related to the probability of war. However, only instances of "[r]ocky changes toward democracy in general (or autocracy for that matter) appear to heighten the likelihood of war involvement" (Ward and Gleditsch 1998, 58-9). Their findings indicate that the process of transition--whether smooth or characterized by reversals--is important for understanding the resulting likelihood of war.

The second hypothesis, which stems from the democratic peace literature, expects the opposite: as states become more democratic and liberal they will become more peaceful. With each increase in democracy, a state is further constrained by cross pressures and the associated political culture of libertarianism. Rudolph Rummel (1983, 1985, 1994) has addressed this issue, and suggests that democracies do not engage in war with other democracies, should exhibit the lowest levels of foreign violence among regime types, and should engage in lower levels of government-related and directed domestic violence.

Some recent studies addressing democratization and war have found supporting evidence for this position. Ward and Gleditsch (1998) discovered that when democratization is characterized by a relatively smooth process, there is a reduced probability that the state will be involved in a war. This study suggests that the direction, magnitude, and variance of change are important in determining the probability of a state engaging in interstate war. Research by Rousseau (1996) finds that democratizing states are not more likely than other states to engage in violence. Furthermore, he determines that autocratization decreases the probability of conflict. The transition process, according to his research, is not an inherently dangerous proposition.

II. Intervention, Democratization, and Peace

One possibility that has only been explored at a cursory level in the research involving democratization and war is the relevance of external intervention. The process of democratization, although largely attributable to internal dynamics, has recently been examined to determine the extent to which external actors and influences have influenced its spread. One study reaches the surprising conclusion that "approaching two-thirds of the democracies existing in 1990 owed their origins, at least in part, to deliberate acts of imposition or intervention from without" (Whitehead 1996, 9). Whether democratic institutions are imposed in the process of decolonization or freedoms are established in the aftermath of war, the population of democratic states cannot be understood without reference to external intervention.

The processes of democratic transition that have occurred in various states since World War II can also be traced to external sources. Many countries that experimented with democratic structures and practices during the Cold War did so under the promise or peril of carrots and sticks. The practice of military intervention has become increasingly justified as a means to spread democracy (Diamond 1992; Lynn-Jones 1988). In fact, some have gone so far as to say that there is an emerging right to democratic governance (Franck 1992). Regardless of the moral or ethical implications of such a change in international norms, it has been observed that the "nonintervention principle" has begun to erode in the post-Cold War period (Kegley, Raymond, and Hermann 1998). As a result, others have begun to consider the possibility of building a global peace through the promotion of democracy (Allison and Beschel 1992; Russett 1993). Furthermore, research indicates that democracies often rely on military intervention as a means of achieving foreign policy objectives such as reforming the governments of target countries (Kegley and Hermann 1996). Whether or not these goals are achievable depends upon the relationship involving democratization, war, and military intervention. Although the first two have been considered with contentious results, the importance of military intervention to date has not factored into the analysis.

Military intervention has been shown to be a moderately effective tool for the promotion of democracy (Kegley and Hermann, 1998). The population of established democratic states has been shown to be attributable in part to external influences. Independent of the long-term effects of such efforts, it also does appear that military intervention in the interest of liberalization, democratization, or "good governance" is possible (Meernik 1996).

It is not clear, however, that a democratic peace is possible through coercion. Given the uncertain results regarding democratizing states, it is important to consider the implications of military intervention on the prospects for peace. In particular, it is necessary to consider the possibility that externally imposed democratic structures could serve to destabilize social institutions, leading to conflict and violence. One hypothesis relating external influence to democratization and peace would be that states which democratize under coercion of military intervention are more prone to conflict. Because such an imposition would lead to rapid modifications to the political and social structures of a state, the regime would be less capable of defending challenges to power. Internally, in keeping with the modernization school of democratic development (see, for example, Lipset 1959) the lack of preconditions or prerequisites to democracy would delegitimize the regime and render it incapable of supporting democratic institutions. The result would be violent opposition that could manifest itself in the form of external aggression. Alternatively, it is conceivable that an intervening state could provide support to a fledgling regime and serve to deflect internal and external challenges to control. The counter-hypothesis, then, would be that military intervention resulting in democratization would reduce the probability of interstate war involving the democratizing state. Because the intervening state would have the capacity and resources to support political change, it could prop up a democratic regime and shield it from internal factions and external threats until a time when democratic structures and norms have sufficiently evolved.

This study seeks to gain more information about the relationship between democratization and war. By including the variable of intervention, which is frequently associated with democratization, the characteristics of a democratizing regime can be clarified. The analysis begins with a replication of the research by Mansfield and Snyder (1995). Using a research design which is conceptually similar but employs different data and covers a different time period, I seek to verify their findings. Then, the relevance of intervention to these results is evaluated. This issue is formulated in terms of probabilities: Does the promotion of democracy through military intervention increase the likelihood of interstate war for the target state? Through logistic regression this question can be addressed.

III. Analyzing the Effect of Intervention on Democratization and War

The political characteristics of states are measured using the 5.1996 version of the Polity III database (Jaggers and Gurr 1996)⁽²⁾. From the measures included in the data, a 21-point scale of political values ranging from high autocracy (-10) to high democracy (+10) can be created from the difference between the democracy {D} and autocracy scores of each state {D-A}. These indices are compiled from a variety of authority dimensions: competitiveness of political participation, regulation of political participation, competitiveness of executive recruitment, openness of executive recruitment, and constraints on the chief executive. From this scale, democratic and autocratic transitions--defined as instances where a state increases or decreases its score at least three points within a five year period--are identified. A three point change signifies a non-trivial modification to the structures and practices that are represented by the authority dimensions described above. The impact of these kinds of modifications, in terms of changing behavior, is what this research attempts to measure. Accordingly, a transition period is coded for each instance of transition (separately for democratic and autocratic events) as well as the ten year period following a transition, unless that period is interrupted by a reversal (transition in the opposite direction of the previous change) or further transition in the same direction. The additional years are intended to capture the subsequent period to a transition when changing norms and structures are institutionalized and stabilized.

The Correlates of War International War (COW-IW) data set is used for the analysis of interstate conflict (Singer and Small 1994). COW-IW includes wars that involve at least one member state in the international system as well as extrasystemic polities, which typically involve colonial or imperial wars. Any act of aggression resulting in more than 1,000 battle deaths is classified as a war. This dataset is more relevant for the analysis of interstate war than other available options such as the COW Militarized Interstate Dispute data. This data includes disputes and conflicts that fail to meet the criteria of 1,000 battle deaths, many of which may not be relevant to the constraining norms and/or structures that are hypothesized to result in the democratic peace (Chan 1997). Because of its conservative coding of instances of war, as well as for its inclusion of extrasystemic wars, COW-IW is utilized in this research.

For the analysis of the targets of intervention, the Overt Military Intervention data are employed (Tillema 1994). This dataset includes 690 instances between 1945-1991 of "combatant or combat-ready military operations conducted upon foreign territory by units of a state's regular military forces" (Tillema 1997). Included in the instances of intervention are the deployment of troops in actions such as alert patrol, offensive maneuver, riot quelling, armed occupation of territory, and battle. Less intense operations such as commando or other small unit raids, aerial bombing, ground-based artillery, and naval gunnery are also included. For the purposes of this research, an intervention period is coded for each nation-year of military intervention as well as the four year period following an intervention. The additional years are intended to capture the subsequent lagged effect that an intervention has on the behavior of the target state.

The probability of a transitioning state engaging in interstate war can be represented by the following logistic equation:

P_i = E ( W_i = 1|A_i + A^T_i ),

where W_i represents participation in an interstate war, and A^T_i denotes the presence of a transition period (either autocratic or democratic). The probability of war for a given observation is expressed as follows:

P_i (War=1) = 1 / 1 + exp^-z,

where Z denotes the linear combination:

Z = B₀ + B₁T_i + B₂T_i + B₃I_i

T_i in this instance represents the presence or absence of transition (1=transition, 0=absence of transition). Because the transition periods of democratization and autocratization are coded separately and are mutually exclusive, T_i can represent either democratic or autocratic transition for a particular logistic regression. I_i denotes the intervention period variable (also coded as 1=present, 0=absent).

The results of this analysis are included in Table 1. These suggest that the effect of the occurrence of military intervention on the likelihood of war is by far the most significant and positive. The odds of war, given military intervention, increase by a factor of 2.72 as shown in the EXP (B) column. The event of democratic transition is also positively related -- a result which replicates the Mansfield and Snyder (1995) study -- although this relationship is quite weak. Surprisingly, autocratization is negatively related to the occurrence of war. In other words, the odds of war during an autocratic transition period decrease by a factor of .74.

Table 1: Empirical Estimates of Bivariate Logistic Models of the Probability of War

Independent Variables	Constant	Estimated Coefficient	Standard Error	EXP (B)	Valid Cases
Overt Military Intervention	-4.2	1.0	.18	2.7	5675
Democratization Period	-3.84	.08	.26	1.08	5675
Autocratization Period	-3.79	-.30	.31	.74	5675

In order to evaluate more clearly the effect of intervention on the process of transition, a selection term is added so that the cases under investigation are only those nation-years in which an intervention period exists (see Table 2). In other words, all cases in this logistic regression are states that are experiencing or recently (within four years) experienced military intervention. When the joint effect of transition and intervention is considered, the results differ: as Table 2 reveals, although autocratization remains negatively related to the occurrence of war, democratization is now also negatively associated (B= -.35). This result is important in that it indicates that the behavior of democratizing states when coerced through military intervention varies from the actions of democratizing states when left to their own devices. The odds of war involvement during a democratic transition period that is influenced by intervention decrease by a factor of .71.

Table 2: Empirical Estimates of Logistic Models of the Probability of War - Examining Intervention Periods

Independent Variables	Estimated Coefficient	Standard Error	Signif.	EXP (B)
Democratization Period	-.35	.39	.38	.71
Autocratization Period	-.21	.37	.58	.81