Institutional Constraints and Transparency.

There are ways in which it is appropriate to consider all of these types of autocratic regimes as fundamentally similar. We accept the liberal argument that these regimes are all less constrained than democracies in the sense that they allow a small elite tremendous leverage in determining national policy. In personalist regimes, one man possesses overwhelming authority to make decisions. Military regimes face somewhat more constraints because decisions need to be approved by a portion of the officer corps. Single-party regimes also may be somewhat constrained because decisions must have at least the tacit approval of party members.

As Bueno de Mesquita et al. (1999) point out, however, the size of the winning coalition in personalist, military, and single-party regimes may be equally small. Leaders in each of these autocratic regimes can, therefore, stay in office by providing private goods to key supporters even if they suffer defeat in war. The size of the “selectorate,” however, differs substantially in these regimes, with single-party regimes possessing large selectorates and personalist dictators and military regimes small ones. Regimes that combine a large selectorate and a small winning coalition may be least constrained of all from going to war because it is more risky for members of the winning coalition to defect in such regimes, since defectors can be easily replaced (Bueno de Mesquita et al. 1999, 797). For these reasons, we believe that decisional constraints are unlikely to be the source of a separate peace among certain types of authoritarian regimes. Indeed, the lack of such constraints could lead to conflict between similar authoritarian regimes.

Similarly, we see few systematic differences in the openness of specific types of authoritarian regimes and find them all to be more opaque than democracies. Because of the need to provide information to party members and to remain consistent with their professed ideological vision, single-party regimes may be somewhat more transparent than military or personalist regimes. None of these regimes, however, possesses the combination of a diversity of information sources, freedom of the press, and political contestation that makes democracies systematically more transparent than all forms of authoritarianism. The logic of the transparency approach, therefore, does not help us much in distinguishing between specific types of authoritarian regimes and is unlikely to account for any separate peace between specific types of autocracies.

War-Fighting Capabilities.

We believe that the institutional argument about democracy and war-fighting capabilities provides important insights for understanding how political institutions could lead to a separate peace among a particular group of autocracies, personalist dictatorships. Our argument, however, emphasizes the weakness of personalist dictatorships, rather than their strength, and suggests that this weakness can generate a political dynamic similar to that anticipated by the institutional constraint argument for the democratic peace. Democracies are more effective in war not only because of the size of their winning coalitions, but also because of the professionalization of their armies and the clear subordination of those forces to civilian authorities (Huntington 1957). With a few exceptions (Mansfield and Snyder 1995) scholars of the democratic peace have underemphasized this aspect of the relationship between democracy and war and have missed the constraints imposed on the political–military capability of states by certain patterns of civil–military relations.

Among autocracies, single-party regimes have proven most successful in maintaining or enhancing the power of the coercive apparatus of the state. Single-party regimes have often proven effective at mobilizing the people and resources of their societies in support of the nation's armed forces and have forged strong civilian control over professionalized militaries. Successful single-party regimes not only maintain the unity and discipline of the armed forces but also transform the Army into an instrument of the Party. An effective military has the potential to make single-party regimes more peaceful toward each other because they recognize the threat that another single-party regime poses. Alternatively, since the war-fighting capabilities of single-party regimes are not directly related to the size of their winning coalitions, their effectiveness in war may not lead to a separate peace among single-party regimes.

The defining characteristic of a military regime is that the military as a unified institution governs the country. Thus, the military retains its cohesion and often increases the resources available to perform its national missions. As Andreski (1980, 3) notes, however, “The more often the armed forces are used internally, the less capable they become of waging a war.” Because they focus on internal security missions, military regimes are likely to be more constrained than single-party regimes. Thus, while the Argentine military proved to be very effective in killing Argentines, it was less effective in fighting the British in the Malvinas War.

We argue, however, that personalist dictatorships are substantially more constrained than either military or single-party regimes. Dictatorial regimes thrive by destroying independent institutions, personalizing all politics, and ensuring that important political decisions are funneled through the one supreme leader (Jackson and Rosberg 1982). Dictators gain allies through the creation of clientelistic patronage networks. Graft and corruption bind individual bureaucrats to the leader rather than to rational, insulated bureaucracies. Those who cannot be coopted by inclusion in the corruption, or who might become future rivals for power, are purged. To stay in power, the dictator must skillfully divide and conquer his potential adversaries in the factional politics endemic to personalist regimes.

No institution presents a greater threat to the power of a personalist dictator than the Army does. Like military regimes, personalist dictators depend on the Army and security forces to repress popular dissent. They understand, however, that the armed forces almost always represent a greater threat to their power than do civilian dissidents. Personalist dictators must be constantly on guard against the possibility of a military coup. Indeed, many personalist dictators are former military officers who took power in coups (Jackson and Rosberg 1982, 32–8).

Dictators guard against this threat by incorporating officers fully in their network of graft and corruption. Officers are promoted and given choice commands because of their loyalty to the supreme leader rather than their skills as soldiers. They report to the dictator rather than through the chain of command, to strengthen the personal ties between the dictator and officers at the expense of the institutional bonds of the military. Multiple security forces and intelligence units proliferate so that the dictator can play one group off against another. These strategies are often very effective at preserving the dictator's power. Personalist regimes last longer than military regimes (Remmer 1989; Geddes 1999a).²

The strategies pursued by dictators to stay in power, however, render military institutions in personalist regimes incapable of doing much more than repressing unarmed civilians. To construct a “coup-proof” regime, dictators must systematically weaken their armed forces (Quinlivan 1999). These institutional constraints make it difficult for dictators to launch aggressive foreign policies because they know that their armies are unlikely to succeed in real combat with disciplined institutionalized military opponents. Because of the constraints personalist regimes impose upon the capabilities of their armed forces, dictators will generally be cautious about using their armed forces in battle. Saddam Hussein may represent an outlier rather than the norm for personalist dictatorships. Thus, personalist dictatorships that face few of the decisional constraints found in democracies are also constrained from participation in international conflict, albeit through different political institutions.

Shared institutional constraints may lead to peaceful relations among personalistic regimes because dictators are aware that their opponents face serious costs when force is used and are likely to be cautious. They may perceive one another as “dovish,” which would reduce the probability that they will engage in wars that flow from security dilemmas (Bueno de Mesquita and Lalman 1992).

Any separate peace among personalist dictatorships is likely to be more tenuous than the one between democracies. The institutional incentives for peace are less profound among personalist regimes. Personalist regimes still experience limited decisional constraints and transparency. Furthermore, if personalist dictatorships are known to be weak, this could make them tempting targets. Even other personalist regimes might consider attacking these regimes, especially if there is a wide gap in the relative material capabilities of the two states. To help differentiate between the decisional constraints and war-fighting arguments, we examine the impact of one specific type of mixed dyad, democracy–personalist dictatorship. If constraints are key, the mutual dovishness of democracies and personalist regimes should lead to peace between them. If war fighting capabilities are key, personalist regimes may be precisely the kinds of easy targets that democracies are most likely to attack.

Shared Values and Peace Among Authoritarian Dyads.

The political institutions of military and single-party regimes do not provide strong arguments for why these regimes might be peaceful toward one another. Indeed, their lack of decisional constraints, transparency, and weak and divided military institutions could encourage conflict among military and single-party regimes. For these regimes, we must look toward explanations focused on shared interests and/or norms to account for a possible separate peace among similar regimes.³

Most military regimes proclaim to be technocratic and apolitical rather than ideological. Despite this, many institutionalized military regimes, especially those concentrated in South America during the 1960s and 1970s, adhere to a well-defined “national security” ideology (Stohl and Lopez 1986). National security ideology maintains an organic concept of the state and proclaims that the military is ideally suited to be the guarantor of the nation, bringing effective government in the national interest. This ideology legitimates aggressive foreign policy “to achieve or defend national goals against existing hostilities and pressures” (Stohl and Lopez 1986, 76). While this vision emphasizes preserving the territorial integrity of the state from external invasion, it focuses more intently on internal threats to the political integrity of the nation. Many of these regimes took power to fight real or imagined communist insurgents and the democratic institutions that might allow communists to achieve power. Shared interests in cooperating against perceived leftist threats could generate a separate peace among military regimes, as illustrated by Operation Condor, in which Latin American military regimes cooperated in hunting down leftist dissidents across national borders. The common socialization shared by most professionalized military institutions could help cement these bonds. Remmer (1998) found that the military regimes of the Southern Cone cooperated extensively on a variety of issues. We test whether these regimes have also avoided engaging in wars or militarized disputes with one another.

We believe that single-party regimes are the most likely kind of authoritarian regime to belong to the same normative community. Marxism–Leninism proclaims that socialists must remain united in their zero-sum struggle against capitalism because, if given the opportunity, capitalists will inflict cumulative losses on socialism. Marxist ideology also stresses the necessity for powerful socialist states to provide assistance to radical and other antiimperialist forces abroad. Since socialist states share a common ideology that stresses unity and brotherhood, one may expect that pairs of single-party regimes will not go to war because they share similar worldviews. While they do not focus on shared values, Oren and Hays (1997) find that developed socialist states were indeed peaceful toward one another during the Cold War.

This hypothesis should be treated with caution, however. While Marxism–Leninism provides a cohesive set of norms that binds these states together, it simultaneously legitimates violent revolutionary means of resolving conflict. Furthermore, the central role and exclusive nature of ideology in socialist single-party states can lead to conflicts about ideological purity. In his classic formulation of the containment policy, Kennan (1947) argued that the Communist Party leadership in the Soviet Union needed both to spread Communism aggressively and to punish satellite states that deviated from Soviet norms to legitimate their dictatorship at home. Any threat to the party's ideological supremacy, from internal or external sources, could be seen as a direct threat to the state. Thus, states governed by single-party regimes may conflict with one another because “they attempt to derive their legitimacy from identification with an ideological corpus” (Linz 1975, 197). Focusing on this dynamic, Bebler (1987) finds that Marxist–Leninist states have been involved in numerous conflicts with one another. Since Marxist ideology contains potentially contradictory ideas, it is difficult to know which ideas will dominate foreign policy.

One additional caveat is important here. China engaged in more disputes with Taiwan than almost any other member of the international system. Each of these cases is coded as a dispute among single-party regimes according to Geddes' guidelines. Clearly, China and Taiwan do not have shared norms and common interests. Therefore, we also examine below the conflict behavior of single-party regimes with Taiwan removed from the single-party regime category, so that we can get a more accurate sense of the importance of shared norms and common interests among single-party regimes.

Finally, it is unlikely that pairs of personalistic regimes will share similar normative values. Because personalistic regimes lack a common ideology, it is difficult to know what shared set of values might unite Saddam Hussein and Mobutu Sese Seko. A defining characteristic of personalistic regimes is that the cult of the leader rather than ideology dominates political life. Therefore, it is unlikely that shared worldviews will unite these regimes.

In sum, we argue that the more sophisticated classification of autocratic regimes presented in this article can lead to a more accurate assessment of whether specific types of autocratic regimes are peaceful toward one another. We anticipate that any dictatorial peace will be less robust than the democratic peace because none of the homogeneous pairs of autocratic regimes simultaneously exhibits the combination of institutional and normative constraints from warring with one another. Nevertheless, shared constraints may provide the foundation for a separate peace among personalist dictatorships, while shared identities and normative values may generate peaceful relations among single-party regimes. Military regimes are in the middle in terms of both shared values and institutional constraints. This leads to the central hypotheses examined in this paper.

The Dependent Variable

The empirical analysis focuses on militarized interstate disputes (MIDs) as the dependent variable.⁴

Independent Variables

Two data sets are used to determine regime type. First, we use Polity IIId to classify democracies between 1945 and 1994, with a score of six on Polity's democracy scale serving as the cutoff point (Jaggers and Gurr 1995; McLaughlin et al. 1998). We do not use Polity to classify authoritarian regimes, however, because, as Gleditsch and Ward (1997) indicate, there is a variety of paths through which states can achieve the same scores on the Polity scales. Regimes with different institutions and structures are treated as though they were functionally identical. Single-party, military, and personalist regimes often receive the same score on the Polity autocracy scale. Thus, studies that have found some evidence of an authoritarian peace, such as those by Oneal and Russett (1997, 281–3), Gleditsch and Hegre (1997, 290), and Raknerud and Hegre (1997, 394–400) may not be measuring truly homogeneous dyads.⁵

Then, as noted above, we use Geddes' (1999a) data set, which relies on a wide range of secondary sources, to classify autocratic regimes as personalist, single-party, or military during this time period. Cases are coded as military regimes if they are governed “by an officer or retired officer, with the support of the military establishment and some routine mechanism for high level officers to influence policy choice and appointments” (Geddes 1999a, 20). Military regimes share the following characteristics: relationships within the Junta and or military council are relatively egalitarian; the regime has some kind of institutions for deciding succession questions, and for routine consultation between the leader and the rest of the officer corps; the military hierarchy remains intact; and the security apparatus is controlled by the military. Geddes (1999a, 18) does not include governments that lasted for less than three years because they represented “periods of holding customary rules in temporary abeyance, struggle over rules, or transition from one set of rules to the next,” rather than regimes, defined as “sets of formal and informal rules and procedures for selecting national leaders and policies.” Since most of these temporary authoritarian interregna involved military governments, her list of military governments may underrepresent the universe of cases of military regimes.

Single-party regimes are defined as “regimes in which the party has some influence over policy, controls most access to political power and government jobs, and has functioning local- level organizations” (Geddes 1999a, 20). The party almost always exists prior to a leader's accession to power. It retains important functions down to the local level. Single-party systems routinely require party membership for most government jobs. While there are some exceptions, such as Taiwan and Mexico, most single-party states have been socialist regimes. To determine the dividing line between single-party authoritarian regimes and single-party dominant democracies, Geddes (1999a, 18) coded cases as authoritarian if the dominant party had never lost control of the executive and won at least two-thirds of all seats in the legislature in all elections prior to 1985.

Regimes were coded as personalistic if “the leader, who usually came to power as an officer in a military coup or as the leader of a single-party government, had consolidated control over policy and recruitment in his own hands, in the process marginalizing other officers' influence and/or reducing the influence and functions of the party” (Geddes 1999a, 20). These regimes lack effective institutions and one ruler dominates politics. To distinguish between military and personalistic regimes, Geddes codes regimes as personalistic when the leader dissolves military councils and other “consultative institutions”; when military officers are forced to retire; the leader forms a “party” to form an alternative base of support; or the leader murders, demotes, or dismisses dissenting officers. A single ruler dominates these regimes and ideology plays little role in politics. Monarchies, another type of autocratic regime in which a single ruler dominates, are not included in this data set.

Regimes that are not accounted for by either of these data sets are included in an “other” category. Many autocracies exhibit important characteristics of more than one category. Regimes Geddes classifies as hybrids are placed in the “other” category in our analysis. There is also a variety of political systems in uncertain processes of political transition that do not fit clearly into either the democratic or autocratic camps. Geddes, for example, does not classify the successor states created after the collapse of the Soviet Union and Yugoslavia because of the fluid nature of the political regimes in many of these states and because insufficient time has elapsed to judge whether their regimes have become institutionalized. These cases are also included in the “other” category here. We include dummy variables for each type of homogeneous dyad, democracies, personalist dictatorships, military regimes, and single-party regimes. Because we hypothesize that homogeneous pairs of states will be less likely to engage in wars and MIDs than heterogeneous ones, we use mixed dyads (including pairs of states in the “other” category) as the excluded category in our analysis.⁶

We also include five control variables that previous research has shown to have an impact on war and dispute involvement. First, previous research has shown that contiguous states are much more likely to engage in conflicts with one another than are states that do not share borders. Therefore, we include contiguity in the models.⁷

Method

We begin by presenting the data for all levels of hostility in simple tabular form (see Table 2). We then use multivariate logit regression to estimate the impact of regime type on conflict involvement (see Table 3). We use logit splines, a technique designed by Beck et al. (1998), to address the problems associated with time dependence in analyzing time-series, cross-sectional data with a dichotomous dependent variable. We add a counter-variable to the model recording the years of peace between a state's involvement in a militarized dispute and a subsequent involvement. The occurrence of a MID resets the counter to 0. We then generate three splines for each logit equation (see Beck et al. 1998; Enterline 1998, 394). The dependent variable in the results reported in Table 3 takes the value of one in any year in which a dyad is involved in a MID of any intensity and 0 when a dyad has not participated in any type of conflict behavior during a given year. We also estimated our models at each level of hostility and found consistent results across all levels of hostility short of war, suggesting that the dynamics of participation in MIDs is not dissimilar to the dynamics of the escalation of disputes among authoritarian regimes. See Table 4 for the results of these regressions.

Authoritarian Regimes and War

Our examination of the relationship between homogeneous authoritarian dyads and interstate wars between 1945 and 1994 shows that no two personalist dictators have fought one another.⁹

Scholars have raised the question of whether the rarity of war and the rarity of democracy in the international system renders the finding of no wars between democracies statistically insignificant (Spiro 1994). This problem is considerably worse in the case of personalist dictatorships and military regimes than for democracies. States initiated wars in only .04% of the dyad years between 1945 and 1994. Personalist dictatorships and institutional military regimes are also rare. Whereas joint democratic dyads constitute 9.8% of the total dyad years during this era, pairs of personalist dictatorships accounted for only 1.27% of the total dyad years. Military regimes were even rarer, with joint military dyads accounting for only .34% of the total dyad years. The absence of war between pairs of personalist dictatorships and pairs of military regimes may, therefore, be the result of random chance.

Authoritarian Regimes and Militarized Interstate Disputes

Table 2 indicates that militarized disputes are also rare among pairs of military regimes or personalist dicatorships, but their proportion of all disputes roughly matches their share of all regime dyad years. What stands out in an examination of all militarized disputes in Table 2 is the comparison of the dispute behavior of democracies and single-party regimes. Single-party dyads have experienced more militarized disputes than democratic dyads even though there are nearly three times as many democratic dyads in the data set. Single-party dyads were involved in 5.84% of all disputes, despite representing only 3.35% of all dyad years. Therefore, a simple bivariate chi-square test of the dispute participation of single-party regimes suggests that they are significantly more likely to engage in disputes with one another than are other dyads.

Since militarized interstate disputes are more common than wars and pairs of each type of regime have engaged in disputes, we can use multivariate logit regression to analyze the relationship between regime type and MIDs. Table 3 presents the logit spline regression results of four models.¹¹

In the first model, the autocratic dyad dummy is negative and significant, suggesting that autocratic dyads are more peaceful than mixed dyads. We found this result surprising given our concern with the significant variation among authoritarian regimes. Nevertheless, this result confirms the findings of Oneal and Russett (1997), Gleditsch and Hegre (1997), and Raknerud and Hegre (1997) and suggests that autocratic dyads broadly conceived are less likely to be involved in militarized disputes than are mixed dyads of democracies and autocratic regimes.¹²

In Models 2–4 in Table 3 and both models in Table 4, dyads made up of two personalist dictatorships or two military regimes are not significantly different from mixed dyads in their propensity to engage in MIDs. The single-party regime dyad is the only one of the homogeneous authoritarian dyads that is strongly significant across each of these models. The sign for this variable is negative, suggesting that single-party regimes are less likely to engage in MIDs with similar regimes than with other types of regimes. This finding contradicts the expectations of Bebler (1987). It also seems to be at odds with the data presented in Table 2, which suggest that single-party regimes are more likely to engage in disputes with other single-party regimes than with other regimes. These results support those of Oren and Hayes (1997), however, especially in model 3, which substitutes the NoTaiwan variable for the single-party dummy. When Taiwan is taken out of the single-party category, the coefficient increases substantially.

The control variables help account for why the single-party regime variable is negative in the logit models when the simple bivariate test suggested a positive relationship. Consistent with previous research, the results in all models in Table 3 indicate that contiguous dyads and those containing a major power are significantly more likely to experience militarized disputes than noncontiguous dyads or those without a major power. Dyads with large disparities in relative power are less likely to experience disputes. Only the alliance variable was not significant.

Single-party dyads were more likely to be contiguous than the norm and two of the five major powers, the Soviet Union and China, were governed by single-party regimes, which should increase the likelihood of conflict among these dyads. Indeed, a substantial majority of militarized disputes among single-party regimes involved either China or the Soviet Union in disputes with contiguous states. Many of these disputes pitted the USSR and China against one another. Almost none of the 90 disputes noted in Table 2 involved conflicts between two of the USSR's East European allies. The peaceful relationship among the Eastern European single-party regimes provides the strongest evidence for a separate peace among such regimes.

These results may mean that the high frequency of militarized disputes among single-party regimes is an artifact of the traditional great power behavior of China and the Soviet Union in their spheres of influence. The relative absence of conflicts among the smaller members of the Warsaw Pact could reflect Soviet hegemony. As Kennan suggested, however, conflicts between China and the Soviet Union and between the Soviet Union and its satellite states may have flowed directly from the character of their political regimes. At the core of the Sino-Soviet split was a struggle for ideological leadership in the socialist camp (Low 1976; Wich 1980; Kramer 1997a). Soviet interventions in the Eastern bloc reflected a desire, as suggested by the Brezhnev Doctrine, to enforce ideological unity in “conditions of unabating ideological war,” to “preserve the indivisible socialist system,” (LaFeber 1993, 258–9; Valenta 1991; Kramer 1997b). In any event, there may be a separate peace among single-party regimes, but the ambiguous roles of China and the USSR in the socialist world are cause for caution in the interpretation of these results.

Only joint democracy was consistently related to a lower frequency of militarized disputes. Democratic dyads were more peaceful despite the fact that they were more likely to be contiguous, as well as to include major powers and states of relatively equal power. The fact that only democratic dyads were consistently more peaceful than the norm, even at the lowest level of hostility where conflicts among transparent democracies are likely to be overrepresented in the MID data, adds further support for the empirical power of the democratic peace.

Furthermore, Model 4 provides some evidence that it is the greater war-fighting effectiveness of democracies that helps account for that peace. The variable for dyads that include a democracy and a personalist dictatorship is positive and significant. A focus on institutional constraints would have suggested that democratic–personalist dictatorship dyads would be more peaceful than other types of dyads. The increased likelihood of conflict between these dyads noted in Model 4 suggests that strong and capable democracies pick weak personalist regimes as targets.

While we have focused our analysis on authoritarian regimes, the finding that democracy–personalist dictatorship dyads are more likely to experience militarized disputes expands our knowledge of the broader relationship between regime type and conflict. Our findings support recent research on the democratic peace that it is the efficiency of democratic regimes in picking fights that they are likely to win and in fighting harder and better than their adversaries that accounts for the distinctive patterns found in the conflict behavior of democracies.