Abstract
Rivalries are likely to persist as long as contentious issues remain unresolved. Due to differing issue characteristics, some issues may be more intractable than others and therefore especially likely to prolong rivalry. In this study, I argue that rivalries rooted in territorial issues tend to be enduring due to broad-bases of domestic support for continuing to pursue territorial claims and loose linkages between territorial issues and particular political leaders, resulting in the persistence of territorial conflict over time despite changes of leadership. Alternatively, ideological and regime-related conflicts tend to be relatively fleeting due to narrow societal salience and close connections between such issues and particular political leaders, facilitating rivalry termination through leadership change. The empirical results reveal that territorial rivalries (as well as positional rivalries concerning influence/prestige) tend to be more enduring than rivalries rooted in ideological or regime-related conflict and that unlike territorial rivalries, ideological rivalries tend to terminate upon irregular changes of leadership.
In October of 1998, Ecuador and Peru came to an agreement concerning access rights to the Amazon basin and border demarcation, bringing an end to over 160 years of rivalry rooted in contending territorial claims. 1 One of the most striking aspects of Ecuador’s rivalry with Peru was its longevity. Rivalry persisted despite generational replacements, numerous changes of administration, and even regime changes. What characteristics of the Ecuador–Peru rivalry account for its longevity? More generally, why are some rivalries particularly enduring?
International rivals are pairs of states that view one another as competitors and enemies (Colaresi et al., 2007: 3–4). Rivals engage in militarized disputes and wars more frequently than non-rivals (Maoz and Mor, 2002: 3–4; Vasquez and Leskiw, 2001). Understanding the causes of rivalry persistence is important given that as long as states remain rivals, the potential for engagement in militarized conflict remains relatively high.
Rivalries are rooted in contending issue claims. The termination of rivalry is unlikely as long as salient issues remain unresolved. Rivalry between Ecuador and Peru, for example, began following the initiation of disagreement concerning territorial delineation following independence (due to the ill-defined nature of colonial borders) and lasted until the final resolution of contending territorial claims in 1998. Due to the centrality of issue disagreements to international rivalry, an examination of issue conflict is necessary in assessing the systematic causes of rivalry persistence.
There is substantial variation in the issues that rivals contend over. Some issues may have certain characteristics that may make issue resolution especially difficult. According to conventional wisdom, territorial issues tend to be especially salient due to tangible and intangible characteristics (Diehl, 1999; Goertz and Diehl, 1992; Hensel, 2000). In particular, intangible factors may result in territory becoming “effectively indivisible” (see Hensel, 2000; Vasquez, 2009: 355). Yet other issue conflicts, such as those centered on ideology or establishing regional or global influence, also contain intangible elements (Hensel et al., 2008). Furthermore, certain non-territorial disputes, such as those in which a state seeks to topple the regime of another state, tend to be highly salient given political leaders desire to retain power (Hensel et al., 2008; Huth, 2000). What then differentiates intractable issues from those that tend to be more amenable to resolution?
Rivalries driven by salient territorial issues likely tend to be enduring due to broad bases of domestic political support for continuing to pursue territorial claims and loose connections between territorial issues and particular governments or regimes. Due to the transcendent nature of territorial issues, government or regime change among territorial rivals does not necessarily result in the de-escalation of conflict and termination of rivalry. Rivalries driven by ideological or regime-related disputes, on the other hand, likely tend to be comparatively brief in duration due to relatively narrow bases of domestic political support for continuing to engage in rivalry and close linkages between such disputes and particular governments or political systems, facilitating rivalry termination through leadership or regime change.
In this study, I first highlight current limitations to theoretical perspectives concerning issue conflict and rivalry termination. I then examine issue characteristics that may tend to make some issues more intractable than others. In the empirical section, I assess the impact of issue variation on the likelihood of rivalry termination for cases of strategic rivalry from 1816–2000 using survival analysis. The results demonstrate that rivalries driven by territorial issues and positional concerns over influence/prestige tend to be more enduring than rivalries driven by ideological or regime conflict. The results also demonstrate in accordance with expectations that unlike territorial rivalries, ideological rivalries tend to terminate upon irregular changes of leadership.
Issue Intractability
From the realist perspective, international politics can be characterized as an unending struggle for power (Morgenthau, 1948). Although states may define their goals in a number of ways, states must acquire power in order to achieve national objectives. From such a standpoint, the professed goals that states seek are relatively unimportant given that all states will pursue power regardless of variation in stated foreign policy preferences.
Different issues, however, may affect different actors and interests (Rosenau, 1967, 1971) and issue characteristics vary across types of issues (Mansbach and Vasquez, 1981; Hensel, 2001). As a consequence, relations in which certain issues are at stake may differ from relations in which a different set of issues are at stake. Empirical research on issue variance has determined that territorial disputes are more likely to lead to arms races (Rider, 2009), tend to produce a higher number of fatalities (Senese, 1996), and are more likely to escalate to war (Senese and Vasquez, 2008; Vasquez and Henehan, 2001) than other types of disputes, demonstrating the utility of examining interstate relations from an issue-based perspective.
Exploring issue variation in the context of interstate rivalry may be particularly fruitful given that issues are at the heart of rival relations. The onset of issue competition can explain the origins of rivalry prior to engagement in militarized conflict (Mitchell and Thies, 2011), issues create linkages across conflict relationships over time (Klein et al., 2006; Mitchell and Thies, 2011), and rivalries generally do not terminate until outstanding issues have been resolved (Bennett, 1997a).
Despite the centrality of issues to rivalry initiation, persistence, and termination, previous research has tended to emphasize non-issue variables in seeking to explain rivalry dynamics. Diehl and Goertz (2000), for example, argue that rivalry initiation and termination tend to occur due to “political shocks” that alter the processes, relationships, and expectations that drive nation-state relations, such as world wars or substantial shifts in the systemic distribution of power. Others focus on variables associated with liberalism such as regime type, economic interdependence, and involvement in international organizations in seeking to predict the likelihood of rivalry termination (Cornwell and Colaresi, 2002; Prins and Daxecker, 2008). Such studies provide pieces of the puzzle in accounting for why some rivalries tend to be enduring while others do not. Nonetheless, due to the fundamental nature of issue competition to the persistence of rivalry, an examination of how issue variation may affect the likelihood of rivalry termination is warranted.
Although some research on issue competition in the context of interstate rivalry has found that territorial conflict is relatively common among enduring rivals (Bennett, 1998; Tir and Diehl, 2002; Vasquez and Leskiw, 2001), more than half of all disputes in the enduring rivalry context are over issues other than territory (Tir and Diehl, 2002) and major power rivalries tend to be rooted in policy disputes (Vasquez and Leskiw, 2001). Which issues, then, tend to form the basis of enduring rivalry? Furthermore, what are the theoretical foundations that might account for why some issues are more likely to lead to enduring rivalry than others?
According to conventional wisdom, territorial issues tend to be particularly salient due to tangible and intangible factors (Diehl, 1999; Goertz and Diehl, 1992; Hensel, 1996, 2000; Vasquez, 2009). In relation to tangibles, land may be viewed as being important due to the presence of resources such as minerals, water, or petroleum, or due to strategic geographic features that provide for tactical advantages. In regard to intangibles, land may be valued due to historical ties, cultural significance, or association with the inhabitants of a disputed area. Intangible factors are often considered to be especially important in accounting for the conflictual nature of territorial issues. If territory were solely tangible, compromises or power-sharing arrangements could potentially be negotiated. When land becomes imbued with intangible symbolic value, territory may become effectively indivisible as land takes on psychological importance beyond its immediate tangible value (Hensel, 2000; Vasquez, 2009: 355).
Yet other issues also have intangible qualities. Ideological issue conflicts centered on contending belief systems, for example, are (at least in part) intangible (i.e. they cannot be perceived by the sense of touch). Competitions over establishing regional or global influence similarly contain intangible characteristics (Hensel et al., 2008: 120). Some non-territorial disputes are highly salient, furthermore, such as those in which a state seeks to topple the regime of another state, placing a leader’s political (and perhaps even physical) survival at risk (Hensel et al., 2008; Huth, 2000). Which types of issues, then, are most likely to prolong rivalry?
Some issues are broadly salient to a state’s citizenry while others are relatively narrowly salient. Issue conflicts that have broad appeal and are not closely linked to a particular government or regime are likely to linger on over time despite changes in leadership or polity. Alternatively, issues that are primarily salient to political leaders and a narrow domestic constituency may tend to be more fleeting due to the enhanced potential that leadership exit will remove the source of contention while allowing for the domestic opposition to gain power and de-escalate interstate conflict.
For states engaged in territorial rivalry, at the domestic level there tends to be broad bases of political support for continuing to engage in acquiring sovereign control over contested areas of territory that become symbols of national identity (Huth, 2000: 100). Political leaders may be hesitant to make concessions on territorial issues fearing removal from office (Allee and Huth, 2006: 222; Huth, 2000: 101). The broad public importance of territorial issues may constrain political leaders and prevent compromise, resulting in the perpetuation of rivalry.
Due to the broad-based salience of territorial issues, government or regime change does not necessarily lead to the resolution of contending claims. As administrations change over time, as long as the public continues to identify with the goal of obtaining disputed territory, territorial conflict will continue. For territorial rivals, government or regime change does not necessarily alter the basis of disagreement (e.g. land may remain undelineated or undemarcated). Territorial conflicts consequently tend to persist over time as new administrations inherit ongoing territorial disagreements. Due to the transcendent nature of territorial issues, rivalries rooted in territorial conflict are likely to be relatively enduring.
H1: Territorial rivalries tend to be enduring.
While some rivalries are driven by territorial concerns, others are driven by ideological or regime-related concerns (Colaresi et al., 2007: 78–80). Ideological conflict centers on contending belief systems (Colaresi et al., 2007: 79). Regime disputes involve a state seeking to topple the regime of another state (Jones et al., 1996), which is at times rooted in ideological contention. Unlike territorial disputes that tend to be of broad public importance, preserving ideological hegemony or preventing the destabilization of one’s regime tends to be more narrowly salient to a state’s political leaders and those who benefit from the status quo or subscribe to the government’s ideology. There may be significant domestic opposition to continuing engagement in interstate rivalry rooted in ideological or regime-related concerns and a domestic faction within a target state of destabilization which, along with the state’s rival, favors leadership change.
Administration or polity change is likely to have significant consequences for states engaged in ideological or regime rivalry. Ideological and regime-related conflicts are closely related to the characteristics of particular governments or political systems. As long as certain leaders remain in power or a certain regime remains in place, such competitions, once established, will likely persist. Government or regime change, however, may remove the source of conflict, resulting in the cessation of hostilities.
Mozambique and Rhodesia, for example, engaged in a brief rivalry from 1975 to 1979 (Thompson, 2001) centered on Mozambican opposition to white minority rule in Rhodesia and support for rebels seeking to topple Ian Smith’s exclusionary regime. Preserving white rule was a highly salient issue for the Smith government and many whites in Rhodesia (Farley, 2008: 102). Yet there was only a narrow domestic base of support for continuing engagement in interstate rivalry. Many Rhodesian blacks, who outnumbered whites by a ratio of 24:1, demanded parity, and similar to the Mozambican government, advocated leadership change (Rotberg, 2002: 49). The toppling of the white minority regime in Rhodesia in 1979 eliminated the basis of disagreement between Mozambique and Rhodesia and paved the way for the normalization of relations. Narrow domestic support for sustaining rivalry among states engaged in ideological or regime-related conflict and the close connection between such issues and particular governments or political systems make it likely that ideological and regime-related rivalries will be relatively brief in duration, lasting only as long as particular leaders remain in power.
H2: Ideological and regime rivalries do not tend to be enduring.
Previous studies have demonstrated that regime change may at times be associated with rivalry termination (Bennett, 1997b, 1998; Cox, 2010; Diehl and Goertz, 2000). Such research, however, has not examined differential effects of leadership or regime change conditional on the primary issue under contention. Irregular changes of leadership in which leaders lose office in a way that is not in accordance with explicit rules and established conventions (Goemans et al., 2009) may result in a significant change of government orientation leading to the termination of ideological rivalry. Territorial rivalry, on the other, may tend to persist despite such changes of leadership.
H3: Irregular leadership change significantly increases the likelihood of ideological rivalry termination.
H4: Irregular leadership change does not significantly increase the likelihood of territorial rivalry termination.
Aside from rivalries driven by territorial or ideological issues, some rivalries are rooted in positional concerns in which states compete at the apex of a power hierarchy over establishing regional or global influence/prestige (Thompson, 1995, 2001). Due to the importance of power in achieving national objectives and providing for one’s security, states are likely to remain locked into positional power struggles once engaged. Rivalries driven by positional concerns are consequently likely to be relatively enduring.
H5: Positional rivalries tend to be enduring.
Non-positional rivalries rooted in policy disputes, on the other hand, likely tend to be somewhat fleeting. States’ policies may significantly change from one administration to the next. Non-positional interstate policy disputes may disappear as government turnover occurs and new political leaders pursue new policies.
H6: Non-positional policy rivalries do not tend to be enduring.
There is substantial variation concerning the issues that underpin rival relations. Certain issues have particular characteristics that may affect rivalry dynamics in systematic ways. Specifically, territorial and positional issues are imbued with qualities that may make rivalries driven by such concerns relatively intractable, while ideological, regime, and non-positional policy issues possess qualities that may make rivalries rooted in such issues more amenable to resolution.
Research Design
Expectations are assessed in relation to cases of strategic rivalry from 1816–2000 (Thompson, 2001). Strategic rivals are states that view one another as competitors and enemies for which there is a possibility of militarized conflict (Colaresi et al., 2007: 25). The strategic rivalry approach does not rely on dispute-density criteria in identifying cases of rivalry. This allows for the possibility that states may become rivals in the absence of militarized conflict due to salient issue competition.
Survival analysis, which enables an examination of causes of event “failure” (such as rivalry termination) is utilized to test expectations. Through survival analysis it is possible to determine whether relevant variables increase or decrease the likelihood of rivalry cessation. Positive coefficients indicate an increase in the likelihood of rivalry termination while negative coefficients indicate a decrease in the likelihood of rivalry termination. Rivalry dyad-year is the unit of analysis and the Cox proportional hazards model is employed due to its flexibility in not imposing a particular distributional form on duration times (Box-Steffensmeier and Jones, 2004).
Rivalries that did not terminate by the year 2000 are right-censored. Cases that began prior to the time frame of analysis are adjusted to reflect the amount of time that has passed since rivalry initiation (see Cornwell and Colaresi, 2002; Prins and Daxecker, 2008). For example, if an interstate rivalry began in 1816 but data are limited to the post-1875 period, the first year of observation is coded as year 60 rather than year 1.
I rely on two typologies to assess expectations concerning issue variance and rivalry duration. The first is a typology of strategic rivals which categorizes cases of rivalry as being driven by spatial, positional, both spatial and positional, or neither spatial nor positional conflict (Colaresi et al, 2007; Thompson, 1995, 2001). Spatial rivals contend for exclusive control over territory while positional rivals contest relative shares of influence and prestige within the international system or a regional sub-system (Colaresi et al., 2007: 79). Some rivals contend over neither spatial nor positional issues. An examination of cases of rivalry that fall into the “neither” category reveals that such states tend to engage in ideological or regime-related competition. Cases include such rivalries as Costa Rica–Nicaragua and Honduras–Nicaragua rooted in opposition to Nicaragua’s socialist Sandinista regime, and Mozambique–Rhodesia and Zambia–Rhodesia centered on opposition to white minority rule in Rhodesia and support for rebels seeking to topple the exclusionary regime.
Dummy variables are created for whether rivals are spatial, positional, both spatial and positional, or neither spatial nor positional in nature (with the neither category serving as a proxy for ideological and regime-related rivalry). Excluding the neither spatial nor positional rivalry variable as the reference category for analysis enables an examination of whether spatial and positional rivalries tend to be more enduring than rivalries driven by ideological (or other regime-related) concerns. 2
The second typology utilized is the Correlates of War identification of types of revisionism associated with militarized interstate disputes (Jones et al., 1996). MIDs are rooted in territorial, policy, regime, or other concerns. Territorial disputes involve making claims to territory. Policy disputes concern the intention not to abide by another state’s policy. Regime disputes are attempts at overthrowing another state’s regime.
The territorial and regime dispute categories can be relied on to test expectations concerning the impact of territorial conflict and regime conflict on the likelihood of rivalry termination. The policy category includes a variety of disagreements that can be further differentiated. Some policy disputes may be rooted in positional competition. Rasler and Thompson (2000) argue that major power rivalries often tend to be driven by positional concerns. Vasquez and Leskiw (2001: 312) subsequently found that minor power rivalries tend to be rooted in territorial disputes while major power rivalries tend to be driven by policy disputes, leading them to speculate that policy disputes among major power rivalries may be the result of positional competition. An examination of positional rivals reveals that policy disputes are quite common. Excluding cases of rivalry in which no militarized disputes occurred, 93.5% (29/31) of positional rivals and 81.4% (35/43) of spatial-positional rivals have engaged in at least one policy dispute.
In order to differentiate between policy disputes rooted in positional concerns versus those rooted in non-positional concerns, policy disputes that occur in the context of positional rivalry are coded as being positional-policy disputes while those that occur outside the context of positional rivalry are coded as being other-policy disputes. The expectation is that positional rivalries in which states engage in policy disputes will tend to be relatively enduring while non-positional rivalries driven by policy disputes will not tend to be enduring.
Each case of rivalry is coded in accordance with the dominant COW issue (the issue that a rival-dyad contended over most frequently). 3 The COW project codes the issue revisionism variable in accordance with the type of issue at stake for each militarized interstate dispute. Yet rivals compete over issues regardless of whether or not there is an active MID. For example, although Ecuador and Peru engaged in militarized disputes over territory on 45 occasions, territorial claims were at stake throughout the duration of their 168 years of rivalry. Dummy variables are consequently created that code the dominant issue as being present for the duration of each rivalry. 4 Excluding regime rivalries as the reference category allows for a comparative assessment of whether rivalries driven by territorial or positional-policy disputes tend to be more enduring than rivalries driven by regime disputes.
An alternative issue typology is the Issue Correlates of War identification of territorial, maritime, and river claims (Hensel et al., 2008). Although the ICOW project has produced a wealth of data concerning such issues, each issue is related to contention over geographic space. Such data do not consequently allow for the testing of hypotheses concerning ideological, regime, or positional competition. The types of strategic rivalry and COW types of revisionism data are therefore employed for the empirical analyses.
Archigos data spanning the years 1875–2000 (Goemans et al., 2009) are used to assess the effects of leadership change on the likelihood of rivalry termination differentiated by rivalry-type. Political leaders can lose power through regular or irregular means, due to death from natural causes, or by being deposed by another state. Rivalry termination may occur among ideological competitors due to irregular losses of power in which a leader loses office in a way that is not in accordance with explicit rules and established conventions, resulting in an alteration in a state’s political orientation. A variable is created that is coded 1 for the year of an irregular leadership change and the year after, and another coded 1 for the year of an irregular leadership change and the two subsequent years (given that the effects of leadership change may not be immediate). 5 Examining the effects of irregular leadership changes across different types of rivalry allows for a determination of whether rivalries driven by certain issues tend to be particularly affected by significant polity changes.
Several control variables are included in the analyses. Morey (2011) has demonstrated that militarized conflicts in which there are a large number of casualties in a short period of time tend to increase the likelihood of rivalry termination. Other studies have demonstrated that world wars at times contribute to the termination of rivalry (Goertz and Diehl, 1995; Diehl and Goertz, 2000). War may result in an unwillingness or inability to continue to engage in rivalry. It is consequently expected that engagement in war will increase the probability of rivalry termination.
Contiguous states may be more likely to become engaged in enduring rivalry than non-contiguous states (Colaresi, 2001; Cornwell and Colaresi, 2002). I therefore include a dummy variable capturing whether rivals are contiguous by land. Major powers are more capable of sustaining the costs associated with engagement in rivalry (such as spending on national defense) than minor powers. I consequently also create a dummy variable to control for the possibility that major power rivalries may tend to be more enduring than major–minor or minor power rivalries (see Morey, 2011). Previous studies have demonstrated that democracy may have a pacifying effect on rivalry (Bennett, 1997b, 1998; Cornwell and Colaresi, 2002; Morey, 2011; Prins and Daxecker, 2008). I therefore include a variable that captures the Polity score for the state with the least democratic political system for each dyad-year in accordance with a “weakest link” approach (see Dixon, 1993; Prins and Daxecker, 2008; Russett and Oneal, 2001).
Previous research on rivalry termination has emphasized the importance of system-level (world war, territorial, and power) shocks (Diehl and Goertz, 2000; Goertz and Diehl, 1995). World war shocks include World War I and World War II, territorial shocks entail large territorial exchanges, and power shocks are significant shifts in the distribution of power. A “shock” variable is created that captures world war shocks from 1914–18 and 1939–45, territorial shocks from 1884–94 and 1956–62, and power distribution shocks from 1859–1971, 1890–1901, and 1989–1992 (see Diehl and Goertz, 2000: 231).
Results
The findings of a basic model of rivalry duration with issue dummy variables are presented in Model 1. The ideological rivalry variable is excluded as the reference category. The results indicate that rivalries rooted in spatial conflict, positional conflict, or both spatial and positional conflict tend to be more enduring than rivalries driven by ideological conflict. The findings for a model of rivalry duration with issue dummy variables and relevant control variables are presented in Model 2. The results again indicate that rivalries rooted in spatial, positional, or both spatial and positional concerns tend to be less likely to terminate than rivalries driven by ideological concerns.
Model 3 estimates the effects of issue variation on the likelihood of rivalry termination with the positional issue variable excluded as the reference category. This enables an assessment of whether spatial rivalries tend to be more enduring than positional rivalries. The results demonstrate that spatial rivalries are not significantly less likely to terminate than positional rivalries. Spatial rivalries are therefore not uniquely enduring. The model also shows that rivalries driven by both spatial and positional issues tend to be more enduring than rivalries rooted in only positional issues, suggesting that rivalries driven by certain combinations of issues may tend to be particularly intractable. 6
Figure 1 illustrates the substantive effects of issue type on the likelihood of rivalry survival based on the Cox regression results from Model 2. 7 The spatial and positional survival curves nearly overlap (and are therefore difficult to distinguish individually in Figure 1) reflecting the similarity of such types of rivalry in relation to the likelihood of rivalry survival. The survival rates for rivalries driven by spatial or positional issues are substantially greater than the survival rates for ideological rivalries, and survival rates for rivalries plagued by both spatial and positional issues are greater than the survival rates for rivalries rooted in only spatial or only positional issues. The hazard ratios from Model 2 reveal that spatial, positional, and both spatial and positional rivalries face hazards that are 29.3%, 29.2%, and 16.4% of the hazard that ideological rivalries face, respectively. 8

Plot of Survival Function, Type of Strategic Rivalry
The models in Table 2 present the findings for the effects of the COW issue variables on rivalry duration. The regime category is excluded as the reference category in Models 4 and 5. The results indicate that rivalries in which territory is the dominant issue tend to be more enduring than rivalries driven by regime disputes. The results also indicate that positional rivalries driven by policy disputes tend to be more enduring than regime rivalries. Non-positional rivalries driven by policy disputes, on the other hand, do not significantly differ from regime rivalries in relation to the likelihood of survival. Model 6 estimates the effects of the COW issue variables on the likelihood of rivalry termination with the non-positional policy variable excluded as the reference category. The results demonstrate that territorial rivalries and positional-policy rivalries tend to be more enduring than non-positional rivalries rooted in policy disputes.
Figure 2 illustrates the likelihood of rivalry survival by issue type, based on the results from Model 5. Similar to the strategic rivalry issue-type data, the COW data illustrates the similarity of the survival curves for territorial and positional-policy rivalries and the difference in survival curves between rivalries driven by territorial or positional-policy issues versus those driven by other issues. The hazard rates from Model 5 indicate that the hazard for territorial rivalries is 35.6% of the hazard for regime rivalries while the hazard for positional-policy rivalries is 33.5% of the hazard for regime rivalries. 9 The results from Tables 1 and 2 provide robust support for Hypotheses 1, 2, 5, and 6. Territorial and positional rivalries tend to be enduring while ideological, regime, and non-positional policy rivalries do not.

Plot of Survival Function, Dominant COW Issue
Cox Hazard Model Coefficient Estimates, Effect of Rivalry Issue-Type on Rivalry Duration
p ≤ .05; ** p ≤ .01; *** p ≤ .001; two-tailed test.
Cox Hazard Model Coefficient Estimates, Effect of Dominant COW Issue on Rivalry Duration
p ≤ .05; ** p ≤ .01; *** p ≤ .001; two-tailed test.
In Table 3, the effects of irregular leadership change on rivalry termination conditional on whether spatial or ideological issues are at stake are estimated. Consistent with Hypothesis 4, Model 7 shows that there is not a positive statistically significant relationship between irregular leadership exit and rivalry termination for spatial rivals. The results concerning the irregular leadership exit variable do not significantly change when control variables are introduced into the model (Model 8) and when the irregular leadership exit variable captures years of leadership exit as well as the two subsequent years (Model 9). 10 The substantive effects of irregular leadership exit on the likelihood of rivalry survival based on the Cox regression results from Model 8 are illustrated in Figure 3. The survival curves for different values of the irregular leadership exit variable are similar for territorial rivals. 11
Cox Hazard Model Coefficient Estimates, Effect of Irregular Leadership Change on Rivalry Duration by Rivalry Issue-Type
p ≤ .05; ** p ≤ .01; *** p ≤ .001; two-tailed test.

Plot of Survival Function, Irregular Leadership Change for Spatial Rivals
Unlike spatial rivalries and consistent with Hypothesis 3, irregular leadership change significantly increases the likelihood of rivalry termination for ideological competitors. The irregular leadership exit variable is statistically significant when included in a basic model (Model 10) as well as when control variables are included (Model 11). 12 The variable similarly remains significant when capturing the year of irregular leadership exit as well as the subsequent two years (Model 12). The substantive effects of irregular leadership change on the likelihood of survival for ideological rivalries based on the results from Model 11 are illustrated in Figure 4.

Plot of Survival Function, Irregular Leadership Change for Ideological Rivals
Research on territory has tended to emphasize intangibility and salience in accounting for the particularly conflictual nature of territorial disputes (see Hensel, 2000; Vasquez, 2009: 355). Such a perspective does not account for why territorial issues tend to be more intractable than ideological and regime-related issues given that ideological issues also contain intangible qualities and regime issues tend to be salient. The results of this analysis suggest that ideological rivalries may tend to be more fleeting than territorial rivalries due to the association between irregular leadership exit and ideological rivalry termination and the lack of a relationship between irregular leadership exit and termination for territorial rivalries.
A few non-issue variables significantly affect the likelihood of rivalry survival. The war independent variable is positive and significant across each model in which it is included. 13 This finding is consistent with research which has demonstrated that large-scale conflicts with a high number of casualties in a short period of time (Morey, 2011) and world wars (Goertz and Diehl, 1995; Diehl and Goertz, 2000) tend to increase the likelihood of rivalry termination. 14 The shock variable is also positive and significant across the general models of rivalry survival (see Tables 1 and 2). This finding supports the perspective that exogenous political shocks tend to increase the likelihood of rivalry termination (Goertz and Diehl, 1995; Diehl and Goertz, 2000).
The democracy variable, which captures the democracy/autocracy score for the less democratic state for each dyad-year, is insignificant across models in which it is included but is positive and approaches statistical significance in Table 2 (p = .11). Evidence concerning the relationship between democracy and rivalry termination has been mixed. Similarly using a weakest-link operationalization, Prins and Daxecker (2008) found level of democracy to be significantly related to the likelihood of rivalry termination. Additional studies have found that joint democracy tends to decrease the likelihood of rivalry survival (Bennett, 1997b, 1998; Cornwell and Colaresi, 2002). Others, however, have found that the relationship between democracy and rivalry termination may not be robust (Colaresi, 2001; Morey, 2011). The results of Tables 1 and 2 suggest that variables other than level of democracy may tend to be more important in predicting the likelihood of rivalry survival.
Contiguity does not significantly decrease the likelihood of rivalry termination when controlling for issue variation. Previous research has assessed whether contiguity or the presence of territorial issues has more of an effect on increasing the likelihood of conflict (Hensel, 2000; Senese, 2005). Consistent with such research, the findings of this analysis suggest that it is the issue under contention rather than whether or not states are contiguous that significantly affects the prospects of rivalry survival.
The major power variable is not significantly related to rivalry termination. Some major power rivalries rooted in positional or both spatial and positional issues are enduring. Yet some minor power rivalries rooted in territorial issues are also enduring. Similar to the contiguity variable, the type of issue under contention tends to be more important than power status in predicting the likelihood of rivalry termination.
The results for the primary explanatory variables are consistent with expectations. Rivalries driven by territorial and/or positional issues tend to be more enduring than rivalries rooted in ideological, regime, or non-positional policy issues. Ideological rivalries tend to terminate upon irregular changes of leadership while territorial rivalries do not. The results are consistent with the notion that territorial rivalries tend to be enduring due to the transcendent nature of territorial issues, while ideological rivalries tend to be relatively fleeting due to the tendency for such rivalries to terminate upon irregular leadership change.
Conclusion
There are a variety of issues that may drive rival relations. Some rivalries are rooted in territorial conflict. Others are driven by ideological or regime-related concerns. Still others are rooted in positional competition or non-positional policy disputes. Some issues may be more difficult to resolve than others. Rivalries rooted in intractable issues are likely to be relatively enduring.
There generally are broad bases of support for continuing to engage in territorial rivalry in which disputed land becomes a symbol of national identity. As long as the public continues to identify with the goal of obtaining sovereignty over contested territory, rivalry will tend to persist through changes of administration as new leaders inherit outstanding disputes. Alternatively, ideological and other regime-related issues tend to be narrowly salient and closely associated with particular political leaders or regimes. Unlike territorial disputes, support for continuing to engage in ideological conflict may tend to be confined to states’ incumbent political leaders and those who benefit from the status quo or subscribe to the government’s ideology. Changes of leadership in which power is lost through irregular means may remove the source of contention among ideological rivals, resulting in the cessation of hostilities.
The empirical analysis indicates that ideological rivalries tend to be less enduring than territorial rivalries and that unlike territorial rivalries, ideological rivalries tend to terminate upon irregular changes of leadership. Previous research on territory has tended to emphasize the highly salient nature of territorial issues due to tangible, and in particular, intangible factors. Yet conflicts rooted in contending belief systems also have intangible characteristics and regime disputes tend to be highly salient due to political leaders’ desire to retain power. This study provides an account of why territorial rivalries tend to be more enduring than ideological and regime-related rivalries despite the intangibility of ideological issues and the salient nature of regime conflict.
The analysis also demonstrates that positional rivalries and rivalries driven by both spatial and positional issues tend to be enduring. Although territorial issues are generally considered to be particularly conflictual relative to all other issues, rivalries rooted in positional issues tend to be as long-lasting as territorial rivalries. This may in part be due to incessant concerns over relative power. Rivalries in which both territorial and positional issues are under contention may tend to be particularly enduring due to the intractable nature of territorial and positional issues individually as well as difficulties involved in resolving conflict when more than a single issue is at stake (see Dreyer, 2010a,b; Mitchell and Thies, 2011).
Further research is warranted on positional competition and rivalries in which more than a single issue is under contention. How are the characteristics of positional issues different from or similar to territorial and other issues? What are the causes of the initiation and termination of positional rivalry? Do rivalries with two or more issues at stake tend to be more enduring than rivalries in which a single issue is at stake regardless of issue type? Answering such questions would enhance our understanding of non-territorial and multi-issue enduring rivalry.
Ideological and regime issues also deserve further attention. There is an absence of theoretical work on the nature of ideological and regime-related conflict. Although this study attempts to begin to fill a void in the literature concerning such disputes, much remains to be known concerning the origins of ideological and regime competition, the process through which such disputes escalate, and how ideological and regime conflicts can be managed.
Several additional types of issues that rivals may at times contend over have received insufficient attention. The Issue Correlates of War database contains data on not only territorial, but also river and maritime issue claims (Hensel et al., 2008). Scholars have only begun to examine ways in which conflict over such issues affects rival relations (see Lektzian et al., 2011; Mitchell and Thies, 2011). The effects of conflict over economic, cultural, and several other issues on rival relations also have not been assessed.
Issues are central to international rivalry. Having an understanding of the ways in which issue variation affects rivalry duration enables us to better predict which rivalries are most likely to endure. The challenge is to then determine the conditions under which rivalries rooted in relatively intractable issues can be successfully managed and peacefully resolved.
Footnotes
1
Ecuador and Peru were rivals from 1830 to 1998 according to the strategic rivalry perspective (Thompson, 2001).
2
There are 64 cases of spatial rivalry, 37 cases of positional rivalry, 49 cases of spatial-positional rivalry, and 14 cases of ideological rivalry.
3
There are 18 cases of strategic rivalry with no MIDs (for which there are COW data). These cases are excluded from the analyses in which the COW issue variables are the key independent variables. There are five cases with an equal number of two different types of disputes. For such cases I referenced the data on strategic rivalry type (Colaresi et al., 2007; Thompson, 1995, 2001) to determine which issue is dominant. The only case for which a primary issue could not be determined is Greece–Serbia, which is consequently excluded from the analyses utilizing the COW issue variables.
4
There are 76 cases of territorial rivalry, 33 cases of positional-policy rivalry, 20 cases of other-policy rivalry, 15 cases of regime rivalry, and 2 cases of “other” rivalry. The two cases of “other” rivalry (Argentina–Paraguay and Brazil–Paraguay) are excluded from the analyses in which the COW issue variables are used as the primary independent variables.
5
The variable does not capture leadership changes through direct removal by another state for two primary reasons. First, there are no instances of such leadership changes (which involve such actions as kidnapping or invasion) among ideological rivalries. As a consequence, empirical results concerning the leadership change variable for ideological rivalries are the same regardless of whether or not the direct removal by another state category is included. Second, territorial rivalries seem to at times terminate upon leadership removal by another state not due to leadership change per se, but rather due to occupation or state death (e.g. the termination of Poland’s rivalries with Czechoslovakia, Germany, and Lithuania in 1939). Nonetheless, as noted in the empirical section, the statistical significance of the leadership exit variable for territorial rivalries does not change when the direct removal by another state category is included in the coding of the variable.
6
Similarly, the probability of militarized conflict may increase when more than a single issue is at stake (Dreyer 2010a,b: Mitchell and Thies, 2011).
7
For figures based on models in which control variables are included, the indicator variables are held at their mode (1 for the contiguity variable and 0 for the major power dyad, war, and shock variables) and the democracy variable is held at its mean.
8
The confidence intervals on the hazard ratios for the spatial, positional, and both spatial and positional variables overlap, indicating a lack of significant difference across such rivalries in relation to substantive effects.
9
The hazard ratio confidence intervals for the territorial rivalry and positional-policy rivalry variables overlap, indicating a lack of significant difference between territorial and positional-policy rivalries.
10
The major power dyad variable is not included as a control variable because there are no instances of major power rivalry in which only spatial (and not also positional) issues are at stake.
11
Models were estimated in which the irregular leadership exit variables also captured instances of leadership exit through direct removal by another state. The variables remained statistically insignificant.
12
The contiguity, major power dyad, and war variables in the models for ideological rivalries are excluded due to a lack of sufficient variation.
13
The war variable is excluded from the ideological rivalry models due to the extremely rare occurrence of war between ideological rivals.
14
I also estimated models including a control variable for militarized interstate disputes that do not escalate to war. The variable was statistically insignificant across each model.
