External Territorial Threats and Military Regimes

External Territorial Threats

To capture the effect of sustained, international security competition on domestic politics, I focus on the combination of interstate rivalry and territorial competition. Interstate rivalry involves a pair of states that regard each other as competitive, threatening enemies in protracted conflict (Colaresi, Rasler, and Thompson 2008; Klein, Goertz, and Diehl 2006). Rivalry produces “a relationship framed in expectations and beliefs that are prone toward exaggerating possibly hostile actions and downplaying the sincerity of possibly cooperative gestures” (Colaresi, Rasler, and Thompson 2008, 13). For example, the relationships between Russia and Turkey or Colombia and Venezuela have been competitive and hostile, generating continuing tension, even though these countries have not engaged in militarized disputes. Learning from repeated confrontations, rival states know that their hostile relationship is likely to continue into the future (Klein, Goertz, and Diehl 2006). Such rivalries, often characterized by mutual threat perception and intense hostility, are the context in which the vast majority of interstate conflicts occur.

Furthermore, scholars agree that territorial issues are more salient and threatening than other issues in world politics (Hensel 2001; Vasquez 2009). Territory is highly valued for its tangible attributes, such as militarily or economically strategic terrain, valuable resources, and the presence of a permanent population (Hensel 2001). Beyond its tangible values, it also carries intangible or psychological values as it is often fundamental to national or ethnic identity or has historical or religious significance. Thus, territorial issues are considered more salient to individuals and the state than other issues, such as policies and regimes. Threats to homeland territories tend to evoke more fear and anxiety among the population than do other issues (Gibler 2012). States are more willing to seek costlier options including military options to achieve their goals over territorial issues, which makes it more difficult to resolve territorial disputes.

Consequently, territorial disputes are more likely to result in violent conflict than disputes over other issues, produce higher fatalities, and become more escalatory (Hensel 1996; Huth 1996; Vasquez 2009). States are highly attentive to the possibility of violent transfers of territory and actively develop war plans based on acquiring or retaining territorial control (Gibler 2012). Arms races are most likely to occur when both rivalry and territorial disputes are present (Rider 2013). Therefore, territorial rivalries, mainly competing over territorial issues, are intense and have strong repercussions on domestic politics and political institutions.

Territorial Threats and Militarization

Sustained external threats, posed by territorial rivals, make states anticipate violent challenges and, in response, develop sufficient defenses, regardless of whether the states harbor expansive foreign policy ambitions. As Rasler and Thompson (2004, 882) note, “[e]ven if a state could avoid the temptation to expand, being in a neighborhood harboring some expansive aspirants meant that one had to develop adequate defenses against the possibility of attack.” Rulers develop large, standing armies in anticipation of such external threats continuing into the future (Huth 1996; Rasler and Thompson 2004). External threats also allow rulers to better extract the resources necessary for militarization (Thies 2005; Tilly 1985). As Gibler (2012) argues, the effect of external threats on militarization is stronger when threats are posed to homeland territories.

High levels of military preparation, in turn, place the military in a politically pivotal role and produce the expansion of military organizations (Gibler 2012; Hintze 1975; Lasswell 1941; Woo 2011). When a country is confronted with external threats to its security, its military is better positioned to demand and obtain greater institutional autonomy in personnel, education, budgetary, organization, and procurement decisions. Rulers delegate extensive power to the military to defend against external threats. Furthermore, the need for military effectiveness tends to produce coherent and united command structures and meritocratic military promotion (Desch 2001). A military equipped with greater resources, autonomy, and cohesion is better able to expand its political influence and intervene in politics (Belkin and Schofer 2003). Thus, territorial threats make countries more prone to military rule, given that the likelihood of military regime is an increasing function of the military’s capacity to intervene in politics.

This may not be the case if territorial threats decrease the military’s willingness to intervene in politics and directly rule. For example, some scholars argue that external threats discourage military political involvement (e.g., Arbatli and Arbatli 2016; Desch 2001; Huntington 1957; McMahon and Slantchev 2015). Desch (2001, 12) argues that “external threats have obvious effects: they threaten the entire state, including the military; they usually produce increased unity within the state; and they focus everyone’s attention outward.” Under external threats, the military may be occupied with its external missions and, thus, be less inclined to participate in domestic politics. In addition, external threats may facilitate power-sharing deals between rulers and militaries by providing more resources to the military and reducing the expected benefits of military rule (Arbatli and Arbatli 2016; McMahon and Slantchev 2015).

However, there are theoretical and empirical reasons to doubt that external territorial threats decrease the motivation for military intervention and rule. ² First, even if external threats can generate rally effects, they cannot shield the government from domestic challenge for an extended time period because those effects are usually short term. In addition, rally effects mostly arise from immediate threats posed by conflict events that tend to be short-lived. ³ However, external threats emanating from territorial rivalries are usually long-lasting. For instance, the average duration of spatial rivalries, mainly concerning territorial issues, is forty-eight years.

Next, external territorial threats cannot resolve the problem of credible commitment inherent in authoritarian power-sharing deals. With the absence of an independent authority to enforce and monitor power-sharing agreement, uncertainty and mutual mistrust are severe in authoritarian regimes (Svolik 2012). Dictators cannot credibly commit to safeguarding the military’s corporate interests. For example, despite empowered militaries in Pakistan, Egypt, Thailand, and Turkey, civilian leaders coming to power after military rule attempted to curtail military prerogatives and privileges and increase their leverage over the armed forces, which provoked violent reactions from the military, generating cycles of military political involvement. Meanwhile, rulers can never be certain that a structurally empowered military will remain in barrack and refrain from intervening in politics because territorial threats increase the military’s organizational power to stage a coup (Florea 2018). As in the security dilemma, they have incentives to coup-proof and extend control over military prerogatives and privileges. These efforts will infringe into the military’s corporate interests and may provoke military officers into contemplating regime overthrow.

Furthermore, these sustained territorial threats may provide additional motivations for military intervention and rule. As in Pakistan, Egypt, and Turkey, the military can perceive its increased role in politics as “the only means to achieve the military’s primary mission of defending the state” (Staniland 2008, 340). The very importance of national security provides the motive for the military to intervene in politics as a guardian of national interests (Finer 2002; Staniland 2008; Woo 2011). Given that the military has a keen interest in a stable state, this is particularly true when the civilian government poorly addresses security concerns or faces a legitimacy crisis caused by economic difficulties. A long exposure to sustained security threats often causes domestic unrest because military buildups undermine the ability of the ruling elite to maintain domestic political support by diverting resources that could be used to distribute benefits to its coalition members and the public (Barnett and Levy 1991). In sum, territorial threats provide a structural condition that makes countries more prone to military rule.

Empirical evidence suggests that external threats lead to military intervention rather than military withdrawal. Recent cross-national studies demonstrate that sustained security threats from international rivalries (including territorial rivalries) or territorial disputes increase the risk of coups (Bak, Rider, and Chavez 2017; Florea 2018; Kim 2018a, 2018b). ⁴ Similarly, the risk of military coups, indeed, tends to be greater in countries such as Pakistan, Egypt, Thailand, and Turkey where the military as an institution has greater political influence.

Furthermore, the effects of territorial threats are likely to proceed and accumulate over time. As discussed, interstate rivalries with unstable boundaries, once formed, are maintained over long periods (Colaresi, Rasler, and Thompson 2008; Klein, Goertz, and Diehl 2006). Both territorial disputes and rivalries tend to “become ensconced in national memories, identities, and political discourses” (Rasler and Thompson 2011, 292). States lock in policies and strategies based on the expectation that mutual confrontations are likely to continue into the future (Klein, Goertz, and Diehl 2006). The longer territorial rivalries are in place, the greater the military’s resources, autonomy, institutionalization, political power, and capacity to intervene in politics. Interstate rivalries operate as a slow-moving, causal process with an incremental impact on institutions (Thies 2005, 454). For instance, Loveman (2001) argues that a political culture of militarism has been reproduced over decades, even centuries, in Latin America through the reinforcement of laws, regulations, rituals, and myths. In sum, I argue that territorial threats both augment and entrench the military’s power.

Militarization undermines democratic rule and fosters authoritarianism. The presence of a politically powerful military impedes efforts toward democratic consolidation as the military offers a credible threat of political intervention (Cheibub 2007). For example, the democratically elected governments in Thailand were several times deposed by the military when the military felt that their organizational interests were at stake. When the military, a repressive agent of last resort, is more powerful, it can more easily repress domestic opposition (Svolik 2013). Bowman (2010) also shows that militarization exerts adverse effects on democracy and development in Latin America. Prolonged threats undermine institutional constraints on leaders and help elites consolidate power in their hands by silencing domestic opposition groups (Gibler 2010). The existence of salient threats and the importance of ensuring national security may lead to the repression of basic civil rights. Existing studies provide supporting evidence that territorial threats hinder democratization (Gibler 2012; Owsiak 2013).

Territorial Threats and Collegial Military Rule

However, I do not expect that a prolonged exposure to territorial rivalries will promote all types of military-led autocracies. As Geddes, Frantz, and Wright (2014) emphasize, there are two types of military rule: collegial military rule and military strongman rule. These differ in both the decision-making procedures and the makeup and influence of the leadership group, although both of them are ruled by current or former members of the military. Collegial military rule implies that political power typically resides within a governing junta involving key members of the armed forces. The junta institutes mechanisms to prevent the leader from concentrating political power around him, and the leader consults with a group of officers representing the junta when making key policy decisions (Geddes, Frantz, and Wright 2014). Collegial military rule is, thus, characterized by rule by the military as an institution. Examples include the three military regimes in Pakistan since 1958, the fifty-year military rule in Myanmar, the regime in Brazil after the military coup of 1964, and the regime in Greece after the military coup of 1967. By contrast, military strongman rule takes the form of “one-man rule,” characterized by uncontested control of executive powers. A single military officer controls the decision-making structure and personalizes the security apparatus, while other officers have little influence on important political decisions. Mobutu Sese Seko’s rule (1960–1997) in Zaire, Idi Amin in Uganda (1971–1979), Ali Abdullah Saleh in Yemen (1978–2012), Rafael Trujillo’s regime (1930–1961) in the Dominican Republic, and the Gaddafi regime (1969–2011) in Libya are notable examples of military strongman rule.

Geddes (2011) argues that the type of military rule depends on the characteristics of the military forces that leaders inherit. After seizing power, new leaders in military dictatorships must deal with other military officers that have the capacity to remove them through a coup. When the military is internally unified and disciplined, high-ranking officers can control junior officers’ behavior, which allows them to make a credible promise that the leader will not be ousted in a coup as long as he sticks to power-sharing agreements. A unified and cohesive military allows leaders and military elites to commit credibly to power-sharing agreements and establish rules and institutions for sharing and rotating power. In addition, coup plotters from professionalized militaries sometimes devise detailed rules about sharing power before staging a coup. However, in factionalized and lowly professionalized militaries where military elites cannot control the behavior of subordinate officers, military elites cannot credibly commit their subordinates to power-sharing agreements with the leader. When faced with such militaries, leaders, thus, choose a different survival strategy: they tend to concentrate power in their own hands and attempt to deter coups by creating political support outside the military, through either paramilitary forces loyal to the leader or a mass civilian party.

This argument suggests that environments rife with sustained territorial threats facilitate the presence of collegial rule by the military as an institution, not the presence of personalist rule by a military strongman. As discussed above, a country’s threat environment significantly shapes the institutional structure of its military. A challenging security situation tends to produce unified and cohesive militaries and discourages personalism in the armed forces (Desch 2001). Faced with security threats, leaders should create a competent military organization by delegating authority and power to military officers, adopting merit-based promotions, and fostering extensive internal communication (Way and Weeks 2014).

Accordingly, personal control over the military, a key component of personalist rule, is difficult to achieve under grave threat environments. Any tactic employed to increase personal control over the military simultaneously erodes the state’s military effectiveness. For example, promoting and appointing officers based on personal loyalty and ethnic affiliation diminishes leadership qualities and discourages the exercise of initiative (Brooks 2003, 162; Pilster and Böhmelt 2011, 335). Given that personalist leaders’ tenure is highly dependent on their ability to undermine the military’s organizational capacity (Way and Weeks 2014), I expect that sustained territorial threats do not contribute to the rule by military strongmen.

In contrast, a threatening security environment engenders the need to increase military effectiveness, which produces relatively professional and powerful military organizations. It hinders political leaders from weakening the military’s political power. A strong and autonomous military serves as an institutionalized veto player, and its political influence is hard to curb when faced with security threats. Hence, sustained territorial threats create fertile conditions for rule by the military because, as Geddes (2011) notes, military rulers can make credible power-sharing agreements with other military elites.

Empirical Implications

Overall, this discussion leads to the following hypotheses. I hypothesize that

Territorial threats do not inevitably produce collegial military rule, but they establish the structural conditions conducive to militarization and the presence of collegial military rule. Prominent examples include current or past military dictatorships in Algeria, Brazil, Egypt, Pakistan, and South Korea, all of which have engaged in long-standing rivalries with their neighbors. As discussed above, I also expect that threats to homeland territory are more salient and immediate than other types of threats. For instance, as Gibler (2012, 22) notes, “recurring disputes over colonial holdings are less likely to affect the population than rivalries over adjacent territories.” States are more likely to engage in military buildups in response to territorial threats than in response to nonterritorial threats (Rider 2013). Accordingly, territorial rivalries will have a greater impact on the military’s capacity to influence politics than nonterritorial rivalries. Contrarily,

because they tend to empower the military as an institution and increase its institutional autonomy and cohesion.

Dependent Variable

To identify collegial military regimes, I rely on regime definitions provided by Geddes, Wright, and Frantz (2014; hereafter GWF). To distinguish between different kinds of dictatorships, GWF (2014, 318) focuses on

whether control over policy, leadership selection, and the security apparatus is in the hands of a ruling party (dominant-party dictatorships), a royal family (monarchies), the military (rule by the military institution), or a narrower group centered around an individual dictator (personalist dictatorships).

GWF (2014, 152) define military regimes as those in which “the dictator consults with other high-ranking officers and can be constrained by them.” Accordingly, GWF’s definition of military regime corresponds to collegial military regimes. Using GWF’s ten-category regime classification, I code “military” and “indirect military” as a collegial military regime.

Following GWF’s recommendation, I code country-years as military strongman rule when they are coded both as personalist by GWF and as military by Cheibub, Gandhi, and Vreeland (2010). Cheibub, Gandhi, and Vreeland’s classification of dictatorships depends solely on the identity of the regime leader without considering institutional configurations or composition of the political leadership. This classification can identify military-led regimes, and the strategy of combining them with the GWF dataset can measure military strongman rule, a subset of military-led regimes.

Key Independent Variable

To capture salient and prolonged threats to a state’s territories, I construct several measures of territorial threats. The first two measures focus on territorial rivalries. For the measure of rivalries, I rely on two widely used datasets: Thompson and Dreyer (2011) ⁶ and Goertz, Diehl, and Balas (2016). In defining rivalry, Goertz et al. emphasize the occurrence of actual militarized interstate disputes while Colaresi et al. employ a perception-based approach to identifying strategic rivalries, relying on leader statements and historical narratives. I employ both measures because both participation in repeated militarized disputes and the perception of military threats affect the need for security and military buildups, thereby enhancing the military’s political influence.

To explore the effect of territorial rivalries, I distinguish between territorial rivalries and nonterritorial rivalries. When I employ strategic rivalry, I use spatial rivalries, primarily concerned with territorial issues, as coded in Thompson and Dreyer’s dataset. I measure Territorial rivalry (strategic) as a dichotomous variable that takes the value one when a country is involved in at least one spatial rivalry in the prior year, and zero otherwise. Nonterritorial rivalry (strategic) is an indicator for countries that experience strategic rivalry but no spatial rivalry. For the Goertz et al. rivalry measure, I create a binary variable Territorial rivalry (Goertz, Diehl, and Balas; GDB) that is coded one when a rivalry’s most frequent reason for militarized disputes is territory. ⁷ Nonterritorial rivalry (GDB) includes remaining rivalries.

The last measure of territorial threats draws on the Issue Correlates of War (ICOW) territorial claims dataset (Hensel et al., 2008). The ICOW dataset indicates whether a state has an active territorial claim with other states during the period 1816–2001. A territorial claim exists if the official representatives of a state “make explicit statements claiming sovereignty over a specific piece of territory that is claimed or administered by another state” (Hensel et al., 2008, 128). I use not only a binary indicator of a state’s territorial claims but also ICOW’s salience index that measures the importance of each claim on a 0–12 scale with six indicators. Hensel et al. (2008) show that as a territorial claim’s salience increases, the risk of militarized conflict over the claim significantly increases.

To further test the cumulative, historical effects of territorial threats, I follow the strategy proposed by Gerring, Thacker, and Alfaro (2012). For each country, I add up years of territorial rivalries from 1900 (or the first year for which data are available) to the present year and apply an annual depreciation rate to give more weight to recent years. To illustrate, the history measure is measured as

H i s t o r y i t = ∑ τ ≥ 0 τ = t − t 0 ( 1 − δ ) τ 1 [ Threat = 1 ] i , t − τ

where t 0 denotes 1900 ⁸ or a country’s first year for which data are available; and 1[Threat = 1] is an indicator function that is equal to 1 if a country-year experiences territorial threats and 0 otherwise; δ is an annual depreciation rate. This measure implies that a history of territorial competition accumulates in continuous years but annually depreciates at a rate δ. For example, δ = 1 percent assigns nearly as much weight to distant years as to recent years, whereas δ = 100 percent assumes a case in which the history does not matter, thereby turning the history into a binary indicator. I choose a slow 10 percent rate and attempt to use alternatively a 50 percent rate to ensure the robustness of my results. The strategy of measuring rivalry history as a stock variable is useful because territorial rivalry is sluggish and, thus, inappropriate for a distributed lag model.

Using the Militarized Interstate Disputes (MID) version 4.01 dataset, I also construct several measures of territorial and nonterritorial conflict to compare threats stemming from territorial rivalries with those emanating from sporadic conflict events. I restrict interstate conflict to militarized disputes involving the use of force (hostility level of 4) or wars (hostility level of 5) because the MID dataset includes many minor or accidental disputes. To identify whether a territorial revision was sought, I use the state-level revtype variables in the MID data. I code an MID as a territorial dispute if territory is a revision motivation for either state and otherwise, as a nonterritorial MID.

Control Variables

For potential confounding variables, I include a variety of control variables that could be related to territorial threats and the presence of military rule. I lag all independent variables, except the post-Cold War indicator, by one year to minimize reverse causation. First, it would be important to control for changes in the geopolitical environment following the end of the Cold War. For example, the number of military collegial regimes peaked in the late 1970s and have continued to drop since then (Geddes, Wright, and Frantz 2014, 318). The end of the Cold War is known to drive this pattern. During the Cold War, the United States strategically supported many military dictatorships, but as its struggle with the Soviet Union ended, its support for military dictatorships also faded away. With that said, my argument may also partly explain the pattern of decreasing numbers of military regimes since the end of the Cold War due to dramatic changes in the international security environment. As Goertz, Diehl, and Balas (2016) demonstrate, international peace has increased since the end of the Cold War because international norms about territorial integrity have been reinforced. I still control for an indicator of the post-Cold War period. However, it is worth noting that the inclusion of the post-Cold War period indicator could hinder the effort to find the total effect of territorial threats on military regimes.

Second, I control for regional political environments as they can affect both a country’s political regime type and the broader security environment. A regional preponderance of democracies may negatively affect the presence of military regimes and improve the regional security environment. At the same time, a regional diffusion among military regimes may exist. Accordingly, I include the regional proportion of military regimes. ⁹ Third, I include an indicator of civil war taken from the Uppsala Conflict Data Program/Peace Research Institute Oslo (UCDP/PRIO) Armed Conflict Dataset. Desch (2001) argues that internal conflicts tend to produce internally oriented military doctrines, requiring the military to make political decisions. Therefore, because internal conflicts encourage the military to take a more active political role, I control for the country’s percentage of country-years at civil war in the previous five years.

Next, I also include the level of economic development and economic growth, which are significant determinants of a country’s regime type and possibly even its conflict propensity. Previous studies have shown that lower levels of economic development and worse short-term economic performance facilitate military intervention (Londregan and Poole 1990). I, thus, include the natural log of real GDP per capita, and the annual percentage change of real GDP per capita, drawing data on GDP per capita from the Penn World Table 8.0. Last, to correct for duration dependence, I include cubic polynomials of years since the last transition to military regime (Carter and Signorino 2010).

Robustness Checks

To ensure the robustness of my results, I perform several robustness checks. Due to space constraints, the results of these robustness checks are discussed here briefly and presented fully in the online appendix.

First, I further explore the possibility that the previous results may suffer from reverse causality. Table A8 examines whether military regimes are more likely than other authoritarian regimes to engage in territorial rivalries or claims. The results provide little evidence for this possibility, regardless of whether I examine onset or incidence. Combined with the finding that even becoming a target of territorial claims increases the likelihood of a military regime, these results cast doubt on the reverse-causality interpretation of the main finding previously reported.

Next, I estimate alternative model specifications and find that the results remain strongly robust to different model specifications. (1) To ensure against a saturated model and examine the degree of posttreatment bias, I reestimate the main models of Table 1 by including only cubic polynomials of time since the last transition to military rule. (2) In addition, I control for the number of years that the country has spent under military rule. Given that military interventions breed other interventions (Londregan and Poole 1990), I explore whether the effect of territorial threats exerts significant influence on the establishment of military regimes after controlling for a previous history of military rule. This would stack the deck against finding support for my argument because a previous history of military rule absorbs much of the dynamic effect of territorial rivalries. (3) I attempt to include mass protests, obtained from Banks and Wilson (2013), or the level of economic inequality, proxied by the Gini index drawn from the World Income Inequality Database. (4) To control for unobserved heterogeneity between countries, I fit models including country random effects, regional fixed effects, or colonial legacies. Results remain similar (see online appendix Tables A5–A8).

I attempt to use alternative measures of collegial military regimes. First, Table A9 uses the original GWF measure of military regimes, which includes military-personal in the category of collegial military regime. Compared with Table 1, the inclusion of military-personal reduces the coefficient estimates of territorial threats in magnitude and their statistical significance, although results remain qualitatively identical. Second, I also create a 3-level ordinal measure of military rule: 0 measuring nonmilitary regimes, 1 measuring military hybrids (military-party, military-personal, or military-party-personal), and 2 referring to “pure” military regimes. ¹⁰ Ordinal probit estimates and multinomial logit estimates show that territorial threats are significantly associated with only “pure” military regimes (online appendix Tables A10 and A11). Similarly, when I merge both collegial military rule and strongman rule and create a combined measure of military-led dictatorships, little evidence suggests that territorial threats are significantly associated with the presence of military-led regimes, which is consistent with the findings in Tables 1 and 3 (Table A12). This result emphasizes the importance of differentiating between collegial military rule and strongman rule.

Fourth, I employ additional measures of a history of territorial threats: (1) using a fast 50 percent annual depreciation rate to calculate a rivalry stock variable; (2) normalizing the rivalry history by dividing it by the country’s total years in the sample so that the rivalry history ranges from 0 to 1; (3) using a natural log of years under threat; (4) choosing the year 1816 or 1945, instead of 1900, as a baseline year to calculate a threat stock. None of these changes alters the central findings (online appendix Tables A14 and A15).

Last, I find that results remain similar when I either attempt to use different thresholds for consolidated democracies, twenty years and thirty years, or use the full sample without excluding any consolidated democracies. Results are presented in online appendix Table A16.