Authoritarian regimes and civil–military relations: Explaining counterbalancing in autocracies

Abstract

How do autocracies structure their civil–military relations? We contend that personalist dictators are more strongly associated with counterbalancing than other authoritarian regime types. Personalists are characterized by weak institutions and narrow support bases, a lack of unifying ideologies and informal links to the ruler. They thus have strong incentives to coup-proof and, as we contend, counterbalancing seems particularly attractive. Quantitative analyses of autocratic regimes’ counterbalancing efforts since the 1960s provide support for this expectation. By showing that institutional coup-proofing significantly varies across autocratic forms of government, we contribute to the literature on comparative authoritarianism and civil–military relations.

Keywords

Autocracies civil–military relations counterbalancing coup-proofing quantitative analysis

Introduction

Coercion, crucial to autocrats’ survival, requires security organizations that can employ violence against rivals and opponents. At the same time, such bodies pose a potential threat to the incumbent owing to their organized capacity for violence (McMahon and Slantchev, 2015). Dictators thus frequently rely on a set of strategies to assert control over their security forces (Feaver, 1999; Quinlivan, 1999). The ultimate goal of these so-called coup-proofing measures is to create ties between the regime and the coercive apparatus so that the latter has a vested interest in regime survival. In Makara’s (2013: 337) words, “more than merely preventing coups, therefore, coup-proofing strategies are intended to bind the security apparatus to the ruling regime to such an extent that the two become mutually dependent on one another.” This article explores whether different autocracies seek to accomplish this in systematically different ways. In particular, we focus on how authoritarian regime types organize their security apparatus and, hence, on how they are associated with institutional coup-proofing, i.e. counterbalancing: the extent to which autocratic states fragment their security forces into multiple military and paramilitary units, which aims at enhancing the reliability and dependence of these forces in contexts where the regular army may lack such traits.

Existing cross-country work shows that democracies engage less in counterbalancing than dictatorships (Pilster and Böhmelt, 2012). Yet, as Geddes (1999: 121) emphasizes, “different kinds of authoritarianism differ from each other as much as they differ from democracy.” Most studies on comparative authoritarianism examining these regimes’ organization concentrated on differences in the existence and impact of political institutions—parties, legislatures, and elections (e.g. Gandhi, 2008; Magaloni, 2008; Svolik, 2012). But a crucial dimension of variation that remains insufficiently examined is civil–military relations and, specifically, the way in which regimes organize their security institutions. Existing research on civil–military relations in authoritarian regimes consists mostly of small-N studies (Bellin, 2004, 2012; Brooks, 2008; Greitens, 2016; Quinlivan, 1999; Talmadge, 2016), which may find it difficult to provide generalizable accounts. Additionally, quantitative studies on the determinants of counterbalancing remain surprisingly scarce and do not examine the existing heterogeneity among authoritarian regimes (Böhmelt and Clayton, 2018; Koga-Sudduth, 2017a; Pilster and Böhmelt, 2012). We thus lack a comprehensive understanding on what explains the distinct patterns of civil–military relations observed in different types of dictatorships.

How leaders organize their security forces has an impact not only on coup risk (De Bruin, 2018; Pilster and Böhmelt, 2011, 2012; Powell, 2012), but also on repression, defection, regime change, military effectiveness and the risk of internal and international conflict.¹ Consequently, understanding the determinants of counterbalancing under autocracy contributes to shedding light on autocracies’ distinct behavior and outcomes. For example, personalist regimes have been systematically shown to be more involved in (and lose) international wars and more likely to experience civil conflict as well as to be forcibly overthrown. However, scholarly accounts of these patterns focus mostly on political institutions and dynamics, including small winning coalitions and the lack of institutional constraints (Peceny and Beer, 2003; Reiter and Stam, 2003), lower audience costs (Weeks, 2008, 2014), low-skilled advisors (Frantz and Ezrow, 2011; Peceny and Beer, 2003), political exclusion and patronage (Gurses and Mason, 2010), and leaders’ background (Colgan and Weeks, 2015). Civil–military relations may be an alternative mechanism explaining personalists’ distinct behavior. Regimes with counterbalanced armies could be more prone to engage in international conflicts because disputes deepen intra-military divisions (Belkin and Schofer, 2005),² or because counterbalancing undermines their military effectiveness (Biddle and Zirkle, 1996; Brooks, 2008; Pilster and Böhmelt, 2011; Narang and Talmadge, 2017; Talmadge, 2015), making these regimes more likely to be targeted (Peceny et al., 2002). As for internal conflict, exclusion and lower military effectiveness might embolden insurgent groups (Powell, 2017; Roessler, 2011), while counterbalancing can lead to splits within the security apparatus during repression campaigns that may escalate to civil conflict (Dahl, 2016; Lutscher, 2016; Makara, 2013; McLauchlin, 2010; Nepstad, 2013). If personalists have higher levels of counterbalancing, this could explain why these regimes have a higher propensity for international and domestic conflict in the first place. However, a theoretical and empirical link between personalism and counterbalancing has not been established yet.

We posit that personalist dictatorships are associated more strongly with counterbalancing than other regime types and develop a theoretical argument that considers both incentives and constraints. First, personalist regimes have stronger incentives to coup-proof as they arise in weak-state and factionalized environments, which reflects a lack of institutional mechanisms to solve intra-regime commitment problems. Second, personalist regimes are characterized by weak and non-binding institutions and narrow support bases, informal links to the ruler, and a lack of unifying ideologies. This prevents personalists from establishing institutional ties and credible cooptation arrangements with the security forces, while it increases the need for personal protection. Third, in these contexts, counterbalancing has two advantages for personalists to establish effective ties to the security forces. On one hand, fragmentation induces a balance between units and creates structural obstacles to the coordination of the armed forces. On the other, it entails the creation of organizations with direct ties to the ruler, outside the regular command-and-control channels, recruited from support groups, and specialized in protecting the regime and its leader. Finally, as we also argue, other authoritarian regimes are less likely to invest in coup-proofing: party-based regimes employ institutional ties and unifying ideologies, military regimes rely on institutional channels to integrate different factions, and monarchies’ control is generally based on traditional legitimacy.

Using Geddes et al.’s (2014) typology of authoritarian regimes, we analyze authoritarian regimes’ association with counterbalancing since the 1960s and find that the degree of fragmentation of security forces is significantly higher in personalist regimes than in other autocracies. This is the first systematic cross-national study of counterbalancing seeking to uncover major trends across autocracies.

Previous approaches: autocracies and counterbalancing

Existing research shows that autocratic regimes engage more in counterbalancing than democracies (Pilster and Böhmelt, 2012), yet the variability across dictatorships remains underexplored. Existing approaches on autocracies offer only partial accounts that focus on one authoritarian form of government, but then overlook other regime types. A first set of works studies the role of historical legacies, structural factors, and ideology to identify which regimes fragment their militaries more. Horowitz (1985: 544–549) argues that counterbalancing is more intense in traditional monarchies where praetorian tribal units are common. Other views suggest that counterforces are more likely present in Communist party regimes, since Marxist ideology painted “professional armies as tools of the old ruling classes and constant sources of counterrevolution” (Huntington, 1957; Rice, 1992: 33f), and because some regimes kept pre-existing party militias that had been active during revolutionary or decolonization struggles (Ahram, 2011; Pachter, 1982). Alternatively, Janowitz (1977) suggests that paramilitary forces are more commonly associated with military regimes seeking to boost internal control.

A second approach claims that the configuration of the security apparatus is shaped by existing threats to the regime. Counterbalanced militaries would thus emerge in dictatorships dominated by high elite-based threats (Greitens, 2016; Talmadge, 2016). As a result, higher levels of counterbalancing would be observed in military regimes, since they are more coup-prone (Kim and Kroeger, 2017; Koga-Sudduth, 2017b; Powell, 2012). This threat-based approach has several limitations, though. First, it ignores how regimes’ characteristics shape these threats and the constraints to implement alternative coup-proofing strategies. Second, personalist regimes are often not distinguished from other non-democracies or considered to be similar to military systems (Talmadge, 2016). Yet, as Geddes et al. (2014: 152) highlight, while many regimes may be headed by military officers, some lack any elite constraints and others are collegially ruled by the military institution. This problem is also present in numerous cross-national empirical accounts of coups in which leaders’ background (or affiliation) is used to distinguish between regime types.³

Personal rule: features, ties to the regime, and counterbalancing

Contrary to the views above, we posit that personalist dictatorships are the ones more strongly associated with counterbalancing than other forms of autocratic rule. Such regimes, as opposed to organizational dictatorships, are characterized by the concentration of power in the hands of one individual (Bratton and van de Walle, 1997; Weeks, 2014), which entails that the ruler has managed to marginalize state institutions and consolidate full control over policy and appointments (Geddes, 1999, 2003). Therefore, a central feature of personalist dictatorships is the weakness (or absence) of formal political institutions regulating elite interactions and conflict. As Jackson and Rosberg (1984: 424) emphasize, “established and effective political institutions are largely absent from regimes of personal rule.” This is not only because many postcolonial states that emerged in the developing world were inherently weak and a resort to violence to settle elite conflicts always loomed in the background (Roessler, 2011), but also because “the fear of potential rivals leads such rulers to undermine these and other institutions that might serve as power bases for challenges” (Geddes, 1999: 132). Subsequently, there is no formal, institutional power-sharing and no rules regulating intra-elite interactions. Instead, personalists build and maintain a dominant coalition, usually comprising powerful individuals with access to the means of violence, based on personal, informal relationships, and discretionary material inducements (Bratton and van de Walle, 1997; Jackson and Rosberg, 1984).

With the absence of regular interactions and formal monitoring mechanisms, problems of information, uncertainty, and mutual mistrust are exacerbated (Svolik, 2012). And while leaders may purge their rivals, other members of the dominant coalition and excluded groups could want to overthrow the ruler. The political landscapes in personalist regimes can thus be characterized by an intense security dilemma between the ruler and other members of his dominant coalition (Roessler, 2016: 87ff): a surprise move permanently eliminating a player from the political scene is always possible. Owing to this commitment problem of maintaining an unstable arrangement and the lack of encompassing ideologies tying members of the dominant coalition and the larger masses to the state, personalist rulers have an incentive to exclude rival groups and rely on ascription and identity as primary indicators of loyalty. The ensuing exclusion of rivals results in support coalitions that are narrow and “more likely to be dominated by a single familial, clan, ethnic, or regional group” (Geddes, 1999: 133). Personal networks based on shared identity and patronage constitute the key source of political loyalty and social mobilization.

Ties to the regime under personalism

These features of personal rule described above, we argue, create both incentives to coup-proof as well as constraints on the leaders’ capacity to establish institutional ties with the security apparatus.

On one hand, a personalist regime’s features create strong incentives for prioritizing internal and personal security and building reliable security forces for two reasons. First, the commitment problem becomes more prevalent in the absence of formal political institutions and monitoring mechanisms. This is exacerbated because strongmen often seize power via a coup, heading minority factions within the military or society that manage to displace the incumbent coalition in the context of conflicts over state control driven by ethnic, religious, or personal rivalries. Consequently, once in power, these leaders face a high risk of being removed by other organized, rival factions from within the military.⁴ As Geddes et al. (2014: 153) stress, “in factionalized militaries riven by ethnic or personal rivalries, however, junta members cannot guarantee the behavior of subordinate officers.” Hence, as opposed to most monarchies and party regimes, respectively, these personalist leaders “inherit” a military (at least partially) controlled by other ethnic or rival groups and lack an encompassing, legitimizing ideological appeal, which creates security dilemmas and concerns about the reliability of the armed forces. Indeed, rulers that came to power through a coup are significantly more likely to face a putsch attempt by those groups that helped them into power in the first place, while the systematic exclusion of rival groups increases the risk of rebellion (Roessler, 2011, 2016). Under such circumstances then, the principal (the leader) is faced with an agent (the army) that it is not only unreliable for the provision of internal security, but that represents a threat to the survival of the leader. Second, personalism and the resulting power concentration bring about an additional security concern. The absence of institutional methods for channeling the demands of social groups and of monitoring mechanisms enhancing transparency and compliance in elite interactions leads to the ruler’s informational insulation and a more intense dictator’s dilemma (Svolik, 2012), which breed fear and paranoia.⁵ Moreover, with a single individual embodying the whole regime, such a leader logically becomes the main and obvious target of any regime overthrowing attempt by any would-be dictator or outside group. This increases the need for individual protection and gives rulers incentives to build security forces in charge of not only controlling the opposition, resisting coup-attempts, surveilling insiders, and collecting political intelligence, but also the personal protection of the ruler.

On the other hand, the personalist-regime’s features create constraints and opportunities that influence the choice of available strategies aimed at establishing effective ties with the security apparatus. These links determine the degree to which the military is interested in the regime’s survival and may take two forms: institutional and personal (Bellin, 2004, 2012). Institutional links are established through formal channels and consist of the effective and credible integration of military elites in collective decision-making bodies that monitor and enforce rent and power-sharing deals (Goldsworthy, 1981; Nepstad, 2013; Svolik, 2012). Contrary to other regime types, under personalism, power concentration and non-binding institutions prevent the establishment of such formal ties that give military elites political influence and a vested interest in regime survival (Roessler, 2011; Talmadge, 2016), while narrow, exclusive support coalitions and the lack of a unifying ideology make it difficult to align the military’s preferences and incentives with those of the regime.

Consequently, as Way and Weeks (2014: 708) highlight, personalist “leaders are induced to choose strategies of survival that emphasize rooting out enemies and disrupting coordination,” but also strategies that are based on identity and personal dependency. Counterbalancing accomplishes this, as it is a technique that establishes personal ties with the security forces, specializes in regime protection, and undermines the collective action capacity of inside rivals.

Ties to the regime under non-personal rule

Non-personal regimes’ use of security fragmentation is lower than that of personalists because they have the capacity to establish institutional ties and promote shared interests and cohesiveness between political and military elites through ideological or historical legitimacy claims. First, formal political institutions in these regimes ease the monitoring and information problems and the security dilemma that underlies them. Such institutions prescribe who may become part of the dominant coalition, how decision-making powers are distributed, integrate military elites into decision-making bodies, and alleviate commitment problems. Second, state ideologies and other, more traditional sources of legitimacy increase the costs for any political actor attempting to seize power and contribute to aligning the interests and preferences of security and political elites.

Party-based regimes rely on institutional ties and legitimizing ideologies to bind the security apparatus to the regime. Institutional links are used to credibly share power (and rents) with military elites and normally comprise a dual-elite system linking the party and the military. That is, they integrate military officers in the government and the party’s decision bodies and grant them material benefits and political access (Barany, 1991; Goldsworthy, 1981; Pachter, 1982; Perlmutter and LeoGrande, 1982). This creates mechanisms for channeling demands as well as institutionalized modes of competition and participation, which can be used to manage factional disputes. Additionally, party regimes normally rely on an encompassing ideological structure that legitimizes the preeminence, authority, and vanguard of the party and, importantly, which aligns the military’s preferences with the regime’s (Barany, 1991; Perlmutter and LeoGrande, 1982).⁶ To strengthen ideological commitment, single-party regimes seek to politicize the military by infiltrating personnel into its ranks, engaging in ideological training of officers and troops, and introducing political commissariats (Barany, 1991; Goldsworthy, 1981; Pachter, 1982; Perlmutter and LeoGrande, 1982). Finally, another argument suggests that revolutionary party regimes that emerge after revolution and state collapse and have their own revolutionary army (Segal and Phipps, 1990) do not need to resort to counterbalancing: “[w]here revolutionary forces build the armed forces from scratch, fill the officer corps with revolutionary combatants, and infuse them with a revolutionary ideology, the security forces are less likely to suffer problems of insubordination and rebellion” (Levitsky and Way, 2013: 11).

Institutional ties and beliefs play a crucial role in military regimes as well. In collegial military regimes, officer corps and military branches share power and are normally represented in a junta (Geddes, 2003; Geddes et al., 2014; Remmer, 1989). Institutional integration of different factions and branches within the military channels officers’ interests, manages factional disputes, alleviates monitoring concerns, and ensures direct participation in decision-making. Under such conditions, parallel organizations are unlikely to be created and the military command chain is not subverted, but also because unity, cohesiveness, and professionalism feature prominently among corporate military preferences (Stepan, 1971; Nordlinger, 1977).

Finally, the role of monarchy in state formation results in traditional legitimacy. This in turn sustains a more unquestionable (and often constitutionally sanctioned) direct control of the king over the military and shapes its historical mission as upholder of the monarchy (Kamrava, 2000).⁷ Furthermore, “tribally based monarchies” rely on launching tribal support networks and thus build more homogeneous militaries that have a shared identity with the monarch (Kamrava, 2000). Lastly, numerous monarchs have cultural practices and institutions that define the members of the ruling group, succession mechanisms, and methods for monitoring the ruler’s actions. These features “helped render monarchs’ promises credible, securing the political elites’ trust and loyalty in the process” (Menaldo, 2012: 711).

Although lower on average than in personalist ones, we do not contend that counterbalancing will be completely absent in other regimes. Some party regimes, especially those from the Communist tradition, maintained pre-existing party militias (Pachter, 1982). In most cases, these were inactive or remained as a reserve force. Some military juntas placed irregular repression in the hands of separate units to avoid elite tensions associated with violent coercion and the treatment of opponents (Geddes et al., 2014: 155–156). And in monarchies, where power is also highly concentrated in one individual, praetorian units for the protection of the monarch are common (Horowitz, 1985).

Counterbalancing

But why would particularly counterbalancing be an attractive strategy for personalist dictators? In contexts were institutional ties cannot be formed and preferences are difficult to align, counterbalancing becomes a crucial, structural element in the attempt to impose stable personal control on the security apparatus under patrimonial and dependent organizations. Co-optation, purges, and stacking, albeit often used in combination, do not represent sufficiently effective and long-term stand-alone strategies to bind the security forces to the leader and ensure the existence of reliable regime-protection forces. Economic co-optation—the delivery of ample military budgets, allowances, and other perquisites—aims at lowering the military’s willingness to act against the ruler (Powell, 2012, 2014). Reshufflings and purges of top officers root out enemies and undermine skilled elites’ ability to coordinate against the incumbent (Horowitz, 1985; Koga-Sudduth, 2017b).

Military spending is a strategy based on altering instrumental incentives of officers, but does not address their underlying preferences, which may flare up in times of political or economic distress or may be impossible to align in contexts characterized by factionalism and ethnic competition (McLauchlin, 2010). Additionally, non-binding institutions reduce the credibility of rent-sharing: the military cannot convincingly commit not to intervene, especially considering that higher resources boost its ability to do so (Powell, 2014; Wang, 1998); at the same time, the ruler cannot commit not to cut discretionary rents should the future balance of power tilt in his favor (Roessler, 2011; Svolik, 2012).⁸ Actually, the ruler has incentives to weaken the military, but cutting spending may trigger a violent response (Leon, 2014). Also, endemic fiscal crises caused by low state capacity and mismanagement make the flow of patronage rents uncertain, which triggers resentment among officers and troops (Dowdle, 2007).⁹

Regarding officer purges, discretionary appointments, and stacking, as Young (1994: 241) notes, “neutralizing imbalances [in military recruitment] constructed over decades without triggering the army’s intervention [is] an extraordinarily delicate task.” Similarly, Horowitz (1985: 539) emphasizes that “homogenization is highly risky” and only possible “in very unusual circumstances.” Conflicts over promotion and autonomy aggrieve the military, while promoting loyalists may alienate junior officers and does not ensure the obedience of the (resented) rank and file, all of which may provoke a preemptive coup (Harkness, 2016; Horowitz, 1985; Koga-Sudduth, 2017b; Makara, 2013; Roessler, 2011).¹⁰ Finally, given the small size of support groups, recruiting sufficiently substantial parts of the rank and file and junior officers based on kinship implies slow change and limited effectiveness in establishing immediate and independent control over the army (Horowitz, 1985: 540).

This is not to say that the strategies above and counterbalancing are substitutes. Actually, they are typically used in combination or in a gradual cumulative manner (Horowitz, 1985). Purges, biased promotions, and cutting spending undermine the military’s capacity and disrupt coordination and thus create opportunities for further consolidating power (Eck, 2015; Koga-Sudduth, 2017b; Svolik, 2012).¹¹ At the same time, counterbalancing “later enables the ethnic restructuring or disarmament of the regular army” (Harkness, 2016: 600). Yet these other strategies fall short of establishing direct and effective control of the army and, by undermining military effectiveness, still leave personal autocrats in need of reliable and loyal (counter-) organizations for internal security and personal protection.

With the creation of parallel organizations, personal leaders create a new reliable and loyal agent, minimize their reliance on a threatening preexisting agent (i.e. the regular army) and, in doing so, re-shape their threat environment. As a result, such strategy entails certain risks and tradeoffs that autocrats ponder. By having strong incentives to prioritize internal and personal security for the reasons discussed above, personal rulers weigh less heavily the reduction in the military’s external effectiveness against foreign non-existential threats (e.g. Biddle and Zirkle, 1996; Pilster and Böhmelt, 2011), and the long-term increase in the risk of rebellions that counterbalancing may bring (Roessler, 2011). Further, as De Bruin (2018) finds, by challenging the military’s interests, counterbalancing can increase the likelihood of coup attempts in the short term. Still, we posit, creating parallel and paramilitary units contributes positively to political survival for two critical reasons that we develop in turn: it entails the creation a new reliable security agent to cope with these domestic threats (coups and uprisings), with which personal and mutual dependency links can be effectively established.

First, counterbalancing entails the creation of a militarized support base made of new loyal pockets that are entrusted with the most critical security roles: defending the regime and its leader owing to the factional rivalries and security dilemmas affecting the regular armed forces.¹² Since personalization breeds insulation and fear, the first type of specialized organization typically observed in personalist regimes is praetorian units fully dedicated to the personal protection of the leader. Additionally, further creations of auxiliary forces may ensue from the need to perform other internal security tasks that entail thwarting and countering potential horizontal and vertical threats. For example, Iraq’s Special Republican Guard was tasked with protecting Saddam, Baghdad, and strategic buildings, while the Fedayeen Saddam specialized in harassing and eliminating opposition members. In Zaire, Mobutu’s Special Presidential Division was in charge of protecting him and eliminating domestic rivals while the Civil Guard was responsible for the maintenance of public security and order.

Concerning horizontal threats (i.e. coups), auxiliary forces, normally located within or near the capital, collect intelligence, monitor, and, if required, confront other civilian insiders and military units. Indeed, the existence of counterweights results in a lower probability of coup success (Böhmelt and Pilster, 2016; De Bruin, 2018; Powell, 2012) as “[a]ny military unit that intends to overthrow a current regime in power has to take into account a likely confrontation with other military and/or paramilitary organizations” (Böhmelt and Pilster, 2016: 159). Moreover, counterbalancing hinders inter-unit coordination—necessary for a successful coup—because counterbalancing “divides a country’s military into rivaling organizations, thereby creating an artificial balance between and structural obstacles for the armed forces” (Belkin and Schofer, 2003; Böhmelt and Clayton, 2018). Such balance involves parallel organizations and command structures, but also prevents any unit from getting too powerful.¹³ For example, in Iraq, besides the pre-existing Republican Guard, Saddam Hussein created the Special Republican Guard and the Fedayeen Saddam in the early 1990s, which were kept separate from and unable to coordinate with the Republican Guard (Sassoon, 2012: 151).

As for vertical threats (e.g. uprisings), parallel military forces eliminate rivals, suppress dissent, intimidate opponents, and fight insurgents. This involves performing irregular internal security roles that the regular military may not be prepared or willing to undertake. Owing to their increased mechanization and reduced infantry, regular armed forces are increasingly ill-suited for the requirements of suppressing dissent, collecting political intelligence, and even facing insurgents (Böhmelt and Clayton, 2018; Carey, et al., 2016; Lyall and Wilson, 2009). Further, the military’s sense of obligation to primarily defend the public and national interests makes it likely that regular armies refuse to carry out domestic repression duties.¹⁴ Paramilitary organizations are more capable of facing vertical threats than the regular army owing to their emphasis on mobility and specialized equipment and training (Horowitz, 1985; Janowitz, 1977; Pilster et al., 2016; Quinlivan, 1999). Certainly, violent crackdowns during protest campaigns are more likely where regimes have paramilitary forces (Koren, 2014). And qualitative studies analyzing armed forces’ responses to protest campaigns such as the Arab Spring show that it was precisely the loyal counterweights which supported their regimes and fought insurgents (Bellin, 2012; Makara, 2013; McLauchlin, 2010; Nepstad, 2013).

Second, performing these tasks above not only requires operational specialization and training, but most importantly, obedience and loyalty. As said, regime protection might require countering coup attempts and performing irregular repressive duties. Lacking a vested interest, the military may refuse to undertake those tasks if interpreted as going against their corporate identity and professionalism or if, owing to their ties to society, they carry social and moral costs (De Bruin, 2018; Nepstad, 2013; Pion-Berlin, et al., 2014; Rudbeck, et al., 2016). Hence, to secure mutual dependence, counterbalancing makes it possible for strongmen to effectively establish personal ties and mutual dependence with parallel and auxiliary units through two mechanisms. On the one hand, such new organizations are not only typically created by the dictator, but also report directly to him and are under his personal control. These new organizations are thus directly bound to the ruler and are dependent on him for their own survival, giving them strong incentives to side with and defend the regime. For example, Zaire’s Civil Guard and SPD were both created by Mobutu in the mid-1980s. Compaoré’s Regiment of Presidential Security in Burkina Faso answered directly to him, and Yemen’s Republican Guard did not report to the Ministry of Defense, but directly to the Office of the Commander-in-Chief, i.e. former President Abdullah Saleh (Knights, 2013: 273).

On the other hand, as Quinlivan (1999: 141) states, “[b]ecause the purpose of a parallel military is to protect the regime, it must be bound to the regime through special loyalties and social relationships.” This is accomplished by creating personal ties to the ruler (while reducing ties with society) and fostering distinct interests between parallel units and the regular military. First, to strengthen both capacity and loyalty, parallel units are usually better paid, trained, and equipped than regular units. Second, besides the leader’s personal control of these forces, mutual dependence is spurred by recruiting commanders and personnel of parallel organizations from support groups—typically the rulers’ own ethnic group, tribe, family, or region. These in-group connections bind more effectively the fate of these organizations with the dictator’s, giving them strong incentives for defending him.¹⁵ Counterbalancing facilitates these staffing practices because “personnel changes are easier to make in new forces than in the army, where existing positions are already filled” (De Bruin, 2018: 1438). Thus, small, elite units can be put under the command of relatives or co-ethnics and be more easily and readily stacked with supporters, especially considering that the support base of personalist regimes normally comprises one or few small ethnic or tribal groups. For instance, Mobutu appointed his nephew and his brother-in-law as commanders of the DSP and Civil Guard, respectively, while troops of both units were mostly recruited from Mobutu’s home region (Equateur) and tribe (Ngbandi). Identical patterns can be seen in Libya under Gaddafi and Iraq under Saddam.

In sum, factionalized or even hostile regular armies, weak institutions, and narrow support groups make it more likely for personalist rules, as compared with other regime types, to prioritize personal protection. Further, among the various strategies available, personalist strongmen are especially likely to resort to counterbalancing, as it makes it possible to create specialized counterweights with personal ties to the ruler. We thus hypothesize that counterbalancing is most likely to be associated with personalist regimes.

Research design

Data, dependent variable, and methodology

The empirical analysis is based on data comprising non-democratic regimes between 1960 and 2010. Authoritarian countries are defined according to Geddes et al.’s (2014; GWF henceforth) dataset, where a state has to be coded as “non-democratic” in a given year to be part of the sample. We only focus on ideal types as captured by GWF but omit countries with small populations, provisional regimes, foreign-occupied territories, warlord regimes, or non-independent countries. We combine these data with De Bruin’s (2018) information on counterbalancing in 65 randomly selected developing countries. The final dataset eventually covers 55 autocracies in 1960–2010 with varying years under study by country, leading to 1990 observations in total. There are 10 states in De Bruin (2018) that are coded as democratic in GWF and, thus, are omitted from the analysis.

The dependent variable we employ in the following main analyses is then taken from De Bruin (2018), who focuses on “counterweights” that have to meet two criteria to be identified. (1) The forces are “independent from military command. Instead, operational control, which refers to the ability to initiate and terminate military operations, rests with the executive, interior ministry, or other government body besides the defense ministry, which controls the military. (2) The force is deployed within sixty miles of the capital, which ensures it has at least the possibility of being able to intercept a coup.” This measure “excludes border guards, rural militia, and other forces deployed outside the capital to address domestic or international security threats” (De Bruin, 2018: 1453). The two main advantages of De Bruin’s data are that uniform criteria for including forces in the data are employed and that there are details on the command and deployment of each security organization, which allows for distinguishing between units created for counterbalancing and those that are not. Moreover, another asset is “the triangulation of over 1,200 primary and secondary sources, which reduces measurement error” (De Bruin, 2018: 1435).

We take the item Counterbalancing, which constitutes a binary variable receiving the value of 1 if the number of counterweights is above the average in a given country-year, i.e. above the sample mean of approximately 1.3 counterweights (0 otherwise). In De Bruin’s (2018: 1441) words, “because coups can be staged (and prevented) by even a small number of men, we should expect the logic of counterbalancing to hold no matter the size of the counterweight.” Hence, a dichotomous indicator is sufficient for the test of our hypothesis. A binary variable is further justified to minimize the potential for bias related to measurement error. Finally, we do not expect counterbalancing to be non-existent in other regime types. For different reasons discussed above, other forms of government also fragment their security apparatuses, but not to the same extent as personalists, which, then, are expected to be above average.

We employ logistic regression models and cluster the standard errors by country to account for intra-group correlations. Temporal dependencies are addressed using Carter and Signorino’s (2010) cubic-polynomial approximation, which is based on the time elapsed since the last counterbalancing investment (coded as 1 by our outcome item). We also take regional influences in some of our models into account as determinants of the “optimal size” of a country’s counterbalancing, e.g. coup-proofing at home might be influenced by the neighbors in the region (Böhmelt et al., 2017). In order to capture these regional influences, some models include region-specific effects. The regions are defined by the Correlates of War Project (Stinnett et al., 2002).

Explanatory variables

To capture or main independent variables, we use GWF’s classification of regime types: (1) single-party regimes, (2) military juntas, (3) monarchies, and (4) personalist dictatorships. This coding focuses on the organization or institutional aspects of regimes, formal and informal ones, to identify personalist regimes and distinguish them from other authoritarian forms of government, which makes these data ideal for our purposes.¹⁶ GWF (2014: 314) highlight that their “definition of regimes emphasizes the rules that identify the group from which leaders can come and determine who influences leadership choice and policy.” Moreover, note that single-party regimes comprise any party-based autocracy if the party-criterion is the dominant one: party-personal, party-military, party-personal-military, oligarchy, and Iran in 1979–2010 are all coded as party-based. Similarly, the military-junta category groups military, military-personal, and indirect military rule. In light of these specifications, our sample selection criteria, and after accounting for missing values, we have information on 55 autocracies (1990 autocracy-years) in 1960–2010. We present models that have Personalist Regime as the excluded reference category and estimations that only include Personalist Regime (with all other authoritarian regimes as the baseline).

We include a number of control variables that may either affect our dependent variable or are correlated with the regime types. These are the primary selection criteria for our confounding variables, while explicitly trying to avoid post-treatment bias.¹⁷ First, we focus on different security challenges (domestic or international), which may influence the degree of counterbalancing as well as the likelihood of regime change. On one hand, countries with a history of international-level threats are likely to shift their attention away from the domestic environment (Pilster and Böhmelt, 2012). On the other hand, while external threats lower the focus on domestic-level security instruments, internal threats often induce a higher investment in them. Potential domestic threats such as civil conflict are unlikely to be an exception here (Pilster and Böhmelt, 2012). To this end, we consider two items. First, using Allansson et al. (2017) and Gleditsch et al. (2002), we include a binary variable on civil conflict incidence (i.e. onset year and ongoing conflict years are coded as 1). Similarly, using Palmer et al. (2015), we coded a dichotomous item capturing a state’s involvement in a militarized interstate dispute in a given country-year.

Further, using time-invariant data from Fearon and Laitin (2003), we incorporate variables on ethno-linguistic and religious fractionalization as well as a measure of oil exports. The latter is a binary variable receiving a value of 1 if fuel exports exceed one-third of a country’s export revenue. As Fearon and Laitin (2003: 81) state, “[o]il producers tend to have weaker state apparatuses than one would expect given their level of income because the rulers have less need for a socially intrusive and elaborate bureaucratic system to raise revenues.”Ethnic Fractionalization and Religious Fractionalization both range in [0; 1] and indicate the likelihood that two randomly drawn individuals in a country are from different ethnolinguistic or religious groups, respectively. These variables are thus supposed to capture threats that “influence the probability that domestic insurgents will challenge the current regime in power” (Pilster and Böhmelt, 2012: 363). In turn, this could affect the likelihood of counterbalancing in a country as well as that of observing one particular authoritarian form of government. However, these are indirect determinants and operationalized in a time-invariant way, making it unlikely that post-treatment bias is an issue.

Empirical findings

Table 1 reports the coefficients of two different specifications: Model 1 merely includes the regime typologies and temporal controls. Model 2 adds the region fixed effects. Model 3 is the naïve estimation as we only consider the control variables, region fixed effects, and temporal corrects. Finally, Model 4 is the main model and includes all right-hand-side variables introduced in the previous section. Personalist Regime is left out as the baseline category in all models. Table 2 mirrors the setup in Table 1 except that we replace the regime typologies by Personalist Regime and thus compare this autocratic form of government to all other typologies. The entries in both Tables 1 and 2 are logit regression coefficients. To facilitate the interpretation of our main variables of interest, Figure 1 shows first differences for each regime dummy, i.e. changes in the probability of Counterbalancing=1 when moving from the reference category to the focal regime typology, while holding all other covariates constant at their median values.

Table 1.

Counterbalancing across authoritarian regime types.

	Model 1	Model 2	Model 3	Model 4
Single-party Regime	−0.627	−0.681		−0.673
	(0.349)*	(0.375)*		(0.402)*
Military Regime	−1.634	−1.512		−1.496
	(0.581)***	(0.680)**		(0.687)**
Monarchy	0.156	−0.602		−0.499
	(0.635)	(0.816)		(0.765)
Ethnic Fractionalization			−0.327	−0.128
			(0.818)	(0.814)
Religious Fractionalization			0.886	0.674
			(1.092)	(1.099)
Oil			−0.298	−0.366
			(0.381)	(0.349)
Civil Conflict			0.478	0.438
			(0.364)	(0.405)
Militarized Interstate Dispute			−0.127	−0.103
			(0.197)	(0.193)
Counterbalancing Years	−1.704	−1.636	−1.653	−1.623
	(0.161)***	(0.153)***	(0.152)***	(0.149)***
Counterbalancing Years²	0.096	0.091	0.091	0.090
	(0.013)***	(0.012)***	(0.011)***	(0.011)***
Counterbalancing Years³	−0.001	−0.001	−0.001	−0.001
	(0.000)***	(0.000)***	(0.000)***	(0.000)**
Constant	2.417	2.266	1.706	2.050
	(0.274)***	(0.380)***	(0.652)***	(0.622)***
Observations	1,990	1,810	1, 810	1, 810
Region Fixed Effects	No	Yes	Yes	Yes
Wald χ²	252.77	247.64	245.63	275.65
Prob > χ²	0.000	0.000	0.000	0.000

Note: Table entries are logistic regression coefficients; standard errors clustered on country in parentheses; region fixed effects included, but not reported.

Significant at 10%; ** significant at 5%; *** significant at 1% (two-tailed).

Table 2.

Counterbalancing in personalist regimes.

	Model 5	Model 6	Model 7
Personalist Regime	0.660	0.816	0.820
	(0.345)*	(0.400)**	(0.410)**
Ethnic Fractionalization			−0.320
			(0.765)
Religious Fractionalization			0.738
			(1.107)
Oil			−0.421
			(0.365)
Civil Conflict			0.375
			(0.374)
Militarized Interstate Dispute			−0.095
			(0.188)
Counterbalancing Years	−1.736	−1.640	−1.630
	(0.168)***	(0.150)***	(0.148)***
Counterbalancing Years²	0.098	0.091	0.090
	(0.013)***	(0.011)***	(0.011)***
Counterbalancing Years³	−0.001	−0.001	−0.001
	(0.000)***	(0.000)***	(0.000)**
Constant	1.778	1.522	1.383
	(0.246)***	(0.348)***	(0.742)*
Observations	1,990	1,810	1, 810
Region Fixed Effects	No	Yes	Yes
Wald χ²	248.54	211.11	241.30
Prob > χ²	0.000	0.000	0.000

Note: Table entries are logistic regression coefficients; standard errors clustered on country in parentheses; region fixed effects included, but not reported.

Significant at 10%; ** significant at 5%; *** significant at 1% (two-tailed).

Figure 1.

First difference estimates for Regime Typologies.

Table 1 stresses that significant differences in the likelihood of counterbalancing exist for the comparison between Single-party Regime and Personalist Regime as well as Military Regime and Personalist Regime. While Monarchy mostly has a negative effect, it is statistically insignificant in Table 1. Table 2 shows that personalist dictatorships differ from non-personalist autocracies significantly, hence supporting our main hypothesis. Specifically, regardless of what variables are considered next to Personalist Regime, this item is consistently positively signed and statistically significant at conventional levels. Hence, personalism is strongly linked to counterbalancing, and the association of this form of government with this structural coup-proofing strategy differs from most other authoritarian regime types in that the likelihood to counterbalance is most strongly pronounced under personalism.

In substantive terms, the chances of counterbalancing in single-party autocracies are lower by 6.26 percentage points than in personalist dictatorships (Figure 1, based on Model 4). This effect is statistically significant as the 90% confidence interval of that first difference estimate [−0.148; −0.002] does not cross 0. We calculated these substantive quantities of interest while holding all other variables constant at their median values along the lines of King et al. (2000). The difference between military juntas and personalist regimes is even larger: the first difference estimate is −0.100 (confidence interval in [−0.201; −0.026]). The confidence intervals of Monarchy and Personalist Regime not only overlap according to Figure 1, but also the confidence interval ([−0.160; 0.099]) of Monarchy crosses 0. As a result, there is no statistical difference between (1) monarchies and personalist dictatorships.¹⁸ In sum, though, the difference between single-party and personalist regimes is statistically significant and the direction of the effect also mirrors our theoretical expectations: personalist dictatorships invest significantly more in counterbalancing than single-party regimes, which are indistinguishable from monarchies or military regimes here. When focusing on Personalist Regime in Model 7 exclusively, the first difference estimate is quite strongly given with a 7.583 percentage-points higher likelihood of counterbalancing than in other forms of authoritarian rule.

Concerning our control variables, none of the substantive predictors is significant at conventional levels. That said, in some of our robustness checks in the Online Appendix, Civil Conflict is positively associated with Counterbalancing at conventional levels. This mirrors other research that reports mixed findings on the relationship between civil conflict and coup attempts (Bell and Koga-Sudduth, 2017) or coup-proofing (Powell, 2017). However, there are interesting spatial and temporal dependencies in our analyses captured by the temporal controls and region fixed effects, respectively. Regarding the former, the likelihood of counterbalancing first decreases sharply with the time elapsed since the last (above average) counterbalancing investment, approximating 0% after about 8 years. The likelihood of counterbalancing remains stable at this level until about 20 years since the last counterbalancing investment, when it increases to about 10% again by year 30. Regarding the latter, the region fixed effects suggest that strong spatial patterns do exist: North Africa and the Middle East are consistently positively and significantly associated with counterbalancing, highlighting that it is particularly states there that invest in counterbalancing.

To assess the robustness of our findings, we also conducted a series of additional analyses reported in the Online Appendix. Here, we briefly summarize the most important ones. First, to consider out-of-sample heuristics, we conducted a 4-fold cross-validation exercise, which we repeated 10 times for Model 7 and then the same model omitting Personalist Regime. We show that our main predictor not only has statistical power, but also predictive power. Second, the sample we use for the analyses above is based on GWF as countries have to be identified as “non-democratic” in a given year. Yet some of the country years we include are actually coded as democratic by other datasets. Employing one of them, we omit country-years that are considered democratic there and include a control that measures the democracy score to consider any remaining bias. Third, we also assess the possibility that regime duration influences counterbalancing. While our main results hold, longer regime durations are positively associated with Counterbalancing. We also report models including further controls capturing the degree of institutionalization of the regime and past coup history. Fourth, we replaced the region fixed effects by country fixed-effects. Dummy variables for the countries in our sample can control for unobserved, time-invariant unit-level effects. We have conducted such an analysis using both De Bruin’s (2018) measure and an alternative outcome variable based on Pilster and Böhmelt (2011, 2012).

Conclusion

Differences in organizational practices in security across non-democracies have not received sufficient attention. Consistent with our theoretical expectations, we find that personalist dictators are more likely to fragment the security forces than other non-democracies. This research makes several important contributions to the literature on comparative authoritarianism and personal rule. First, we further our understanding of civil–military relations in autocracies by highlighting that different types of ties between the security apparatus and the regime may be established under different regime types. Second, identifying a link between personalism and counterbalancing contributes to improving our understanding of the distinct behavior and political outcomes of these regimes. Personalist regimes—jointly with party-based regimes—are the most prevalent form of non-democracy, representing over 40% of existing autocracies. Personalist regimes have been shown to be more conflict-prone and to produce worse economic outcomes. Thus, our findings can have policy implications by contributing to identifying early warnings of personalization. The creation of parallel units and a more fragmented security apparatus is a characteristic mostly linked to personal rule, which may deeply influence these governments’ propensity for international and domestic conflict, and even the pursuit of weapons of mass destruction.

Several avenues for further research exist. Our theoretical claims may be used as a baseline to more thoroughly study the interrelations between coup-proofing measures as well as their timing and intensity. Second, given the different patterns of investment in coup-proofing, our results inform the debate on autocracies’ military effectiveness or the “autocratic peace.” Third, the limitations of this work could set the ground for a research agenda to move forward on the issues that currently prove difficult to address. For example, the theory highlights several avenues for the strong link between personalism and counterbalancing. However, the quantitative research design cannot shed light on which mechanism is the crucial one, the driving force behind the observed macro-level relationship, although knowing this may be important from a policy and scholastic perspective. Qualitative work may thus want to address the micro-foundations of our arguments in a more nuanced way than our research design permits. While the theory comprises several illustrations, more detailed and focused qualitative work analyzing how strongmen emerge and gradually shape their security forces is needed. Finally, our statistical models address associations between variables, not causation. While it is beyond the scope of this paper to successfully link macro-level theories to individual-level assessments of motives and preferences in a causal way, a detailed empirical analysis seeking to address thoroughly the underlying causal mechanisms could further improve our knowledge of the relationship between personalism, authoritarian regimes, and coup-proofing tactics.

Supplemental Material

CMP836285_Online_Appendix – Supplemental material for Authoritarian regimes and civil–military relations: Explaining counterbalancing in autocracies

Supplemental material, CMP836285_Online_Appendix for Authoritarian regimes and civil–military relations: Explaining counterbalancing in autocracies by Abel Escribà-Folch, Tobias Böhmelt and Ulrich Pilster in Conflict Management and Peace Science

Footnotes

Acknowledgements

We thank the journal’s editors, Caroline Hartzell, Glenn Palmer, and Idean Salehyan, and the anonymous reviewers, as well as participants at the workshop “The International Dimension of Vertical threats and Regime Security in Authoritarian Regimes” (University of Heidelberg) for useful comments on earlier drafts.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

ORCID iD

Tobias Böhmelt

Supplemental material

All data, replication materials, and instructions regarding analytical materials upon which published claims rely, along with the Online Appendix, are available online through the SAGE CMPS website:

Notes

References

Ahram

(2011) Proxy Warriors: The Rise and Fall of State-sponsored Militias. Stanford, CA: Stanford University Press.

Allansson

Melander

Themnér

(2017) Organized violence, 1989–2016. Journal of Peace Research 54(4): 574–587.

Arbatli

(2016) External threats and political survival: Can dispute involvement deter coup attempts? Conflict Management and Peace Science 33(2): 115–152.

Barany

(1991) Civil–military relations in communist systems: Western models revisited. Journal of Political and Military Sociology 19(1): 75–99.

Belkin

Schofer

(2003) Toward a structural understanding of coup risk. Journal of Conflict Resolution 47(5): 594–620.

Belkin

Schofer

(2005) Coup risk, counterbalancing, and international conflict. Security Studies 14(1): 140–177.

Bell

Koga-Sudduth

(2017) The causes and outcomes of coup during civil war. Journal of Conflict Resolution 61(7): 1432–1455.

Bellin

(2004) The robustness of authoritarianism in the Middle East: Exceptionalism in comparative perspective. Comparative Politics 36(2): 139–157.

Bellin

(2012) Reconsidering the robustness of authoritarianism in the Middle East: Lessons from the Arab Spring. Comparative Politics 44(2): 127–149.

10.

Biddle

Zirkle

(1996) Technology, civil–military relations, and warfare in the developing world. Journal of Strategic Studies 19(2): 171–212.

11.

Böhmelt

Clayton

(2018) Auxiliary force structure: Paramilitary forces and pro-government militias. Comparative Political Studies 51(2): 197–237.

12.

Böhmelt

Pilster

(2016) The impact of institutional coup-proofing on coup attempts and coup outcomes. International Interactions 41(1): 158–82.

13.

Böhmelt

Ruggeri

Pilster

(2017) Counterbalancing, spatial dependence, and peer-group effects. Political Science Research and Methods 5(2): 221–239.

14.

Bratton

van

Walle

(1997) Democratic Experiments in Africa: Regime Transitions in Comparative Perspective. New York: Cambridge University Press.

15.

Brooks

(2008) Shaping Strategy: The Civil–Military Politics of Strategic Assessment. Princeton, NJ: Princeton University Press.

16.

Brown

Fariss

McMahon

(2016) Recouping after coup-proofing: Compromised military effectiveness and strategic substitution. International Interactions 42(1): 1–30.

17.

Carey

Colaresi

Mitchell

(2016) Risk mitigation, regime security, and militias: Beyond coup-proofing. International Studies Quarterly 60(1): 59–72.

18.

Carter

Signorino

(2010) Back to the future: Modeling time dependence in binary data. Political Analysis 18(3): 271–292.

19.

Colgan

Weeks

(2015) Revolution, personalist dictatorships, and international conflict. International Organization 69(1): 163–194.

20.

Dahl

(2016) Military Defection. PRIO Policy Brief 18.

21.

De Bruin

(2018) Preventing coups d’état: How counterbalancing works. Journal of Conflict Resolution 62(7): 1433–1458.

22.

Dowdle

(2007) Civil wars, international conflicts and other determinants of paramilitary strength in Sub-Saharan Africa. Small Wars & Insurgencies 18(2): 161–174.

23.

Eck

(2015) Repression by proxy: How military purges and insurgency impact the delegation of coercion. Journal of Conflict Resolution 59(5): 924–946.

24.

Fearon

Laitin

(2003) Ethnicity, insurgency, and civil war. American Political Science Review 97(1): 75–90.

25.

Feaver

(1999) Civil–military relations. Annual Review of Political Science 2(1): 211–241.

26.

Frantz

Ezrow

(2011) The Politics of Dictatorship: Institutions and Outcomes in Authoritarian Regimes. Boulder, CO: Lynne Rienner.

27.

Gandhi

(2008) Political Institutions under Dictatorship. New York: Cambridge University Press.

28.

Geddes

(1999) What do we know about democratization after twenty years? Annual Review of Political Science 2(1): 115–144.

29.

Geddes

(2003) Paradigms and Sand Castles. Ann Arbor, MI: University of Michigan Press.

30.

Geddes

Wright

Frantz

(2014) Autocratic breakdown and regime transitions: A new data set. Perspectives on Politics 14(1): 313–331.

31.

Gleditsch

Wallensteen

Eriksson

Sollenberg

Strand

(2002) Armed conflict 1946–2001: A new dataset. Journal of Peace Research 39(5): 615–637.

32.

Glentworth

Hancock

(1973) Obote and Amin: Change and continuity in modern Uganda politics. African Affairs 72(288): 237–255.

33.

Goldsworthy

(1981) Civilian control of the military in Black Africa. African Affairs 80(318): 49–74.

34.

Greitens

(2016) Dictators and their Secret Police: Coercive Institutions and State Violence. New York: Cambridge University Press.

35.

Gurses

Mason

(2010) Weak states, regime types, and civil war. Civil Wars 12(1–2): 140–155.

36.

Harkness

(2016) The ethnic army and the state: Explaining coup traps and the difficulties of democratization in Africa. Journal of Conflict Resolution 60(4): 587–616.

37.

Horowitz

(1985) Ethnic Groups in Conflict. Berkeley, CA: University of California Press.

38.

Huntington

(1957) The Soldier and the State: The Theory and Politics of Civil–Military Relations. Cambridge, MA: Belknap Press of Harvard University.

39.

Jackson

Rosberg

(1984) Personal rule: Theory and practice in Africa. Comparative Politics 16(4): 421–42.

40.

Janowitz

(1977) Military Institutions and Coercion in the Developing Nations. Chicago, IL: University of Chicago Press.

41.

Kamrava

(2000) Military professionalization and civil–military relations in the Middle East. Political Science Quarterly 115(1): 67–92.

42.

Kim

Kroeger

(2008) Regime and leader instability under two forms of military rule. Comparative Political Studies 51(1): 3–37.

43.

King

Tomz

Wittenberg

(2000) Making the most of statistical analyses: Improving interpretation and presentation. American Journal of Political Science 44(2): 347–361.

44.

Knights

(2013) The military role in Yemen’s protests: Civil–military relations in the tribal republic. Journal of Strategic Studies 36(2): 261–288.

45.

Koga-Sudduth

(2017a) Coup risk, coup-proofing and leader survival. Journal of Peace Research 54(1): 3–15.

46.

Koga-Sudduth

(2017b) Strategic logic of elite purges in dictatorships. Comparative Political Studies 50(13): 1768–1801.

47.

Koren

(2014) Military structure, civil disobedience, and military violence. Terrorism and Political Violence 26(4): 688–712.

48.

Leon

(2014) Loyalty for sale? Military spending and coups d’état. Public Choice 159(3–4): 363–383.

49.

Levitsky

Way

(2013) The durability of revolutionary regimes. Journal of Democracy 24(3): 5–17.

50.

Lutscher

(2016) The more fragmented the better? The impact of armed forces structure on defection during nonviolent popular uprisings. International Interactions 42(6): 350–75.

51.

Lyall

Wilson

(2009) Rage against the machines: Explaining outcomes in counterinsurgency wars. International Organization 63(1): 67–106.

52.

Magaloni

(2008) Credible power-sharing and the longevity of authoritarian rule. Comparative Political Studies 41(4–5): 715–741.

53.

Makara

(2013) Coup-proofing, military defection, and the Arab Spring. Democracy and Security 9(4): 334–59.

54.

McLauchlin

(2010) Loyalty strategies and military defection in rebellion. Comparative Politics 42(3): 333–350.

55.

McMahon

Slantchev

(2015) The guardianship dilemma: Regime security through and from the armed forces. American Political Science Review 109(2): 297–313.

56.

Menaldo

(2012) The Middle East and North Africa’s resilient monarchs. Journal of Politics 74(3): 707–722.

57.

Narang

Talmadge

(2017) Civil–military pathologies and defeat in war: Tests using new data. Journal of Conflict Resolution, in press.

58.

Nepstad

(2013) Mutiny and nonviolence in the Arab Spring: Exploring military defections and loyalty in Egypt, Bahrain, and Syria. Journal of Peace Research 50(3): 337–349.

59.

Nordlinger

(1977) Soldiers in Politics: Military Coups and Governments. Englewood Cliffs, NJ: Prentice Hall.

60.

Pachter

(1982) Contra-coup: Civilian control of the military in Guinea, Tanzania, and Mozambique. Journal of Modern African Studies 20(4): 595–612.

61.

Palmer

D’Orazio

Kenwick

Lane

(2015) The MID4 Dataset, 2002–2010: Procedures, coding rules and description. Conflict Management and Peace Science 32(2): 222–242.

62.

Peceny

Beer

(2003) Peaceful parties and puzzling personalists. American Political Science Review 97(2): 339–342.

63.

Peceny

Beer

Sanchez-Terry

(2002) Dictatorial peace? American Political Science Review 96(1): 15–26.

64.

Perlmutter

LeoGrande

(1982) The party in uniform: Toward a theory of civil–military relations in communist political systems. American Political Science Review 76(4): 778–789.

65.

Pilster

Böhmelt

(2011) Coup-proofing and military effectiveness in interstate wars, 1967–1999. Conflict Management and Peace Science 28(4): 331–350.

66.

Pilster

Böhmelt

(2012) Do democracies engage less in coup-proofing? On the relationship between regime type and civil–military relations. Foreign Policy Analysis 8(4): 355–72.

67.

Pilster

Böhmelt

Tago

(2016) The differentiation of security forces and the onset of genocidal violence. Armed Forces & Society 42(1): 26–50.

68.

Pion-Berlin

Esparza

Grisham

(2014) Staying quartered: Civilian uprisings and military disobedience in the twenty-first century. Comparative Political Studies 47(2): 230–259.

69.

Piplani

Talmadge

(2016) When war helps civil–military relations: Prolonged interstate conflict and the reduced risk of coups. Journal of Conflict Resolution 60(8): 1368–1394.

70.

Powell

(2012) Determinants of attempting and outcome of coups d’état. Journal of Conflict Resolution 56(6): 1017–1040.

71.

Powell

(2014) Regime vulnerability and the diversionary threat of force. Journal of Conflict Resolution 58(1): 169–96.

72.

Powell

(2017) Leader survival strategies and the onset of civil conflict: A coup-proofing paradox. Armed Forces & Society, in press.

73.

Quinlivan

(1999) Coup-proofing: Its practice and consequences in the Middle East. International Security 24(2): 131–165.

74.

Reiter

Stam

(2003) Identifying the culprit: Democracy, dictatorship, and dispute initiation. American Political Science Review 97(2): 333–337.

75.

Remmer

(1989) Military Rule in Latin America. New York: Unwin Hymen.

76.

Rice

(1992) Problems of postcommunism: The military under democracy. Journal of Democracy 3(2): 27–42.

77.

Roessler

(2011) The enemy within: Personal rule, coups, and civil war in Africa. World Politics 63(2): 300–346.

78.

Roessler

(2016) Ethnic Politics and State Power in Africa. The Logic of the Coup-Civil War Trap. New York: Cambridge University Press.

79.

Rudbeck

Mukherjee

Nelson

(2016) When autocratic regimes are cheap and play dirty: The transaction costs of repression in South Africa, Kenya, and Egypt. Comparative Politics 48(2): 147–166.

80.

Sassoon

(2012) Saddam Hussein’s Ba’th Party: Inside an Authoritarian Regime. New York: Cambridge University Press.

81.

Segal

Phipps

(1990) Why communist armies defend their parties. Asian Survey 30(10): 959–976.

82.

Stearns

(2011) Dancing in the Glory of Monsters. New York: Public Affairs.

83.

Stepan

(1971) The Military in Politics: Changing Patterns in Brazil. Princeton, NJ: Princeton University Press.

84.

Stinnett

Tir

Schafer

Diehl

Gochman

(2002) The Correlates of War Project Direct Contiguity Data, version 3. Conflict Management and Peace Science 19(2): 58–66.

85.

Svolik

(2012) The Politics of Authoritarian Rule. New York: Cambridge University Press.

86.

Talmadge

(2015) The Dictator’s Army: Battlefield Effectiveness in Authoritarian Regimes. Ithaca, NY: Cornell University Press.

87.

Talmadge

(2016) Different threats, different militaries: Explaining organizational practices in authoritarian armies. Security Studies 25(1): 111–141.

88.

Thompson

(1998) The Marcos Regime in the Philippines. In: Chehabi

Linz

(eds) Sultanistic Regimes. Baltimore, MD: The Johns Hopkins University Press, pp. 206–229.

89.

Wang

(1998) Arms transfers and coups d’état: A study on Sub-Saharan Africa. Journal of Peace Research 35(6): 659–675.

90.

Way

Weeks

(2014) Making it personal: Regime type and nuclear proliferation. American Journal of Political Science 58(3): 705–719.

91.

Weeks

(2008) Autocratic audience costs: Regime type and signaling resolve. International Organization 62(1): 35–64.

92.

Weeks

(2014) Dictators at War and Peace. Ithaca, NY: Cornell University Press.

93.

Wrong

(2002) In the Footsteps of Mr. Kurtz. New York: Harper Collins.

94.

Young

(1994) Democratization in Africa: The contradictions of a political imperative. In: Widner

(ed.) Economic Change and Political Liberalization in Sub-Saharan Africa. Baltimore, MD: The Johns Hopkins University Press, pp. 230–250.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.31 MB