Mass revolts against authoritarian regimes produce highly uncertain and complex environments. It is often unclear to demonstrators and regime incumbents alike whether protests will continue to grow, how the regime will respond, and what results mass protests will bring about. Moreover, popular unrest does not inevitably lead to democratic or liberal outcomes, even when incumbents fall and regime transitions do unfold. Footnote 1
Indeed, a few years into the most recent wave of popular uprisings—the Arab Spring—studying regime trajectories in countries such as Tunisia, Egypt, Libya, Syria, and Yemen still seems like shooting at a moving target. Yet it is safe to say that pacted, democratic transition is not the norm of regime change in that region. Quite to the contrary, contemporary politics in the Middle East presents us with a laboratory of sorts for the study of political transformation beyond democratic transition and a messy picture emerges: authoritarian resilience (Bahrain), authoritarian recalibration (Egypt), and uncertain democratization (Tunisia). As the cases of Syria, Yemen, and Libya show, mass uprisings and their trajectories do not only produce uncertain transitions, but also failing states and violent domestic conflict. Understanding trajectories of popular uprisings is important not only from a scholarly perspective, but also for anticipating and mitigating the very practical humanitarian and security challenges posed by transitions gone wrong.
What explains the stark variation of regime transition and conflict trajectories in a region where countries share political, economic, and cultural commonalities and where similar mass protests erupted? It is intuitively compelling to recall that the military is a central actor in violent conflicts, including those that emerged from Arab Spring uprisings. But—with the exception of Eva Bellin, Milan Svolik, and a few others—scholars have underestimated the military as an authoritarian institution that is decisive to an autocrat’s survival under imminent threat. Footnote 2 Prior to the 2011 Arab uprisings, regimes in the Middle East established themselves as coercive states, the incumbents of which rested upon the support of large-scale military and “high-scope” security apparatuses. Footnote 3 We hold that the military’s response to popular mass uprisings is not only central to an incumbent’s political survival, but also to the pathways of political transition and violent conflict.
In fact, the military has not been absent from the literature on regime transitions more broadly, as is made clear by extensive studies on coups d’état. Scholars of civil-military relations, such as Alfred Stepan, Felipe Aguero, and Michael Desch, have also studied the military’s “return to the barracks” during democratic consolidation, that is, at a time when democratic transition is already well on its way. Footnote 4 The military’s interaction with civil unrest has been less explored, perhaps with the exception of works by D.E.H. Russell and, more recently, Terrence Lee and David Pion-Berlin with collaborators. Footnote 5 What these works share, however, is a tendency to view the military as an institution acting coherently, either as part and parcel of a ruling regime or a revolutionary actor working against it.
We are concerned here with military behavior at the time when an anti-authoritarian uprising takes place, that is, at the crucial moment of potential authoritarian breakdown rather than an ongoing democratic transition. We suggest a new way of thinking about how the military shapes the trajectory of popular contestation, and authoritarian regime breakdown or recalibration during unrest, which has largely gone unnoticed in the literatures on civil-military relations and authoritarian regime resilience. During such critical moments, autocrats do not rely on the military as an institution, but rather on a diverse set of actors wielding coercive power. When these military actors step into the limelight they face a critical choice. In reference to Albert Hirschman’s language, we see military agents as deciding to exit the regime by remaining in the barracks or going into exile; choosing to resist by fighting for the challenger or initiating a coup d’état; or remaining loyal to the regime and using force to defend it. Footnote 6 The regime’s survival, the efficacy of mass mobilization, and the prospects of domestic violent conflict hinge on patterns in how these military agents respond to orders to repress domestic unrest.
We argue here that existing studies of civil-military relations in authoritarian regimes fail to fully explain the dynamics of military behavior during social unrest. In response to these empirical puzzles and theoretical challenges, we propose to account for both structure and agency in military organization. Accounting for the nature of military hierarchy and control helps explain the military’s degree of cohesion, but disaggregating “the” military and parsing the interests of different agents in that apparatus is crucial for explaining exit, resistance, and loyalty patterns at the start of an uprising and as it continues. In short, we argue that higher officers are not only military personnel, but also political elite members; decisions regarding whether to be loyal or not hinge on their expectations of regime change and eventually their capacities to dislodge embattled incumbents. Soldiers, in contrast, look at military service as a job—but one in which hierarchical control impedes exit or resistance when opportunities of desertion increase.
Based on substantial field research on military behavior during unrest in three countries that saw only scant attention in treatments of civil-military relations—Syria, Bahrain, and Yemen in 2011—we present an empirical narrative for patterns of loyalty and defection that not only explain different degrees of cohesion in militaries that share similarities in organizational features, but also different conflict trajectories. Although the armies in all three countries were distinguished by patrimonialism, we did not witness a uniform rallying-around-the-flag behind embattled incumbents, but rather very different loyalty patterns—and hence different conflict trajectories. Our aim here is theoretical advancement in the literatures of authoritarian regime change and civil-military relations, rather than theory testing. Studying individual cases and process tracing has proven particularly valuable in thematic areas where empirical phenomena have been underexplored and theoretical development limited. Footnote 7
In the remainder of this article, we situate our argument in the literatures on civil-military relations—more broadly and with a reference to the military’s role in the Arab Spring—before establishing our theoretical contribution. We suggest analyzing the interests and capacities of different agents within the military hierarchy, namely commanding officers and rank-and-file soldiers, accounts for loyalty dynamics at the start of an uprising and as it unfolds. After reflecting on our empirical material and logic of case selection, the three following sections discuss evidence from Bahrain, Yemen, and Syria to provide support for our broader argument. The conclusion summarizes our argument, reflects on how our framework sheds light on military responses to current uprisings, and suggests implications for further research in related fields of political science.
Theorizing Military Behavior during Popular Unrest
The events in the Middle East and North Africa in 2011 have triggered a new surge in military studies among experts of both the region and mass mobilization. Scholars such as Theodore McLauchlin, Eva Bellin, Zoltan Barany, Derek Lutterbeck, and Michael Makara have turned to existing studies in civil-military relations to hypothesize that organization-level variables influence military actions. Footnote 8 A common denominator of these early analyses is to use traits of military organization and recruitment to explain loyalty or insubordination. A military’s patrimonial nature versus institutionalization, or cohesion versus fragmentation, is said to affect the coercive apparatus’s support for an autocrat during unrest. Experts of Middle Eastern military politics have identified armies in Syria, Yemen, Bahrain, and Libya as “communal” or “patrimonial” armies, while the militaries in two other Arab Spring cases—Egypt and Tunisia—are considered “institutionalized” forces, owing to their professional organization and recruitment mechanisms.
It has been argued that a coercive apparatus is more likely to withhold support during crisis when it is sufficiently institutionalized. Applying these explanations to institutionalized armies convincingly explains the 2011 coup d’état in Egypt and the Tunisian military’s reluctance to defend Ben Ali. These militaries had substantial corporate interests and, in Hicham Bou Nassif’s term, a “republican ethos” that not only resulted in the military’s rapid disaffection with an incumbent’s deployment order, but also ensured collective action against that incumbent. Footnote 9 In contrast, communal armed forces were described as employing personal loyalties to tie military men to authoritarian rulers, be they religious (Syria, Bahrain), tribal (Yemen, Saudi Arabia, Libya), or ideological (Iran) ties. Since conflicts would reinforce existing communal divides and because commanders and their subordinates receive “selective favoritism and discretionary patronage,” Footnote 10 patrimonial militaries would rally around the flag and support embattled incumbents. Indeed, Richard Snyder suggested that lack of military autonomy ensures regime dependence and loyalty, Footnote 11 and more recent research on civil war and post-conflict state-building, such as by Lee Seymour and Jesse Driscoll, has shown how patronage relations within communal military organizations influence loyalties. Footnote 12
Yet a comparative look at patrimonial armies presents us with a number of unanswered questions about these militaries’ behavior. Among patrimonial militaries, we witness differing degrees of military cohesion and variation in how communal armies behave over time in a protracted conflict. First, if communal variables determine military actions, why do we see different patterns in the breakdown of military cohesion across communal armies? During Syria’s unrest, the military saw most commanders remain loyal while subordinates across a range of army units defected, in the vast majority of cases doing so individually. This led to a pattern of horizontal defections. When similar unrest broke out in Yemen, we witnessed corporate insubordination of individual units: certain military commanders and subordinates within the same unit either defected or remained loyal, resulting in a pattern of vertical defection. Finally, Bahrain witnessed dual loyalty of both commanders and soldiers. Second, if slow-moving institutional attributes determine military actions, what explains divergent patrimonial military responses to unrest across short horizons? For instance, although both officers and subordinates in Syria remained loyal through the first few months of the uprising, soldiers’ defections increased by late 2011 and led to dramatic defection cascades in spring 2012.
When do defections occur? How do soldiers leave their units? And will they go on the run individually, or organize against political incumbents? Answering these questions is important not only for a better understanding of military cohesion, but also for conflict trajectories and modes of political transition more broadly. The pattern of defection influences how an uprising develops. Individual, mass desertions are likely to result in the hollowing out of the armed forces and the emergence of militias, and hence violent domestic conflict. Mutinies and collective insubordination can lead to civil war, but they may also result in a stalemate between rivaling military factions and hence more orderly, negotiated conflict management, including a coup d’état or democratic transition. Moreover, insights on the timing of defections will signal when a popular mass uprising will turn into potentially violent internal military conflict. Rapid collective military insubordination will most likely lead to the ousting of embattled incumbents, but also the end of anti-authoritarian collective action. Late episodes of military insubordination, in contrast, will face a more sustained and organized popular revolt.
Military Responses in Principal-Agent Hierarchies
In response to these challenges in theorizing military action during domestic unrest, we follow the suggestions by Siddharth Chandra and Douglas Kammen to disaggregate the military and analyze the interests, strategies, and actions of individual officers and soldiers as distinct agents. Footnote 13 Looking at agency in military behavior during the Arab Spring was suggested by Sharon Nepstad and Hicham Bou Nassif, Footnote 14 and we develop their approaches further. Individual agency approaches have proved particularly insightful in studying politics during crisis, and our Middle Eastern cases show that analyzing “the” military as a corporate actor can be misleading when studying episodes of loyalty versus defection.
The military institution is not a unitary actor, but rather encompasses “a shifting set of actors who share a central identity but who have malleable allegiances and potentially divergent interests.” Footnote 15 The immediate question for individual military personnel in crises such as the Arab uprisings is whether to shoot at the people—as a consequence of an incumbent’s order or in anticipation of it. Footnote 16 And there is often variation within the military regarding how soldiers respond to these orders. We see three possible options for military personnel. Exit is a largely passive reaction where a soldier remains in the barracks or goes into self-imposed exile. Although the soldier departs from the current ruling coalition and loyalist armed forces, he does not actively support the regime’s challengers. Resistance entails the soldier defecting to fight for the challenger or seize power through a coup d’état. Finally, loyalty is the soldier using force to defend the autocrat against the challenger.
Civil-military hierarchies influence patterns in soldier exit, resistance, and loyalty during unrest; specifically, we suggest there are two principal-agent relationships at work. First, there is a contract between the civilian autocrat as principal and his high-level military commanders as agents. But, as Susan Shapiro has recognized, “actors are not just principals or agents, but often both at the same time—even in the same transaction or hierarchical structure.” Footnote 17 Second, then, there is the relationship between the military commanders as principals and their military subordinates as agents.
Incumbent Contracts Commander
A theory of decision-making in the civil-military hierarchy rests on Peter Feaver’s principal-agent model. Footnote 18 According to Feaver, the civilian principal contracts the military agent to defend the nation from external or domestic threats. However, the military agent chooses to either “work” or “shirk.” Working entails subordination to civilian authority, that is, loyalty; shirking involves insubordination, that is, exit or resistance.
An expanded principal-agent framework aids in explaining military behavior during authoritarian crises. At the outset of an uprising, an incumbent contracts his commanders to suppress unrest. Footnote 19 Contracting these higher officers creates a “moral hazard problem” in the authoritarian elite bargain. Footnote 20 The incumbent principal has the utmost interest in his commander agent fulfilling the repression order—the autocrat’s political survival likely hinges on it. Wielding coercive power, however, the military commander has a degree of autonomy in deciding whether to be a loyal agent. On the one hand, higher officers may rally around the flag and support an embattled autocrat. Doing so would allow them to receive the limited rewards offered by grateful incumbents, such as material benefits or greater influence in the elite coalition. Yet mass uprisings also present officers with an unrivalled opportunity to replace power-holders. Commanders may just as well seize power or serve as brokers in a transition process, which also provides them with substantial moral rewards for ending repressive dictatorship. Indeed, their proactive involvement to oust an incumbent may generate its own benefits because “it is reasonable to expect that the richest rewards go to the most resourceful actors contributing to a transfer of power.” Footnote 21
Both loyalty and resistance therefore have potential value for commanders. Does exit? As Pion-Berlin et al. describe, “in the terrain between coups and subordination lies a space where soldiers disobey orders without intervening politically.” Footnote 22 The authors argue that militaries may have material, ideational, and institutional incentives that make remaining quartered a desired option during unrest. Yet under autocracies we expect this to be less likely. Self-imposed exile and remaining in the barracks are extremely dangerous under dictatorship, where rule of law is weak and autocrats may exact brutal punishments. As a result, exit is as risky as resistance but offers a lower pay-off for commanders should regime change occur.
A commander’s decision between loyalty and resistance is difficult because outcomes are highly uncertain during uprisings. Findings in social psychology prove helpful in accounting for how individuals make decisions when expected gains and risks are in the future. Such expectations are sensitive to the time horizon of the gains, with individuals preferring “high-probability, low-payoff gambles in the near term and low-probability, high-payoff gambles in the long term.” Footnote 23 Indeed, if an officer chooses loyalty, he can expect the incumbent to offer a limited reward (in comparison to full power) in the near-term. Moreover, the loyalty decision itself renders the granting of a reward more likely because the chances increase for incumbents to survive the uprising. Based on this logic, there are only narrow circumstances under which officers consider a low probability of success, but high payoff, power grab. We highlight two factors that would lead commanders to reconsider a loyalty decision: a regime’s shortened horizon and commanders’ autonomous coordination among one another.
First, commanders may reconsider their calculations when they are confronted with a dramatically shortened time horizon for the regime. At the start of an uprising, the incumbent’s likelihood of falling is uncertain. But there are clearer signs regarding the leader’s ultimate longevity. In particular, when an autocrat is advanced in age and has compromised health at the time of a domestic crisis, commanders may view the regime as having a short horizon in any potential future. As a result, commanders will not gain much from a loyalty decision, as the time to deliver rewards is short when succession is in view. As Hale uncovered when analyzing regime horizons during the “Color Revolutions,” “massive street rallies are costly to suppress, and the more blood that will likely be shed in doing so, the more likely it is that the military will hesitate to engage in violence on behalf of a lame-duck leader.” Footnote 24 Similarly, in reference to the Shah’s fall in the 1979 revolution, Bruce Bueno de Mesquita and Alastair Smith concluded that “no leader can commit to reward supporters from beyond the grave.” Footnote 25
If the regime’s longer-term duration affects an officer’s preference for or against immediate regime change, opportunities to coordinate with other commanders influence how likely he is to act on those preferences. Recent research by Naunihal Singh and Andrew Little has underscored that the primary obstacle to executing a coup d’état is coordination. Footnote 26 A commander has an interest in moving in the same direction as the majority of officers to avoid punishment and reap rewards. Coup leaders can win fellow officers over by signaling that the coup is a fait accompli, that is, making it appear that most officers agree with the move or abstain from resisting it. The same logic applies to military reactions to civil conflict when loyalty or resistance is made a fact. Commanders gain information on fellow officers’ preferences and likely decisions when they are able to coordinate autonomously among one another, beyond the direct observation of the incumbent. When such coordination options exist, officers are more likely to follow the apparent majority of commanders’ preference—for loyalty or resistance. Yet, coordination among officers is a necessary but insufficient condition for resistance. Coordination might well lead to assured loyalty, in particular when the political incumbent observes the coordination efforts and there are no signals of an incumbent’s weakness.
Commander Contracts Subordinate
Once a commander makes his agent decision, he becomes a principal and orders his subordinate to either use force against or join him in fighting with the opposition. Footnote 27 A subordinate’s considerations are quite different from those of the military commander. First, subordinates have less at stake in a given regime. Military service is a job, primarily defined by a salary and limited prospect of professional advancement. Hence, exit is a valuable option for soldiers and low-ranking officers who do not suffer much from losing their military salary, and who do not gain much personally from replacing the political incumbent. Second, in comparison to higher officers, the subordinate’s physical well-being is generally put to greater risk. Subordinates shoot and are shot at because they are closer to the opposition they are ordered to fight. As the subordinate gains less from the current regime, we do not expect his preferences to be linked to calculations about the regime’s time horizon. Instead, the rank and file are influenced by conflict dynamics on the ground, including their emotions, identity, relationship with demonstrators, public discourse, a commander’s charisma, and more.
On the other hand, due to the hierarchical chain of command, soldiers and low-ranking officers are less likely to be able to follow their true preferences than are commanding officers. Owing to barriers to autonomous organization among soldiers and low-ranking officers, mutinies are less common than officer coups. In Feaver’s language a subordinate is likely to “work” and not “shirk” if his commander has chosen loyalty and maintains strong monitoring and punishment capacities. Simply speaking, the more a commander is able to monitor his contracted subordinate, the more likely a soldier is to falsify his preferences and remain loyal. Footnote 28 In contrast, if monitoring capacities are weak and punishment threats not credible, the subordinate’s preference will translate more directly to the actions he takes.
Protracted Unrest
A reassessment of the factors underpinning agents’ decisions is necessary once an uprising turns into protracted conflict. As perceptions of regime stability alter, higher officers’ risk/reward considerations do as well. Despite a possible change in the commander’s preference for regime change or stability, however, his initial decision cannot be revoked easily. If a higher officer has chosen to challenge the incumbent, he cannot easily revise his decision and be readmitted to the loyal elite coalition. If he has initially supported the incumbent, he will also find it difficult to resist and join the opposition, even if the regime’s survival likelihood declines; officers associated with shooting orders will be made accountable if the regime falls. Thus, a commander’s decision at the outset of protracted regime crisis will likely persist.
The situation is different for the rank and file. Subordinates can invoke the chain of command as preventing insubordination. Commanding officers might attempt to implicate their subordinates in atrocities and so prompt rebels to turn on potential deserters; yet, these soldiers could still escape post-conflict justice if they argue they had a gun to their head. Subordinates therefore are not as bound to their initial decisions and have incentives to reassess the conflict’s changing course. If the evolution of the conflict as well as initial resistance and loyalty decisions weaken monitoring and punishment, we expect defection cascades among low-ranking officers and subordinates similar to what Timur Kuran has found in revolutionary dynamics. Footnote 29
Arab Uprisings and Research on Military Behavior
The remainder of this article highlights the plausibility of our framework by comparing varied military behavior during uprisings in Bahrain, Yemen, and Syria 2011–2012. At the outset of unrest in Bahrain, both commanders and subordinates remained loyal to the incumbent, which resulted in a dual loyalty pattern and the uprising’s suppression. Yemen’s uprising saw both higher officers and subordinates defect from particular units; these vertical defections pitted resisting and loyal military units against one another in a low-level armed conflict. In Syria, the popular uprising witnessed a largely loyal officer corps; yet, about one year into the conflict subordinates increasingly chose both exit and resistance, producing horizontal defection, that is, individual defections of officers and soldiers across military units.
Our following empirical discussion serves as an analytical narrative of these loyalty and defection patterns in Bahrain, Yemen, and Syria. This variation in patterns of exit, resistance, and loyalty is intriguing, given the many characteristics these countries share in common. They witnessed popular uprisings at the same time—early 2011—and of similar scale. All were authoritarian countries nested within the Middle East and, moreover, prior to the Arab Spring scholars described each country’s military as “patrimonial.”
Despite these commonalities, we witnessed variation in how militaries held together, broke down, and even turned on the regime during the Arab uprisings. This suggests that the existing arguments that patrimonial militaries are strong autocrat supporters during unrest have limited explanatory power, and that we can enhance our understanding of regime durability by investigating how military hierarchies vary across authoritarian regimes.
Figure 1 Variation in exit, resistance, loyalty patterns by case
We highlight how variations in principal-agent relationships distinguish these patrimonial militaries by drawing on fieldwork in Bahrain and Yemen, as well as interviews conducted between 2012 and 2015 with Syrian military personnel now based in Jordan, Lebanon, and Turkey. We conducted a total of 114 interviews, in Bahrain (15), Jordan (46), Lebanon (23), Yemen (17), Turkey (3), and Washington, DC (10). Our interlocutors were primarily former military personnel, though we also spoke with civilian witnesses and experts. We recruited interviewees via respondent-driven chain referral (snowballing) and relied on multiple initial contacts to diversify our population of respondents. We bolster this interview data with evidence drawn from news media and secondary-source research on the Arab uprisings.
Our empirical insights illustrate how variation in commanders’ coordination capacity and perception of a regime’s longevity regardless of the uprising, as well as the stringency of command and control, help explain military responses to contentious politics under authoritarian regimes. By selecting cases that share patrimonial organization, we investigate a more intriguing phenomenon than institutionalized armies’ behavior in the domestic arena. First, loyalty is widely believed to be the norm in these patrimonial military institutions, and our insights help to provide a better understanding of the actual mechanisms at work in the military hierarchy that sometimes produce resistance within patrimonial armies. Second, most contemporary authoritarian regimes in the Middle East, Central Asia, and Sub-Saharan Africa have developed a measure of communalism in recruitment logics and military organization, which allows us to invoke greater explanatory power for studying military behavior during social unrest more globally.
Bahrain: Dual Loyalty and a Truncated Uprising
On February 14, 2011, six thousand protestors called on Bahrain’s ruling Al-Khalifa family to deliver on reform promises. Demonstrators were met with a swift response by the Ministry of Interior (MoI), Bahrain Defense Forces (BDF), and National Security Agency (NSA). Footnote 30 Yet, on February 21 and 22, and on March 2, more than 100,000 people—nearly one-fifth of the Bahraini population—were on the streets again. By March 9, there were calls not only for regime reform, but also removal. Through one month of domestic unrest, and before receiving aid from the Gulf Cooperation Council’s Peninsula Shield Forces, the Bahraini military and security forces remained resolute in containing the demonstrations and in supporting Al-Khalifa.
As Bahrain’s popular uprising unfolded, military commanders had an interest in the status quo. Reform and certainly regime change would cost high-ranking officers from the ruling family their leadership positions. Moreover, commanders assessed that the regime’s long-term durability was high—based on the king’s young age and Saudi Arabia’s continuous support—and thus individual defections were extremely risky. Subordinates followed their commanders’ orders to repress domestic unrest, as the coercive apparatus’s monitoring and punishment capacities were substantial on the island of Bahrain. Soldiers followed their commanders. As the loyalty-loyalty pattern predominated, Bahrain’s uprising was suppressed; the regime’s stability has not been jeopardized despite periodic unrest since 2011.
Incumbent Contracts Commander
When unrest unfolded BDF commanders assumed that Al-Khalifa would serve their needs better than an alternative regime. Most BDF major commands were headed by members of the Al-Khalifa ruling family, including Commander-in-Chief Khalifa bin Ahmed al-Khalifa, Chief of Staff Dajj bin Salman al-Khalifa, Royal Guard Rapid Intervention Force Commander Khalid bin Hamad al-Khalifa, and more. Although demonstrators did not demand military reform in Bahrain, regime change or significant reforms would threaten command appointments based on ruling family ties. Being a commander for King Hamad also came with personal privileges, such as real estate assets. Footnote 31
Most importantly, commanders had cause to believe that, even outside the uncertain effects of the uprising, their relatively lucrative position would stand firm in the foreseeable future. King Hamad turned sixty-one years old amid the 2011 unrest and was in reasonably good health, ruling out expectations of an immediate succession scenario beyond the uprising. During the unrest, Saudi Arabia also sent a clear signal that it continued to support regime stability in Bahrain. By February 23, Bahrain’s King Hamad traveled to Saudi Arabia and was seen off by the Saudi Arabian Assistant Minister of Defense and Aviation as well as the Inspector General for Military Affairs. Subsequently, on March 10, the Gulf Cooperation Council pledged twenty billion USD in aid to Bahrain and Oman. On March 14, the Kingdom sent one thousand troops across the 16-mile causeway to protect key infrastructure in Bahrain. Senior Bahraini commanders were able to read Saudi Arabia’s stance on Bahrain from the start of the unrest.
Although the Bahraini military was linked to the regime through familial ties, commanders were able to coordinate among one another as well, with the result that the majority supported a loyalty decision. Bahrain established a Supreme Defense Council (SDC) in 1973; following the king’s declaration of a State of National Safety, on March 15, 2011 the SDC recommended the establishment of a National Safety Council (NSC) as well. Footnote 32 Headed by the BDF Commander-in-Chief, the NSC was composed of both military and domestic security service representatives to provide an avenue for commander coordination during the unrest. Indeed, during twelve NSC meetings in only one month, “each of the participating agencies presented its evaluation of the unfolding situation in Bahrain, briefed the other agencies on the measures it had undertaken and outlined its proposals regarding future measures that should be taken to restore order in the country.” Footnote 33 Commanders’ loyalty preferences could be revealed through such private forums, providing the context for higher officers to coordinate their efforts. Bahraini commanders had an interest in regime stability as well as confidence that the regime would remain resilient regardless of the demonstrations.
Commander Contracts Subordinate
Like their commanders, the preponderance of subordinates chose loyalty amid unrest. Studying Bahrain’s military shows that two groups of subordinates existed in 2011: those who uniquely benefited from Al-Khalifa rule and thus had an interest in regime stability, and those who did not. Among the former were Al-Khalifa family members in preparation to become commanders, as well as Bahrain’s mercenary soldiers. Family members at the lower-officer level had an increased likelihood of becoming a commander than they would under an alternative regime. Mercenaries also had an interest in the status quo. Recruits were brought from India, Oman, and Yemen, under the supervision of British nationals. By 2011, many came from Baluchistan, Pakistan, Syria, and Jordan as well. Footnote 34 These recruits were said to gain a salary ranging from 500–700 BD (1,300–1,800 USD) monthly. Footnote 35 In addition, foreign recruits were provided overtime pay, free housing, and compensation to their families should they die. Footnote 36 Many security force members were not born in Bahrain; but some, particularly those working in the BDF, gained citizenship for themselves and their families once in the country. For these individuals regime change would certainly result in losing their jobs—demonstrators explicitly called for removing mercenaries from the military and security services.
In contrast, most Bahraini subordinates did not benefit uniquely from the status quo. Being in the military was a steady job with average benefits; however, military service was not a privileged position in Bahrain and salaries were not extremely lucrative. Footnote 37 Moreover, there has been a clear bifurcation between Bahrainis and non-native Bahrainis in the security forces. In the years leading up to the uprising, native BDF members felt superior, particularly to low-level recruits from Pakistan. In addition, some Bahrainis felt as though money that could aid their remuneration and benefits was channeled to the mercenary recruits. For instance, non-native military members were provided with free housing, while native Bahrainis were not. Footnote 38
At the start of Bahrain’s uprising, one group of military subordinates appeared to have a stake in regime stability, yet other subordinates’ interests were likely more affected by other personal concerns. In the end, the vast preponderance of subordinates chose loyalty and used force to repress demonstrations. Evidence suggests that the Bahraini military had established mechanisms that enabled commanders to effectively monitor and punish subordinates during the uprising, leading subordinates to “work” despite some preferences otherwise. A former BDF officer related that there were some nine to fifteen intelligence personnel within each BDF unit in the years ahead of the uprising. These intelligence officials were of Bahraini, Syrian, and Jordanian origin and wrote regular reports about subordinates’ behavior. In addition to formal monitoring, the regime used family networks to indirectly coerce security force members into submission. If a soldier stepped out of line, he would receive a phone call from his own family advising him to modify his behavior. Footnote 39 In addition, the regime pursued a strategy of composing units of both Bahrainis and non-natives. Footnote 40 By decreasing the number of Bahrainis in any one unit, the regime sought to limit collective action among Bahrainis. This achieved two purposes: keeping those with some connection to demonstrators more removed from them, while ensuring supervision of those more alien to demonstrators but who might have shirked dangerous assignments. A similar configuration was likely in effect within the military. There are some reports, for instance, that Pakistani subordinates would not use torture against prisoners unless Bahraini supervisors were in the room monitoring them. Footnote 41
Perhaps most importantly, the regime did not hesitate to punish dissenters within the coercive apparatus. Outside the military, police officer Ali Al Ghanami deplored the security forces’ response to the February 2011 demonstrations, which resulted in his 12.5-year jail sentence. Former BDF officer Mohammed Albuflasa came out early in the demonstrations, calling for unity and reform at a prominent rally. His remarks circulated, as he was a Sunni Muslim emphasizing universal, Bahraini concerns. Albuflasa was soon arrested and imprisoned for six months, and has testified to being tortured during that time. In another account, a demonstrator was arrested and taken to a BDF base, where he was tortured. He reported having seen more than one dozen BDF personnel torturing a member of the BDF for not using force against protesters. Footnote 42 Finally, threats of punishment were credible in Bahrain due to its geography as a small island; defection was disincentivized by the lack of geographical safe havens. Commanders and subordinates had incentives to remain loyal in Bahrain, a pattern that truncated the Al-Khalifa regime’s crisis.
Yemen: Vertical Defections and Regime Transition
The Yemeni uprising started on January 15, 2011. Footnote 43 Initially a gathering of youth groups in Sana’a, protests swept across the country in the following days. The regime’s containment strategies grew increasingly uncompromising and led, on March 18, to the killing of at least thirty demonstrators in Sana’a. Soon after, a substantial part of the armed and security forces defected. Most important was the defection of General Ali Mohsen al-Ahmar, commander of the 1st Armored Division, known in Yemen simply as the Firqa (military division).
In the following months, Ali Mohsen’s Firqa stood against loyalist forces, including the Republican Guard under the command of the president’s son Ahmed Ali Saleh. Popular protests in Yemen turned into a protracted armed conflict between rivaling factions and culminated in an attempt on president Saleh’s life on June 3. The Gulf Cooperation Council-brokered takeover by Saleh’s former vice president, Abd Rabbo Mansour Hadi, terminated the immediate political crisis. Yet popular discontent, secessionist movements in the North and South, and increasing al-Qaeda activity contributed to instability throughout 2012 and 2013. Yemen witnessed extensive defections and subsequent conflict, providing clear evidence counter to the assumption that patrimonial militaries will uniformly support incumbents.
Incumbent Contracts Commander
The Yemeni case offers strong support for the expectation that higher officers will choose resistance when prospective elite change benefits their strategic interest, irrespective of a patrimonial organization of the officer corps. Those higher officers uniquely benefiting from President Ali Abdullah Saleh’s rule remained loyal to his regime, while those who had become less dependent on his patronage chose resistance. When Yemen’s uprising broke out commanders had reason to believe that regime change was not an impossible scenario. Irrespective of the immediate crisis, the succession question had already been openly discussed in the years prior to the uprising due to Saleh’s advanced age. This presented commanders with an uncertain expiration date for rewards associated with a loyalty decision.
With regime change possible, many higher officers questioned their interest in regime re-stabilization. In early March, Abdullah al-Qadhi, commander of Ta’iz military district and member of the Sanhani core elite, defected. Footnote 44 Ali Mohsen—the president’s king-maker and second most powerful man in the elite—and his Firqa were the main military force in support of the mutiny. Apart from the Firqa, almost all air defense brigades across the country defected. Various air force brigades, including the 4th brigade (Sana’a), 101st mountain infantry brigade (Sana’a), 122nd infantry brigade (Hodaydah), 314th armored brigade (Hadramawt), 1st artillery brigade (Sa’ada), and 127th infantry brigade (Amran), defected from the central command. Footnote 45 The core loyalist forces consisted of the Republican Guard, the Central Security Forces (a paramilitary riot police), and the air force.
Along with Ali Mohsen, various members of the political and military elite chose resistance against Saleh because their own stake in power had been compromised prior to the Yemeni uprising. Footnote 46 For decades, politics and the coercive apparatus in Yemen were run by a tribal confederation of the northern highlands, primarily of Saleh’s own Sanhan tribe. Following Saleh’s take-over in 1978, a small elite clique consisted of Mohammed Saleh al-Ahmar, the president’s brother; Ali Mohsen; and other members of important families, such as Mahdi Mukwallah, Abdulillah al-Qadhi, Mohammed Isma’il al-Qadhi, and Ahmed Faraj. Yet, since the mid-1990s the state and military’s patrimonial organization had been reinvented through the introduction of the second generation of Saleh’s family members. Some of the most sensitive leadership positions came to be occupied by the president’s nephews, Yahya, Tareq, and Ammar, or his brothers, Mohammed and Ali. The president’s son, Ahmed Ali, assumed leadership of the Republican Guard—the best-equipped and most efficient military force, numbering approximately 30,000 men. A simmering power-struggle between the Saleh camp and other members of the Sanhan elite produced rumors of the assassination of Mohammed al-Qadhi and Ahmed Faraj in 1999 and an attempt on Ali Mohsen’s life in his northern military headquarters in 2010. Footnote 47 Most importantly, the position of Ali Mohsen and his elite faction was clearly threatened by the grooming of Ahmed Ali, since 2004, as the most likely successor of his father. Footnote 48
Ali Mohsen and his collaborators preferred resistance to exit for two reasons. First they had established channels of communication among each other, and beyond the president’s control, prior to the uprising. While the Saleh family members in the military and security establishment gathered in the High Central Command, Ali Mohsen established an alternative command structure that facilitated coordination among defectors, including the Firqa; the leaders of the Islah-affiliated tribal militias, Hamid al-Ahmar and Mohammed al-Yadumi; head of the Political Security Office, Ghaleb al-Qamish; and the commander of the 310th Armored Brigade in Amran, Hamid al-Qushaybi. Footnote 49 Second, defection of commanders around Ali Mohsen promised these resisters substantial benefits and political power. Indeed, it appears as though they correctly assessed their ability to expand privileges under a new regime. According to political observers, Ali Mohsen became the most powerful man behind the scenes in the immediate post-Saleh era. Footnote 50 President Abd Rabbo Mansour Hadi pushed “security sector reforms” through in 2012, which in fact weakened his predecessor’s camp by dismissing Saleh family members from top posts in the military and security apparatuses, thereby strengthening the Mohsen-camp. Footnote 51
In sum, while horizontal officer coordination in Bahrain produced corporate loyalty, coordination among dissenting elites in Yemen ultimately led to their defection. The ability to make defection a fact was possible because defectors perceived Saleh’s fall as likely and rewarding.
Commander Contracts Subordinate
The Yemeni case also provides evidence that at the early stages of an uprising a subordinate’s decision-making is often contingent upon his superior’s. The Yemeni military consisted of tribal forces more closely resembling militias than a regular army. Unlike Syria or Bahrain, the social composition of the armed forces in Yemen was quite homogeneous, not only regarding the higher officer corps but also the rank and file. The latter were recruited primarily from tribal confederations in the northern highlands, especially from around the capital and Amran province. As the uprising progressed, the standoff between loyal and resistant units was therefore between socially similar factions, at the officer and subordinate levels. Soldiers remained subordinate not due to unique communal ties, but rather their material interests as well as the absence of personal risk during their deployments.
Yemeni subordinates had an interest in following their officers because they were economically dependent on them; hence, officers enjoyed great leverage and control over their respective units. Soldiers received a 100–150 USD regular monthly salary, barely enough to sustain a living. Footnote 52 Moreover, the state’s payment had been uneven, favoring units closer to the president. For instance, the elite Republican Guard had an official monthly budget of twelve billion Yemeni rials, whereas the equally large Firqa received only four billion. Footnote 53 Ali Mohsen’s Firqa was a de facto private militia of 30,000 to 40,000 men, some called to duty only when their commander deemed necessary. The Firqa’s soldiers, some reportedly under age, were loyal to Ali Mohsen because he provided them with financial resources beyond their regular salaries.
A second phenomenon explaining soldiers “working” for their commanders was the particular type of tribal conflict that unfolded in the aftermath of the uprising. In March 2011, two major armies were facing one another in and around the capital, with a prospect of all-out civil war. The two sides claimed control over parts of Sana’a and a battle line was drawn along Zubayri Street in the heart of the capital. Yet, despite the major stakes at play and attempts on the lives of both sides’ leadership, a particularly low level of violence characterized the ensuing conflict. Footnote 54 One observer described the 2011 standoff as a protracted turf battle with flares of low-level fighting, but an unspoken agreement between the two sides to avoid major damage in life and property. Footnote 55 When some commanders chose resistance in Yemen, their subordinates followed. They did so because they were economically dependent on their officers and able to avoid physically harming fellow tribesmen. Material concerns led soldiers to formulate a true preference to follow their commanders.
Syria: Horizontal Defections and Civil War
Syria’s revolt started with demonstrations in the southern city of Dera’a, drawing thousands to the streets from mid-March 2011 onward. Footnote 56 Isolated protests were also reported in Homs, Hama, and Aleppo. The uprising provoked a harsh reaction by security forces; as a consequence, pockets of militant resistance emerged in September 2011. The defection of a number of low-ranking officers soon resulted in the establishment of the Free Syrian Army (FSA), a loosely-knit network of militias. Footnote 57
In the course of the regime’s military initiative in spring 2012, and rebel attacks in Damascus and Aleppo in the summer of that same year, Syria slipped into civil war. The rebels gained limited international support and the regime’s forces lost control over swaths of the country. Yet, despite widespread expectations regarding the regime’s fall, the military consolidated its position considerably one year later. This is in part owed to support from Russia, Iran, and the Lebanese Hezbollah, but also as a result of the lack of vertical military defections, such as those seen in Yemen, or resistance by the officer corps in the form of a coup d’état, as witnessed in Egypt in 2011 and 2013.
Incumbent Contracts Commander
Prior to the uprising, few observers of Syrian politics were doubtful of officers’ readiness to support President Bashar al-Asad. Under his father Hafez al-Asad, the armed forces were deployed to end a Muslim Brotherhood revolt, culminating in the Hama massacre in 1982. Recruitment patterns in the Syrian higher-officer corps confirmed sectarian bonds to a political regime dominated by the Alawis, the faith of the Asad family and large parts of the political elite. Footnote 58 At times, ninety percent of the higher officers were Alawis, who comprised only ten percent of the population. Footnote 59
On the other hand, the sectarian nature of the Syrian army itself is not a sufficient condition to explain higher officers’ loyalty. In fact, it was Hafez al-Asad’s own brother, Rif’at, who staged a coup attempt in 1984 relying on substantial support from the Alawi officer corps. Bashar al-Asad had forged personal ties with the higher officer corps as he entered the political arena in Syria through military channels and became commander of the Republican Guard and supervisor of military intelligence. In 2002, he replaced the army’s Chief of Staff Ali Aslan with Hassan Turkmani and enforced leadership changes throughout the intelligence apparatus. Footnote 60 Even long-time Minister of Defense Mustapha Tlass, who had facilitated Bashar’s ascent to power in 2000, was forced to resign in 2004. The president’s brother-in-law, Assef Shawkat, became a dominant figure to oversee the security establishment, and his brother Maher assumed command of the Presidential Guard.
Higher officers’ loyalty proved advantageous during Bashar’s ascent to the presidency, but also when the military’s active engagement was necessary to keep him in office in 2011. At the outset of the uprising, commanders had reason to expect the regime to remain strong, since Bashar al-Asad had recently reconfigured the elite coalition and, at the age of forty-six, had hopes to stay in office for some time. To our knowledge, no higher officer defected in the first six months of the Syrian uprising. Once the revolt continued, scores of low-ranking officers defected, including lieutenants, captains, and majors. Patterns of loyalty and resistance in Syria support our argument about the behavior of commanders. They had a strong preference for maintaining the status quo, not only because of sectarian bonds among the Alawi minority, but also because of the strong personal relationship that Bashar had forged in his ascent to power and commanders’ assessment that he was a leader who would remain in power. Defections were extremely rare among higher officers and mostly identified individuals who had been sidelined prior to their decision to rebel. Footnote 61
There is also evidence that officers’ decisions were maintained in a protracted regime crisis. The bomb attack on the regime’s operative military leadership, on July 18, 2012, resulted in the killing of Bashar’s brother-in-law Assef Shawkat, Minister of Defense General Daoud Rajha, former Chief-of-Staff Hassan Turkmani, and Minister of Interior Mohamed al-Shaar. The attack dealt a serious blow to the regime. Yet neither the proven vulnerability of the regime’s inner circle as the uprising progressed nor the insurgents’ success on the battlefield at that time prompted the officer corps to disintegrate once initial loyalty decisions had been made.
Commander Contracts Subordinate
Resistance and exit among subordinates were rare occurrences during the early months of the Syrian uprising. Some anecdotal evidence hints at individual soldiers’ refusal to shoot at demonstrators, but there were no reports of substantial defections among subordinates and security personnel during the first three months of the Syrian revolt. At the initial stages of the uprising, the monitoring and punishment capacities within military units remained largely intact, facilitating a strong loyalty-loyalty pattern in the decision-making of officers and their subordinates. Yet Syrians remained defiant, kept demonstrating, and organized militant resistance from late summer 2011 onwards.
Once an uprising evolves into a protracted crisis, there is a greater propensity among subordinates to “shirk.” Initially dominant loyalty-loyalty patterns disintegrate, as soldiers become increasingly aware of the regime’s vulnerability and compromised monitoring capacities. Our interviews with former Syrian service members confirm that the number of subordinate defections increased considerably throughout the first half of 2012. Footnote 62 In order to prevent defections, loyalist forces had established a system of punishment threatening defectors’ family members. Yet this strategy proved increasingly difficult to sustain throughout 2012 for three reasons: the regime could not maintain control over large parts of the Syrian territory; the establishment of refugee camps in Turkey and Jordan and permeable borders created a safe haven for defectors (for both resistance and exit); and a network of activists operated within Syria to facilitate defections.
Our interviews indicate that many soldiers had entertained defection for a long time, waiting for an opportunity to overcome the regime’s monitoring. Footnote 63 Their reports imply that group defections had been attempted, but failed owing to the regime’s close stationing of units in their barracks; hence, soldiers defected individually. Soldiers were contacted when on home or sick leave; safe houses were organized where defectors would spend two to three nights after their defection; and, most importantly, defectors’ families were transferred abroad—concurrent to the defection itself—in order to prevent retaliatory punishment. Although its readiness to suppress resistance and exit was as strong as in Bahrain, the Syrian regime was unsuccessful in coercing loyalty as the conflict progressed. In sum, there is strong evidence in Syria that the military’s loyalty-loyalty pattern disintegrated as the revolt progressed; although the vast majority of higher officers remained loyal, soldiers increasingly felt threatened in their positions and found ways to evade monitoring and leave their positions throughout 2012. Footnote 64
Conclusion
Militaries are not unified actors. Commanders and subordinates have very different interests, which often come to the fore during mass uprisings. And yet these personnel’s decisions affect one another’s owing to hierarchical military relationships. Existing arguments regarding military decision-making during uprisings are unable to explain puzzling phenomena in the Arab Spring, particularly variation in patterns of exit, resistance, and loyalty within militaries, because they often start with the assumption that the military is a unified actor. Our principal-agent framework more accurately theorizes the interactions between authoritarian incumbents, military commanders, and subordinates during such crises. We also reflect on the fact that decision-making contexts change over the course of a protracted revolt.
These findings have important implications for the study of civil-military relations in authoritarian regimes. Deployment orders to suppress domestic, popular discontent are triggers for military insubordination because commanders and soldiers have reason to reconsider their oath of loyalty when asked to shoot at their own people. While scholars have been primarily interested in social unrest as a catalyst for coups d’état, Footnote 65 we reach beyond these approaches by accounting for the different ways in which military insubordination manifests itself, as well as the way in which initial military decisions determine regime transitions and conflict trajectories. This allows us to make an important contribution to the emerging literature on military cohesion in violent conflict. Footnote 66
Looking at recent events in our cases, not only do we see variation in regime change—survival in Bahrain as a consequence of the military’s corporate loyalty, versus leadership replacement in Yemen, versus civil war in Syria—but our framework also allows us to account for the paths countries followed in the years after the Arab Spring. Although both Syria and Yemen have slipped into civil war, their different conflict scenarios have been shaped essentially by variation in loyalty patterns in civil-military relations. In Syria, horizontal atomized defections of lower-level officers and soldiers helped undermine the army’s fighting capacities over time and prop up a fragmented, violent uprising, but it did not lead to the disintegration of the military’s organizational infrastructure. Indeed, Al-Asad’s regime has become too weak to retain control over the country’s full territory, but it also remains strong enough not to lose the war. As a result, Syrians suffer through a prolonged civil war without regime change. In Yemen, by contrast, rifts within the political and military elite led to vertical defection patterns, and hence created splits between rival forces. This hastened the government’s collapse, and President Saleh’s fall in 2011 was also accompanied by the virtual dissolution of the state’s coercive apparatus and a “militia-ization” of the military. The post-Arab Spring Houthi uprising in 2013 therefore did not only trigger civil war, but the very destruction of the regime.
Current events in the Middle East show how our framework travels to domestic unrest scenarios other than those described in greater depth here. An understanding of principal-agent relationships within militaries elucidates the Iraqi army’s startling collapse in the face of advances of the Islamic State in Iraq and Levant (ISIL) since January 2014. Although the Iraqi Security Forces outnumbered ISIL fighters, had received substantial U.S. training, and faced a “lower tempo of enemy operations through June 2014 compared to the earlier Anbar [Fallujah] offensive,” Iraqi forces fled the battlefield and allowed ISIL to occupy the city of Mosul in northern Iraq in June 2014. Poor logistics, corruption, and fear of the enemy undoubtedly influenced individual emotions and military planning ahead of the operation. But the disintegration of the Iraqi forces fighting ISIL was reportedly due to commanders ordering their subordinates to exit on the eve of confrontation. Footnote 67
We find that classical accounts in the literature on civil-military relations, which focus on broad patterns of military organization, are unable to explain a military’s response to popular uprisings and subsequent conflict trajectories. Under authoritarian regimes, it is not organizational features per se that guarantee loyalty, but rather the nature and quality of principal-agent relations mediated through communal ties in the military hierarchy. Taking a wider perspective, our findings therefore support approaches in studies of the military in politics that take military organization as seriously as agency within that hierarchy, such as by Wendy Hunter, Samuel Decalo, Sharon Nepstad, and Hicham Bou Nassif. Footnote 68 Studies on the causes of coups d’état and coup-proofing, for instance, have just begun to more prominently account for variation in military hierarchy. Footnote 69 Disaggregating “the” military will lead us to reconsider the very meaning of “civil-military” relations.
Studying the role of the military in domestic conflict promises inspiring insights for various fields in political science. Scholars of civil-military relations will gain a better understanding of the impact that domestic shocks have on the military apparatus itself: on its composition, size, effectiveness, and internal cohesion. And scholars of civil war will find an interest in the degree to which military disintegration (as a consequence of defection) impacts on a state’s capacity to apply counterinsurgency measures; hence, our findings will help us better understand under which conditions violent insurgencies are more likely to succeed.
Most importantly, scholars of regime transition in comparative politics will need to account for the military once it is deployed to suppress popular contestation, and an understanding of the internal dynamics of military behavior is absolutely crucial to understanding immediate regime trajectories. We expect this field of study to gain in importance over the coming years. Events in the Middle East and North Africa, which came to be known as the Arab Spring, but also other remarkable episodes of popular mass uprisings around the globe—in countries as diverse as Ukraine 2013–2014, Turkey 2014, Thailand 2014, Hong Kong 2014, Burkina Faso 2014, and Guatemala 2015—have come to characterize a new era of global popular contestation. With formal state institutions and electoral politics increasingly delegitimized in favor of street protest, the diverse military and security agents are likely to play resurgent roles in politics.
Supplementary Materials
-
Online Appendix: Research Methodology
