Abstract
A wealth of research in comparative politics and international relations examines how the military intervenes in politics via coups. We shift attention to broader forms of military involvement in politics beyond coups and claim that terrorist violence and the threat of terror attacks provide a window of opportunity for military intervention, without taking full control of state institutions. We highlight two mechanisms through which terrorism influences military involvement in politics: (1) government authorities demand military expertise to fight terrorism and strengthen national security and “pull” the armed forces into politics, and (2) state armed actors exploit their informational advantage over civilian authorities to “push” their way into politics and policy-making. A panel data analysis shows that domestic terror attacks and perceived threats from domestic and transnational terrorist organizations increase military involvement in politics. We illustrate the theoretical mechanisms with the cases of France (1995–1998 and 2015–2016) and Algeria (1989–1992).
Introduction
Since the 9/11 terrorist attacks, the military has played a far larger role in US politics than ever before, leading some experts to talk about an increasing politicization of the US military (Owens, 2011; Ulrich and Cook, 2006). Similarly, the French government reinforced military means and prerogatives to counter terrorist threats after the terror attacks in Paris and Nice, including the deployment of 13,000 troops, an increase in the defense budget and much wider autonomy for the military. These examples illustrate greater military intervention in politics following terrorist attacks, which in turn can change the civil–military equilibrium, and may hinder civilian control over the military and undermine the quality of democracy.
Surprisingly, previous research has not systematically examined the link between terrorism and military involvement in politics (MIP). Aksoy et al. (2015) show that terrorism increases the probability of coups in autocracies, which is consistent with other studies stressing that domestic and external security threats raise the risk of coups (e.g. Ezrow and Frantz, 2011; Goemans, 2008; Marinov and Goemans, 2014; Piplani and Talmadge, 2016; Powell, 2012; Svolik, 2013; Bove and Nisticò, 2014a; Bove and Rivera, 2015). However, even though coups marked the 20th century, these events are rare and affect autocracies disproportionately. More importantly, coups represent only one strategy that military actors have at their disposal to intervene in politics. To quote Finer (1975: 4), “the modes of military intervention are as often latent or indirect as they are overt or direct (. . .) the level to which the military press their intervention varies; they do not always supplant the civilian regime. Often, they merely substitute one cabinet for another, or again simply subject a cabinet to blackmail.”
This article focuses on the relationship between terrorism and military intervention in politics, beyond coups. Whereas coup plotters seek to obtain power and take over the government, we focus on MIP and define it as military actions that fall short of an outright attempt to acquire full control of the government. 1 MIP ranges from coup events to increasing levels of military autonomy that often imply greater influence on governments’ decisions in policy areas within the defense domain (i.e. military expenditures, the structure and organization of the security sector), and sometimes extend to drafting constitutions, redefining justice, and appointing specific actors in key state institutions (e.g. Feaver, 1999; Huntington, 1957; Linz and Stepan, 1996; Pion-Berlin, 1992). Any such change in MIP is relative to a country-specific level of involvement that is usually a function of geopolitical, institutional, and idiosyncratic factors (Brooks, 2008; Levy, 2016).
Our theory highlights that military actors have motives to intervene in politics and their ability to do so depends on the presence of an opportunity structure. To a varying degree and extent, military actors generally have incentives to intervene in politics and influence policymaking, particularly in defense and foreign policy arenas (Brooks, 2019; Recchia, 2015; Talmadge, 2015). Because terrorism threatens national security—including a state’s population, infrastructure, and sovereignty—and has deep psychological effects among the population, civilian authorities use the armed forces to signal their willingness and commitment to stop terrorism and secure the country from violence by domestic and transnational terrorist organizations. Accordingly, we argue that terror attacks and the threat of terrorist violence can open a window of opportunity that facilitates the involvement of armed state actors in politics. We argue that an increase in terrorist attacks and/or threat violence affects MIP through two different mechanisms: (1) government authorities demand military expertise to fight terrorism and strengthen national security and “pull” the military into politics, and (2) state armed actors exploit their informational advantage over civilian authorities to “push” their way into politics and policymaking. In the empirical analysis, we use a global panel data set (1984–2004) and find that domestic terror attacks and perceived threats from domestic and transnational terrorist organizations are associated with higher levels of MIP. We also illustrate the causal mechanisms of our theory with the cases of Algeria and France.
This article adds to the existing literature in two ways. First, we examine how terrorism affects a military involvement that can influence government policy, without taking direct control of executive power. Theoretically, most previous works focused on coups but overlooked that the armed forces often influence policymaking via other means. In this way, we connect with the qualitative civil–military relations literature and propose a third threat-based explanation—terrorism—beyond civil and international wars to explain increase in MIP. From a policy perspective, this approach is relevant because related studies show that increasing military involvement in politics can affect civilian control over the military, and ultimately undermine democratic institutions and the quality of democracy (e.g. Karl, 1990; Linz and Stepan, 1996; Pion-Berlin, 1992; Wagstaff, 2013).
Previous research
The civil–military relations paradox is as theoretically simple as politically crucial: state armed actors in command of state survival and national security have the muscle to prey on society, influence the policymaking process, and remove executive authorities by force or the threat of force. To quote Feaver (1999: 214), “[b]ecause the military must face enemies, it must have coercive power, the ability to force its will on others. But coercive power often gives it the capability to enforce its will on the community that created it.” An extensive literature has analyzed the conditions under which military actors intervene in politics and how civilians can maintain control over the armed forces. Interestingly, research traditions differ in the kind of military participation in politics they have focused on, and both cross-national and case-based studies have neglected whether and how terrorism affects such involvement.
Cross-national research shows that internal and interstate wars and popular unrest often motivate coups (Bell and Sudduth, 2017; Goemans, 2008). Other studies examine the causes of coups in autocracies, helping to understand how different conflicting dynamics between regimes and opposition actors within and outside the ruling elite motivate military actors to use force or the threat of force against dictators (e.g. Aksoy et al., 2012; Bove and Rivera, 2015; Svolik, 2013). Yet most studies overlook that state armed actors can intervene in politics in many ways, and that coups are only one strategy to influence government policies. 2 As Feaver (1999: 216) puts it, “even if a society achieves adequate levels of assurance against utter collapse at either extreme, battlefield defeat and coup, there is a range of problematic activities in which the military can engage [. . .] Thus, ‘solving’ the problem of coups does not neutralize the general problem of control on an ongoing basis.”
Case-based research highlights that the military not only intervenes in politics via coups, suggesting that military actors have several degrees of institutional autonomy and influence military spending, the organization of the security sector, as well as defense and foreign policies more widely (e.g. Finer, 1975; Huntington, 1957; Levy, 2016; Pion-Berlin, 1992; White, 2017). Gelpi and Feaver (2002) find that the presence of veterans in the US political elite—their proxy for the civil–military gap over time and the prevalence of military experience and opinions—reduces the propensity of the United States to initiate militarized disputes. “Once a dispute has been initiated, however, the higher the proportion of veterans, the greater the level of force the United States will use in the dispute” (Gelpi and Feaver, 2002: 779). Similarly, Flynn (2014) shows that military actors occupying key positions within the US government shape defense spending priorities in favor of their respective branches (Flynn, 2014). Other studies likewise suggest that the army retained important influence on many civilian governments after democratic transitions, underlining that military actors imposed “reserved domains” to maintain control on relevant policymaking processes (e.g. Linz and Stepan, 1996; Pion-Berlin, 1992). Recent literature points to “the existence of unofficial loci of power within military organizations” (Hundman and Parkinson, 2019: 20).
While existing research has made progress in explaining MIP, our review suggests that the literature has several limitations. Cross-national studies have largely focused on coups, without exploring MIP beyond coups. In turn, case studies have shown that MIP can take many forms and that the military can influence policymaking processes without perpetrating a coup. Yet there is still a lack of comparative research on broader forms of MIP. Finally, within either research tradition, the potential role of terrorism for military participation has been largely overlooked. A partial exception is Aksoy et al.’s study (2015) on terrorism and coups in autocracies, 3 but we still know little about whether and how terrorist activity affects other forms of military intervention.
The argument
We analyze civil–military relations through the lens of the principal–agent theory, with political authorities seen as principals and the armed forces as agents (Feaver, 2009; Svolik, 2013). We argue that a set of actors’ motives and an opportunity structure shape a civil–military relations equilibrium. A first set of motives to intervene in politics stems from the military view of its role as guardian of the state. The military’s raison d’être is to protect the state from violent direct action by domestic and foreign actors like insurgent armies and rival states. As Huntington (1957: 63) puts it, “the responsibility of the [military] profession is to enhance the military security of the state.” A second set of motives is linked with the corporate interests of the military. The bureaucratic politics theory depicts policymaking as the product of multiple interactions between competing interest groups (Halperin, 1974). The timing and content of policy decisions are shaped by officials’ efforts to protect or promote their interests and overstate their own priorities in competition with other agencies. Like other government entities, the military strives to protect or extend its operational autonomy and advance its own ends (Allison and Halperin, 1972). Unsurprisingly, the surge of terrorism over the last decades has led military actors in many countries to emphasize the threats posed by terrorist groups and highlight their role in protecting the state from terrorism.
Beyond motives, an opportunity structure that facilitates intervention and increases military autonomy in policymaking is also necessary (Finer, 1975). 4 Military actors have the power and capabilities to enforce their will and remove executive leaders using violence, although a large proportion of coups are bloodless and do not require the use of violence (Powell and Thyne, 2011: 251; Singh, 2014). However, coups can be costly and military actors do not always have incentives to seize power through the use of force or threat thereof. Indeed, military forces often refrain from using force against political leaders because of the fear that political factionalism could provoke divisions within the army (Nordlinger, 1977). 5 Unlike coups, other forms of intervention require a context or conditions—an opportunity structure—facilitating participation in politics by the armed forces. In the remainder of this paper, we detail how terrorist attacks and/or the threat of terrorism provide such opportunity structure, which makes intervention in policymaking more likely.
Terrorism as a security threat
Much research highlights that the armed forces are the sentinels of the state and that civil and international wars severely threaten the territorial integrity of the state and the national interest. Threat-based explanations stress that domestic and external security threats affect civil–military relations and encourage military participation in politics to a different extent (Huntington, 1957; Lasswell, 1941; White, 2017). The general consensus in the civil–military literature is that external threats should decrease MIP, while higher internal threats should lead to higher levels of MIP (Desch, 2001; Piplani and Talmadge, 2016; Staniland, 2008: 326–327). High levels of internal threat may lead to what Stepan labels “new professionalism,” one in which the military have an expanded and more politicized role (Stepan, 1973: 1978). Consistently, recent work finds that civil wars make coups more likely, while international conflicts less likely (Bell and Sudduth, 2017; Piplani and Talmadge, 2016).
Along similar lines to Stepan’s (1978) work, we focus on terrorism as a security threat. We contend that terrorism threaten a state’s population, infrastructure, and sovereignty. Terrorism delivers a message beyond its immediate targets and victims, with significant psychological effects among the population. It represents a violent tactic intentionally devised by individuals or subnational groups to achieve a specific goal, based on intimidating and terrorizing a wider audience beyond its direct victims (Krueger and Malečková, 2003). Terrorism pursues a range of goals, including government and regime change, territorial change, policy change, social control, and maintenance of the status quo, among other areas (Kydd and Walter, 2002). It is a form of “coordinated destruction” that often hinders a state’s national security (Tilly, 2003: 14, quoted in Sambanis, 2008: 178) and it has significant detrimental effects on the target state’s economy and political stability (Sandler, 2014).
Terrorism—most notably after 9/11—constitutes a major concern for national and international security policies (Mueller and Stewart, 2012). In the United States, the Bush administration considered terrorism as a primary threat to national security: “Given the potential catastrophic consequences of terrorist attacks employing weapons of mass destruction, Administration decision-makers felt that the nation could not afford to sit back, wait for attacks to occur, and then respond. The nation was mobilized; combating terrorism and crippling Al Qaeda became top national priorities” (Pearl, 2007). Similarly, after the 2015 Paris attacks, François Hollande declared “France is at war” 6 and the Prime Minister Manuel Valls warned “the European project can die very soon if we are not able to respond to security challenges and not giving ourselves the means to fight terrorism.” 7 This mechanism is not unique to advanced democracies; for example, in recent years the militant Islamist terrorist group Boko Haram has spread terror in Nigeria and other African countries, forcing authorities to declare three northern Nigerian states in a state of emergency and declare war against terrorism. 8 We acknowledge that these declarations and statements might be strategic and can be used instrumentally for attaining particular political objectives, but they suggest how terrorism might create openings for an increase in MIP.
Terrorism and MIP
We argue that terrorist violence and terrorist threats provide a window of opportunity, and thus we expect to see an increase in MIP as a result of an increase in terrorism and perceived threats. We highlight two analytically distinct but empirically intertwined mechanisms through which terrorism can increase levels of MIP, which we label “pushing” and “pulling.” 9 A first mechanism is that political authorities pull the armed forces into politics. The “management of violence” is the hallmark of military actors and their technical expertise and skills distinguish them from any other actors from within society (Lasswell, 1941). Importantly, under contexts of political violence by terrorist groups or when the perceived threat is high, government authorities have incentives to demand the knowledge and skills of the specialists of violence, and both the principal and the agent share preferences about the need of a more active role in policymaking by the armed forces. Military actors find themselves with the opportunity to influence policymaking processes without the use of force and can do so because civilian authorities demand military actors’ expertise to strengthen national security.
Civilian leaders under security threat environments are frequently dependent on military actors’ information and capacity to prevent violent threats and strengthen the security of the state. 10 Consequently, state armed actors in the face of terrorist violence can acquire significant political autonomy and influence decision-making processes. The negative externalities of terrorism on a leader’s survival and increasing popular support for military-oriented counterterrorism policies generate strong pressures for civilian authorities to provide the military with additional powers to fight terrorism, facilitating state armed actors to increase their autonomy vis-à-vis civilian authorities. Put differently, political authorities often consider the risks and costs of terrorism and they frequently demand the expertise of military actors who then increase their involvement in politics and intervene in policymaking, following their own organizational interests.
A second mechanism involves the military pushing its way into politics. Under some environments, government authorities have different policy preferences than the military, but the military has strong incentives to intervene. The military expertise with an informational advantage over the principal combined with the principal’s inability to systematically assess insecurity environments may allow the military to take actions that increase their degree of involvement in politics (Finer, 1975: 74). Military actors are the experts on violence, and have a strong comparative advantage regarding intelligence capacity, information gathering about security threats, including threats from terrorist groups, as well as coercive capacity. Politicians rely on military intelligence and information acquisition capacity to define specific threats, as well as coordinate and implement policies aimed at reinforcing homeland security. Because the principal’s decisions largely depend on the agents’ knowledge and many of the actions of the agents (the military) are hidden from the principal (political authorities), state armed actors can easily exploit this informational asymmetry to shape a government’s decisions and advance their own views about how to respond to terrorism. 11
Military actors often overstate the risk of conflict and the threats posed by competing polities and rival actors, including terrorist groups. To a varying degree and different effects, the armed forces not only draw attention to the constant presence of violent threats, but also underline the degree and proximity of these threats (Brooks, 2008; Feaver, 2009; Levy, 2012; Staniland, 2008). They normally “stress the continuing nature of the threats to the security of the state and the continuing likelihood of war,” as well as “emphasize the magnitude and immediacy of the security threats” (Huntington, 1957: 65). Shortly after becoming the Chief of Staff of the US Army in 2015, for example, General Mark Milley continued emphasizing the threat of terrorism and the need to “sustain counterterrorist and counterinsurgency capabilities.” 12 Importantly, the emphasis on terrorist threats has become increasingly common due to the use of terrorist tactics in civil wars and rival states’ support for terrorist groups (Asal et al., 2012).
Hypotheses
Our argument leads to a number of observable implications about the effects of terrorism on MIP. As discussed above, terrorist violence facilitates state armed actors to get involved in domestic politics and influence a state’s policy in different areas. So far, however, we have not made any distinction between domestic and transnational terrorism. Without denying differences about the nature of domestic and transnational terrorism that are likely to have implications to explain other outcomes (Savun and Phillips, 2009), we expect the same effect of domestic and transnational terrorist violence on MIP. The main difference between these forms of terrorist violence is related to nationalities of the perpetrators and the victims. In fact, whereas domestic terrorism includes perpetrators, targets, victims, and audiences in the same state, transnational terrorism involves actors from different states (Enders et al., 2011). Our general argument should apply equally to both, largely because domestic and transnational terrorist violence hinder homeland security and provide military actors with opportunities to intervene and influence policymaking, without taking direct control of executive power. We summarize this reasoning in Hypothesis 1:
Hypothesis 1: Domestic and transnational terror attacks increase military involvement in politics.
Military actors do not only consider observed episodes of domestic and transnational terrorism. Nordhaus et al. (2012) point out that the risk of fighting in an international war affects a state’s military spending. They estimate the ex-ante probability that a state will become involved in an international war and show that the threat of an international war increases military expenditures. Accordingly, we argue that civilian authorities and military actors also consider the ex-ante probability that the state will be the target of a terrorist group. The armed forces do not only want to reduce the repercussions of terrorist violence once it occurs, but to prevent any potential terrorist event. Furthermore, intelligence gathering and counterterrorist strategies are severely limited by the “clandestine and opaque nature of terrorism and terrorist threats” (Conrad et al., 2017: 763), leading military actors to magnify the risks of potential terror attacks due to the lack of information about the strength and purposes of terrorist organizations. Unsurprisingly, for example, “[d]uring the height of the Iraq War, US government sources frequently cited the lack of information about the terrorist groups as a key reason for the persistence of the terrorist threats” (Conrad et al. 2017: 763). This reasoning leads to Hypothesis 2:
Hypothesis 2: The threat of domestic and transnational terrorism increases military involvement in politics.
Illustrative examples: France and Algeria
We now examine two examples to illustrate and elucidate the mechanisms highlighted in our theory. We selected two cases with a different baseline level of MIP: France—a democracy with a solid and well-established norm of civilian control of the military, which should stay out of politics—and Algeria, an authoritarian regime in which the military has influenced politics through both violent and nonviolent means since independence. 13 While both cases are characterized by an increase in MIP, they illustrate the two mechanisms of the theory. While France helps illustrate both “pushing” (1997–1998) and “pulling” (2015–2016), Algeria illustrates our “pushing” mechanism (1989–1992).
France
As in many democracies, French civil–military relations display professional armed forces respectful of civilian control (Vennesson, 2003). Unlike other democracies, however, the French military organization has traditionally displayed low degrees of military autonomy (Irondelle, 2008). Known as la grande muette [the big silent one], the French military has been careful to avoid any kind of political statement in public since De Gaulle’s strong reaffirmation of the norm of civilian control in 1962, a result of the progressive disbandment of the French officers’ corps after World War II, which culminated in the 1961 failed putsch in Algiers (Alexander and Bankwitz, 1994; Ruffa, 2017: 404). Traditionally concerned with external operations, the French military can be asked to intervene in domestic contexts in exceptional circumstances. 14 Since the 1970s, France developed a judicial approach to counterterrorism, preferring the use of judges over the one of the military, which traditionally played a marginal role in the fight against terrorism (Guittet, 2008). 15
Pushing (1995–1998). Between 1995 and 1998, we observe in our data an increase in terrorist attacks and threats in France, among others the attack at St. Michel RER station, the one at the Port-Royal metro station, and the killing of Préfet Erignac in Corsica. At the same time, the International Country Risk Guide (ICRG) score of military involvement in politics (on a scale between 0 and 6) increased from 1.1 in 1995 to 1.9 in 1997 and then to 2.1 in 1998, which is an unusually high increase in an advanced democracy.
In February 1996, President Chirac announced the end of conscription and a set of profound transformations toward a smaller, more agile, and technologically advanced professional military, which would have substantially reduced the military’s budget and its autonomy. 16 This triggered widespread discontent and concern among the military (Rigouste, 2014). Unseen for decades in the French context, the military reacted publicly and systematically highlighted the terrorist threat to argue against budgetary cuts, “pushing” to defend their interests and autonomy (Irondelle, 2008). Between 1996 and 1998, several high-ranking officers published a series of interviews with the press, explicitly linking the increased terrorist threat with the need of making greater use of the military, Gendarmerie, and army in particular both abroad and internally. 17 For instance, General Henri Paris wrote that any outbreak of violence would require the use of the armed forces, instead of the police, to take care of the raised terrorist threat (Paris, 1998). Other officers called for strengthening the Vigipirate program with a more extensive use of the Army and the Gendarmerie to “fight terrorists with military means.” 18 Moreover, some officials questioned the decision to cut the military budget because it hindered the military capacity to prevent and fight terrorism. 19 To quote a retired Army General, “1996–1998 was a messy time in which we tried to show the importance of the military for anti-terrorist purposes because we did not want to end up being micro-managed as in UN peacekeeping missions and we wanted to contain the budgetary cuts.” 20 This ended in the unique gesture of the Chief of Army General Mercier to release an interview with Le Monde, in which he argued against budgetary cuts (Inciyan, 1997). 21
This “pushing” of the military in a democracy is successful only when the opportunity structure allows for it. Before the increase in the terrorist threat in 1996, the opportunity structure was severely constrained: the French defense budget had declined continuously and the decision to end conscription was part of the process of reducing the military, foreseeing the cutting down of more than 2000 workplaces. Even the conservative government was pursuing a cut down of the relative role of the military (Daho, 2017; Irondelle, 2011). The increase in terrorism in 1995 changed the preexisting opportunity structure since the military could justify its claim against budget cuts, and by making it more attractive for politicians to increase the military’s role (and budget) and signaling a focus on security to voters, which was reflected in parliamentary debates in both chambers. 22
Chirac and Prime Minister Millon responded strongly to terrorism, endorsing a militarized approach, which is reflected in an increase in our measure of MIP: France launched operation Vigipirate at the second highest alert level, delayed its entrance into Schengen by more than a year, and deployed the highest number of gendarmerie and army personnel ever mobilized domestically until then. 23 This strategy continued and increased after the advent of the cohabitation between the conservative pro-military President Chirac and the socialist Prime Minister Jospin, traditionally pacifist and critical toward the military, which led to the second increase in 1998 (Dionnet, 2004: 145). Against all odds, Jospin in 1996 radically transformed his party’s stance and transitioned to tougher views on the fight against terrorism, which became one of the core tenets during his time as Prime Minister. 24 Chirac and Millon, and even Jospin, were all perceived to be extremely close to the military between 1996 and 1998 (Gregory, 2000). 25 The military used the psychological power of terrorist threat as opposed to other motives to increase their relative autonomy (Inciyan, 1997; Isnard, 1999). A retired French General summarizes the motives behind these policies: “politicians felt pressed to show solidarity with the military given the tough time.” 26
The new opportunity structure emerging from increasing terrorist threats and voters’ security concerns together with the lobbying by the military led to policies that increased military’s autonomy. 27 In a first phase, budget cuts were stopped and all services were involved in a wide range of internal counterterrorist activities. 28 In particular, the Gendarmerie was given much ampler power, which triggered a heated debate with the police. 29 In an interview, the Defense Minister defended the Gendarmes’ more assertive role, reminding the public that they were not supposed to tend cows but protect French citizens against terrorists. 30 In a second phase, from 1998 terrorist attacks declined, but the military budget did not decrease and the role of the military for internal security increased even further, including for the first time the monitoring of sensitive sites by the military in the entire territory. 31 Jospin’s Defense Minister Richard claimed that the military were opening a second front at home with their wider influence also in domestic affairs. 32 Many experts regretted the dangers of a “militarized security” control internally. 33
Pulling (2015–2016). In a situation where President Hollande’s popular support was at an all-time low in 2015, the terrorist attacks against Charlie Hebdo, Bataclan, and Stade de France (2015) and Nice (2016) changed the opportunity structure by giving him the possibility to obtain “a flair of energetic statesman posture” 34 through tough internal security measures (Rigouste, 2014). Accordingly, the president argued that “the country was at war” and mobilized the Gendarmes, the military, and the reserve in operation “Sentinel.” 35 More than 13,000 troops were deployed domestically—of which 6000 only in the Paris region—amounting to 12% of the army’s active duty members and making it the largest domestic operation ever launched in France (Tenenbaum, 2016). The so-called most warrior-like President of the 5th Republic, Hollande gave prominence to the military’s role not only by launching a domestic operation but also by systematically choosing to convey the Defense Council over the Internal Security Council and by reinforcing military operations abroad. The combined effect lead to an increase in military autonomy.
Already overstretched with several military operations abroad, however, the military establishment reacted critically to Hollande’s decision to launch operation Sentinel and the military was reluctant to assume this role and remained skeptical. 36 Nevertheless, Sentinel progressively increased its autonomy along several dimensions, including an increase by 15% of the French operational force, which had remained frozen since 2002; an increase in tactical autonomy, including the right of the army to conduct inspections and use force in the street; 37 and an increased access to the president who meets high-level officers of the Joint Chief of Staff every seven days, instead of the usual meeting every 41 days. 38 This increased autonomy was confirmed by the fact that the Ministry of Interior became “increasingly jealous of its prerogatives” and was reluctant to share information with those in charge of operation Sentinel. 39 Likewise, when Hollande mobilized the reserve he maintained the same regimental structure, which allows the military to maintain more autonomy. 40 Importantly, both the military and gendarmes were provided with some of the judiciary police function, a decision considered to be highly controversial. 41 Aware of the increased autonomy, the newly elected President proceeded to change the organizational structure to increase control. 42 Finally, Sentinel greatly improved recruitment rates and the legitimacy of the military to exceptionally high levels, and surveys suggest that the French public felt reassured by the military presence. 43 Thus, a reluctant military was “pulled” into politics and found itself with increased autonomy, resources and bargaining power. 44
Taking both phases together, while still clearly operating within the boundaries of civilian control, terrorism seems to have increased MIP in the French case via the “pushing” (1996–1998) and the “pulling” (2015–2016) mechanism, respectively. In both phases, the French military increased its degree of autonomy and indirectly influenced political decisions.
Algeria (1989–1992)
We now turn to the case of Algeria. Before the military entered violently into Algerian politics with the 1992 coup, we observe a period (1989–1992) with an increase in both our terrorism index and our ICRG indicator. The Algerian military “pushed” for increasing its involvement in politics and overemphasized the threat posed by the terrorist attacks that occurred between 1989 and 1992, but avoided a coup until 1992 (Aït-Aoudia, 2015). 45
For 30 years, the military constituted the backbone of support to the Front de Libération Nationale (FLN), the single dominant ruling party and was considered the “arbiter” (Lenze, 2016: 38) or the “real power” (Cook, 2007: 27) in Algeria. The military directly chose the President and had informal veto power in many other political decisions (Lahouari, 2002, cited in Lutterbeck, 2012: 31). The military traditionally oversaw the political process from the background through informal influence (Aït-Aoudia, 2015; Martínez, 2000, 48–51). 46
From the mid-1980s, when oil revenues and support to the FLN declined, the political situation in Algeria started to change (Entelis, 1992: 46). The government attempted to address some of the public’s concerns through economic reforms and eventually conceded political reforms, including a new constitution that allowed the formation of political associations. Just two weeks after its approval in 1989, the Islamist movement founded the Front Islamique du Salut (FIS) and the masses took to the street to show their support for the democratization process and the FIS. The Constitution also limited for the first time the power of the military and removed the clause acknowledging the role of the military in the development of the country.
Facing a threat to its own power position and its role as defender of stability, the military promptly reacted and pushed to increase its involvement in politics. Alongside the violent repression of mass protests (Martínez, 2000), the military used systematically the surge of terrorist attacks occurred from 1989 until the coup in 1992 as a justification for its increasing interference in politics. 47 Both in the public and military press such as El Djeich, the military explicitly connected their prominent role in defending Algerian democracy to specific terrorist attacks. To illustrate, in June 1989 in Ouargla, a child died when a band of criminals burned down a house. This event was immediately labeled as terrorism by the military establishment and connected to the strong need to defend democracy and political stability (Lahouari, 2002). A few weeks after this event, the Parliament approved the state security court regime and shortly thereafter in early 1990, the state of siege, entailing severe restrictions to the right of association and imposing a curfew. 48
As early as 1989, high-ranking officers and politicians with a military background started issuing more explicit political statements, portraying the military as the defenders of the country against a terrorist insurgency (Martínez, 2000: 164–167). 49 For instance, the minister of national defense, Major General Khaled Nezzar, supported the military’s resolve to “respond to any organized excesses that might jeopardize the national unity of the country (. . .) [and] would not hesitate to intervene to re-establish order and unity so that force remains in the hands of the law” (Nezzar, 1999).
In line with its traditional hands-off approach, the military initially remained ambivalent toward FIS and its leader Abassi, even after FIS won the local elections in 1990 and the first round of national elections in 1991. But when Abassi started to be more openly critical toward the military in early 1991 and it became clear that FIS was negotiating a deal with the President for limiting the military’s influence, the military became critical and verbally aggressive against FIS. 50 A key member of the Joint Chiefs of Staff, General Chelloufi, reacted very strongly: “I will not be tolerant of those who use democracy in order to return to dictatorship once they have achieved power” (Cook, 2007: 53).
In addition to overemphasizing the terrorist threat in general, and the narrative connecting terrorists with the military role as defender of stability, from June 1991, the military linked all terrorist attacks to the FIS. 51 For instance, when the commander of an army brigade in Cheraga was killed and shortly thereafter a policeman was kidnapped and then murdered in Lakhdaria by what was allegedly a criminal group, the military reaction was strong and accused again “terrorists from within the country, who threaten to weaken our values and our democracy.” 52 Similarly, when in November 1991, the jihadist group led by Tayeb El-Afghani launched an attack killing eight militaries, the military connected it to FIS contentious strategies. At that time, the military argued that the military response was absolutely necessary for defending the country from subversive groups. 53 Eventually, however, the military used violent coercion to contain the rise of the FIS: it cancelled the second round of voting, annulled the first round, Benjedid was removed from office and, shortly after, the FIS was declared illegal.
In sum, the case of Algeria between 1989 and 1992 illustrates nicely the “pushing” mechanism of our theory on MIP. First, the Algerian military systematically overemphasized terrorist attacks and threats to justify their autonomy and disseminated a narrative that portrayed them as the guardian of the state against those who destabilized the country with terrorist means. Second, it linked its most prominent political enemy to this terrorist threat. In both phases, terrorism was crucial to justify the military pushing its way into politics. Taken together, the illustrative examples from France and Algeria shed light on our mechanisms.
Data
We assess our hypotheses using subjective and objective data of MIP for all independent states from 1984 to 2004. These measures are regressed against the observed domestic and transnational terror attacks and our estimations of the ex-ante probability of domestic and transnational terror events, along with a set of controls.
Dependent variables
We use an indicator for military involvement in politics from the International Country Risk Guide (ICRG), a system based on a set of 22 components grouped into three major risk categories: political, financial, and economic. Political risk comprises 12 components. We use the military in politics component, classified on a scale between 0 and 6. 54 The assessments are made by ICRG editors based on subjective analysis of the available information on each country and are informed by a range of criteria, such as the size of the military, spending and budget levels, the participation of the military in the executive and legislative branches/orders of government, a country’s democratic tradition, and a country’s democratic traditions, among others. These values are based on several pre-set questions to ensure consistency between countries and over time (Howell, 2011). 55 ICRG risk subcomponents have been extensively employed in many studies in political science, economics, and finance (Acemoglu et al., 2001; Bove and Nisticò, 2014b; Dreher et al., 2017; Pinkowitz et al., 2006). This is a reliable measure for the analysis at hand. Consider, for example, one democracy, South Korea, and one dictatorship, Indonesia. South Korea embarked on a stable democratic transition in the 1980s, following a military dictatorship. Yet the country has never been completely immune from a degree of military interference in politics. During the Chun Doohwan administration (1981–1988), 8.7% of the members of the National Assembly and 20.8% of the members of the government had a military background, whereas the Roh Tae-woo administration (1988–1993) saw these shares slightly decreasing to 5 and 18.5%, respectively (Saxer, 2004). The average level of military involvement in politics was 3.9 out of 6 throughout this period, while it peaked to 4.2 in 2004. According to Saxer (2004), in this period 32% of the members of the National Defense Committee in the National Assembly were former military officers and President Roh Moo-hyun’s first defense minister was a former general. Similarly, during the final years of the authoritarian government in Indonesia, President Suharto employed about 14,000 officers outside the armed forces. In fact, about 50% of the provincial governors and over 30% of district heads had military backgrounds (Clear, 2005). Indonesia’s ICRG score is between 5 and 6, higher than military dictatorships in Guatemala (between 4.2 and 4.8) or Niger (between 3.2 and 4.2) in the same period.
Key independent variables
For our main independent variables, we use information on terrorist incidents from the Global Terrorism Database (GTD), distinguished between domestic and transnational terrorist events (Enders et al., 2011). Terrorism is defined as “the premeditated use or threat to use violence by individuals or subnational groups against noncombatants to obtain a political or social objective through the intimidation of a large audience beyond that of the immediate victims” (Ibid., 321). As discussed above, we expect military involvement to be affected by domestic terror attacks, in which the venue, target, and perpetrators are all from the same country, as well as by transnational terrorist events, involving victims and perpetrators of different nationalities and/or crossing interstate borders (Ibid.). We employ counts of all terror attacks—domestic and transnational—and assess whether there are substantive differences between these two types of terrorism. Data on the occurrence of transnational terrorist incidents are taken from the International Terrorism: Attributes of Terrorist Events (ITERATE) Database (Mickolus et al., 2007). 56
Our second hypothesis relates perceived terrorist threats and the risk of MIP. Following Nordhaus et al. (2012), we estimate the ex ante probability of domestic terror attacks to capture a “threat perception effect.” 57 We build on Wilson and Piazza (2013) to calculate the prospective probability that a country will become the target of domestic terrorism. Wilson and Piazza model domestic terrorism as a function of democracy, the gross national income per capita, population size, area, income inequality, regime durability, state failure, and dummies for the Cold War, civil wars, and international wars. 58 We also include lagged dependent variables to model inertia in the use of terrorism and address additional temporal dynamics. Using the statistical results of a simple probit model, we estimate the predicted probability of domestic terrorism for each country every year.
Similarly, we replicate Gelpi and Avdan’s model to get an estimate of the prospective risk of international terrorism for each country. They use a set of directed dyads, where the dependent variable is coded “1” if a terrorist incident was recorded against a target state in each year from a group operating in the origin state and zero otherwise. Gelpi and Avdan (2018) show that a simple model that includes only the distance between states and the prior history of terrorism “yields forecasts that are nearly as accurate as the predictions of a model with more than 20 covariates” (Ibid.:19). We thus employ this parsimonious model to estimate transnational terrorist attacks within each dyad every year, and then create an average predicted risk of transnational terrorism for each target state.
Control variables
Much research shows that domestic and external security threat environments increase the risk of coups. We therefore incorporate a dummy variable for international wars to account for the independent effect of other concurrent major threats. Moreover, we follow Aksoy et al. (2015) and create two mutually exclusive variables that indicate whether a country is involved in a civil war crossing the 25 battle deaths threshold per year (low-intensity civil war) or the one thousand-battle-deaths threshold (high-intensity civil conflict). Whereas the high-intensity dummy variable picks up severe civil wars, the 25-deaths threshold “accounts for the numerous conflicts that involve organized insurgents but do not result in a high number of yearly casualties” (Ibid.:443).
To ensure that our variable is not capturing military coups, we exclude all country-year observations with successful coups using data from Powell (2012). Yet as the risk of successful coups could also affect the measurement of our dependent variables, we add a binary variable for attempted (but failed) coups at time t, and a dummy variable indicating whether there was a coup attempt (regardless of whether it succeeded or failed) in the preceding five-year period. We also include a country’s GDP per capita since the risk of coups is lower in richer societies using data from the World Bank. Although this does not necessarily suggest an effect on military involvement, it is also plausible that civilian authorities in wealthier states achieve a higher level of control over the military. We also consider a country’s population size since larger countries tend to be regional or global powers, which could be linked to the status and prestige of the armed forces. Furthermore, we control for the possibility that democracies develop a higher level of civilian control over the military using the Polity2 variable from the Polity IV dataset (Marshall et al., 2017).
Regime age and leadership’s tenure are also likely to affect MIP, as the longer the length of a regime or the tenure of a leader, the lower the risk of military participation in politics. We thus control for regime durability, the natural log of the number of years since the most recent regime change or the end of a transition period (see Marshall et al., 2017 for more detail). Similarly, we include the log number of years that a chief executive has been in office from Cruz and Scartascini (2016). Our model also incorporates the type of authoritarian regimes from Geddes et al. (2014), i.e. military, monarchic, personal, and single party. MIP might be driven by rent-seeking such as land use and legal and illegal business opportunities, and case studies suggest a strong positive relationship between corruption levels and the role of the military in politics (Majeed and MacDonald, 2010). We thus include the level of corruption of a country, from the ICRG dataset. Finally, we address the issues of spatial dependence by controlling for the average level of military in politics in contiguous states to directly account for spatial contagion. 59
Results
We run two-way fixed-effect models (i.e. with country- and year-fixed effects) that use a within-transformation, allowing us to control for the likely omission of country-specific determinants of MIP and terrorism, and for unobserved common trends. By exploiting within-country variation, it also allows us to assess the impact of changes in terrorism on changes in MIP over time within the same country. We transform all positive nondichotomous variables into logs to simplify the interpretation of the coefficients, scale down the variance, and reduce the effect of outliers. 60 Table B1 in the Appendix reports the summary statistics.
Table 1 reports the impact of different forms of terrorist violence on the ICRG military in politics measure. Model (i) is a “naive” estimation that merely comprises the controls, next to the country and year fixed effects. Hence, Model (i) neglects that actual terrorism and the risk of terrorist violence can also affect MIP. Models (ii)–(viii) include our variables of interest, omitting controls to demonstrate that our findings are not affected by their inclusion (Clarke, 2009). Before turning to our variables of interest, we summarize the results for the controls, based on estimates shown in Table 1. MIP decreases with a country’s level of democracy, as the coefficient for the polity score is negative and statistically significant. The length of a political regime is negatively associated with MIP, which is consistent with studies showing that regime duration decreases the risk of coups. Corruption and military regimes are associated with higher levels of MIP, whereas the coefficient of single-party regimes is negative. Coup attempts and both low- and high-intensity civil wars increase MIP. Interstate armed conflicts bears a positive but statistically insignificant coefficient, possibly related to the reduction of international wars in the last decades. Finally, per capita GDP, population, leadership’s tenure, coups in the previous five years and the level of MIP in the neighborhood do not reach statistical significance at conventional levels. Note that our estimates are quite conservative, and our empirical strategy likely absorbs much of the effects of the controls in either the country or the year fixed-effects.
Terrorism and Military Involvement in Politics (ICRG).
$* p < 0.10, ** p < 0.05, *** p < 0.01$. Standard errors in parentheses are clustered at country level. Two-way fixed-effects OLS.
Observed terror attacks are positively signed and highly significant in models (ii)–(v), lending strong support for Hypothesis 1. The coefficients allow for a direct interpretation and can be interpreted as elasticities. Therefore, a 10% increase in the total number of terrorist attacks (Terrorism) corresponds to an increase in the degree of military involvement of 0.3%, with similar results when we use a dummy variable rather than the count of terrorist incidents. Similarly, domestic and transnational terrorism have coefficients of comparable magnitude. Models (vi)–(viii) reveal the effect of terrorist threats, as measured by the probability of terrorism and probability of domestic and transnational terrorist incidents. The probability of aggregate terrorism is significant and when we increase this indicator by 10% we see an increase of 0.5% in the outcome variable, holding all other variables constant. Neither the probability of domestic terrorism nor the odds of transnational terrorism are statistically significant.
In Table 2, we present models that incorporate our variables of interest, the control variables (shown in the Appendix, Table B2) and two-way fixed effects. Comparing uncontrolled results and results with controls offers interesting insights. As seen, our main results do not substantively change by the inclusion of controls, and the statistical significance of the coefficients are mostly unaltered. Actual terrorism does affect the way military exerts institutional influence. Yet whereas domestic terrorism is still positively signed and significant, transnational terrorism does not affect MIP. Moreover, we find again that our proxies for the threat of domestic and transnational terrorism are not significant at convention levels.
Terrorism and Military Involvement in Politics (ICRG).
$* p < 0.10, ** p < 0.05, *** p < 0.01$. Standard errors in parentheses are clustered at country level. Two-way fixed-effects OLS. Control variables include: Population (ln), GDP per capita (ln), Polity score, Regime durability (ln), Leader tenure (ln), Corruption (ln), Personal regime, Military regime, Single-party regime, MilPol in neighborhood, Coup attempt, Coup attempt (last 5 ys), Low-intensity civil war, High-intensity civil war, Interstate conflict.
We assess the robustness of our findings with a variety of additional model specifications. We report these results in the Online Appendix, sections B3-B8. A potential objection about the use of the ICRG measure is that the data generation process entails experts’ subjective judgmental decisions that may produce biased information. Therefore, we also employ two objective and observable measures of MIP. The first measure is taken from White (2017), who has recently introduced the Military Participation in Government (MPG) Dataset. It captures the proportion of a state’s cabinet, state council, or equivalent that is made up of military officers. The second one is a dummy indicator, which takes value “1” if a state’s defense minister is a military officer for the duration of his term—with no indication of formal retirement when they assumed office—and “0” otherwise (from Cruz and Scartascini, 2016). The estimates using alternative measures for military involvement in politics are consistent with results reported above and are discussed in the Online Appendix.
Second, the ICRG measure could be also treated as a categorical and ordered variable. Using an ordered probit model yields estimates that are virtually identical to those reported in Table 1. Third, we estimate probit models with random effects, as well as rare event logit models to assess the robustness of our findings. Fourth, whether terrorism successfully increases the level of MIP may depend on a country’s political environment, in particular its regime type. Indeed, someone may argue that democracies are less sensitive to the pressure of military actors in presence of terrorist violence; however, we find no evidence that regime-type moderates the effect of terrorism on MIP. The models in the Online Appendix increase the confidence in the main results and show that our findings do not depend on specific decisions regarding to research design and operationalization of the dependent variable.
To sum up, we find that observed political violence by terrorist groups is associated with an increase in military participation in politics. At the same time, the threat of terror attacks also matters, and we find a positive and significant relation between the probability of terrorism and MIP. Interestingly, neither the probability of domestic terrorism nor the odds of transnational terrorism seem to be associated with MIP, whereas it is only the probability of aggregate terrorism that is statistically significant.
Conclusion
Much research has explored the determinants of coups but little macro-quantitative research examines how the military intervenes in politics, without deposing political leaders. This is puzzling given a large tradition of case-based research suggesting that state armed actors often acquire significant levels of institutional autonomy and can influence policymaking processes in relevant areas. Furthermore, the salience of terrorism in the contemporary world has generated a great deal of scholarly attention on its causes, but its political repercussions remain poorly understood, particularly compared to research on the political consequences of civil and international wars. This study contributes to the literature by examining the relationship between terrorist violence and MIP. We argued that terror attacks and terrorist threats modify the opportunity structure that facilitates an increasing involvement of the armed forces in politics. Our empirical analysis provided strong support to these claims, showing that terrorism and perceived threats from terrorist organizations are associated with higher levels of MIP. Future research could explore the effects of terrorism on other agencies within the security sectors, and also strive to better integrate the different research avenues within the civil–military relations field (see Brooks, 2019: 15).
Understanding how terrorism affects civil–military relations is important as terrorism can have negative externalities on the quality of democracy and democracy promotion. We have shown that terrorist violence facilitates state armed actors’ participation in politics, who can increase their relative role in policy-making processes. Related research has found that military actors tend to support harsh repressive measures to fight terrorism and strengthen homeland security (Conrad et al., 2017). It is perhaps not surprising that terrorist violence is linked to different forms of state-sponsored repression, regardless of regime type (Wagstaff, 2013). Furthermore, civil liberties restrictions and some repressive strategies aimed at reducing terrorist threats generally find support among citizens (Davis and Silver, 2004). Taken together, these studies suggest that terrorism and MIP interact in a way that can hinder states’ respect for human rights. In the context of the so-called war on terror, counterterrorist strategies promoted by foreign powers—most notably the United States—often draw on prescriptions endorsed by military actors who often privilege security and stability over respect for civil liberties and political rights (Shah, 2011; Wagstaff, 2013). Thus, increasing military involvement in foreign policy as a result of the “war on terror” can have detrimental effects on democratic institutions.
Supplemental Material
EJT866499_supp_mat – Supplemental material for Beyond coups: terrorism and military involvement in politics
Supplemental material, EJT866499_supp_mat for Beyond coups: terrorism and military involvement in politics by Vincenzo Bove, Mauricio Rivera and Chiara Ruffa in European Journal of International Relations
Footnotes
Acknowledgements
The authors are listed in alphabetical order; equal authorship is implied. We are grateful for valuable comments and discussions with Risa Brooks, Tobias Böhmelt, Erica Frantz, Danielle Lupton, Nikolay Marinov, Maline Meiske, Gerry Munck, Brian Phillips, Florian Reiche, Pascal Vennesson and the participants to panels and seminars at the Department of Politics and International Relations at Oxford University, the International Studies Association 2019 in Toronto and the Hertie School of Governance.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental material
Supplemental material for this article is available online.
Notes
Author biographies
References
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.
