5. Dictatorial Survival Strategies in Challenging Conditions : Factionalized Armed Supporters and Party Creation
When Benin became independent in 1960, three parties dominated politics, each rooted in a region and its ethnic groups. The same regional loyalties factionalized the newly created army. Most officers came from the south. They had started in the ranks during French rule and were promoted rapidly in order to indigenize the officer corps at independence. Most enlisted men came from the north (Decalo 1976, 55–57). As of 1965, when Benin’s second coup occurred, the army was a few years old and had only 1,700 men. It had 43 indigenous officers and 12 French ones (Bebler 1973, 12–13). During the first decade after independence, divisions between older officers, who were rapidly promoted as the new nation built its army, and younger ones trained in military schools, whose promotions were soon blocked by budget crises, reinforced and overlaid regional factionalism within the army (Decalo 1976).
After the first coup in 1963, soldiers replaced the civilian president with a different civilian and returned to the barracks. Officers themselves took power after the1965 coup. Factionalism undermined Benin’s first military dictatorship, which ended with a coup in 1967 that brought to power a new military regime of mid-ranking officers from a different ethnic group. The second military dictatorship attempted to deal with army factionalism by appointing an ethnically balanced cabinet. The new cabinet included captains, lieutenants, and NCOs in the dictatorship’s inner circle to make sure that all army interests were represented. The leadership also dismissed some southern senior officers and promoted some from the north in an effort to equalize opportunities (Bebler 1973, 20–23). Nevertheless, another faction-based coup ousted this regime two years later. Officers then tried to reunify the military around an inclusive decision-making body with collegial leadership, but were unsuccessful. They returned power to civilians in 1970 because they could not find a successful formula for power sharing within the officer corps.
Benin’s history in the 1960s is a story of military dictators’ repeated failure to consolidate their rule because the factionalized officer corps could not provide stable support. Decalo describes the army as a “patchwork of competing personalist/ethnic allegiance-pyramids centered around officers of all ranks in which superior rank or authority was only grudgingly acknowledged” (1979, 234).
A seizure group that includes many members with control over armed force should be able to achieve an especially advantageous power-sharing arrangement with the leader because their threats to oust the dictator are highly credible. Responding to these threats, early military dictators in Benin agreed to oversight by broadly representative groups of officers, but to no avail. These strategies failed because officers included in the regime’s inner circle could not prevent rogue coups led by other officers formally subordinate to them. Despite successive dictators’ efforts to secure stability by consulting with representatives of many military factions, some officers always remained dissatisfied and quick to oust the current dictatorship. These failures were caused not by the inability of dictators’ allies to make credible threats to oust the dictator if he fails to share, but rather by their inability to make good on promises of support when hewas sharing. In that situation, a dictator cannot make himself safe by sharing more and more because dictators, like all other political leaders, face budget constraints.
This destructive game of musical chairs among military factions ended a few years after the 1972 coup that brought Major Mathieu Kérékou to power. Kérékou initially followed the same strategy as earlier military dictators. He dismissed all senior officers, appointed a cabinet of junior officers, and consulted a military ruling council representative of major military factions. Despite Kérékou’s effort to consult all factions, the military remained unable to provide a stable base of support. Kérékou survived several coup attempts during his first two years in power (Decalo 1976, 76–84). Then, in 1974 Kérékou began creating an organized civilian support base, the Benin People’s Revolutionary Party (RPB), to counterbalance the military and help stabilize his rule. Six “close friends of the president” made up the new party’s politburo (Martin 1986, 68). Over time, Kérékou gradually increased civilian participation in inner-circle decision-making. Their support enabled him to remove the most threatening rival officers from posts from which armed challenges could be launched (Martin 1986, 68–75). Adding a loyal civilian support group to balance the factionalized military stabilized Kérékou’s rule. He retained control until 1990, when widespread popular opposition forced him to democratize.
In this chapter, we explore two theoretical ideas highlighted by Benin’s history. First, armed supporters can drive a hard bargain with the dictator when they maintain unity but not otherwise; factionalism prevents them from making credible promises to support the dictator if he shares. To be viable, a powersharing bargain between the dictator and his allies must be credible on both sides: allies must believe the dictator’s promise to share, and the dictator must believe the allies’ promises to support him if and only if he shares. The allies’ promise of support is not credible, however, if they cannot commit their own subordinates to honor the bargains they make. The second theoretical idea is that dictators can in some circumstances change the balance of forces within the ruling group by empowering new political actors. Kérékou did this when he organized a new support party and brought its leaders into the inner circle of the dictatorship.
In Chapter 4, we analyzed how and why dictators exclude rivals from their inner circle. Here we focus on the dictator’s strategic inclusion of new players he expects to be more malleable and less dangerous than the original members. We also consider the conditions under which he has the ability to exercise this option.
In the first part of this chapter, we describe the especially difficult situation facing dictators whose power initially depends on an armed support base, or what Alex De Waal (2015) calls specialists in violence. In the next, we spell out our argument about how the interaction of dispersed armed force and factionalism in the seizure group can limit the dictator’s options for bargaining with these supporters.1 The third section considers the dictator’s incentives to bring into the inner circle political actors who are more dependent on him for benefits and protection than are specialists in violence (military officers or others who command armed subordinates). It explains why civilian support groups are less dangerous to dictators than armed factions, and it discusses party creation as a strategy for mobilizing organized civilian support to counterbalance unreliable armed supporters. The fourth section provides evidence that party creation in dictatorships that lack support parties when they seize power contributes to a strategy designed to concentrate more power and resources in the dictator’s hands, reducing his need to rely on members of the original seizure group. The last sections show evidence that fewer coups against dictators occur after party creation and that dictatorships that create parties post-seizure last longer than otherwise similar regimes lacking party organization.
the strategic context
Dictators who achieve power through armed force face an especially difficult survival problem. Consider the dilemma faced by a dictator who comes to power in a coup.2 Besides the support of fellow plotters, he probably has the initial support of some civilian elites who were fed up with the ousted incumbent but who would lack the capacity to save the new leader from ouster if his armed supporters turned against him.
At the outset he often has quite a bit of popular support as well, and for the same reason: dissatisfaction with the overthrown incumbent. This popular support, however, is unorganized, opportunistic, and superficial. “The people” or “the street” may support a coup in order to rid themselves of an incompetent, brutal, or disreputable leader, but that support will evaporate if the economy fails to improve quickly or if any other disaster strikes. Much of the citizenry can swing rapidly from support to opposition. Egypt’s recent experience illustrates the volatility of unorganized popular support: Egyptian protestors forced a military-supported dictatorship from power in 2011, but in 2013 helped oust a democratically elected president and return the military to power.
The coup leader needs widespread support from other officers to survive in power when popular opposition arises, as it inevitably does. Military dictators have the support or acquiescence of many other officers at the time of seizure (otherwise their coup would have failed), but no way of guaranteeing that support in the future. Even at the beginning, support from the rest of the military might be quite superficial. As shown in earlier work (Geddes 2003), the incentives facing military officers when their colleagues initiate a coup create a first-mover advantage similar to that in “battle of the sexes” games. This means that if a small group of officers makes a successful first “coup move,” such as seizing the airport and presidential palace, the rest of the military tends to go along, whether they sincerely want the intervention or not. So, the fact that a seizure of power has occurred implies temporary acquiescence by the rest of the officer corps but very little in terms of sincere or long-term support. In short, military dictators cannot count on the support of fellow officers tomorrow even if they have it today, and today’s support may be shallower than it appears at the time the military seizes power.
Within the dictatorial inner circle, support for the dictator may also be short lived. Even though the dictator was a brother officer – and often a longtime ally and friend of other high-ranking officers – the day before he was selected as supreme leader, their interests diverge after his selection just as those of civilian dictators and their supporters do. Officers in the support group want to ensure large military budgets and their continued monopolization of force. They do not want their budget reduced in order to hire more security police or provide patronage jobs for civilian supporters of the dictator. They oppose the creation of presidential guards or people’s militias to protect the dictator because such forces challenge their own monopoly of force and the credibility of their threats to oust the dictator if he fails to share, as well as depleting the budget. They have an interest in consultation within the officer corps before major policy decisions. They want to avoid the concentration of power, resources, and discretion in one man’s hands, as well as the favoritism and deprofessionalization within the military that often accompany it. Dictators who succeed in concentrating resources and discretion in their own hands, that is, in personalizing power, threaten both the military as an institution and individual officers, since they control promotions, postings, forced retirements, and access to profit opportunities (both legal and illegal). At the extreme, such dictators control life and death through their personal control of the security apparatus, and military officers have no special immunity from security police. The same logic applies to other dictators brought to power by specialists in violence.
Since coups have ousted most dictators, especially those who come from the military, it is obvious that officers can be dangerous. Weapons, know-how, and the command of troops are widely dispersed in armies and in some insurgencies. This reality creates a hazardous environment for dictators, because coups require only a small number of individual plotters to execute them. Indeed, coups involving fewer than twenty men have occasionally succeeded.3 Only a minority of coups involve consensus among the whole officer corps. Instead, small conspiratorial groups of officers carry out most coups using the firstmover strategy (Nordlinger 1977). In short, many different small groups of officers could stage a coup with reasonable prospects for success.
the interaction of dispersed arms and factionalism
Though much of the literature on autocracies has emphasized dictators’ credibility problem,4 it is not the main impediment to successful sharing bargains when arms are widely dispersed within the ruling group. Military dictators can increase the credibility of their promises to share in some ways not available to civilians, for example, by retiring from active duty.5 Once they no longer command other officers, do not determine promotions, and cannot decide which officers will command which garrisons, leaders’ ability to deter coups depends almost entirely on their ability to satisfy the rest of the military’s policy and budget demands. This makes their promises credible. Military dictators can also increase their credibility by leaving internal security services within the military chain of command, thus limiting their ability to spy on, intimidate, and murder other officers.
The more serious impediment to successful power-sharing bargains is that armed supporters’ promises not to oust as long as the dictator shares are never completely credible because they cannot always prevent “rogue” coups or armed ouster by other specialists in violence.6 These are coups by factions, often led by lower-ranked officers, that could be defeated if the rest of the armed forces mobilized against them, but the dictator cannot count on the rest of the army doing so. The first-mover advantage built into the incentives facing officers means that a faction that makes a credible first coup move without being met by violent opposition can overthrow the government because the rest of the armed forces will acquiesce to this coup just as they did to the one that brought the current leader to power.7 All dictators face some risk from armed supporters, but the less control commanding officers have over lower-ranked officers – that is, the less disciplined and unified the armed support group is – the less ability officers in the inner circle have to make enforceable bargains with the dictator that would reduce the risk.
An alternative way to express this problem is to note that militaries and other armed groups are not unitary actors, though some more closely approximate unity than others. In more professionalized militaries and unified insurgent groups, individuals’ future career success is inextricably bound to following the orders of their commanders. Professional success depends on obedience to orders from the day that youths enter military school. Subordinates obey superiors regardless of personal, ethnic, or political loyalties. The likelihood of a rogue coup succeeding is low in unified militaries, and the cost of a failed attempt very high because serious breaches of discipline end careers. Coup leaders can also face court-martial, jail time, or execution. In this kind of military institution, commanding officers can count on lower-ranked officers to obey orders, which transforms a large group of individuals into something approximating a unitary actor. Unitary actors can make credible promises.
Factionalized militaries were common between 1946 and 2010, however, especially in newly independent countries. A factionalized army, “far from being a model of hierarchical organization, tends to be an assemblage of armed men who may or may not obey their officers” (Zolberg 1968, 72). In factionalized forces, discipline is less predictably enforced because personal, partisan, or ethnic loyalties can cross-cut military hierarchy. “[S]oldiers often [have] a stronger sense of commitment to their unit commander than to the army” (Crouch 1978, 27). Promotion and protection for lower-ranked officers, NCOs, and soldiers depend on faction leaders, not just on compliance with orders and military norms. Routes to higher rank other than the slow but predictable rise via increasing seniority are more available. Consequently, lower-ranked officers may disobey the orders of higher-ranked officers if they conflict with those of faction leaders. Support for coups can be a short-cut to rapid promotion, and punishment for coup attempts is not always severe. Some coup attempts result only in demotion, apparently because dictators fear the consequences of imposing harsher punishments in factionalized militaries. Others result in dismissal, but when changes at the top occur later, as they often do in countries with factionalized militaries, dismissed officers may be reinstated.
Officers included in the inner circle in this kind of setting cannot commit their subordinates to abide by the sharing bargain as they would be able to in a more unified military institution because faction leaders command the obedience of their members but do not offer unconditional obedience to hierarchical superiors. If factions become dissatisfied with their share or oppose a policy decision, they have a reasonable chance of bringing off a successful coup.
If the dictator has reason to doubt higher-ranked officers’ ability to prevent factions from launching rogue coups, it makes little sense for him to share and consult because doing so will not protect him. Instead, he will renege on sharing agreements and use the resources saved to pursue other strategies for deterring coups.
the strategic creation of new political actors
Dictators facing a factionalized armed support base use several strategies for trying to deter coups and thus reduce their vulnerability. They spend heavily on the military. They promote loyal officers and retire opponents or appoint them as ambassadors to faraway places. They resist resigning from active service so that they can maintain personal control of promotions and command assignments. They strengthen security police and try to take personal control of them. They create paramilitary forces and presidential guards led by relatives and recruited from their home regions to counterbalance the regular military.
These strategies show how much dictators fear military supporters. Though these strategies may sometimes be useful, they can also backfire because other officers try to defend their own positions of power, which depend on the credible threat to oust the dictator. Officers resent and sometimes resist policies that undermine that threat. Promotion of the dictator’s friends over the heads of competent officers violates military norms and leads to anger. Officers who have been passed over for promotion, dismissed, jailed, or exiled have led many coups and insurgencies. Officers may also resent the creation of paramilitary forces. In other words, these strategies are not risk free. They are highly visible to officers and may trigger the outcome they were designed to prevent.
Bangladeshi dictator General Mohammad Zia ul-Haq, for example, created a paramilitary force soon after achieving power and remained active duty chief of staff for several years to try to control the officer corps, but every time he intervened in promotions, mutinies broke out. He frequently transferred officers, sent them for new training, and raised salaries (Codron 2007, 14). Despite these measures and many executions of rebellious soldiers, he “did not manage to make the military a safe constituency to back his rule” (Codron 2007, 15). He began organizing a support party “to create a civilian power base when he failed to achieve the united support of the armed forces” (Rizvi 1985, 226).
Creating a civilian support organization, as General Zia did, is a subtler and often safer strategy for counterbalancing the military. Paul Lewis (1980) interprets Paraguayan dictator General Alfredo Stroessner’s reorganization of the Colorado Party into a support vehicle in exactly this way. After describing Stroessner’s lavish spending on the military and attempts to manage its factions, Lewis notes:
Other Paraguayan presidents [also] tried to buy military support and surrounded themselves with trusted officers. In the end, however, they failed to keep the greedier or more ambitious soldiers in line. Stroessner . . . achieved a real advantage over his predecessors [by] fashioning . . . a dominant single-party regime, based on a purified and obedient mass organization . . . This instrument . . . makes it risky and unprofitable for [officers] to conspire against him. (1980, 124–25).
Samuel Decalo describes the civilianization of the military regime in Niger similarly: “Faced with continuous factionalism within both the Supreme Military Council (CSM) and his cabinet, Kountché progressively disencumbered himself of his most threatening . . . officer colleagues” (1990, 277). Kountché could not afford to challenge other officers openly. “[H]e was forced rather to ‘work his way around them by mobilizing the masses’” (p. 278).8 Egyptian military dictator Gamel Abdel Nasser established the National Union party “to strengthen his personal power and weaken the RCC [Revolutionary Command Council, dominated by the Free Officers]” (Perlmutter 1974, 143).
Organizing civilians aims at reducing the dictator’s dependence on the military. A civilian support base can change the calculations of potential coup plotters by reducing their chances of a credible first coup move. Civilian support for dictators, even if superficial and manipulated, can deter coups because officers do not want troops on their way to encircle the presidential palace to confront crowds of fellow citizens. As a Guatemalan officer explained to an interviewer after an aborted attempt to overthrow a military dictator, “with civilians standing in front of the artillery tanks, the commander didn’t want to cause civilian casualties” (Schirmer 1998, 218).
Many coups are bloodless. That is because potential coup leaders choose times when they expect little opposition. Since the Russian Revolution, officers have understood that asking troops to fire on their fellow citizens can lead to indiscipline, desertions, and mutiny.9 The military can of course defeat unarmed or lightly armed civilian demonstrators, but orders to beat or fire on civilians risk provoking defiance among troops, which would undermine the military institution and officers’ political power base, so officers exercise caution in what they demand of soldiers. The strategy of organizing a mass civilian support base – a new support party – helps dictators survive because of officers’ strong preference for unopposed coups. The “fear of having to deal with massive civilian opposition” deters military plotting (Brooker 1995, 111).
As an example of how deterrence can work in practice, consider this sequence of events in Paraguay. Two top officers central to General Stroessner’s military support base publicly criticized his decision to sign a treaty, which they claimed would compromise Paraguayan sovereignty. Such open criticism was extremely rare during Stroessner’s rule. Insiders interpreted it as a sign of widespread military disaffection, indicating danger of a coup to replace Stroessner. In response, Stroessner mobilized his civilian support vehicle, the Colorado Party, in a massive campaign. Letters supporting Stroessner poured in to the newspapers. The party organized a pro-Stroessner demonstration in the capital. They used posters, fliers, full-page newspaper ads, and sound trucks in every neighborhood to publicize the event. Local party activists contacted people in person. The party assembled more than 1,400 cars and trucks to transport people in and out of the capital. On the day of the demonstration, the vehicles deposited people at party headquarters where they were all given red party T-shirts and large pictures of the dictator to brandish during the demonstration. The party provided free lunch. All public officials and their families were required to attend. In these ways, a demonstration of 50,000 people, “enormous by Paraguayan standards,” turned out to support the dictator (Lewis 1980, 148). The military dropped its opposition to the treaty and made no coup attempt (Lewis 1980, 148–50).
The ability to mobilize mass demonstrations and overwhelming votes in support of the dictator makes organized civilians useful to dictators. Dictatorships provide organized civilian supporters with access to mass communications and government-controlled transportation, as in the Paraguayan example above. They supply party cadres with the carrots and sticks to make sure that ordinary people turn out for big demonstrations or vote when told to. Some autocratic parties also earn popular support by distributing benefits to ordinary citizens, providing good economic policy, and making opportunities for education and upward mobility available to people whose futures looked bleak before.
Parties’ usefulness to dictators does not depend on supplying benefits to citizens, however. Many of the toothless parties created after armed seizures of power are incompetent, abusive, corrupt, or simply inconsequential for most people. Cadres may sell the things they are supposed to distribute, and they may use their party positions to exploit their fellow citizens. According to Mobutu Sese Seku’s Commissioner of Political Affairs, for example, cadres of the party created after Mobutu’s seizure of power “treat the population with arrogance . . . [and] love to threaten [people] with arrest for any reason at all, large or small” (Callaghy 1984, 168). Nevertheless, these party cadres can turn out masses of citizens for votes and demonstrations of support by threatening to block their future access to important services or to turn the names of those who fail to participate over to security police. One of the main tasks of Mobutu’s party was organizing mass marches to demonstrate the people’s “unfailing attachment and support for the Father of the Nation” (Callaghy 1984, 324).
Party militants develop a vested interest in the dictator’s survival since the dictator supplies them with benefits in return for support. The dictator’s control of state revenues makes this strategy possible. Party officials and activists often draw salaries. They have preferential access to jobs in the state bureaucracy and schooling for their children. They have insider opportunities to form businesses subsidized by the government and to manage or even take ownership of expropriated businesses and land. Their connections help them to get lucrative government contracts and profit from restrictions on trade. They have the possibility of rising in the party to achieve the political power and, usually, wealth associated with high office. Party militants’ stake in regime maintenance derives from these advantages. Even where party activists enjoy no current benefits, their connections open up future possibilities for rewards and upward mobility (Svolik 2012). These benefits explain why dictators never have difficulty recruiting civilians into their support parties.
Dictatorships often also task party members, especially local party officials, with reporting suspicious behavior, hostile attitudes, and the presence of strangers in villages or neighborhoods. Party militias can be used to set up roadblocks to impede the movement of weapons and to patrol at night looking for clandestine meetings and other suspicious activity. The effectiveness of such mass spy networks varies a lot from one dictatorship to another, but in some, the pervasiveness of spies and informers makes it very hard for potential plotters to find ways of meeting and communicating with each other. In this way as well, civilian support parties can reduce the likelihood of coups.
Although many dictators who achieved power without party support create one after the seizure, most do not. About 40 percent of groups that established dictatorships after 1946 were organized as parties beforehand. If monarchies are excluded, military officers led about four-fifths of the nonparty seizure groups. Of those cases in which groups not organized as parties seized power, 38 percent of the time a party was later created to support the dictatorship and 9 percent of the time the dictator coopted and allied post-seizure with an existing party that had been organized during an earlier regime. In the latter scenario, the dictator often reorganized and purged the party, refashioning it into a personal support machine. In the other 52 percent of cases, the dictator never created or coopted a party – as would be expected if party creation is attractive to dictators who depend on the support of an officer corps riven by factions but not to those whose military support base is more unified and thus more stable.
evidence that post-seizure party creation aims to counterbalance factionalized armed supporters
The argument that dictators create parties after seizures of power in order to counterbalance factionalized armed support groups unable to make credible sharing bargains implies that a number of relationships should be observable in the real world. If authoritarian party creation is a strategic choice by dictators to protect themselves from armed rivals, we should see that dictators themselves initiate most party creations. We cannot observe that directly because public announcements may be untruthful or may not make clear who sought to create this new institution. We should, however, observe the following:
Most newly created dictatorial support parties should be led either by the dictator himself or one of his relatives or close allies.
If dictators create parties to counterbalance their armed rivals, we should also find that party creation brings with it lessened military influence on policymaking. As an example, consider events in the Somalian dictatorship led by Major General Siad Barre. When a group of colonels ousted the elected government in 1969, they formed the Supreme Revolutionary Council (SRC) of twenty-five officers to rule the country and invited Barre to lead it. Barre had to consult with other officers on the SRC to formulate policy. After surviving several coup attempts, in 1976 the dictator created a new ruling party (the Somali Revolutionary Socialist Party, SRSP).10 Decision-making was formally transferred from the military SRC to the seventy-five-man executive committee of the party, which included Barre’s civilian supporters as well as some of the officers from the SRC. The SRC was disbanded, ending Barre’s formal consultation with the military but retaining the most powerful officers in the party executive committee. After this first step toward military marginalization, the military had to share decision-making with selected civilians. Barre then further concentrated power in the party’s five-member politburo, which included Barre and his son-in-law, who headed the internal security service. Over time, Barre replaced officers in the politburo with civilians, further marginalizing the military. Though the SRC was revived in 1980, it functioned thereafter as a parallel structure to the SRSP, with Barre very clearly the key regime decision maker (Adibe 1995, 8). In general, we expect that:
Dictators create authoritarian support parties as part of an effort to marginalize armed supporters from policy-making.
Elections to confirm the dictator as president of the nation help to strip armed supporters of their role as king-maker and -breaker. Elections create an appearance of popular support aimed at undermining the feeling among officers that what the military gives it has the right to take away. Parties help incumbents “win” such elections even if no opposition candidates are permitted. Party activists, who are often public employees, spread the regime’s messages, distribute T-shirts and benefits, hold campaign events, and make sure that citizens turn out to vote and vote for the right candidate (if they have a choice). When legislatures have the task of anointing the president, the selection of legislative candidates achieves high importance. Ruling party executive committees typically choose candidates, usually in consultation with the dictator himself. Because parties help dictators control elections, we should expect party creation to predate elections to confirm the dictator.
Parties should be created before elections that confirm the dictator as national executive.
If the military were unified and able to bargain effectively, the dictator would be unable either to redistribute resources toward new civilian supporters unilaterally or to add new actors to the dictatorship’s inner circle at will. Consequently, dictators infrequently propose party creation in countries with unified armed forces.11 It is instructive therefore to observe one of the few times when a dictator misunderstood the strategic situation and tried to create a party despite the existence of a fairly unified army. General Gustavo Rojas Pinilla, the leader of a military regime in Colombia (1952–58), announced the creation of a new party, the Third Force movement. The idea received an unenthusiastic response from other officers, and languished unimplemented until the following year, when Rojas brought it forward again. But after meeting for several hours with an incensed group of more than a hundred other officers, Rojas let the party quietly disappear (Szulc 1959). They had made the threat to oust credible.
If more united seizure groups have greater ability to resist the personalization of rule, and hence party creation, then in military-led regimes, we might also expect to see more parties created in dictatorships led by junior officers compared with those led by senior officers. If the military is united and its hierarchy intact, junior officers who lead successful coups will turn power over to a senior officer immediately. This happens because a senior officer can command the cooperation of officers who do not share the goals of those who led the coup and because many officers would object to the violation of military norms involved in a junior officer becoming president and thus commander-in-chief of higher-ranked officers. The ability of junior officers actually to take power after a coup suggests severe factionalism. It indicates that factional loyalties have undermined the military’s conventional emphasis on hierarchy and discipline.
Military dictatorships led by junior officers should create more support parties than those led by higher-ranked officers.
Finally, we think it unlikely that party creation would be the only effort a threatened dictator would make to try to safeguard himself from coups. We expect that a dictator who fears rogue coups would also invest in internal security agencies to spy on possible plotters and that he would create new, more loyal armed forces to protect himself from army attempts to oust him. If party creation is part of a broad strategy for reducing reliance on the military as a base of support, then we might expect to see that dictators who create parties are also more likely to take personal control of the security forces and to establish paramilitary forces to counterbalance the military than are dictators who can rely on a united military support base.
Dictator control over internal security services should be more likely in dictatorships that create a new support party than in those that do not.
The establishment of new paramilitary forces to protect the dictator should be more likely in dictatorships that create a new support party than in those that do not.
In what follows, we use our data set (unless otherwise noted) to examine whether these expectations match empirical reality.
Is Party Creation Usually Initiated by the Dictator?
In three-quarters of dictatorships in which a party was created after the seizure of power, we find that the dictator or a close relative led the newly created party, as would be expected if he controls party creation. The dictator always delegated the leadership of the new party to a close ally if he did not keep the post for himself or a relative.
Is Post-Seizure Party Creation Part of a Military Marginalization Strategy?
If dictators initiate party creation as part of a strategy to concentrate power, then party creation should be part of a process of marginalizing other officers from policy-making. To examine whether the process visible in Somalia is more general, we investigate how the creation of a new support party influences bargaining between the dictator and other officers by examining three related measures of military marginalization: leadership rotation within the military ruling group, consultation with other officers about policy decisions, and military representation in the cabinet. Because this argument pertains to military-led regimes, we test these expectations on dictatorships that gained power through armed seizures of power (coup, rebellion, uprising, or foreign imposition) in which the military selects the dictator. We further restrict the analysis to regimes that did not inherit a regime support party so that we can examine the extent to which the creation of a new party influences military marginalization. Among these regimes, 42 percent create a new support party, while the majority rule without a political party.12
The strongest indicator of collegial decision-making is the regular rotation of the presidency among officers. We expect party creation to be associated with less leadership rotation since it helps the dictator to reduce the power of his armed supporters. Regular leadership rotation is relatively rare: it occurs (at some point) in only 9 percent of military-led regimes that came to power in armed seizures (and lacked an inherited support party). We test the relationship between party creation and leader rotation using a model with regime-case fixed effects to isolate the influence of creating a new support party.13 The model thus controls not only for cross-country variation in factors such as level of development and colonial legacy, but also for regime-specific features such as how the regime seized power, the prior experience of regime elites, and the institutional environment in which elites operate. We also control for two factors likely to influence the creation of new parties: whether the current dictator is the first leader of the regime, and the change in the international environment after the end of the Cold War to encourage the creation of “democratic-looking” political institutions such as parties. Figure 5.1 shows that post-seizure party creation (New party) reduces the likelihood of the regular rotation of the presidency by roughly 2.5 percent for military leaders who seized power without a party.
We next look at two other measures of military marginalization: lack of consultation with officers about policy decisions and civilianization of dictatorial policy-making. We capture the former using country specialists’ assessments of whether the dictator consults regularly with other officers. We measure the latter by looking at the composition of cabinets. Military representation in the cabinet is measured as whether the most important members of the cabinet – other than the defense minister – are active duty or recently retired military, police, or security officers. We define the “most important” ministries as the prime minister (if one exists), the ministry of interior or state (which in most countries controls the police, internal security agencies, and voting), and others that are particularly important in the country context (e.g., the ministry that deals with oil in oil-exporting countries). These forms of military marginalization are quite common: in more than half of these military-led regimes (56 percent) the dictator makes most decisions without regular military consultation; and in most (77 percent), cabinets have little military representation.
To analyze the effect of party creation on these measures of military marginalization, we again estimate a linear probability model with regime-case fixed effects, and controls for the Cold War period and first regime leader. This approach compares periods prior to party creation with periods after party creation within the same regime, and then pools these comparisons into one average estimate. It accounts for cross-country variation in economic and cultural factors as well as regime-specific features that affect elite decision-making. Figure 5.2 reports the results, showing that party creation is associated with increases in both the probability that dictators eschew consultation with other officers and the likelihood that officers are excluded from the most important cabinet posts.
We also expect that if dictators create parties as part of an effort to marginalize the military, they should often be established in the run-up to elections that confirm the dictator as national leader, either directly or through the election of a legislature tasked with doing so. To assess this, we again restrict our analysis to dictatorships that came to power via armed force and lacked a support party when they seized power. In order to model the creation of a new support party, the sample includes only the observation years in which the dictatorship lacked a support party in the previous year.
We use data on national-level elections in which the incumbent, his party, or his chosen successor appears on the ballot to identify election years. We test the likelihood of party creation in the year before or year of elections relative to other years of the same regime.14 As shown in Figure 5.3, we find that dictatorships are much more likely to create parties in the year of or before leader elections (31 percent) than at other times (4 percent). We also tested this prediction in a logistic regression model with control variables (Cold War, seizure type, a polynomial of years with no new party, and first leader), with a similar result. Finally, we tested a linear model with regime-case fixed effects to isolate variation over time within dictatorships. Again, we find that dictatorships are more likely (18 percent) to create new parties in the run-up to elections.
Can More United Militaries Deter Party Creation?
To test the hypothesis that a more unified officer corps can better deter party creation, we use two individual traits of autocratic leaders as proxies for factionalism/unity within the officer corps: their age when they seized power and their rank just prior to the seizure of power.15 As explained above, we think that the pre-seizure rank of the first dictator can be used as a proxy measure for factionalism. We reason that dictatorships led by highly ranked officers may depend on either factionalized or united officer corps for support (since a faction may be led by a high-ranked officer), but that those led by junior officers all rely on a factionalized military base. The future dictator’s rank before the seizure of power is exogenous to promotions and other decisions made by the dictatorship. We use a parallel logic with regard to the age of the first dictator. Youthful military dictators are either low-ranked officers or higher-ranking officers in newly created armies. We expect new armies to have more problems with factionalism and indiscipline. We restrict the analysis to first leaders in regimes that came to power in armed seizures of power after 1945; there are 135 dictators in this group.
Figure 5.4 shows the share of dictators in each category that created a new party. The left two bars show that dictators forty years old or younger at the time they seized power were almost twice as likely to create a new party as those older than forty.
Next, we divide these dictators into four categories: civilians before seizing power and military officers of different ranks.16 The four bars on the right show that two-thirds of low-ranking military dictators create new parties, while just over one-quarter of mid- and high-ranked officers do. Just over half of the civilian dictators to whom officers delegate power create post-seizure parties. In other words, young or low-ranking military dictators, who reflect factionalism in the armed forces, are even more likely to create parties than civilians.
Next, we examine whether these patterns persist when we control for potential confounders. To do this, we estimate a model that compares first regime leaders with one another, while controlling for seizure type (rebellion, uprising, and coup, with foreign imposition as the baseline category), indicators for whether the regime before the one being coded was a democracy or a military regime (with nonmilitary dictatorship as the reference category), and a variable measuring whether the dictator seized power before 1990. We cannot test dictator age at the time of seizing power in the same model as variables for officer rank because age and rank are highly correlated: more than 80 percent of low-ranking officers who seize power are aged forty years or younger, while almost 90 percent of mid- and high-ranking officers are older than forty.
Figure 5.5 shows the results. The estimates in the left panel show that older dictators are less likely to create new parties than younger ones, while the right panel shows that low-ranking officers are more likely than high-ranking officers to do so. If leader age and officer rank are good proxies for military factionalism, then these findings suggest that dictatorships launched by factionalized militaries are more prone to personalization. Previous democratic experience reduces the likelihood that a new dictatorial support party will be created, as does earlier experience of military rule, relative to earlier experience of civilianled autocracy.
Finally, as another proxy indicator of less professionalized military forces, we look at the newness of the military institution, the idea being that discipline and norms about hierarchy probably take some time to develop. In most previously colonized countries, the officer corps was created at around the time of independence, so we might expect to see more post-seizure party creation in new nations with new indigenous officer corps. To investigate this possibility, we examine the calendar time trend in post-seizure party creation by estimating a nonlinear model with a cubic calendar time polynomial and controls for how the regime seized power (coup, rebellion, and uprising, with foreign imposition as the reference category), leader’s time in office (log), and whether the dictator is the regime’s first leader.17 Figure 5.6 shows that early in the post–World War II period, when indigenous militaries in many countries were still quite new, dictators had the highest propensity to create new parties. Party creation drops to a low point in the mid-1970s, once the period of decolonization was largely finished, and remains low for the next forty years, with a slight rise as the Cold War ended and newly independent countries emerged from the Soviet Union. Some of these countries lacked indigenous officers and had to create new officer corps at independence just as the earlier wave of newly independent countries did, and leaders in some of the post-Soviet dictatorships created new parties.
Is Party Creation Part of a Strategy to Reduce the Dictator’s Vulnerability to Coups?
If dictators organize new support parties as part of a broad strategy to reduce the likelihood of coups, as suggested above, we would also expect them to engage in other coup-proofing strategies. Coup-proofing could include taking personal control of internal security in order to monitor potential plotters. It could also include establishing paramilitary forces such as presidential guards, which are often recruited from dictators’ home regions. Such paramilitary forces help to defend dictators by remaining loyal during coup attempts by regular military officers. In the dictatorships in which a party was created after the seizure of power, the dictator also takes personal control of internal security 72 percent of the time, compared with 27 percent of the time for dictators of regimes never supported by a party. Dictatorswho establish new support parties are three times as likely to establish paramilitary forces to protect them (52 percent) than are dictators who do not form post-seizure parties (17 percent).
As in the tests above, these relationships hold after we account for crosscountry variation in a model with fixed effects: creating new parties is associated with a 25 percent increase in the probability of personalizing the security apparatus and a 33 percent rise in the chances of forming a paramilitary force loyal to the dictator.
These relationships indicate that dictators initiate party creation as part of a strategy to reduce dependence on the regular military when their original armed support base was too factionalized to make commitments of stable support in exchange for power sharing credible. Party creation after armed seizures of power thus paradoxically contributes to the personalization of dictatorial rule. Parties originally organized to lead a revolution or the struggle for independence may develop enough internal unity and discipline to constrain the dictator, but parties created by sitting dictators rarely do so because the dictator himself chooses and dismisses their leaders and controls the resources they need to maintain themselves.
post-seizure party creation and dictatorial survival
We concur with other analysts in seeing dictatorial ruling parties as autocratic survival tools (Gandhi and Przeworski 2007; Gandhi 2008; Magaloni 2008). We believe that post-seizure parties, like inherited parties, prolong dictatorial survival beyond what it would have been without them, but we see parties created after armed seizures of power as playing a quite different role in bargaining among elite actors than that played by parties that lead seizures of power. Post-seizure parties reduce intra-elite conflict by helping the dictator to concentrate power at the expense of armed supporters. The reduction in elite conflict increases dictatorial longevity.
Beatriz Magaloni’s (2008) influential argument suggests that parties extend the life of dictatorships because they make possible credible intertemporal promises by the dictator to continue sharing spoils if allies continue supporting him. She suggests that parties can solve dictators’ credibility problem if the dictator delegates control over appointments to high offices, including the dictatorship itself, to the party. When the party controls access to office, the dictator has reason to fulfill his promises because he knows he can be ousted, and his allies have reason to remain loyal because they can expect higher offices in the future. She refers to this form of power sharing as delegation by the dictator to the party, which implies that the dictator’s own interests are served by it.
That seems a dubious assumption. Such agreements prolong regime survival while limiting both the dictator’s time in power and his resources while in power. Thus, they do not appear to serve the dictator’s interests. We suggest that dictators agree to such arrangements when they are weak relative to the rest of the dictatorial elite. That is, they agree to limit their own discretion when their choice is between becoming dictator under constraints and not being dictator at all. We can identify two empirical conditions that contribute to the dictator’s relative weakness: (1) very recent accession to leadership, which we discuss in greater detail in Chapter 8 and (2) the wide dispersion of armed force among factions within the ruling group. Weakness forces the dictator to grant this form of power sharing. If power has already begun to be concentrated in the dictator’s hands when the party is created, or if creating a new party helps the dictator reduce his dependence on armed supporters, he has less reason to delegate powers that may increase regime survival but not his own time in office. After most seizures of power through force, the dictator does not delegate control over highest offices to the party.
With the notable exception of the PRI regime in Mexico, nearly all dictatorships in which the dominant party actually controls access to high office were brought to power by parties originally organized before the seizure of power to win elections, lead revolutions, or fight for independence. In very few cases in which parties were created after the seizure of power does the party control access to high office.
Instead, as shown above, the dictator himself or one of his relatives leads the newly created party most of the time. In 45 percent of regimes in which the dictator creates a post-seizure support party, he also controls appointments to the party executive committee.18 Controlling appointments to the party executive committee is symptomatic of much broader control. A contemporary observer of Mobutu (who created a post-seizure support party), for example, reports: “He controls and distributes all offices, all the posts, all advantages linked to power. All revenue, all nominations, all promotions ultimately depend on presidential good will.”19 Where the dictatorship’s support party pre-dates the seizure of power, by contrast, the dictator controls appointments to the party executive committee in only 28 percent of cases.
The dictator who controls appointments to the party executive committee not only cannot make credible promises to share; he does not want to do so. Instead, he tries to keep his supporters insecure about the future so that they will compete with each other and work hard to demonstrate their loyalty. In Rafael Trujillo’s Dominican Republic, for example, all officials had to sign undated letters of resignation before taking office. Legislators who displeased Trujillo simply disappeared from the legislature and sometimes from the world of the living without anyone seeming to notice. Multiple legislators “resigned” within a month after nearly every election (Galíndez 1973). Cabinet ministers could find out they had lost their jobs by reading it in the newspaper (Hartlyn 1998). The statutes of the party Trujillo created after seizing power gave him the unilateral right to make decisions about who occupied these posts (Galíndez 1973). Under Mobutu, “state-party personnel are completely dependent on him for selection, appointment, and maintenance in power . . . The powers of appointment and dismissal that Mobutu wields create constant uncertainty for all officials, which helps to maintain their loyalty to him” (Callaghy 1984, 180).
In many of the regimes in which the dictator’s intertemporal commitment problems have not been solved, both dictators and dictatorships nevertheless last a long time. Mobutu lasted thirty-seven years in a very turbulent political environment, and Trujillo for thirty-one. Dictatorships that achieved power by force but then later created post-seizure parties last more than twice as long on average as otherwise similar regimes without support parties. To sum up our argument, even parties that do not deliver benefits beyond a relatively small group and do not control access to highest offices can still prolong both dictator and regime survival.
Next, we compare the effect on dictatorial survival of ruling parties established before and after seizures of power. We test a linear probability model with country and year fixed effects, controls for regime duration, and indicator variables for party history: pre-seizure electoral party, pre-seizure rebel party, no party (reference category), and new party (post-seizure creation). This analysis accounts for the fact that some regimes were still in power at the end of the sample period in 2010. The results show the following average yearly regime collapse rates: 10.1 percent probability of breakdown for dictatorships without a support party; less than half that, 4.5 percent, for those in which a party was created post seizure; 3.8 percent for regimes led by parties first organized to run in elections before the initiation of dictatorship; and only 1.9 percent for regimes led by parties organized to lead insurgencies. Comparison between the breakdown rate for regimes lacking support parties and those that create a party post-seizure provides compelling evidence that party creation prolongs survival because regimes that lack support parties and those that create one post-seizure have similar origins: both typically achieved power by force (most often via coup).20 Apparently, autocracies supported by parties that have not solved the dictator’s commitment problem – that is, parties that we know did not control the dictator’s access to office because he achieved it before the party came into existence – nevertheless last quite a long time.
Figure 5.7 displays these same relationships as comparisons among regimes without support parties and regimes supported by parties created at different times for various purposes. In all models, dictatorships with support parties are less likely to end. In models that include fixed effects (circle symbol), postseizure party creation is associated with a bit more than a 10 percent decrease in the likelihood of regime collapse compared with having no support party. Parties originally created before the seizure of power to run in elections or lead insurgencies reduce the likelihood of regime breakdown even more, but not by a great deal more.
These comparisons show that armed forces by themselves often do not provide a reliable support base for autocracy. Consequently, dictators who are delegated power by a seizure group that did not need to be organized as a party in order to achieve power can improve the odds of retaining it by creating a support party to counterbalance the potential volatility of military support.
the effect of post-seizure party creation on the likelihood of coups
If our argument about party creation is correct – that it is pursued by dictators propelled to power by factionalized armed seizure groups in order to lessen their risk of ouster by force – then we should see fewer coups and coup attempts in regimes that create parties than in those that opt not to. To evaluate this expectation, we look at the incidence of coups in dictatorships without support parties at the time of the seizure of power. We compare the likelihood of coups in dictatorships that initially lacked a support party but later created one with those that never organized supporters into a party at all.
We begin by looking at the differences in the baseline rate of coup attempts – both failed attempts and successes – between regimes that had created parties and those that never did.21 Coup risk declines on average during the first few years in all kinds of dictatorships, as dictators try to remove their least reliable supporters from troop commands, and the least stable dictatorships collapse and exit the sample. Figure 5.8 shows that during the first two decades in power, regimes that never create a party are at greater risk of both coup attempts (shown in the left panel) and successful coups (shown in the right). This implies that dictatorships that never create a party are particularly susceptible to coup conspiracies.
Because we have argued that the creation of a mass-based civilian support party deters officers from staging coups, we expect to see a larger difference in coup attempts than in successful coups. A comparison of the left and right panels of Figure 5.8 shows larger differences between the two lines for attempted than for successful coups. Both panels also show that the coup risk for dictatorships that have created support parties remains stably lower for more than a decade after party creation.
In contrast, there is a large spike in coup attempts after the first decade in power for regimes that never create a party. Typically, armed seizures of power are followed by retirements of officers ranked above those who take power and a purge of officers who supported the ousted government. Rapid promotions for the cohort of the active coup plotters and those just below them follow. These changes in the officer corps create a cohort of coup beneficiaries, but over time many of these officers will retire or be dismissed, creating opportunities for younger officers to gain command of troops. Coups depend on troop command, and these junior officers become the coup plotters of the future. In this way, the normal seniority-based promotions inherent in military careers interact with the ambitions and criticisms of junior officers to produce the spike in coup attempts that begins after about ten years in power unless a wellestablished support party deters them.
The difference in trends for regimes with and without ruling parties is important for showing that party creation has consequences. If the trend lines moved in similar ways in each set of regimes, we might wonder if preexisting differences accounted for the difference in coups. However, the trends diverge, suggesting the limitations in the coup-proofing strategies that can be deployed by dictators dependent on all-military support bases.
For further evidence that dictatorships with parties created post-seizure are less vulnerable to coups, we note the success rate for coups conditional on attempts: while 22 percent of coup attempts in regimes with post-seizure parties succeed, nearly 29 percent succeed in regimes that never established a support party. After five years in power, the difference is greater: only 20 percent of coups in regimes with post-seizure parties succeed, while the share that succeeds in regimes with no party increases to 35 percent.22 This suggests that once we account for all the factors that lead to coup attempts in the first place, dictators who have created a new support party have greater ability to thwart them. This result most likely reflects the strong correlation between post-seizure party creation and investment in other coup-proofing strategies, especially the dictator’s establishment of new paramilitary forces recruited from especially loyal regions. Republican guards, presidential guards, and other kinds of specially recruited paramilitary forces are usually stationed near the presidential palace and tasked explicitly with defending him from armed challenges.
Next, we test a series of parametric models to evaluate whether creating a party reduces coup risk. We employ a nonlinear estimator with random effects for each regime. First, we test a specification with a minimum of control variables: time since last coup (logged), whether the dictator is civilian (as opposed tomilitary), and how the regime seizedpower (coup, rebellion, foreign imposition, or popular uprising).23The top set of estimates in Figure 5.9, shown as large diamonds, is from this model, with successful coups as the dependent variable. Next, we add a set of standard control variables often thought to affect coup risk: GDP per capita (log, lagged), economic growth (lagged two-year moving average), oil rents (log, lagged), military expenditure (log, lagged), civil and international conflict (lagged), and protests (lagged). In the third specification we return to the minimal model but use all coup attempts (failed and successful ones) as the dependent variable. In the final model, we add controls to the all attempts model. In all four tests, post-seizure party creation is associated with a lower incidence of successful coups; however, the estimate for New Party is only statistically significant at the 0.10 level for the all attemptsmodels.
To further explore how the creation of new support parties influences coup risk, we estimate a conditional logit model that compares the coup risk of individual leaders before and after new party creation.24 This analysis looks only at the sixty-seven leaders in dictatorships that initially lacked a support party but later created one; that is, it looks only at rulers who held power during periods both with and without a support party. Importantly, by focusing on the variation over time for individual dictators, we can rule out alternative explanations based on differences between leaders, regimes, and countries. The results from this test (not shown) also indicate that coup risk is lower after new party creation.
Thus far in this chapter, we have combined reshuffling and regime-change coups when examining how party creation protects the dictator from ouster. Our theory about why dictators who achieved power by force later create parties to organize civilian supporters, however, claims that party creation helps the dictator concentrate personal power over other members of the ruling coalition. That idea leads to the expectation that new parties should lengthen dictators’ time in power by deterring coups aimed at replacing leaders (reshuffling coups), but not necessarily those aimed at ending the regime. Members of the ruling group who want a change in leadership but not the end of the regime organize most leader-change coups. Regime-change coups, however, are usually organized by factions of the military excluded from the inner circle of the dictatorship. If party creation is a dictator’s strategy for increasing his own power relative to that of other members of the dictatorial ruling group, as we have argued, it should protect him from leader-shuffling coups but not necessarily from regime-ending coups.
Figure 5.10 shows the results from a series of random effects models similar to those in Figure 5.9. The first two estimates are from models of leadershuffling coups that treat regime-change coups as right-censored events, while the latter two estimates are from models of regime-change coups that treat reshuffling coups as right-censored events. These tests indicate that post-seizure parties are associated with fewer leader-shuffling coups but not with fewer regime-change coups.
This evidence is consistent with our theory that post-seizure party creation is a strategy to protect dictators from their erstwhile allies in the armed forces. That is, party creation protects dictators from coups led by ambitious regime insiders eager to take the dictator’s place without ending the regime. Postseizure party creation is less reliably helpful for deterring coups aimed at ending the regime. We know from the findings reported in Figure 5.1 that the creation of new parties contributes to the survival of regimes as well as individual dictators, but the results shown in Figure 5.10 imply that the extra regime durability associated with party creation comes not through coup deterrence but through some other mechanism. We discuss this other mechanism, the civilian side of how newly created parties contribute to regime durability, in Chapter 6. Those coups that do end the entire regime – rather than just replace the dictator – occur most often when junior officers from excluded ethnic groups successfully oust the regime leader and allied senior officers. Thus, the evidence from differentiating regime-change from leader-shuffling coups suggests that the kind of parties created by dictators post-seizure do not successfully coopt ethnic groups excluded from executive power and the senior officer ranks.
To conclude this section, dictators who organize post-seizure parties are less vulnerable to coups than dictators in regimes unsupported by a ruling party, as our argument detailing the motivations for post-seizure party creation implies. Further, the evidence suggests that post-seizure party creation helps deter coups emanating from regime insiders rather than those plotted by groups of soldiers excluded from the ruling group.
conclusion
Dictators who achieve power through force of arms can face special difficulties in consolidating their rule because many of their supporters control sufficient weapons to oust them. This chapter has focused on the consequences for intraelite bargaining of the interaction between the dispersion of armed force across members of the ruling group and the group’s division into multiple factions. When many members of the seizure group – and the ruling coalition it becomes – command armed forces sufficient to threaten the dictator with ouster, they can achieve an effective power-sharing bargain with the dictator if they can maintain their own unity. If, however, deep factions divide an armed seizure group, those included in the dictator’s inner circle cannot credibly commit their subordinates to support the dictator if he shares power and spoils. Consequently, power-sharing bargains cannot be maintained.
When the dictator cannot secure his hold on power by agreeing to share with the rest of the seizure group, he is better off keeping a larger share of the spoils and other benefits of office so that he can invest in other strategies. We suggest that dictators who depended on armed supporters to achieve power, but who cannot count on those supporters for holding onto it, often try to counterbalance their armed supporters with unarmed ones. To accomplish this, they organize civilian support networks and appoint their leaders to the dictatorial inner circle. The addition of civilian supporters to the ruling group changes bargaining within the group by increasing the diversity of interests and further reducing the unity of the inner circle; this in turn undermines the bargaining power of members of the dictatorial inner circle relative to the dictator. It thus contributes to the personalization of dictatorial rule. These civilian support organizations are usually called parties.
The evidence shown in this chapter is consistent with the argument that factionalism within an armed seizure group increases the likelihood of postseizure party creation. We also offer evidence that post-seizure party creation usually reflects a dictator’s interest rather than the collective interest of the ruling group: dictators usually assume a new party’s leadership themselves; new parties tend to be established in the run-up to elections that “legitimize” the dictator’s occupation of national executive office; and dictators often control appointments to the executive committees of ruling parties created post-seizure. We further show that post-seizure party creation is associated with the marginalization of military influence within the dictatorial inner circle.
Finally, we provide several kinds of evidence that post-seizure party creation is an effective dictatorial survival strategy. It is associated with both longer dictator tenure in office and longer regime survival. Party creation seems to protect dictators from coups, as would be expected if it were a strategy for reducing the dictator’s vulnerability to ousters launched from an unreliable military force. After controlling for other factors known or believed to affect the incidence of coups, post-seizure party creation is associated with a reduced incidence of both coup attempts and successful coups.
Further investigation shows that post-seizure party creation affects the incidence of leader-change coups but not regime-change coups. This is what would be expected if party creation is a strategy used by dictators to safeguard themselves from armed insiders rather than a group strategy for regime defense. The findings in this chapter thus explain how party creation extends the survival of individual dictators: it is associated with a reduction in the likelihood of coups that would replace one dictator with another insider. They do not, however, explain how new dictatorial support parties contribute to regime survival. We turn to that question in the next chapters.
Post-seizure party creation thus seems paradoxical within the usual way political scientists think about authoritarian parties and support coalitions. The creation of a mass party through which some benefits are channeled from the political center to ordinary citizens seems to imply the broadening of the dictator’s support coalition. At the same time, however, such broadening among the mostly powerless accompanies a disorganization and eventually a narrowing of the support coalition among the powerful, as threatening military supporters are replaced by less powerful civilian ones. In most cases, much of the military becomes marginalized after post-seizure party creation. Civilian party leaders handpicked by the dictator replace military members of the dictatorial inner circle, but they have much less ability to constrain or oust the dictator than do armed supporters. Post-seizure party creation thus transforms the dictator from a relatively equal bargainer within a group of others similar to himself into an arbiter among competing support factions, reducing his dependence on all of them and enhancing his individual discretion over resources and policy.