|Party Creation as an Autocratic
Many dictators called sultanistic by Juan Linz,2 for example Mobutu in the former Zaire or Trujillo in the Dominican Republic, create parties to support their rule; so do many whom observers call military, for example, Eyadema in Togo or the military presidents in El Salvador. This paper seeks to explain why dictators whose real power seems to depend on the military or security apparatus often create parties after they seize power.
In authoritarian seizures of power since 1946, about 40 percent of seizure groups were organized as parties before they achieved power. Although the most famous of these gained power in revolutions (e. g., the communists in Russia and China) or foreign occupations (as in Eastern Europe), most originally achieved office in competitive elections. Once in office, governments were transformed into autocracies either by banning opposition parties or in other ways intimidating or disadvantaging the opposition such that it lost any real chance of winning future elections. In about 60 percent of autocratic seizures of power, however, the seizure group had not been previously organized as a party. Excluding monarchies, almost all of these seizures were led by military officers. In 43 percent of these, a party was later created to support the dictatorship. In nine percent, the dictator later allied with a pre-existing party that had been created during a prior democratic or authoritarian regime; this was the strategy chosen by Musharraf in Pakistan. These parties were coopted, reorganized, and often purged to turn them into support machines for the dictator. In about 49 percent, no party was ever created or coopted.3
This paper explains why some dictators who did not need a party in order to achieve power later created a party while others did not. To begin thinking about this question, consider the dilemma faced by the kind of dictator who comes to office in a coup – the usual means for those not organized as a party before seizure. Although virtually all authoritarian installations are supported by important civilian allies at the time of seizure, these civilian groups would not have been able to achieve power in instances of seizure by coup without the military. Furthermore, civilians have been notoriously unable to discard their military allies as soon as they tire of them, which often happens quickly. Thus it seems reasonable to think of these regimes as dependent on armed support for their survival, at least at the beginning.
Dictators in this kind of regime have the support of the military at the time of seizure, but no way of guaranteeing that support in the future. Even at the beginning, support from the rest of the military might be quite fragile. As I have shown in earlier work (Geddes 2003), the incentives facing military officers create a first-mover advantage similar to that in a Battle-of-the-Sexes game, meaning that if a small group of officers makes a successful first coup-move, the rest of the military will go along whether they sincerely support the intervention or not. So the fact that a seizure of power has occurred implies very little in terms of solid military support for it. 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 more fragile than it appears.
Empirically, most dictators, especially those who come from the military, have been ousted in coups carried out by military officers. It is obvious that the military can be dangerous to any ruler because officers control weapons and men, and they face minimal collective action problems (compared to civilians) if they decide to rebel. A feature of the situation that has received less attention is that it sometimes takes only a small number of men to stage a coup. Most coups are carried out by small conspiratorial groups of officers using what I term the first-mover strategy (Nordlinger 1977). Only a minority involve consensus among the whole officer corps, the other strategy available in games like Battle-of-the Sexes. At the extreme, coups involving a couple dozen men led by a sergeant have been successful. To be safe, then, a military dictator might have to satisfy the demands not only of top military brass, but of all officers with command of troops.
Civilians can also threaten the dictator’s survival, but the task of ousting the autocrat is much harder for them, both more costly and riskier. They face huge collective action problems as they have to persuade many others to join them in a risky venture and to remain loyal often over extended periods of time. If they contemplate violent overthrow, they have to find the resources to acquire arms and training. All these aspects of plotting are easier for officers, not least because coups organized by small groups during short periods of time have a reasonable prospect of success, while it usually requires large numbers of civilians, often organized over a long period of time, to overthrow a dictator. Consequently, the groups dictators usually fear most are military factions. Because officers have the means to oust dictators without broad-based organization, the military can often force dictators to share spoils and choose policy positions it favors because the threat to oust is so very credible.
Dictators have several strategies for trying to deter coups and thus reduce their dependence on the rest of the military. They promote loyal officers and appoint opponents as ambassadors to far away places. They try to keep military budgets high. They resist resigning from active service so that they can maintain personal control of promotions and command assignments. They strengthen security services and often take personal control of them. They create paramilitary forces loyal to themselves to counterbalance the regular military. A few, like Duvalier in Haiti, kept military weapons and ammunition locked in the basement of the presidential palace. These strategies show how much dictators fear the military, and they can be useful, though they can also backfire. The military tries to defend its position of power, which depends on the credible threat to oust. They resent and sometimes resist policies that undermine that threat. Many coups are led by officers who have been passed over for promotion, dismissed, or exiled. One of the main grievances mentioned as a cause for coups is the creation of paramilitary groups that challenge the military’s monopoly on the use of force.
Party creation, I argue, is a subtler and perhaps safer strategy aimed at reducing the dictator’s dependence on the military.
Theories about Parties
Schattschneider’s (1942) view of parties as organizations dedicated to achieving power through control of government and Anthony Downs’s (1957) very similar definition of parties as “coalitions of men seeking to control the government”4 can be applied to the parties that have led authoritarian seizures of power as well as to electoral parties, though Schattschneider and Downs did not have that in mind. As noted above, many dictatorships arise when parties win elections and then change the rules after they control the executive, legislature, and the courts. Often, opposition parties are legally banned or repressed, but in other cases they are so disadvantaged by the incumbent’s near-monopoly over campaign resources and control over the media, party registration, and permission to hold campaign events that they have no chance of winning future elections. Whatever the specific method used to consolidate their hold on government, parties that initially achieved power through fair elections fit standard political science theories about why parties are formed. Nothing in the logic of the theories implies that parties would necessarily play by the rules after achieving power. That is simply an implicit assumption in most theories developed in the context of U.S. politics.
Parties that achieve power through the violent ouster of the previous government also fit standard theories about why parties exist,5 though they do not pursue the strategies for achieving power posited by Downs since they need not attract the support of a majority of citizens. Instead, when out-of-power party leaders choose policy positions, they face a trade-off between appealing to extremists who are willing to invest in the party if it reflects their preferences even if its likelihood of actually achieving power is very low and appealing to larger numbers of citizens with more moderate policy preferences.6 Once in office, leaders are even less constrained by the preferences of the median citizen. There is no puzzle about why such parties exist, however. They are organized to get control of government, just as Schattschneider observed.
Only a few scholars of authoritarianism have investigated the creation of support parties. The question why parties are created after seizures of power has received little attention, I suspect, because most observers have not realized how frequently it occurs. As far as I know, explanations of post-seizure party creation have only been proposed for specific instances.
In one of the best of these, Beatriz Magaloni (2006) suggests that the creation of the Partido Revolucionario Institucional (PRI) in Mexico was authoritarian leaders’ solution to the deadly competition that had plagued them since the revolutionary seizure of power. At the end of the revolution, much of the country was controlled by local armed caciques who had fought for the revolution, but could not be controlled by the central government once the revolution was over. Competition among the top revolutionary generals for the presidency was both deadly to individuals and destabilizing for the regime. The creation of the party and its enforcement of rotation in office and a six-year term limit for the presidency gave rivals an incentive to wait their turn rather than plotting assassinations or leading rebellions against the sitting president. I find this a very plausible explanation for the unusual presidential institutions of the PRI government.
It is these specific institutions – regular rotation and term limits -- that moderated elite competition and helped the regime survive, however, not the existence of the party by itself. The PRI in Mexico enforced adherence to these institutions, but very few other authoritarian parties rotate leadership on a regular schedule. The other two autocracies that enforced term limits were the military regimes in Brazil and El Salvador. Both relied on support parties, but in both cases, term limits were enforced by the military not the party. In both the latter regimes, highest level decisions, including the identification of the next president, were made by top officers. Furthermore, the warlordism that characterized Mexico after the revolution has occurred after only a few other seizures of power, so the need to limit competition among armed territorial magnates cannot be a general explanation for party creation.
In an important recent paper, Magaloni (2008) generalizes the explanation of the creation of the PRI to argue 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. The most important point Magaloni makes is that dictators can trade spoils for support at any one time, but their promises about the future are not credible because they can always do better in the short run by reneging on their agreements. She suggests that parties can solve dictators’ credibility problem if the dictator delegates control over who occupies offices, including the dictatorship itself, to the party.7 If the party controls access to office, the dictator has reason to fulfill his promises, and his allies have reason to remain loyal because they can expect higher offices in the future. I think this is a good explanation of how the PRI in Mexico worked and that it fits some other very well institutionalized dominant party systems, but I do not think it fits most of the regimes originally brought to power through coups. It does not fit most of the regimes in which parties are created after the seizure of power precisely because the dictator in these regimes does not delegate control over highest offices to the party.
Quite the opposite. In two-thirds of these cases, the party is led by the dictator himself or a close relative. It is always led by someone seen as a close ally at the time he is chosen. The dictator personally selects the members of the ruling committee of the party in most of these cases, and he chooses or vets candidates for the legislature. He often appoints those who occupy high offices in the bureaucracy. He not only cannot make credible promises; he does not try to do so. Instead, he tries to keep his supporters insecure about the future so that they will try hard to demonstrate their loyalty. In the Dominican Republic, for example, legislators who displeased Trujillo simply disappeared from the legislature and sometimes from the world of the living without anyone seeming to notice. Cabinet ministers could find out they had lost their jobs by reading it in the newspaper (Hartlyn 1998). Many of these regimes in which the dictator’s intertemporal commitment problems have not been solved nevertheless last a long time.
An alternative explanation of party creation is suggested by Bruce Bueno de Mesquita et al. (2003). Within the theoretical framework they develop, party creation would be defined as a decision by the dictator to broaden the Selectorate. They define the Selectorate as the group that has some influence on choosing who rules and that receives some benefits from the government. In recent work, Bueno de Mesquita and Smith (2007) suggest that autocracies have a tipping point that depends on the initial size of the support coalition. Threats cause autocracies supported by narrow coalitions to narrow further, but cause those supported by broader coalitions to become more inclusive. Data limitations prevent careful testing of this argument, but I consider its plausibility below.
Ellen Lust-Okar (2005) has explained why economic elites cooperate with dominant parties in the Middle East, why people compete to run for seats in powerless legislatures, and why most citizens vote for the dominant party. She thus explains many of the puzzles of electoral authoritarianism. This is a major contribution to understanding authoritarian politics, but she has not addressed the question why those who seized power decided to form a mass party organization in the first place.
more public goods, including secure property rights, that contribute to better economic performance. Escriba Folch (2005) makes a similar argument about the effects of authoritarian legislatures. Wright (2007) shows, however, that legislatures Other scholars have focused on a related subject, legislatures, that might have implications for party creation. Carles Boix and Milan Svolik (2007) argue that dictators permit legislatures when they are threatened by challengers and need the support of economic elites in order to survive. They see legislatures as concessions by rulers to socio-economic elites whose support is needed in order to maintain the regime and fora within which bargains between elites and rulers can be enforced. This argument seems to fit contemporary China and the constitutionalization of monarchies in early modern Europe, but not the many autocracies in developing countries in which much of the private sector was expropriated soon after the seizure of power. Jennifer Gandhi and Adam Przeworski (2005; Gandhi 2003) make a related argument that dictators who come to power in societies with strong preexisting political forces must offer concessions in order to remain in power. Like Boix and Svolik, they see the legislature as a forum within which policy concessions can be made and enforced. As a result, they expect autocracies that have legislatures to supply only contribute to better economic performance in military and single-party regimes. In personalist dictatorships and monarchies, legislatures have no effect on economic performance. He suggests that instead they serve as the forum within which spoils are divided among elites.
Scholars who investigate why many dictators allow legislatures to function do not address the question why parties, but it might be the case that authoritarian support parties are created in order to organize regime supporters in the legislature, as posited by Aldrich (1995) and Schwartz (1989) to explain the creation of institutionalized majority coalitions (parties) in democracies. Parties might in some circumstances serve as commitment devices within the legislature. If legislators will remain dependent on the party beyond the length of a legislative term, they have more reason to stick to their side of bargains to support the dictator in return for policy concessions. Generally speaking, however, it is the credibility of the dictator’s commitments that needs shoring up, not that of the much less powerful legislators, as Magaloni (2008) shows. More important, many authoritarian legislatures have little influence on policy. Instead, as Lust-Okar (2005) has reported, deputies in authoritarian legislatures report that policy making is not their job; their job is to deliver pork and other benefits to their local constituents. The possibility that authoritarian parties are created to solve commitment or other problems that arise within authoritarian legislatures will be considered again below.
My argument builds on some of the insights summarized above. Like Bueno de Mesquita et al. (2003) and Magaloni (2008), I focus on the rivalries among potential leaders of the authoritarian regime rather than on threats to regime survival that arise from societal or external opponents because most threats to autocratic survival come from erstwhile elite supporters. In contrast to Bueno de Mesquita et al., I do not assume institutional stability. Quite the contrary, because in the early post-seizure period, authoritarian institutions tend to be changeable and informal, norms fluid and uncertain.
Autocracies supported by parties that have not solved the dictator’s commitment problem last, on average, almost as long as those that have solved it. With the exception of the PRI regime in Mexico, previously organized parties led the seizures of power that resulted in nearly all the regimes in which the dominant party controls access to high office. In very few of those in which parties were created after the seizure of power, however, does the party control access to high office. Table 1 shows the average regime survival time for those regimes in which parties were created after seizure compared with those in which parties were never created. Regimes that were not led by parties at the time of the seizure but that created them afterward last an average of 18.1 years. Otherwise similar regimes that neither created nor allied with a pre-existing party to create a support organization last about a third as long. If we compare regimes in which a party was created with those in which none was, limiting the sample to countries that had experiences with both, differences in average longevity remain very large, suggesting that they are not caused by factors such as level of development, region, or prior history of the countries.
Individual leaders in regimes that created parties lasted more than twice as long on average as those in regimes that neither created new nor allied with old parties, even though a few of the regimes in which parties were created enforced term limits.
The observation that regimes with weak parties can also last a long time leads to two questions. First, what is it about these rubberstamp parties that increases the durability of authoritarian regimes? Second, if parties increase regime and dictator longevity, why doesn’t every dictator create one?
How Rubberstamp Parties Deter Coups
My answer to the first question is that reliance on the military as the sole base of support is always risky. Military dictators can increase the credibility of their own promises to share by, for example, retiring from active duty – which many officers are required to do before becoming junta president. Once they no longer command others, do not determine promotions, and cannot decide which officers will command which garrisons, however, presidents’ ability to deter coups depends almost entirely on their ability to satisfy the rest of the military’s policy and budget demands. The rest of the military’s promises not to oust are never completely credible, however, because the high command cannot always prevent what might be called “rogue” coups. These are coups by small groups of officers that could be defeated if the rest of the military mobilized against them, but usually they do not. The first-mover advantage means that a military faction that makes a credible first coup-move – empirically, seizes the presidential palace, airport, radio, and TV stations without being met by violent or widespread opposition – can overthrow the government because the rest of the military will acquiesce. All dictators face some risk from the military, but the less able commanding officers are to control lower officers, that is, the less professionalized the military is, the less able they are to make bargains that reduce the risk.
An alternative way to express this problem is to note that militaries are not unitary actors, though some more closely approximate unity than others. Individual officers’ motivations for seizing government vary from high-minded commitments to improving the welfare of the poor to concern about the exclusion of their own ethnic or religious group from the current government to naked personal ambition to be dictator. In more professionalized militaries, individuals’ future career success is effectively tied to following the orders of their commanders, which transforms a large group into something approximating a unitary actor. In less professionalized militaries, however, discipline is less predictably enforced, personal, and sometimes ethnic, loyalties play a larger role in determining officers’ behavior, and routes to higher office other than the slow but predictable rise through the ranks seem more available. Coups are the preferred method of ousting government for both professionalized and unprofessionalized militaries, but rogue coups are more likely in less professionalized.
A party organized to support the dictator can reduce this risk if it can organize large numbers of ordinary people into apparent demonstrations of support, whether sincere or not. Most coups are bloodless. That is because potential coup leaders choose times when they expect little or no opposition. Ever since the Russian Revolution, everyone has understood that asking troops to fire on their fellow citizens tends to lead to indiscipline, desertions, and mutiny. Officers try to avoid these situations.
A rubberstamp party is costly to maintain since party leaders and activists have to be paid with jobs and other benefits, but it is a much less risky base of support than the military precisely because it cannot easily oust the dictator. A rubberstamp party does, however, have one major similarity with the kind of party described by Magaloni: it is a giant patronage machine. Those elite and midlevel supporters who receive benefits from it have a vested interest in the dictator remaining in power even if they cannot be sure they will get a share in future spoils because their chances are better as allies of the current dictator than they would be if someone else with whom they are not allied seizes power.
These supporters have access to mass communications and transportation, and they have the carrots and sticks to make sure that ordinary people turn out for big demonstrations or votes when told to. Some autocratic parties distribute benefits to ordinary citizens, provide good economic policy, and make opportunities for education and upward mobility available to many people whose future looked bleak before. But many of the rubberstamp parties are incompetent and corrupt. Cadres may sell the resources they are supposed to distribute, and they may use their party positions to exploit their fellow citizens instead of helping to build mass support for the regime. Nevertheless, they can turn out masses of citizens for votes and demonstrations of support by threatening to block citizens’ future access to important goods or to turn in the names of those who do not turn out to security police.