Black Knights and Mass Uprisings in Authoritarian Regimes

Transcription

1 Black Knights and Mass Uprisings in Authoritarian Regimes Michael Aagaard Seeberg Assistant professor, PhD Department of Political Science and Public Management University of Southern Denmark Campusvej Odense M, Denmark Jakob Tolstrup Assistant professor, PhD Department of Political Science Aarhus University Bartholins Allé Aarhus C, Denmark tolstrup@ps.au.dk Note to the reader: This is very preliminary work, and we welcome all kinds of comments. In the current manuscript, we make a descriptive analysis, but we are planning to do time series analysis also. We wish to get a better understanding of what conditions black knight support in time series analysis and also when black knight support has an effect on regime outcome. We primarily seek to understand determinants of black knight support in this paper. Also, we are planning to supplement the descriptive analysis with illustrative cases. We are particular curious to discuss other conditioning factors of black knight support that we should look into as well. The manuscript has been written quite recently, and there might be cosmetic errors in the text. Apologizes for that! We look forward to see you in a week. Michael & Jakob Please do not cite without author permission

2 Abstract During the latest decades, works on autocratic survival and failure has proliferated. While for long only internal determinants of authoritarian stability and breakdown were addressed, scholars now increasingly pay attention to the international dimension as well. One aspect, in particular, has been emphasized: the role played by the so-called black knights (external actors that bolster autocracies and/or undermine democratization processes). While these black knight effects have been thoroughly dealt with in case-studies, no systematic longitudinal large-n analysis of the phenomenon exists. In this paper, we attempt to rectify this imbalance through a statistical analysis of black knight support to authoritarian regimes during periods of mass uprisings, making use of a new dataset on Nonviolent and Violent Campaigns and Outcomes (NAVCO 2.0). Initially, we explain why black knight assistance is likely to matter for autocratic survival in the context of mass uprisings, and we spell out the relevant factors that induce external actors to provide such assistance. Second, we map out sender and target characteristics of black knight support across a period of 50 years ( ). 2

3 Introduction Since the late 1990s, works on autocratic survival and failure has proliferated (for an overview, see Art 2012). A host of factors such as authoritarian regime type (Geddes 1999; Hadenius and Teorell 2007; Cheibub, Gandhi, and Vreeland 2010), coercive capacity (Bellin 2004; Way 2005), institutional factors like elections, parties, and parliaments (Brownlee 2007; Magaloni 2008; Gandhi 2008) and incumbent economic control (Smith 2004; McMann 2006) have shown to be decisive for authoritarian survival rates. In addition to disentangling domestic determinants, scholars now also increasingly work on the external dimension of authoritarian persistence and breakdown. While for long this literature emphasized only the external factors that weaken authoritarian rule such as economic sanctions (Escribà-Folch and Wright 2010), Western democracy promotion (Magen, Risse-Kappen, and McFaul 2009; Levitsky and Way 2010) or diffusion effects (Brinks and Coppedge 2006; Ambrosio 2010), scholarly interest for the external factors that strengthen authoritarianism now seem to be on the rise. One key aspect in this subfield is the so-called black knights that is, external actors, be they democratic or authoritarian, great powers or regional powers, states or international organizations, that act as guardians of autocracy or challengers of democracy in specific contexts. In particular, scholars have emphasized how authoritarian great powers hinder democratization and stabilize authoritarianism in their respective neighborhoods (Ambrosio 2009; Tolstrup 2009; Levitsky and Way 2010; Vanderhill 2013; Tolstrup 2014). But also, democratic external actors may behave as black knights, evident from, for example, US support to military dictatorships in Latin America in the 1960s and 1970s (Muller 1985) or its long-time support for autocratic Egypt (Brownlee 2012). While these black knight effects have been thoroughly dealt with in case-studies, no systematic longitudinal large N-analysis of the phenomenon currently exists. 3

4 In this paper, we zoom in on one particular aspect of black knight support the assistance provided to authoritarian governments whose internal stability is threatened by violent or nonviolent mass uprisings. 1 Making use of privileged access to a new dataset on Nonviolent and Violent Campaigns and Outcomes (NAVCO 2.0), we map out the senders and targets of black knight support across a period of 50 years ( ) in an attempt to broaden our understanding of who the black knights are, which types of regimes they tend to support, and why they act the way they do. The paper proceeds as follows: first, the literature on black knights is briefly reviewed. Second, we tease out four mechanisms through which black knights can bolster authoritarian regimes threatened by mass uprisings (strengthening the coercive apparatus of the incumbent, deterring defection, dampening popular discontent and deterring further mobilization, and counterbalancing pressure from other external actors), and we highlight four types of conditional factors that are thought to induce external actors to provide assistance to dictators in need (regime type congruence, geographical proximity, density of linkages and the perceived geopolitical and economic value of the incumbent regime, and contextual factors). Finally, we dig into the empirical data, mapping the frequency of black knight assistance and the dominant characteristics of both the external actors providing it and the states receiving it. What do we really know about black knights? Since the late 1990s, research on authoritarianism has proliferated. Scholarly works have spawned from conceptual considerations of differences between various regime types (Geddes 1999; Hadenius and Teorell 2007; Cheibub, Gandhi, and Vreeland 2010) and theoretical contributions on the dynamics of autocracies in general (Wintrobe 1998; Haber 2006; Gerschewski 2013) to 1 We use uprising, demonstration, mobilization and campaign interchangeably. 4

5 thorough case-studies of particular regimes (see, e.g., Magaloni 2006; Sassoon 2012) and large-n studies of factors contributing to autocratic stability and breakdown (examples include Smith 2004; Ulfelder 2005; Gandhi 2008; Reuter and Gandhi 2011; Powell 2012). In addition to the above-mentioned publications that primarily focus on internal determinants of autocratic persistence and collapse, focus on the international dimension of authoritarianism has lately also come under scrutiny. Studies have investigated the effects of economic aid (Wright 2009), sanctions (Escribà-Folch and Wright 2010), democracy promotion (Adesnik and McFaul 2006; Seligson and Finkel 2009; Magen, Risse-Kappen, and McFaul 2009; Levitsky and Way 2010), and diffusion (Brinks and Coppedge 2006; Bunce and Wolchik 2011; Ambrosio 2010). However, external factors need not only play into the hands of those challenging the autocrat in power. In fact, some of the impulses coming from the outside may help authoritarian regimes stabilize and strengthen power on the longer term or bolster their capacity to defend themselves during times of crisis. This is exactly what the literature on the so-called negative external actors or black knights emphasizes. We understand black knights as external actors, be they democratic or authoritarian, great powers or regional powers, states or international organizations, that act as guardians of autocracy or challengers of democracy in specific contexts. Acting as a black knight is not something that is intrinsic to particular international actors, but must rather be understood as a specific role that any actor might resort to towards a particular target state at a particular point in time (cf., Tolstrup 2009). In the literature on black knights, scholars have mainly focused on the support of authoritarian great powers like Russia (Ambrosio 2009; Tolstrup 2009; Levitsky and Way 2010; Vanderhill 2013; Tolstrup 2014) and China (Bader, Grävingholt, and Kästner 2010; Bader 2013) in their respective neighborhoods. But also autocracies like Venezuela (Vanderhill 2013) and South Africa during Apartheid (Onslow 2009) as well as international organizations like the Shanghai 5

6 Cooperation Organization (Ambrosio 2008) and ASEAN (Diamond 2008, ) have been brought forward as examples of black knights. In addition to authoritarian regimes and international organizations dominated by such states, democratic external actors also behave as black knights, evident from, for example, US support to military dictatorships in Latin America in the 1960s and 1970s (Muller 1985) or its long-time support for autocratic Egypt (Brownlee 2012), and from French support to former colonies like Cameroon and Gabon (Levitsky and Way 2010, ). Thus, so far black knight effects have been thoroughly dealt with in case-studies, but apart from works on sanctions-busters (Early 2011), 2 no systematic longitudinal large N-analysis of the phenomenon currently exists. Hence, our knowledge of black knights and their effects on authoritarian regimes seems to be very much a product of a specific time-period, the post-cold War period, and a few dominant cases. But if we are to better understand who the black knights are, why black knight support materializes, and not least how it affects regime developments in autocracies we need to broaden the perspective considerably. In this paper we do just this. Making use of privileged access to a new dataset on Nonviolent and Violent Campaigns and Outcomes (NAVCO 2.0) that covers a period of 50 years ( ), we can, capture one important aspect of black knight support: the political assistance provided to authoritarian rulers challenged by violent and non-violent mass uprisings. Albeit based on a crude measure, this data allows us to do two things. First, we can map out the senders and targets of black knight support in both the pre- and the post-cold War period. This provides us with a first overview of the dominant black knights throughout the past 50 years, and it help us establish a candidate dataset to which additional information and cases can be added. Second, it allows us to take the first step to investigate the factors that induce and constrain external actors to engage in black knight activities. 2 Originally, the term black knight was coined to describe foreign patrons that undermine the effectiveness of sanctions by undergirding the target state (Hufbauer, Schott, and Elliott 1990, 12). 6

7 How can black knight support make a difference? Mass anti-regime uprisings constitute a significant challenge to authoritarian stability a challenge that incumbents must necessarily respond to. However, authoritarian regimes are not equally successful in doing so. Sometimes rulers are strong enough to quell demonstrations with violence, sometimes conflict is dampened through cooptation means, and at times all efforts to withstand opposition simply fail and rulers are subsequently forced from power. A number of factors such as regime type and resource dependency (Ulfelder 2005), the loyalty of the coercive apparatus (Bellin 2012), the size and the approach of protestors (Chenoweth and Stephan 2011) and the degree to which the regime controls regional and local authorities (Lorentzen 2013) account for such differences. In the following we explain how one additional determinant, black knight assistance, may work to strengthen authoritarian incumbents vis-à-vis their internal challengers. It is argued that in case of massive popular upheavals, black knights can improve the odds of authoritarian survival in four ways: it can strengthen the coercive apparatus of the incumbent, help deter state and military defection, help dampen popular discontent and deter further mobilization, and not least counterbalance the effect of political and economic sanctions from other external actors. First, black knights can strengthen the coercive capacity of the regime facing popular upheavals. The most obvious way is through the provision of outright military assistance, be it in the form of deployment of troops and military specialists or through the provision of weapons and surveillance technologies. Both foreign troop deployment (Peksen 2012) and arms-import (Blanton 1999) have been documented to increase state repression levels. Thus, it seems fair to assume that governments receiving military assistance from abroad are better equipped to face the challenge of 7

8 popular upheavals, allowing for more determined and effective responses in the form of violent crackdowns. Second, a more indirect approach is to help the incumbent minimize the risk of disloyalty spreading among the elites and the rank and file of both the coercive and the state apparatus. When mass uprisings erupt, political elites may be tempted to defect to the opposition (Reuter and Gandhi 2011), military elites to reestablish order through a coup d état (Powell 2012), and the rank and file in the military or the police may refuse to follow orders (Bellin 2012). Black knights can reduce such risks in two ways: they can provide additional financial resources to the regime, allowing it to buy off elite supporters and raise salaries to police officers and soldiers; and through political, economic and perhaps military support the external actor can signal that it stands shoulder to shoulder with the incumbent regime, reducing the fear among regime supporters that the incumbent will not be able to withstand pressure. Third, black knights can help dampen popular discontent. Just like state and military elites and rank and file personnel can be bought off, popular frustration can be dampened if social spending is boosted. Even though this may not quell discontent with flawed elections, economic mismanagement, or other triggering events, it does help to reduce socio-economic grievances, which in itself is often a strong impetus for protesters (Goldstone 2001). In addition, black knights can reduce the risk that more people join the rebellion through the aforementioned bolstering of the security apparatus and by reproducing the regime s framing of events; discrediting the unruly demonstrators and supporting the firm hand that effectively disperses the crowds. By raising the costs associated with anti-regime activities and by helping the regime shape popular perception of events, the risk that more people join protesters is reduced. Finally, black knights can counterbalance mounting pressure from other external actors. During mass uprisings, Western nations and organizations often increase the pressure on the incumbent 8

9 regime, demanding democratic concessions or that its leaders relinquish power. Black knights can counterbalance external criticism, dismissing it as non-objective and illegitimate interferences in domestic affairs, potentially cancelling out, or at least taking some of the bite off condemnations (see, e.g., Fawn 2006). In case Western powers, or any other external actors for that matter, introduce sanctions or withhold benefits, a black knight can criticize (and perhaps even veto) the decision and reduce the economic and political costs by offering financial compensation or by increasing political cooperation to re-boost regime legitimacy (Early 2011). Summing up, black knights can indeed protect authoritarian regimes against the challenge of popular upheavals via several channels. But the presence of black knight support does not necessarily mean that the incumbent regime will be successful in thwarting or dampening opposition. Just like Western attempts to promote democracy may have unintended consequences or simply not be strong enough to spur change, so is black knight assistance not a guarantee of regime survival. Sometimes interference further triggers anger and defiance amongst the people and elites, making them turn against, not support, the incumbent regime. At other times, intervention does work according to intentions but is simply crowded out by more powerful domestic or international drivers of change. But the probabilistic claim still stands: Authoritarian regimes that enjoy strong black knight backing are better equipped for facing the challenge to stability that arise when people take up arms or peacefully take to the streets in the hope of removing the dictator. When is black knight support most likely? Not all authoritarian regimes facing mass uprisings are privileged with black knight bolstering, and not all autocrats that are granted external protection will receive the same degree of support. So the next question is: What determines such differences? What spurs external actors to take on the role of black knights and come to the rescue of authoritarian regimes in their time of need? 9

10 The obvious answer is that external actors only offer black knight support to authoritarian regimes if there is something to gain from it, and if the costs associated with doing so are not too high. Below we argue that four types of factors in particular shape the cost-benefit analysis of external actors: regime type congruence, geographical proximity, density of linkages and the perceived geopolitical and economic value of the incumbent regime, and not least contextual factors. As a first and indeed very crude rule, we can expect external actors to want other states to be mirror images (Bader, Grävingholt, and Kästner 2010). 3 Ceteris paribus, democracies hold preferences for democracies, while authoritarian regimes favor authoritarianism. Hence, in comparison to their democratic counterparts autocratic external actors will simply have greater incentives to take on the role of black knight and protect other non-democratic regimes. That said, recent research has shown that authoritarian regimes differ as much between them as they differ from democracies (Geddes 1999; Hadenius and Teorell 2007; Cheibub, Gandhi, and Vreeland 2010). In fact, similar types of authoritarian regimes less frequently fight wars with each other (Peceny, Beer, and Sanchez-Terry 2002; Weeks 2012) and like-minded regimes more often grant each other political, economic and military support (Colombo 2012). Hence, we expect that the higher the congruence between the regime type of the external actor and that of the authoritarian regime threatened by mass uprisings, the more likely it is that black knight support is offered. Relatedly, calculations of costs and benefits associated with black knight intervention are likely to be shaped by geographical factors. As Huntington (1991) originally suggested and as the Arab Spring confirms, authoritarian breakdowns often spread from one country to another like a wave. External actors can, rightfully or not, be led to think that the outcome (regime persistence, 3 The argument is that democracies and autocracies work differently. While democracies are committed to generating public goods in order to win votes, autocracies focus, predominantly, on maximizing private goods, i.e., rents used for personal enrichment and cooptation-strategies. Given these basic systemic differences cross-cutting cooperation is simply more difficult than cooperation between more similar regimes. We should therefore expect that regimes following similar incentive structures, all else equal, will be able to cooperate more smoothly. 10

11 breakdown or perhaps contracted conflict) of popular upsurge in one authoritarian regime will affect developments in other countries as well, and for this reason they will be more inclined to intervene and help the challenged ruler quell the rebellion before it inspires and enflames discontent elsewhere. Fear of negative spill-over effects will obviously be largest in countries that share borders with or belong to the same region as the affected autocracy. Consequently, we expect black knight support to be conditional on geographical proximity. The closer, the external actor is to the instability unfolding in the challenged authoritarian regime, the more prone it is to intervene and offer its assistance. Third, black knight assistance is likely to be conditioned by the density of existing linkages between the external actor and the challenged authoritarian regime as well as the current state of bilateral relations between them. According to the leverage-linkage theory, proposed by Levitsky and Way (Levitsky and Way 2005; 2010), Western external actors are not only more effective but also more interested in promoting democracy in countries with which a dense set of economic, political and social links exists. Likewise, we can expect linkages to matter for the propensity of external actors to offer black knight assistance (Tolstrup 2013; Tolstrup 2014). If sender and target are highly economic interdependent, share cultural and historical bonds, and have well-developed political and geostrategic relations we should expect external actors to be more willing to grant black knight assistance. Basically, this is about how the external actor perceives the value of the incumbent regime. In authoritarian regimes, ruling elites often act as powerful veto-players when it comes to deciding whether bilateral connections with external actors should be extended or constrained. An incumbent that shows geopolitical abidance and willingly grants privileged access to domestic resources and markets is of course perceived by the external actor to be of greater value than one that does not offer such benefits. Consequently, we should expect black knights, be it democracies or 11

12 authoritarian regimes, to be more likely to offer assistance to incumbents that behave acquiescently in geopolitical and economic questions. Finally, we argue that the likelihood of black knight intervention is shaped by contextual factors such as time period, characteristics of the popular upsurge, and questions of legitimacy related to international bandwagon effects. As to the time period, we concur with Charles Boix (2011) argument that the structure of the international context affect the propensity of great powers to intervene in other states to protect either democracy and autocracy from breaking down. In periods with intense great power rivalry, he argues, democratic great powers will be less inclined to protect democracies and less troubled by defending authoritarianism if it serves the primary purpose of balancing other great powers. Consequently, we expect black knight assistance, in particular among the democratic external actors, to be more widespread during the Cold War period, when bipolarity and geopolitical competition dominated the foreign policy agenda of most nations, than during the post-cold War period of liberal hegemony (Levitsky and Way 2002). In addition to the broader international context, particular factors concerning the character of the mass mobilization and the reaction of other states to it is likely to shape the propensity of external actors to take on the role of black knights. Intervening on behalf of authoritarian incumbents can be considered a costly endeavor, in particular in terms of international prestige. Just like dictators may find it difficult to justify why they need to repress their own population, black knights will face the same problem in relation to explaining their motives to the international community and at home. However, two factors might diminish such concerns. First, if protesters resort to violence, crackdown from the side of the government, and also support of such actions from black knights, will simply be perceived as more legitimate than if protesters peacefully show their discontent with the current regime (Chenoweth and Stephan 2011). Likewise, legitimacy costs are likely to be lower if other external actors also support the authoritarian incumbent, in particular so if international 12

13 organizations that represent several states are among the supporters (for a similar argument related to economic sanctions, see Bapat and Clifton Morgan 2009; Bapat et al. 2013). Recapitulating, an external actor is more likely to engage in black knight support if the regime affected by popular upsurge is of a similar regime type, is located in the same region, has dense and beneficial ties with the leadership in power, if the popular upsurge is violent, unfolds during the Cold War, and not least if other external actors in general, and international organizations in particular, also support the incumbent authoritarian regime. The more favorable conditional factors present, the more likely external actors are to offer black knight support, and the more likely authoritarian incumbents will be able to quell or dampen popular upsurges. Methods and Data The main aim of this paper is to transgress the boundaries inherent in case study research and provide an overview of global incidents of black knights interventions across a longer time-period. The newly released Nonviolent and Violent Campaigns and Outcomes (NAVCO 2.0) dataset (Chenoweth and Lewis 2013) provides such an opportunity. In the following, we describe this dataset and the main variables examined in the subsequent analysis. The NAVCO dataset provides detailed and nuanced data on violent and non-violent mass uprisings (so-called campaigns ) across a period of fifty years ( ). It includes numbers related to 250 mass demonstrations, which is considered to be the full population of known cases in this time period. The data includes a total of 1741 country years with mass demonstrations throughout the period. 150 of these mass demonstrations were violent. In addition to the violent/non-violent distinction, the data includes dense structured information on the mass demonstration, such as the composition of the demonstration, the coherence of the group of demonstrators, the size and diversity of the group, and not least regime responses. 13

14 But how then are the events defined? When is a demonstration massive? And how are violent mass demonstrations distinguished from non-violent ones? To count as an incidence of mass uprisings at least 1000 citizens contentiously and in a coordinated manner must take it to the streets. This means that it must be more than just spontaneous bursts with no coordination. Mass demonstrations are coded non-violent if the participating civilians are unarmed and do not directly threat or physically harm the well-being of the opponent (variable prim_method ). Examples are strikes, boycotts protests and sit-ins. A violent mass demonstration is defined as violent if it is armed resistance and civilian or guerrilla challengers deliberately and regularly relies on violent means, which includes the use of force to physically threaten, harm, and kill the opponent (Chenoweth and Lewis, 2013: ). However, not all mass demonstrations included in the NAVCO dataset serves our purpose, as we are only interested in the kind of mass demonstrations that truly constitute a threat to the incumbent authoritarian regime (variable camp_goals ). Thus, we include only the mass demonstrations, in which demonstrators demands regime change, significant institutional reform, territorial secession, or greater autonomy. Thus we exclude mass demonstrations calling for policy change (this could be environmental protection, corporate responsibility or labor rights), because they do not intend to alter the fundamental rules of the regime. Moreover, we do not focus on mass demonstrations pursuing anti-occupation campaign goals because the threat is towards an external source and not the dictator. We do touch on campaign success and whether black knight support actually makes a difference in general. The campaign is considered to be successful if it reaches its goals within one year after its peak of activities. However, in this analysis we do not disentangle what factors related to black knight support is most important for unsuccessful mass demonstrations, but only the factors conditioning when a black knight tends to support. 14

15 Finally, as we exclusively look at black knight support to authoritarian incumbents, we have excluded all the mass demonstrations that take place in democracies. We rely on the often used democracy-dictatorship measure constructed by Alvarez et al. (1996) and updated by Cheibub et al. (2010). We use this measure because it covers the entire period and is constructed as a dichotomous measure and not out of a re-coding of a continuous measure. It makes for a conservative scoring of democracies, as it only counts a case as democratic if there has been a successful alternation in power. 4 We have extended the overall regime type coding provided by Cheibub et al. (2010) not to lose cases of black knight support where the numbers can be added intuitively. A number of regimes were threatened by mass demonstrations during times, where the regime was officially assigned to an empire. These regimes did not have full self-governance, and do not qualify for the criteria set by Cheibub et al. (2009). However, we have assigned an authoritarian regime score during the years under the rule of an empire. 5 To measure black knight involvement we rely on the variable regime support. 6 This raw data codes primary, secondary and tertiary external supporters. In a few instances of mass demonstrations, two or more states are listed together as primary, secondary or tertiary supporters. 7 The coding of regime support has primarily focuses on major and regional powers, thus, taking note of the three most important state supporters. A regime supporter is coded if other regime have 4 This, on the other side, means that it might score a case authoritarian even that it has free and fair elections, but has seen no alternations (such as Botswana). 5 This goes for the following cases (period in brackets): Belarus ( ), Cyprus ( ), Estonia ( ), Georgia ( ), Ghana ( ), Guinea-Bissau ( ), Kyrgyzstan (1990), Latvia ( ), Lithuania ( ), Malawi (1958), Nigeria ( ), Oman, , 1969), Palestinian Territories ( , , , , , 2006), Russia (1991), Rwanda ), Serbia ( ), Slovakia ( , Slovenia (1990), Tunisia ( ), Uganda ( ), Vietnam ( ), Yugoslavia ( , ). 6 We thank Chenoweth and Lewis for giving us access to the country specific data on the regime support dimension. 7 As a rule of thumb we have assigned the support status to the most populous country among the supporters and dismissed the others. This goes for the following mass demonstrations: In the mass demonstration in El Salvador in 1981, Venezuela was coded secondary supporter and Costa Rica dismissed; Nicaragua in 1980 (United Nations coded tertiary supporter and The US and Panama dismissed); Nicaragua in 1980 (United Nations coded tertiary supporter and The US and Panama dismissed). El Salvador in 1981 (Guatemala coded tertiary supporter and Honduras dismissed). Somalia in (France coded tertiary supporter and other UN affiliated countries dismissed). 15

16 voiced their confidence in the target regime, defended it from criticism abroad, vetoed or opposed resolutions in the UN Security Council (Chenoweth and Lewis, codebook). This makes for a conservative test of the propositions stated above. This weak form of support mostly legitimating (or morally supporting) the target regime is a least likely case for our argument. If moral support is sufficient for the target regime to weather the storm of a mass demonstration, then we would, all else equal, expect more comprehensive forms of support, such as economic or military support, to also have an effect when it occurs. We do probably not capture all forms and incidences of support. Thus, if we find patterns in regime support, we should expect it to be more outspoken if we could capture all kinds and incidences of external support. But question is if the assigned support is only moral? In practice the support could as well be combined with military, economic and further political help. But, it is as a minimum moral support. This questions if the least likely logic holds in all cases. With these qualifications of mass demonstrations, we end up with a total of 891 country years with mass demonstrations throughout the period. Other than that, we have added information to the existing data to shed light on the four factors, which we expect to shape the cost-benefit analysis of external actors support to target regimes: regime type congruence, geographical proximity, density of linkages and the perceived geopolitical and economic value of the incumbent regime, and finally contextual factors. Regime type congruence is first and foremost a question of similar regime type. Are the target regime and the black knight both autocracies? As explained above we here rely on Cheibub et al. (2010). We also look at congruence in subtypes of authoritarian regimes. That is, we disentangle different types of authoritarian regimes using the coding constructed by Geddes et al. (2012). In line with Geddes, we distinguish between party-dictatorship, military-dictatorships, monarchydictatorships or personal dictatorships. There is a number of subtype coding s available, but we use Geddes et al., because it is concerned we types of authoritarian rules and not types of authoritarian 16

17 elections or other characteristics. Geddes et al. s types are markedly different, and if regime subtypes congruence matter, we expect these types to be most important to distinguish between. Moreover, the coding is extensive covering the entire period from Geographical proximity between the target state and Black Knight is measured by two factors, neighborliness and regions. To measure neighborliness of two states, we use numbers from the Correlates of War Direct Contiguity data (v. 3.1) provided by Stinnett et al. ( Correlates of War Project. Direct Contiguity Data, ). We define neighborliness as two states separated by a land or river border. The distinction between regions follows Teorell (2010). The density of linkages and the perceived geopolitical and economic value of the incumbent regime are measured by four proxies: colonial legacy based on Teorell (2010), alliance partnership from the Correlates of War Formal Alliance data set (v. 4.1) (Gibler 2009; see also Singer and Small (1966); Small and Singer (1969)), 8 oil reserves and economic wealth of the target regime (Finally, we use Ross data on oil revenue and a country s GDP as crude measures of economic interdependence between the sender and target regime [THIS PART IS PENDING AND NOT YET IN THE ANALYSIS]). We focus on two contextual factors. To capture variation in the international context we create a dummy variable, coded 0 if a mass demonstration occurs during the Cold War era ( ) and 1 if it occurs during the post-cold War era ( ). We also investigate a spatial geographic character of the international context, namely whether the black knight is a major/great power in the world system. We use the Correlates of War State System Membership List (v ) 8 The Formal Alliance data set distinguished between four types of alliances; Defense of one or more states in the alliance; neutrality toward one or more states in the alliance; promise of non-aggression; and understanding that the state would consult with one or more states in the alliance if a crisis occurred. We collapse the four categories into a dummy measuring whether sender and target have some form of a formal alliance or not. 17

18 to make this distinction on international status of the Black Knight (reference: Correlates of War Project. State System Membership List, v2008.1, 2008). 9 Mapping Black Knight Interferences So what do the analysis of the data on mass demonstrations in authoritarian regimes show? The descriptive statistics first and foremost shows, that black knight support matters. Out of 794 reported country years, mass uprisings were unsuccessful in the sense that the incumbent authoritarian regime weathered the storm and remained in power (see table 1). Out of the 722 unsuccessful cases, there were 379 unsuccessful mass campaigns where the target regime had external support. Thus, in a little more than half of the cases where the target stood the pressure from the masses (52.49 %), it was supported by a black knight. Only 26 target regimes fell, even that they had at least one external supporter (6.4 %). Table 1. Number of times target regime weathered the storm of a mass uprising by presence or absence of external support Success of mass uprising unsuccessfull successfull Total Note: numbers are in bold, column percentage in italic. Regime support No support External support Total ,2 93,6 90, ,8 6,4 9, The following countries are listed as major powers from : USA, United Kingdom, France, Soviet Union (until 1990), Russia (since 1991), China and Germany (since 1991) and Japan (since 1991). 10 Unfortunately, not the entire set of country years (891 in total) are evaluated with regards to the success of the campaign. 18

19 Mass uprisings, violence and international context This overall finding motivates our primary inquiry. Given that black knight support seems to matter, when is it most likely to occur? We will start by painting the general picture, and then dig into particularly interesting black knights and try to disentangle what might induce them to take on this role and provide assistance to dictators in need. The analysis is structured by the four types of factors hypothesized to shape the likelihood of black knight support: regime type congruence, geographical proximity, and density of linkages and the perceived geopolitical and economic value of the incumbent regime. We relate these three factors to the contextual factors (pre-/post-cold War and violent/non-violent campaign) throughout the analysis. As expected, the contextual factors weigh heavy on the calculations of external actors. The most dominant form of campaigns directed against authoritarian regimes is violent ones. Out of a total of 891 autocratic country years with mass uprisings 700 (or 79.2 %) were primarily violent. Likewise, in almost half (47.1 %) of all campaigns was the incumbent regime supported by at least one state or international organization, and such support was by far most frequent if the campaign was violent (85 %). 3 out of 5 mass uprisings happened during the Cold war era ( ) and the subsequent 2/5 evolved during the post-cold War era ( ). As hypothesized, external support was most likely in the former period (64 % of the regimes under pressure by the masses during the Cold War had external support and % had support during the post-cold War era) and by far most frequent, if the campaign was violent (87.7 % of the support in the Cold War was related to violent campaigns and 80.8 % in the post-cold War era) (see table 2). 19

20 Table 2. Number of mass uprisings by time period and primary method of the campaign. Primary method of campaign Primary method of campaign Violent Nonviolent Total Violent Nonviolent Note: numbers are in bold, column percentage in italic. Time period Cold War Cold War Total ,7 80,8 85, ,3 19,2 14, It is perhaps not so surprising that the Cold War saw most mass uprisings given that it spans over a longer time period than the post-cold War era. However, looking into the average number of conflicts per year, we can check to see if this pattern is more manifest. During the Cold War, the average number of campaigns per year is close to 12. During the post-cold War era the number is 13.2 conflicts per year. Thus, the Cold War period saw more mass uprisings in autocracies only in absolute numbers. As to the frequency of mass uprisings across the two periods, the 1970 s, 1980s and the 1990s stand out as particularly active decades with on average 17.4, 23.3, and 25 mass mobilizations per years. 11 These decades are also by far the most violent, with on average 15.5, 18.3, 17.7 violent uprisings respectively per decade. 12 Thus, it seems that the Third Wave (Huntington 1991) that radically expanded the number of democracies across the globe was also a period in which protesters were also more likely to take up arms to challenge incumbent dictators. 11 In comparison, the 2000s has been remarkable silent with 8 mass mobilizations a year on average. The 1950s saw 4 mass mobilizations per year on average, and the 1960s In contrast, the 1950s saw 2.6 violent uprisings per year on average, the 1960s slightly more 9.3 on average and finally the 2000s had 6.3 violent campaigns on average per year. 20

21 Table 3. Average number of mass mobilizations by regime support and primary method of the campaign No support External support Decade Violent uprising Nonviolent uprising Violent uprising Nonviolent uprising Total average of uprisings 1950s 1,6 1,5 1,6 1, s 4,8 1 4,5 1 9,9 1970s 7,2 1,8 8,3 1,3 17,4 1980s 8,4 2,6 9,9 2,7 23,3 1990s 9,8 6,2 7,9 1, s 3,8 1,3 3,3 2 8 Pinning down the black knights Now, let us take a closer look at who the black knights actually are. We begin with the international organizations that throughout the fifty years analyzed have provided assistance to authoritarian regimes challenged by mass uprisings. Surprisingly, international organizations supported incumbent regimes in only 3.8 % of all 891 country years with mass uprisings. In by far most cases, support was granted only if nation states also provided assistance (only 4 times did international organizations support a target regime alone), if the campaign was violent (80 %), and if it took place during the Cold War (76.4 %). Zooming in on the particular international organizations, we see that Western-based organizations are no less likely black knight supporters than other kind of organizations; in fact, on the contrary, they are the ones most often providing assistance to dictators in need (see table 4). Surprisingly, the European Union is the most frequent supporter. It sought to bolster an authoritarian regime fifteen times, followed by ASEAN (ten times), and NATO (five times). 13 Only two times did the United Nations or the Arab League involve themselves in a conflict of mass 13 In addition, ISAF supported dictators twice. 21

22 demonstration by supporting the target regime. The EU defended an authoritarian regime under pressure from violent campaigns seven times in the Cold War era and two in the post-cold War era. NATO, the Arab league, UN, and ASEAN only supported autocracies threatened by violent campaigns and with the exception of the UN only during the post-cold War era. Thus, the picture of the EU as the perpetual white knight in international politics perhaps needs to be reconsidered (cf., Tolstrup 2014). Table 4. Number of times international organization supported target regime, by regime support and primary method of the campaign. Cold War post-cold War International Violent Nonviolent Violent Nonviolent organization uprising uprising uprising uprising Total no support EU NATO 5 5 Arab League 2 2 UN ASEAN Turning to the black knight assistance of nation states, we see, contrary to our expectations, that the most frequent form of support was by a single state see table 5 (single states account for 19.5 % of all support to authoritarian regimes under pressure, 12 % received help from two states, and in another 11.8 % of the cases of black knight support cases three states or more were involved). However, taking the other contextual factors into account, a pattern more in line with our hypotheses emerge. Black knight assistance from sole actors is most frequent if mass uprisings were violent (74 %) or took place during the Cold War (76.4 %). Assistance from sole actors has thus seen a step drop in numbers from the former to the latter period. Whereas two state support has 22

23 dropped from the Cold War to the post-cold War era (from 82 cases to 25 cases), the opposite seems to characterize support from three states or more. This has gone up from 46 cases in the Cold War era to 59 cases in the post-cold War era. Three state support in fact is the most likely form in the Post-Cold War era vis-à-vis single state or two state support. Table 5. Number of times one, two or three states supported target regime, by regime support and primary method of the campaign. Cold War post-cold War International Violent Nonviolent Violent Nonviolent organization uprising uprising uprising uprising Total One state Two states Three states Combining the two contextual factors, 58.6 % of all instances of single state support were during the Cold War related to violent campaigns. In contrast, 50 % of all support from three states was during the post-cold War era and related to violent campaigns. Thus, state support is by far most likely, if the target regime is under pressure from a violent campaign, but the tendency to support alongside other states is slightly more likely in the post-cold War period, at least by three or more states. But who then are the nation states that have acted as black knights? Taken together, we have a total of 748 instances of black knight state support, whether that is primary, secondary or tertiary. The United States stand out as by far the most frequent supporter of authoritarian regimes under pressure from mass uprisings see table 6. Out of the 748 instances, it accounts for 186 instances (close to 25 %). The Soviet Union accounts for 99 cases, Japan 54, France 49, China 48, Cuba 37, the UK 30, Thailand 27, and Russia 13 instances. 23

24 Table 6. Number of times particular black knights supported a target regime, by regime support and primary method of the campaign. Cold War post-cold War Violent Nonviolent Violent Nonviolent State uprising uprising uprising uprising Total The US Cuba United Kingdom France The Soviet Union/Russia China Japan Thailand In three out of four times of the cases with US support, Washington acted as the primary supporter, it predominantly supported regimes under pressure from violent campaigns (164 instances or 88.2 %), and mostly so during the Cold War (126 instances, or 76.8 % of the 164 violent campaigns). France supported 47 target regimes under pressure by violent campaigns, 24 during the Cold War and 23 during the post-cold War era, the Soviet Union supported 84 violent campaigns and 15 nonviolent. Out of Japan s 54 instances of support, 48 were related to violent campaigns. 23 of these were during the Cold War and 25 during the post-cold War era. Russia only supported together with other countries from its region India, Pakistan, Iran, Armenia, Azerbaijan and China with the exception of Greece. These numbers indicates that support from states was more frequent if it also was perceived as a great power. In fact, in % (454 instances) of all 748 instances of state support, a major power was involved see table 7. Of the cases with support from only one state, 83.3 % involved a major power. 24

25 Table 7. Number of times a major power supports a target regime by time period and primary method of the campaign. Primary method of campaign Violent Nonviolent Time period Cold War post-cold War , ,3 Total Note: numbers are in bold, column percentage in italics Black knight support from great powers was more frequent during the Cold War era (60.4 %), and in line with our expectations regarding the impact of the international context, they were most likely to do it alone during this period (47 % of all support by major powers during the Cold War was done alone without any coalition partner). During the post-cold War era, the great powers seem more inclined to offer support if they are backed by a coalition of states (only 25.8 % of the major powers stood alone during the post-cold War era). Turning to the hypothesis concerning regime type congruence, we find that only in in 47.7 % of the black knight cases was the external actor also autocracy (see table 8) % of all authoritarian support occurred during the Cold War era and % of the support during the Cold War era was related to violent campaigns. Looking at the two dimensions violent/non-violent and Cold war/post-cold War together, the most frequent form of support from an authoritarian black knight to an authoritarian regime, was if the campaign was violent during the Cold War (55.46 % of the 357 instances) followed by violent campaigns during the post-cold War era (35.57 % of the

26 instances). Somewhat more surprisingly, democratic regimes acted as black knights in 56.6 % of the cases of support. Table 8. Black knight support from authoritarian and democratic regimes by time period and primary method of campaign. Primary method of campaign Violent Nonviolent Authoritarian black knight Democratic black knight post-cold Cold War post-cold War Cold War War , ,7 82, , ,3 17, Total Note: numbers are in bold, column percentage in italic. Congruence in subtypes of authoritarian regimes matters in 28 % of the 357 total instances of authoritarian support see table 9a and 9b. The kind of dictatorship and the kind of authoritarian supporter were, thus, the same type in a total of 100 cases of regime support, whether that was party-dictatorship, military-dictatorships, monarchy-dictatorships or personal dictatorships or had a primary, secondary or tertiary support status. 26