The force from below: popular mobilization Democratization never just happens. Someone has to take action to ' install, or protect, democratic institutions. In the most immediate sense these actions are usually taken by political elites: that is, by key decision 1 makers in the governmental organizations of the state. It has often been noted, however, that another forceful actor, the populace or "demos'p •■ itself, sometimes enters the scene, pushing for or resisting reforms. To what extent did this force from below affect the third wave of democratization? In this chapter I first review the literature and empirical evidence to this effect, and then, in order to uncover the causal median-isms at work, I turn to case studies of three critical incidences of democratization wrought through peaceful demonstrations. The literature and empirical results In the founding texts of the strategic approach, as noted in Chapter 1, " democracy appeared to have been brought about in the context of demobilized masses (O'Donnell and Schmitter 1986). This view, which was primarily based on experiences from Southern Europe and Latin America, was, however, challenged empirically in later accounts* i of the same world regions (Bermeo 1997; Collier 1999). Moreover, the contrast seems even sharper in relation to the subsequent collapse of i authoritarian regimes in Asia, Eastern Europe and Sub-Saharan Africa. In these instances collective action undertaken by the mass public appears to have been a widely occurring phenomenon, with alleged democracy-enhancing effects (Bratton and van de Walk 1997, pp. 83-4; Geddes 1999, p. 120; McFaul 2002, pp. 222-23; JBunce 2003, pp. 171-78; Schock 2005). Evidence from the "colored revolutions" in Eastern Europe and Central Asia points in the same direction i (Beissinger 2007; Tucker 2007). Anecdotal evidence also suggests that democratization in both Western Europe and Latin America in the early , twentieth century followed in the wake of social unrest and popular ; | f jj j-^g literature and empirical results 101 i 1 mass action (Acemoglu and Robinson 2006, pp. 67-68,71-73; also, see 1 Tilly 2004). From a theoretical perspective, dtis is what we should expect if the "social forces" approach to explaining democratization were to prove < correct, although this literature has never been very specific on causal i1 rnechanisms (Slater 2009). Strike activity should thus be one form of ' popular mobilization predicted to impact on democratization, particu-i larly within the strand of this tradition that emphasizes the importance of organized labor (RueschemeyereJij/. 1992; Collier 1999). But an effect of more general forms of protest activity undertaken by other groups in 11 society, including both violent clashes and peaceful demonstrations, ! could also be conjectured (Foweraker and Landman 1997; Gill 2000; Wood 2000; 2001; Schock 2005). Although less attention has been paid to the subject lately, there also seems to be a growing awareness of an older tradition daiming that popular mobilization may not be unreservedly beneficial for democracy. Bermeo (2003a), for one, takes note of the critical importance of labor mobilization and strike activity in undoing f democracy in Latin America during the 1970s (also see Armony 2004). In light of these observations there are surprisingly few large-w studies of the possible effect that popular mobilization may exert on democratization. To the best of my knowledge only two other global studies relate to the effect of popular mobilization (Lipset et al. 1993; Przeworski et al. 2000), but neither of them makes this assessment in dynamic models explaining regime change. I thus concur with ■ Coppedge's (2003, p. 125) verdict that "the true impact of political !> mobilization ... remains an open question." To resolve this question, I turn to Table 5.1. Confirming the expecta-tions of the social forces tradition, large numbers of peaceful anti-■i government demonstrations facilitated upturns toward democracy during the third wave. It should be kept in mind that this variable, p much as all the other time-varying determinants tested, is lagged one : !• year. What I observe is thus not an upsurge of popular protest that is an integral part of the democratization process itself. What is being captured i i is instead the impact of popular mobilization in one year on the propensity to democratize the following year, all else being equal, which lends stronger support to a causal interpretation of its impact. This estimated short-run increase in the rate of democratization was .038 per demonstration, 1 whereas the long-run equilibrium level of democracy increased by .475 per demonstration. The result is completely insensitive to the all 100 102 Table 5.1 Popular Mobilization The force from below Short-run. General Upturns Downturns Long-run Peaceful demonstrations .038'** .039*** -.001 .475*** (.009) (.007) (.006) (.142) Riots -.010 -.004 -.007 -.129 (.010) (.007) (.007) (.124) Strikes -.043* -.019 -.024 -.541 (.026) (.019) (.016) (332) * significant at the .10-Ievel, * * significant at the .05-leveI, * * * significant at the .01-Ievel, No. of observations = 3,795; no. of countries = 165; mean years observed per country = 23.0. Note: Entries are unstandardized regression coefficients with panel-corrected standard errors in parentheses. All models also include two lags of the dependent variable and the determinants in Tables 2.1, 3.1 and 4.1 as controls. All explanatory variables.in the table have been lagged one year. robustness checks applied in die Appendix C.1 This confirms, on systematic evidence, that during the third wave popular mobilization played a more influential role for the outcome than "transition paradigm" theorists initially acknowledged. However, I do not observe homogenous effects of all forms of popular mobilization. Riots (i.e., violent clashes involving the use of physical force) did not exert any impact on democratization. Although the general effect is negative and marginally significant, nor did strikes (aimed at national government policies or authority). Thus, although the effect of demonstrations is consistent with the more general "social forces" approach to explaining democratization, I find no systematic evidence in favor of a special role played by labor through the organization of strike activity {cf. Foweraker and Landman 1997; Collier 1999). Moreover, pace Bermeo's (2003a) insightful analysis of the Latin American experience of the 1970s, no forms of popular mobilization appear to work as triggers of downturns toward autocracy. The result is also insensitive to the exclusion of rwo relatively extreme influential outliers: Bolivia in 1980 and Argentina in 1983. Even more convincingly, the result holds when the square root of the number of demonstrations is used instead. sill' 'ill |i' The literature and empirical results 103 'tjj \ i, Is the peacefulness of anti-government demonstrations critical, or are we simply observing a form of popular contention that operates on a .;i > sufficiently large scale to undo authoritarian regimes? The fact that >\! violent riots have no effect speaks in favor of the former interpretation, ■ i »but the lack of an effect from peaceful strike activity, in favor of the i latter. Another test of which interpretation is correct would be to apply ..- ? a direct measure of violent popular contention, namely domestic armed 'conflict. In effect, there are theoretical ideas, and even some sketchy empirical evidence, of a positive link between the incidence of civil war and democratization. One suggestion is that the parties to a civil conflict could choose to invite the people as arbitrator: to hold an election and ;,'■! pass their power to a democratically elected government as a mechanism for conflict resolution (Wantchekon and Neeman 2002; Wantchekon 2004; Sollenberg 2008; cf. Wood 2000; 2001). In accordance with this proposition is Leonard Wantchekon's (2004, p. 17) observation that nearly 40 percent of all civil wars that took place 'between 1945 and 1993 resulted in an improvement in the level of democracy. Similarly, Bermeo (2003b, p. 159), by a rough estimate, i states that at least half of the democracies founded after 1945 that still ■::'. exist "emerged either in the aftermath of a war or as a means of bringing ■:. an ongoing war to an end." While these studies suggest a positive .:' impact of civil wax. termination, a different mechanism suggested is that armed conflict in and of itself helps solve collective action problems in attempts (or threatened attempts) to overthrow an authoritarian regime (Acemoglu and Robinson 2006). This would thus suggest a positive impact of even the outbreak of civil war. In Table 5.2, I test the democratizing impact of domestic warfare, here being defined as conflicts between the government and internal opposition groups incurring at least twenty-five battle-related deaths (Gleditsch et al. 2002). As the results show, while there was a negative impact, particularly on downturns, of having an ongoing civil conflict of this type in the past two years, there was no significant impact of either war termination or outbreak in the previous year. In other words, neither the Acemoglu and Robinson (2006), s nor the Wantchekon (2004) conjectures are supported for the third ! wave period. This, in turn, lends support to the contention that the effect of demonstrations not only stems from the fact that they are a j| form of popular mobilization, but also from the fact that they are peaceful. 104 Table 5.2 Domestic Armed Conflict The. force from, below ' Short-run General Upturns Downturns Long-run Civil conflict: Ongoing at t-2 -.083* -.016 -.068 ** -.996* (.048) (.033) (.031) (.583) Outbreak at t—1 .100 .087 .012 1.19 (.091) (.061) (.062) (1.10) Termination at t—1 -.053 -.028 -.024 -.627 (.108) (.075) (.065) (1.30) * significant at the. 10-Ievel, * * significant at the .05-leveI, * * * significant at the .01-level. No. of observations = 3,450; no. of countries = 155; mean years observed per country = 22.3. Note: Entries are unstandardized regression coefficients with panel-corrected standard errors in parentheses. All models use yearly change in the level of democracy as the dependent variable, including two lags of the level of democracy and the determinants in Tables 2.1, 3.1, 4.1 and 5.1 as controls. Why are peaceful demonstrations conducive to democratization? But why? Why do popular demonstrations foster democratization, and why is it critical that they are peaceful? The most promising attempt to tackle these grand questions, together with relevant case study evidence, can be found in the work of Kurt Shock on "unarmed insurrections."2 According to Schock (2005, pp. 40-44), the main explanation lies in what Lee Smithey and Lester Kurtz (1999) call die "paradox of repression": that an unarmed challenge may be sustained and even promoted in the face of brutal state force directed against it. There are several reasons why this dynamic may occur. Non-violent protest requires no special technology or equipment, nor is it critically dependent on the physical fitness of its implementers. It thus has "the potential to allow 2 Thompson (2004) documents case study evidence from fifteen "democratic revolutions" in Asia and Eastern Europe. The main thrust of his argument, however, lies in detailed descriptions of events, the conditions under which uprisings occur and their characteristics. As opposed to Schock (2005), no coherent account of why popular mobilization helps topple authoritarian regimes is developed. Peaceful demonstrations 105 the maximum degree of active participation in the struggle by the highest proportion of the population" (Schock 2005, p. 40). Moreover, harsh repression against peaceful demonstrators may have a mobilizing effect by spreading a sense of victimization, or even martyrdom, of innocent people. It may further or even spawn elite divisions by questioning the legitimacy of the regime, and it risks leading to mutinies or defections within the military and security forces, when diey are ordered to shoot at innocent people. Finally, unarmed protest raises the prob-' ability of third-party involvement, such as by transnational social movements, international organizations and foreign governments, by highlighting the fact that the incumbent regime rests on physical force rather then voluntary acquiescence. This in turn may further mobilize activists to join the unarmed cause. In all these respects, violent methods sharply differ. Armed rebellion requires weaponry and "has historically been limited to young, physically fit, ideologically indoctrinated or mercenary males" (Schock 2005, p. 40). Thus, the share of the population that may join an armed insurgency is comparatively limited. Using violence against the regime, moreover, is to fight the government on its own turf, where the challengers are likely the inferior party. It also justifies the state's use of repression in the name of "law and order" and "national security." Finally, violence may alienate potential supporters in the population, solidify the regime elites by creating a common enemy, and polarize diird parties (Schock 2005, p. 44). Although the contrast is less sharp, all these dynamics also help explain the difference between situations where peaceful mobilization occurs and where it does not. By provoking large numbers of people, by exacerbating elite divisions and mobilizing third parties against the regime, a democratizing outcome is made more likely in the wake of peaceful demonstrations. But why then does the authoritarian regime sometimes give in to the demands of challengers, sometimes not? Schock (2005, pp. 49-52,143-44), on the one hand, stresses internal features of die challenging movement - such as being decentralized but still coordinated within an umbrella organization - that may facilitate its resilience in the face of repression. Given that this resilience has kicked in, on the other hand, the critical feature is whether the resistance can "tap into the state's dependence relations." Although the conceptual language differs, here Schock basically adheres to Haggard and Kaufman's notion of "authoritarian bargains" discussed in Chapter 3: 106 The force from belou In any society, the state directly depends on segments of its own populace to rule. If any of these segments, such as military personnel, police officers, administrators, or workers in energy supply, transportation, communications, commerce, or other key sectors, refuse or threaten to refuse to carry out their duties, the state's power is significandy undermined. (2005, p. 53) To these domestic support groups, Schock (2005) adds the potential "indirect dependence relations" that the regime may sustain with third-party forces on the international scene. In both instances the key assumption is that an authoritarian state explicitly or implicitly draws its support from some groups or sets of actors, and that the withdrawal of that support undermines regime survival and opens the prospects of democratization. Again it seems fair to assume that this mechanism is less likely to occur in the absence of popular mobilization. In sum, then, Schock (2005) argues that "unarmed insurrections," incorporating a large set of non-violent activities (among which I only measure the occurrence of peaceful demonstrations), may lead to democratization by sustaining the challenge in the face of repression, thereby spawning elite divisions and mobilizing third-party support, and by disrupting the material and other support bases of the regime. The case studies that provide Schock with supportive evidence in favor of these propositions are the political transformations of the Philippines in the 1980s, South Africa in the late 1980s and early 1990s, as well as the transition toward democracy in Nepal and Thailand in the early 1990s. As. shown in Figure 5.1, three of these cases fit fairly well to the pathway criterion: The Philippine transition, which occurred over a couple of years, ranks 13th in 1984,19th in 1986 and 11th in 1987.3 South Africa in 1990 is ranked 14th and Nepal in 199122nd among all 3,795 country years in my estimation sample. The case that does not fit is Thailand in 1992, where the transition back to democracy was so swift as to occur in the same year as the people mobilized against the military dictatorship installed in the previous Although Marcos was forced to step down in February, 1986, surrendering power to a transitional government headed by Corazon Aquino, there were no clean elections or a ratified constitution until 1987. Moreover, already in 1984 a critical election was held where the opposition for the first time ran on a united ticket (see, e.g., Thompson 1995). In accordance with this, both Freedom House and Polity score the Philippine transition as occurring gradually over these consecutive years (with the largest shift occurring in 1986-87). ■If: peaceful demonstrations I If ' 107 4- E 2 LEtiil!ppJDesJfl82.pi]„.s__________ ■■-.19B6I1B] a 11 [22] South Africa 1990 [14] PhllirjpJnesJ[984[13] -10 -5 0 5 10 Peaceful demonstrations at t -1 (conditional) 15 Figure 5.1 Case studies of the impact of peaceful demonstrations on upturns Note: The graph is a partial regression (or added-variable) plot of the conditional relationship between the number of peaceful demonstrations, lagged one year, and upturns (n = 3,795). The figures within brackets are the rank order of each case in terms of the pathway criterion. year. Since I impose a one-year lag in my estimates of the effect of popular mobilization, I will thus not incorporate Thailand among the supportive case study evidence that follows. The Philippines As already mentioned in Chapter 3, the event that first triggered massive popular mobilization in the Philippines was the assassination of Marcos' main political rival, Benigno Aquino, upon his homecoming from US exile in August 1983. "In the days following the assassination," writes Schock (2005, p. 73), "hundreds of thousands of people filed by Aquino's open coffin, and an estimated two million people from all socioeconomic strata gathered to witness Aquino's funeral procession." From October that same year, weekly anti-Marcos demonstrations were staged in the Makati business district, but also freedom marches, so-called "parliaments of the streets," occurred irregularly, involving labor, peasant, 108 The force from below i student and teacher associations, as well as women's, human rights groups and gatherings of the urban poor. Another organizational specialty was taken up in January 1984, so-called welgang bayan, or "people's strikes," based on a general workers' strike but more comprehensive, "as: all stores are closed, all public transportation is stopped, and community members construct barricades to stop the operation of private vehicles"' (Schock 2005, p. 75). Such large-scale strikes were organized on several occasions throughout 1984, and on a particularly momentous scale in May and June, 1985 (Schock 2005, pp. 76, 83). By November that year, popular protest had deprived Marcos of almost all legitimacy. In an attempt to shore up his rule, he decided to call snap presidential elections for February 7, 1986 (originally scheduled to be held in 1987). This decision proved to be the last nail in the Marcos regime's coffin. Backed up by the powerful Catholic Church and an independent electoral watchdog organization called NAMFREL, the opposition's main 1 candidate, Corazon Aquino, managed to claim victory at the polls. Since the official and state-controlled electoral commission claimed Marcos the winner, the task to adjudicate in the conflict was passed to the similarly controlled parliament. On February 16, as the parliament certified Marcos' fraudulent victory, Aquino led a rally of approximately two million people, proclaiming victory for herself and condemning Marcos. She also announced a renewed non-violent ; disobedience campaign to be launched on February 26, the day after Marcos' planned inauguration, involving a general strike and a boycott of all crony-controlled banks, media outlets as well as a refusal to pay utility bills. This campaign was preempted, however, by an abortive military coup attempt by two ministers from Marcos' own administration -the generals Enrile and Ramos. When their plotted coup was revealed on February 22, they barricaded themselves in two military camps jusl outside of Manila and announced their defection from the Marcos government and their support of Aquino. Manila's Archbishop Sin, who had spoken out against the regime since the establishment of martial law in 1972, now urged people to support the mutiny, drawing tens of thousands of supporters to surround the camps and protect the military rebels with their bodies. This effectively blocked Marcos' attempt to crush the mutiny by military means. By February 25, hundreds , of thousands of civilians were gathered outside the military camps, at the same time as Aquino set up a parallel government to peaceful demonstrations 109 '■• denounce Marcos' rule. By this time the US administration under 11'President Reagan, a close friend of Marcos, officially withdrew its '■' support for Marcos' rule. The following day he fled the country [ (Schock 2005, pp. 77-79). All essential parts of Schock's theory of how popular mobilization may topple authoritarian regimes were in place in the Philippine case. The protest activities evidently mobilized large numbers of people, and ' from most segments of society. They exacerbated elite divisions, parti-', ' cularly the split between crony and non-crony business elites, and within the military (as the unraveling of the Enrile-Ramos mutiny ■ made clear). Finally, they tapped deeply into the state's dependence relations, both within and outside the country. The strikes and boycotts I undermined the already recessionary economy, destabilized the busi-i ness climate and in the end caused massive capital flight. Not the least important, the protest activities eventually triggered the withdrawal of ! US support for Marcos (Schock 2005, pp. 88-90). South Africa South Africa had since independence in 1910 been ruled on the principles of racial segregation, later consolidated under the apartheid system, denying political rights and civil liberties for the black majority population. A series of popular struggles throughout the 1980s, first ignited by the mass shootings in Soweto in 1976, however, finally brought this authoritarian regime to its end. Three umbrella organizations are generally considered as having provided leadership for the uprisings: the United Democratic Front (UDF), launched in 1983 to oppose a new constitutional proposal by the government; the Congress of South African Trade Unions (COSATU), formed in 1985 in response to a state of emergency leading to heightened repression of labor; and, of course, the African National Congress (ANC), founded in 1912 but banned in 1960, which thus mostly provided an underground support base in events unraveling up until repression was lifted in 1990 (Schock 2005, pp. 57-59, 64, 67-68). The South African resistance movements drew upon a wide range of non-violent (and, as we shall see, some violent) actions to oppose the regime. Two waves of protest may be identified. The first, unsuccessful in the short term, cmminated in two states of emergency in 1985 and 1986. The second, occurring in 1989, is the one that lends South Africa the 110 The force from below status of a pathway case in my analyses. During the first wave, the UT)F organized, apart from large-scale demonstrations, a series of boycotts on housing rents, services, schools and local business oudets, while COSATU, apart from labor strikes, organized "stayaways," in which' workers and students "would stay at home in support of a variety of economic and political demands" (Schock 2005, pp. 61, 65). When thef first state of emergency in 1985 only fuelled the uprising and also led to bloodshed and riots, a second, nationwide and more effective state of emergency was installed in 1986. This brought an end to the demonstrations, but not to the stayaways and the rent and consumer boycotts (Schock 2005, pp. 62-63; Price 1991, pp. 266-67). Although the UDF* was banned in 1988, its detained activists organized a hunger strike including more than 600 prisoners in February, 1989, eventually leading to their release (Schock 2005, p. 63). This sparked the second wave of protest throughout the country, called the "new Defiance Campaign," "collectively encompassing hundreds of thousands of demonstrators, and effectively neutralising the state of emergency" (Zunes 1999, p. 159). Together with a general strike in August that same year, "essentially: shutting down commerce in Pretoria, Johannesburg, Durban and East London, and severely crippling industry in the Western Cape". (Zunes 1999, p. 156), this marked the beginning of the end of the authoritarian regime. After P.W. Botha had stepped down due to a stroke, F. W. de Klerk was first elected new party leader of the incumbent National Party in February 1989, and then national president in September (Wood 2000, p. 180). In early February 1990, de Klerk declared that the state of emergency was to be lifted, the ANC and UDF unbanned, and political prisoners, among them Nelson Mandela, to be released. This formed the start of a protracted negotiation between the ANC and the government, eventually leading to free and fair elections with universal.suffrage in 1994 (Schock 2005, pp. 56, 68). The year in which South Africa enters my statistical results as a pathway case, that is in 1990, is thus not the year of a completed transition to democracy, but the year of partial political opening (i.e., "liberalization," in terms of the old transition paradigm parlance). It is thus noteworthy for the gradualist strategy toward measuring democracy I employ that Schock's mechanisms are still in full sway in the South African case. The popular uprising in South Africa clearly mobilized large segments of the population, for the most part by being able to sustain protest {peaceful demonstrations 111 1«! Bii activities in the midst of harsh repression, and generated substantial third-party support, both from white South Africans and from abroad, most notably Great Britain and the US. It also precipitated elite divisions, most importantly the divide between the hardliner "securocrats" and the softliner "internationalist reformers" within the government i(Schock 2005, pp. 68, 89). The former group, closely associated with !p. W. Botha, comprised personnel within the major security agencies land emerged as the most powerful government body after the second state of emergency in 1986. On the other side of the division were officials associated with economic policy and foreign relations, both of whom started to see negotiations and more radical democratic reform as the only way out of the government's predicament. The election of F.W. de Klerk, who stood on the latter's side, thus "marked the ascendancy of the internationalist-reformers over the securocrats" (Price 1991, pp. 252, 275-76). Finally, popular mobilization undermined the South African state's legitimacy and resources, most directly by exploiting the apartheid system's dependence on black labor, but also indirectly by reinvigorating international support for economic sanctions (Schock 2005, pp. 85-87).4 Nepal As compared to both its Philippine and South African counterparts, the transition toward democracy in Nepal was a swifter affair. After a short democratic interlude in the late 1950s, Nepal was governed as an absolutist monarchy from 1960, although under the trappings of the so-called panchyat system, with consultative bodies at various levels that, however, were completely controlled from the royal palace , (Schock 2005, pp. 121-22). The trajectory of the pro-democracy movement in the country took a new direction from the late 1980s, when the Nepali Congress Party, which prevailed as an underground organization despite being banned under the constitution, forged an alliance with the several communist parties of the country. Begimiing on February 18, 1990, this movement of forces staged an organized unarmed rebellion against the monarchy and its panchyat system. 4 As Elisabeth Wood (2000, chap. 6-7) has persuasively shown on the basis of extensive interview data, this also caused the withdrawal of regime support froin the South African economic business elite, which instead started to push the incumbent Nationalist Party rulers to initiate democratic reform. 112 The force from below The first peaceful demonstration gathered some ten thousand people; in central Kathmandu, but also in several other cities around the coun-try, and was met with harsh police brutality, including arrests, deaths and injuries. A general strike was called the following day, again leading to clashes between unarmed demonstrators and police forces. On February 25, a second wave of demonstrations swept across the country, "ushering in a period of terror unprecedented in recent Nepalese history." Approximately one thousand protestors were arrested, some of whom were subsequently tortured, and the government used spei1 daily trained thugs to attack the demonstrators (Schock 2005, p. 124). After these events, according to one observer, "the level of participation began to dwindle," but the pro-democracy movement then turned to other more covert protest measures instead (Parajulee 2000, p. 87). Most notable among them were strikes, since it is harder for the regime to punish people for what they do not do than for what they do, and so-called "blackouts," whereby all lights were turned off between 7.00 and 7.30 p.m., "signaling solidarity and widespread support for the movement" (Schock 2005, p. 124). Through these and other inventive measures, the pro-democracy movement managed to regain momentum in a new series of large-scale protest demonstrations by early April. In the midst of "continuous strikes and demonstrations," King Birendra on April 6 attempted to appease the masses by sacking the prime minister and promising limited reform (Parajulee 2000, p. 90). In response, people poured into the streets of Kathmandu and most other major cities in the largest demonstration ever to have occurred in the country, involving hundreds of thousands of people. While there was shooting and occasional violence, demonstrators mostly stayed peaceful. Shortly before midnight diat same day the Icing issued a proclamation, broadcast over Radio Nepal, to lift the ban on political parties (Schock 2005, p. 125). Within days the king began negotiating with the opposition, leading to the formation of a new interim government, the dissolution of the pancbyat system, and the establishment of a constitutional amendment system (Parajulee 2000, pp. 93-94). A new constitution that circumscribed monarchical powers under a parliamentary system was promulgated in November, followed by parliamentary elections that were generally considered free and fair in 1991 (Schock 2005, p. 125; Parajulee 2000, pp. 105-16). All components of the mechanism linking popular mobilization to democratization are again present in the Nepalese case. The .....Bflf li'M I'lill I [•Veaceftil demonstrations 113 II l'>l I''Pill'' , , i! i p I 1 VI iiVii1 pro-democracy forces were clearly able to sustain their campaign in ' ' 11 '{ ll| the face of repression, and "anti-government sentiment was only inten-, 'i| | j sified by the violent repression of unarmed demonstrators" (Schock 'i'l % l'ijl'|li2005, P- 140). The popular uprising also drove a wedge between 1 ! '] ' more liberal reformist and the hardline conservative factions within Hi |il11 'jl'f (the regime. The foreign minister, belonging to the former group, resigned I H I,from the cabinet - in part due to the violent repression used by the egime - by late March, and a cabinet reshuffle in early April was | implemented to purge the government of softliner ministers. Also -riembers of the National Panchyat (the party-less parliament) condemned ' l ''"'j 11" ii | the use of violence, pointing to yet another rupture within the elite p" | 1 (Parajulee 2000, p. 89). In Schock's own words, "[m]ass political 'l ! *l [' i if| action and elite defection formed a combination that was potent i | 1, ' 1 in the toppling of the regime - to the extent that the regime was i ,Ji I'm' overturned despite the fact that the military remained loyal through-l' i 1 1 out the crisis" (2005, p. 140). ,' \i Not least important, the unarmed popular insurrection undercut the '' ' , I -I | regime's critical dependence relations. On the domestic scene, this occurred most prominently by the withdrawal of support from govern- I ' !j j «||j | ment employees, such as lawyers, doctors, nurses and university pro-i! ,, fessors (Schock 2005, p. 136; Parajulee 2000, pp. 83-88). i |,] I | Internationally, the pro-democracy movement consciously exploited i 111 Nepal's dependence on foreign development aid. As the violent repression of unarmed citizens represented grave human rights violations, the 'i (l*' popular uprising mobilized third-party support from transnational NGOs such as Amnesty International and Asia Watch. Eventually the US, Germany and Switzerland "publicly condemned the repression of ' the government, threatened to withdraw aid, and privately pressured ! * f [ i the government to negotiate with the pro-democracy movement" !i \ (Schock 2005, p. 137).5 '.'ii 'II r • i 3 In a sample survey of a broad range of pro-democracy activists and government II I ( | j|H 11 representatives in Nepal, Parajulee (2000, chap. 6) documents the widely held belief that donor countries, particularly the US and Germany, as well as , ' international human rights organizations and the international media, played a ' I'1 ( 'A [ significant role in the success of the Nepalese transition to democracy. Although 1 the "Indian connection," mostly through India's imposition of sanctions against Nepal in 1989, was also an international factor of importance (Parajulee 2000, '"' |!| chap. 5), that external influence took place before the popular uprising started, 'i , 1 and thus cannot account for its success (cf. Schock 2005, pp. 136-37). 114 The fores from hrfnw ' I] Summary Schock's theory and case study evidence obviously does not answer all , questions related to why and how popular mobilization affects democratization. An important puzzle that arises from my statistical findings is why only demonstrations, not strikes, appear to have a signil n impact. As the cases of the Philippines, South Africa and Nepal all ' ',(. clearly document, strike activity was a crucial part of the popular i insurgency, particularly when it came to severing the authoritarian regimes' dependence relations. Schock {2005, pp. 38-40) does not distinguish between demonstration and strike activity, but views both as an integral part of the non-violent action repertoire. Why then do the i j number of demonstrations in my results affect democratic upturn-irrespective of the amount of strike activity? And why does strike activity appear inconsequential? As noted in Chapter 3, one of Haggard and Kaufman's (1995) key 1 >l pathways through which economic crises affect democratization is by ' • the mobilization of popular protest. This suggestion is actually also in ■ keeping with how their case studies line up in my statistical findings. The three economic crisis cases depicted in Chapter 3 (that is, apart from the '.[ Philippines: Bolivia, Argentina and Uruguay), as well as the cases of f Peru in 1978 and Brazil in 1985, are all among the top twenty-five l| pathway cases for my effect of peaceful demonstrations on upturns. As Haggard and Kaufman persuasively argue, poor economic performance in these cases helped trigger the mobilization of protest against the ' regime, first based on economic grievances but along the way turning the protest agenda toward broader civil liberties and political rights issues (1995, pp. 60-64). However, they never distinguish explicitly between strikes and demonstrations, and hence cannot help explain why strikes in my data are not related to democratization. Except for the Philippines the Haggard and Kaufman cases are also^ included in Collier's (1999, p. 23) list of "recent democratization" ' cases, where she deemed "pro-democratic labor action" consequential for the outcome. Adding two of the cases on her list, Portugal in 1976 I and Spain in 1978, with the corresponding pathway ranks of 6 and 12, respectively, there seems to be even more prima facie case study evi- i dence that popular mobilization is causally linked to democratization. Again, however, Collier restricts her attention to labor mobilization, Wi |i feaceful demonstrations 1 15 if,; i ii r i finding strikes to be "the cutting edge" of popular mobilization's effect ,pl on democracy (1999, p. 185). This again speaks to the need for an -i. ' account of the mechanisms linking popular protest to democratization 1 that explains why participation in labor strikes is not systematically ^'Jf.1 related to democratization.6 j i Another remaining puzzle concerns the role played by violence. As I 'J, noted above, Schock (2005) explicitly aims at explaining why the non-I'jljl' violent character of these uprisings matters, and my statistical finding Pl'l 'with respect to both riots and civil conflicts tends to confirm this claim. But among the three cases under study, only the Nepalese was free from j '''Jj popular violence. In the Philippines the armed wing of the Communist f'V- Party had been waging a revolutionary guerilla war against the regime i|ij 11| since the early 1970s, and in South Africa the ANC had been committed ,' I to armed rebellion for decades. Was it really then the unarmed insurrec-j I tion that toppled the authoritarian regime in these cases, or could the '' "'' parallel armed rebellion have contributed with a possible "radical flank 'i-'i effect"? l' i'!' Schock argues that the armed strategy of the Philippine rebels effec-\, ' tively inhibited widespread support for their existence, and that it, by |11 '|| contrast, encouraged US support for the Marcos regime (2005, pp. 71, J1','1 90). The unarmed insurrection on both accounts differed, since it mobi->\'\>t lized momentous support from within the country while eventually II 1 deterring external regime support from the US. In the South African I' 11, case, moreover, the ANC never formed a real military threat to the state 1 , (2005, p. 66)7 Nevertheless, Schock (2005, pp. 158-60) concedes that it cannot be ruled out that the armed insurgents in the Philippines and 1 South Africa did play a supportive role in the toppling of the authoritarian regime, although they apparently were unable to beat the regime in military terms. 'i 6 Apart from poor data qiiality, which is of course always a possible culprit, timing could be at least part of the answer. Both the Philippine and the South African 1 cases would seem to suggest that labor mobilization partly precedes the broader 11 demonstration activities. This pattern is also what Joe Foweraker and Todd Landman (1997, chap. 5) found in their more detailed mobilization data from Spain, Chile, Brazil and Mexico. In accordance with this hypothesis, if T lag the ; v strike variable two years instead of one, so that it in the data precedes the 11 >i demonstrations variable, its coefficient turns positive (although still not , I significant). , i 7 It is also noteworthy that according to the civil conflict measure I apply (see Appendix A), armed resistance ended in South Africa in 1989, that is, the year ■ in; i before the political opening. 116 The force from belou A final puzzle concerns timing. Apart from the Nepalese case, the cases under study were sustained mobilization campaigns occurring over several years: in the Philippines for around 30 months, in South Africa for over a decade. What explains these differences? And how cari; future statistical work on this topic make their findings more sensitive to country-specific lag specifications? Having mentioned these remaining puzzles, the combination of my ■ statistical findings with the causal mechanisms accounted, for by Schock's case study evidence lend support to a truly causal interpretation of the relationship between peaceful demonstrations and democratization. Conclusion To sum up, I have in this chapter found empirical evidence to the effect that peaceful demonstrations were a positive trigger of democratization 1 during the third wave. Case study evidence supports the notion that this occurred by increasing the likelihood of divisions between hardliners and softliners within the regime elite, by incorporating larger segments of society in a more sustained protest through the paradox of repres-: sion, and by tapping into the state's economic and other dependency relations. I have, however, found no systematic support for a link between riots, strikes or domestic armed conflict and democratization. 1 There is a strong force from below, when peaceful. 61 Exogenous shocks and authoritarian J regime types: institutional contingency One of the latest turns in the field of comparative democratization has been an increased attention paid to different types of authoritarianism. There is a growing literature on the dynamics of different forms of autocracies, including die "hybrid" regimes located in the gray zone between democracy and autocracy (Geddes 1999; 2003; Diamond 2002; Levitsky and Way 2002; Schedler 2002a; 2002b; 2006). In this chapter I shall contribute to this body of knowledge by testing the extent to which different authoritarian regimes have different propensities to democratize, both in and of themselves and in response to exogenous shocks. Since different institutional traits are what distinguish different regime types, I shall thus explore the institutional contingency of democratization. As opposed to mostprevious studies, however, I shall not be concerned with the institutional configurations distinguishing between democracies, such as forms of government and the electoral system (see, e.g., Persson and Tabellini 2003; Cheibub 2007) or other power-sharmg institutions (Norris 2008). Instead I focus on the consequences of authoritarian institutions, mostly argued to be a neglected subject in the field (Snyder and Mahoney 1999). In contrast to the theoretical eclecticism of previous chapters, I shall in this chapter take a more deductive approach. My theoretical point of departure will be Geddes' (1999; 2003) seminal treatment of the logic of military, one-party and personalist dictatorships.1 Extending the argument into some other authoritarian regime types omitted by Geddes, I first derive expectations regarding how types of authoritarian regimes affect democratization, and how they respond to exogenous shocks. I then present my estimation strategy, followed by the findings. 1 A more recent theoretical framework for understanding authoritarian regimes that I will not address in this chapter is Gandhi and Przeworski's (2006; 2007; also, see Gandhi 2008), the most important reason being that they mainly claim to explain the emergence of different institutional setups under autocracy, not the iatter's effect Dn democratization. 117