Clouds with silver linings: how mobilization shapes the impact of coups on democratization
Abstract
There is a long-standing debate over the impact of coups on democratization. Some argue that coups can help promote transitions to democratic rule. Yet, others contend that coups often spur increased repression and autocratization, undermining hopes of democratic reform. We argue that both democratic and autocratic changes are more likely after a coup and that popular mobilization plays a crucial role in shaping the post-coup trajectory. Democratization is more likely when coups occur in the presence of significant popular mobilization. A coup reveals cracks within a regime, and the combination of pressure from within and threat from below during popular mobilizations fosters greater incentives to promise democratic reform. In the absence of popular mobilization, autocratic rule is more likely, especially when a coup is successful. We test our argument on the combined effect of popular mobilization and coups on changes in democracy in a global dataset, considering the specific dates of events and institutional changes, the outcomes of coups, and using decay functions to capture persistent effects. The analysis provides strong support for our argument, with the key findings robust across a number of alternative tests. Our analysis underscores the value of examining variation in the context of coups to understand their likely political consequences.
Introduction
There is a long-standing discussion on the impact of coups on democratization and political change.1 Although coups often end democratic rule, some researchers have argued that coups can provide important opportunities to remove autocratic rulers and foster democratic change (Collier, 2009; Marinov and Goemans, 2014; Miller, 2012, 2014, 2016, 2021; Snyder, 1992; Thyne and Powell, 2014). The 1974 coup in Portugal, for example, ended long-standing autocratic rule and paved the way for a democratic transition. Recent coups such as Mali in 2020 and Sudan in 2021 have been accompanied by promises of elections and reform. Others, however, emphasize that coups are most often followed by increased repression and autocracy, even if autocratic rulers are ousted (Derpanopoulos et al., 2016, 2017; Lachapelle, 2020). After a failed 1969 coup attempt in Equatorial Guinea, for example, President Nguema unleashed a wave of repression and suspended the constitution to rule by decree, making the country even more autocratic (Jackson and Rosberg, 1982: 246).2 While some have recognized the heterogeneous impact of coups on political change (Tansey, 2016; Zengin, 2021), we still lack an understanding of what causes this variation.
In this manuscript, we seek to reconcile these disparate findings and argue that a missing link is popular mobilization. We assert that the threat from popular mobilization plays a crucial role in shaping the post-coup trajectory. The aftermath of coups tends to be chaotic; rulers often have limited control after a successful coup, and failed coups can reveal serious vulnerabilities for an incumbent. Coup attempts boost intra-regime tension, increase the likelihood of new coups, decrease expected regime life-span, and heighten the probability that the incumbent is exiled, jailed, or executed. The existence (or absence) of popular mobilization affects leaders’ incentives to democratize or autocratize as a way to overcome the towering challenges created by the coup. Regime divisions combined with pressure from below can promote democratic reforms, as concessions to a protest movement can help navigate the political crisis. Moreover, regime splits improve the prospects for effective popular mobilization. Thus, when coups and popular mobilization coincide, the regime should be more susceptible to offer concessions and democratic reforms to try to hold on to power. Furthermore, an incumbent facing elite divisions may also seek to establish alliances with a mobilized opposition, thereby increasing the potential leverage of the opposition and its ability to demand democratic reform. Finally, popular mobilization can strengthen soft-liners within a regime favoring democratic reform. In the absence of popular mobilization, however, an incumbent is likely to respond to regime fragility and elite divisions by tightening autocratic rule and expand the use of repression.
We extend prior research on coups, mobilization, and political change by showing how elite divisions combined with mobilization from below can promote democratic change as well as when coups are more likely to strengthen autocracy. We provide a first systematic evaluation of how popular mobilization affects political change after coups and show results in line with our theoretical expectations. Our analysis considers the full range of possible changes after a coup, that is, any change toward more democracy or autocracy, unlike existing studies restricted to changes between regime categories. We develop measures based on the precise dates of events and changes, and use decay functions to capture delayed effects of coups and mobilization. This mitigates potential problems arising from the loss of the ordering of events in aggregation and allows us to account for the sequence of events. We first present robust evidence showing that the likelihood of changes in either direction increases substantially in the aftermath of a coup, highlighting the importance of understanding the heterogeneous effects of coups d’état. We then show that coups in the context of popular mobilization are likely to spur democratic change and that the impact of popular mobilization on democratic change almost doubles in the presence of a coup. In contrast, in the absence of popular mobilization, successful coups are likely to foster autocratic change, while failed coups do not show a systematic impact on autocratization. The findings lend substantial support to our theoretical argument and highlight how popular mobilization can sway post-coup trajectories toward more democracy. Indeed, combined with popular mobilization, the clouds from coup attempts are likely to have silver linings, with brighter prospects for future democratic reform.
The ambiguous nature of coups
Coups and democratization
Research on coup outcomes has traditionally focused on the role of coups in democratic breakdown (Onwumechili and Carle, 1998; Svolik, 2015), but recent research has highlighted how coups can also promote democratization in autocracies. Collier (2009) argues that a coup can remove an entrenched dictator, potentially opening up for reform and democratization. More systematic empirical studies by Thyne and Powell (2014) and Miller (2016) find that coups notably increase the likelihood of a transition to democracy. Moreover, Miller (2012, 2021) argues that violent turnovers between autocratic rulers make subsequent democratization more likely, and coup-initiated transitions are the most common path to democracy. This line of research tends to emphasize coups successfully ousting an autocratic regime as a launchpad for democratization. It identifies three key channels linking coups to democratic reforms. The first emphasizes how international pressure affects leaders in need of foreign aid, trade, and investment (Marinov and Goemans, 2014; Thyne and Powell, 2014). Promises of aid or removing sanctions are increasingly conditional on demands for democratic reform and respect for human rights after the Cold War. Marinov and Goemans (2014) suggest that the pressure for elections implies that only actors expecting to win elections will stage a coup.
The second mechanism pertains to the need to enhance legitimacy after usurping power. A leader may try to legitimate a coup by claiming popular consent, but this is difficult to do without opening up for elections (Svolik, 2012; Thyne and Powell, 2014). Leaders who have seized power in a coup face a high risk of challenges from rivals (Bueno de Mesquita, Siverson, and Woller, 1992; Londregan and Poole, 1990; Singh, 2014; Thyne and Powell, 2014). Offering democratic reforms to broaden public support provides a possible way to try to curtail risks.
The third mechanism arises from the likely post-tenure outcomes if a leader loses power. Only 2 percent of leaders who lose power through regular exit are imprisoned or killed in the following year, but the share is over 1/3 (36.6%) for leaders who lose power through irregular means (Miller, 2021: 51). From this perspective, implementing democratic reforms is often better than alternatives—or in the words of Miller (2021: 6), “a salvation rather than sacrifice.”
Coups and autocratization
Even if coups can open up for democratic change, they can also be followed by democratic backsliding and autocratic consolidation (Bermeo, 2016; Kieh and Agbese, 2004; Marinov and Goemans, 2014; Onwumechili and Carle, 1998; Svolik, 2015; Waldner and Lust, 2018). Derpanopoulos et al. (2016, 2017) note that coups in an autocratic regime are most likely to lead to more autocracy and repression.3 Tansey (2016) concludes that coups rarely lead to high-quality democratic rule, even when they are followed by elections and competition. Furthermore, successful coups have been linked to increased repression, weaker judicial institutions, and increased corruption (Bennett et al., 2021; Lachapelle, 2020).4
All coups—whether successful or failed—reveal deep divisions within the regime and highlight critical and enduring challenges (Miller, 2021). Coups increase the likelihood of new coups (Londregan and Poole, 1990; Powell, 2012). A higher risk of coup and revealed elite or regime divisions may also encourage increased popular mobilization. Coups rarely install durable autocracies, and most coup leaders face high insecurity and short expected tenures. Less than 70 percent of coup leaders remain in power until the following year, fewer than 50 percent survive for 2 years, and not even 25 percent make it to 6 years. Leaders may try to enhance control through increased repression, purges, and efforts to discourage opposition, potentially leading to more autocratic political institutions (Bove and Rivera, 2015; Easton and Siverson, 2018; Lachapelle, 2020; Tansey, 2016).
The determinants of the post-coup trajectory
Our brief review of the literature demonstrates the ambiguous effects of coups on political change and prospects for democratization, appearing to spur increased repression and autocratization in some cases and elections and democratic reform in others. One possible explanation is that coups simply have highly erratic consequences and no consistent effect on democratization, reflected in the divergent results and high sensitivity to specifications in existing empirical studies. A more compelling alternative is to consider plausible factors that might account for this heterogeneity and identify when one set of these mechanisms is likely to dominate. We posit that popular mobilization plays a key role in shaping post-coup trajectories. Powell and Thyne (2011) define coups as “illegal and overt attempts by the military or other elites within the state apparatus to unseat the sitting executive” (p. 252). Coups represent clear threats to leaders “from within” elites, but leaders can also face important challenges “from below” through popular mobilization. By one estimate, popular uprisings are second only to coups in removing autocratic rulers (Svolik, 2012).5 Large-scale popular mobilization has in practice posed greater challenges to leaders than violent uprisings.6 Furthermore, many studies show that non-violent movements increase the likelihood of democratic transitions compared both to cases with violent mobilization and no mobilization (Chenoweth and Stephan, 2011; Kim and Kroeger, 2019; Rivera Celestino and Gleditsch, 2013; Teorell, 2010; Ulfelder, 2005).
Coups and popular mobilization are, however, not independent alternative threats to political leaders; they often occur at the same time and may reflect common conditions and causes. In our data, we find that 13 percent of all coups occur within the same calendar year as popular mobilization, and about 40 percent occur within a 4-year window of popular mobilization. Popular mobilization against an unpopular ruler can legitimate attempts to seize power and increase elite divisions. Both forces were present in Egypt in 2011, when the army removed President Mobarak following a popular uprising. A number of empirical studies confirm that coups are more likely during popular mobilization (Casper and Tyson, 2014; Frantz and Ezrow, 2011; Johnson and Thyne, 2018; Powell, 2012; Wig and Rød, 2016; Yukawa et al., 2022). Coups can also boost popular mobilization, encouraging more actors to participate in anti-government demands (Brancati, 2014). Coups and popular mobilization are often examined separately, even though political change often reflects a combination of pressure from below and from within, and elite splits are rarely exogenous to popular mobilization and dissent.7
Just as coups and protest may have related origins, mobilization can fundamentally shape the prospects for democratization after a coup. Looking at each component in isolation risks exaggerating the apparent independent effect of individual factors. Although we have insightful case studies relating popular mobilization and elite divisions in democratization and political change, we lack comprehensive comparative studies of their combined impact.8 An important exception is Miller (2021) who argues that the effect of popular mobilization on democratic transitions is contingent on violent shocks disturbing the autocratic equilibrium, including coups.9 We extend Miller (2021) and consider the combined effect of popular non-violent mobilization and coups on polity changes. We argue that popular mobilization can make democratization more likely after coups, but coups are unlikely to foster democratic change in the absence of popular mobilization.
Coups and regime vulnerability
Coups introduce a host of challenges to the regime. Both failed and successful coups increase the likelihood of new coup attempts, decrease the expected life-span of a regime, and increase the likelihood that the incumbent ends up in exile, is jailed, or executed (Chambers and Ufen, 2020; Tansey, 2016). One year after losing office, 31.5 percent of leaders are in exile, 13.9 percent are in jail, and 12.5 percent are dead (Miller, 2021: 71–72). This situation is further complicated by lack of oversight; coups make it unclear who is siding with whom. Consequently, intra-regime tensions, fractionalization, and competition will likely be exacerbated (Chambers and Ufen, 2020; Miller, 2021; Tansey, 2016). Leaders in power after successful or failed coup attempts will need to consolidate power to reduce the likelihood of future attempts (Bennett et al., 2021). In the following section, we detail how popular mobilization can influence the likelihood that leaders will seek to resort to repression or concede to reforms in order to address the challenges after coup attempts.
Post-coup challenges during mass mobilization
Existing research provides clear support for a positive impact of mass mobilization on autocratic breakdown and democratic change (Bayer et al., 2016; Dahlum et al., 2019; Hollyer et al., 2015; Karatnycky and Ackerman, 2005; Kim and Kroeger, 2019; Rivera Celestino and Gleditsch, 2013; Teorell, 2010; Ulfelder, 2005). A protest movement demonstrates both willingness to mobilize and the ability to overcome barriers to collective action. Moreover, popular mobilization often brings together broad coalitions and wide social networks (Chenoweth and Stephan, 2011). This empowers the opposition and strengthens the masses relative to ruling elites (Gleditsch and Ward, 2006; Kim and Kroeger, 2019; Rivera Celestino and Gleditsch, 2013). Popular mobilization can spur democratic change in the wake of deposed autocrats, but an incumbent can also initiate democratic reforms when confronted with popular mobilization. Work on democratization has stressed how reforms often arise as a response to changes in the balance of power between the ruler and the ruled (Acemoglu and Robinson, 2001), and ruling elites have strong incentives to promise democratic reform when the regime is threatened. Democratic reforms may reduce their power, but can allow them to retain political influence, and are more attractive than to be overthrown (Kim and Kroeger, 2019). Leaders losing power in competitive elections face better post-tenure fates than leaders ousted irregularly (Debs, 2016; Kim, 2017; Kim and Kroeger, 2019; Miller, 2021).
The likelihood of success clearly depends on movement characteristics, such as size and inclusiveness (Chenoweth and Stephan, 2011; Dahlum, 2022), but is also shaped by the strength and unity of the regime. Miller (2021) suggests that even broad-based movements have limited ability to unseat an authoritarian regime when the regime is united and willing to make use of physical force. However, coups reflect elite divisions and deepen cracks within a regime, expanding opportunities for effective popular mobilization (Miller, 2014, 2021). Popular mobilization in the context of a coup further increases the revolutionary threat and yields incentives to implement democratic reforms. An incumbent might try to undermine and repress a movement, but repression might backfire and increase support for mobilization, both among the general population and individuals within the regime (Chenoweth and Stephan, 2011; Davenport et al., 2022; Hess and Martin, 2006; Rasler, 1996). Orders to repress can turn the security forces against the incumbent, spelling the end of the regime with the loss of coercive capacity. Such loyalty shifts are more likely to occur when regime divisions are highlighted by a coup. Thus, repression is at best a high-stakes gamble for a divided regime facing popular mobilization. In sum, elite division and popular mobilization can reinforce each other, and coups in the context of popular mobilization is likely to boost mechanisms connecting mass mobilization to democratic change.10
Democratic reforms can also be introduced by incumbents to navigate intra-regime tensions. In the high-risk and non-transparent situation after a coup, incumbents may try to reach out to the broader population to bolster support and secure their position. An incumbent might try to increase their popularity and decrease the risk of a new coup by offering democratic reforms, since coup attempts are staged in part depending on the popularity of incumbents (Wig and Rød, 2016). For example, when efforts by the military to force Mugabe to resign in Zimbabwe in 2017 initially failed, the army encouraged people to take to the streets and demonstrate against Mugabe, permitting protest that would not previously have been accepted. However, once protest has been allowed and people sense that further concessions could be possible, preventing future protest is likely more difficult. Popular mobilization in the post-coup context can also bolster soft-liners who are willing to contemplate democratic reform and increase their influence within the regime. Based on the above discussion, we propose the following hypothesis on coups and democratization:
H1. Coups are more likely to foster democratic change when accompanied by popular mobilization.
The 1974 coup removing the Estado Novo regime in Portugal illustrates the pervasive uncertainty following a coup and how popular mobilization can promote political change and democratization. The 25 April military coup by the Armed Forces Movement called for an end to ongoing wars in overseas colonies and reform of civil-military relations—democracy was not a goal at the outset (Huntington, 1991: 4). However, the coup was followed by large-scale mobilization, celebrating the end of dictatorship, and set the stage for a transition to democracy after 48 years of autocratic rule. The outcome was by no means obvious at the time. Many feared that the coup makers—which included officers aligned with the Communist Party—would introduce a Marxist dictatorship (Graham, 1992; Linz and Stepan, 1996). Although the coup was instrumental in removing the regime, many accounts accord popular mobilization a key role in securing the democratic transition (Accornero, 2013; Fernandes, 2015; Miller, 2021). Palacios Cerezales (2017), for example, highlights how the anti-communist rallies and demonstrations in 1975 empowered moderate leaders and reversed the balance of power within the military, paving the way for a turn to electoral democracy.
Coups in the absence of mobilization
In the absence of mass mobilization, we expect autocracy and autocratic survival strategies to dominate. Without a threat from mass mobilization or a clear need to seek broader external support, both leaders that have come to power through a coup and surviving incumbents are more likely to focus on containing potential coup-makers and minimizing threats from within the elite (Shen-Bayh, 2018). Likely tactics such as purges or efforts to accommodate through selective rents will often be associated with greater autocracy.
Furthermore, instability and chaos following coups can provide leaders an opportunity to accumulate more power, and leaders not constrained by popular mobilization are more likely to take advantage of these. Coup attempts are often motivated by personal ambitions rather than any clear policy or political differences (Decalo, 1990; Miller, 1971; Onwumechili and Carle, 1998). Moreover, they are normally carried out by high ranking officials, with extensive networks and ties to other influential actors (Geddes, 1999; Singh, 2014). Coup-makers thus have both incentives and opportunities to consolidate power by reaching out to powerful actors rather than the masses.11
Although many of these arguments focus on the consequences of coups in non-democracies, coups pose a significant threat to democracies, in particular unconsolidated democracies (Onwumechili and Carle, 1998; Svolik, 2015). Our discussion and the mechanisms highlighted are also relevant to the potential ability of mass mobilization to prevent autocratization following coup attempts. Popular mobilization can play a pivotal role in preventing democratic breakdown and autocratization. It is easier to concentrate power through a coup attempt in the absence of popular mobilization. Moreover, popular mobilization can increase elite incentives to preserve democratic institutions and support peaceful competition for political power.
Successful versus failed coups
Our previous H1 on democratization also implies that coups in the absence of mass mobilization are likely to foster autocratic change. But extant studies demonstrate that coup outcomes also matter, and many studies suggest that specific changes such as increased corruption, repression, and reduced judicial constraints follow successful coups but not failed coups (Bennett et al., 2021; Lachapelle, 2020). We propose that there are greater opportunities to autocratize in the aftermath of a successful than a failed coup. Successful and failed coups present very different challenges to the ruler. A leader in place after a failed coup is likely to be notably weakened, and the coup demonstrates that elites perceive that a “leader is weak enough to be removed by the use of force and/or they are dissatisfied enough with the status quo that an attempt is worthwhile” (Easton and Siverson, 2018: 597). Efforts to purge and punish those responsible after a failed coup can backfire. It may be difficult for leaders to identify who may have sympathized with the coup attempt, and efforts to purge potential rivals can alienate elites and the security apparatus, potentially increasing the risk of a new coup attempt (Easton and Siverson, 2018: 600). Similar problems also arise from other countermeasures. For example, efforts to centralize power can antagonize the security forces. Successful coups are to some extent a display of power, demonstrating the ability to seize control and at least tacit support of the security forces. Centralizing power is less risky for a successful coup maker than an incumbent weakened by a failed coup. Thus, in the absence of mass mobilization, we expect autocratic changes to be more common following successful than failed coups.
H2. In the absence of mass mobilization, successful coups are more likely to foster autocratic change than failed coups.
Timing
We have argued that coups often have distinct short- and long-term consequences. A coup may depose a leader immediately, but it may take some time before changes in institutions are introduced. The instability induced by a coup can have persistent effects for some time after the initial event on incentives to democratize and autocratize. Tusalem (2010) demonstrates a persistent legacy for coup risks, with a higher risk of a new coup noticeable 5 years into the post-coup period.
Most extant studies of coups or mobilization on political change focus on the immediate effects and are limited to changes in the current year or following 2 years (see, for example, Derpanopoulos et al., 2016; Rivera Celestino and Gleditsch, 2013; Thyne and Powell, 2014).12 However, for many outcomes, past events can also have important lingering and cumulative effects, even if the most recent events tend to be most important (Reid et al., 2021). Dahl (2016) and Dahlum et al. (2019) identify enduring effects of protest movements on the likelihood of democratic change. In some cases, popular mobilization may quickly be followed by democratic change, but in other cases, it may take considerable time before the impact of contention is manifested in political change.13 One round of protests may fail to unseat a regime, but can help establish a sustained challenge and trigger subsequent events that result in greater democracy after some time. Research on organized violence stresses how social norms and mobilized networks continue to play a crucial role after active conflict and contribute to the high risk of renewed conflict (Hegre et al., 2017; Reid et al., 2021). Similar legacy effects are likely to apply to popular mobilization as well, in the sense that past protest movements can affect perceived threats and subsequent government responses well beyond the initial active protest period. Protests in South Korea in the 1980s, for example, are often seen as crucial for the subsequent transition to democracy, even though this did not take place until the 1990s (see, for example, Kim, 2000). Consequently, looking only at the immediate consequences of campaigns can understate the full impact of mobilization on political change. A longer time perspective allows us to capture the potential enduring impact of events as well as incomplete or interrupted changes and enduring outcomes, as in Egypt following the fall of Mubarak, with less-than-fully-competitive initial elections and a subsequent military coup. Indeed, Miller (2021) finds that half of all post-coup democratic transitions fail within 10 years.
We expect coups and popular mobilization to have a combined impact on the prospects for political change even when they do not occur exactly at the same time. When coups are preceded by previous popular mobilization, the prior history of mobilization will guide an incumbent’s risk assessment and the choice of tactics. Prior mobilization can influence the incumbent’s expectations about successful repression as well as the potential negative consequences of repressive strategies. By contrast, the likely threat of popular mobilization is lower without a prior history of collective action, even if a coup might generate grievances. However, synergistic effects on democratization from a coup and popular mobilization are less likely the longer the distance between events. We use decay functions to allow for coups and mobilization to have interactive effects beyond when events occur simultaneously and how the length of time between events modifies the conditional impact.
Empirical strategy
We now proceed to examine empirically our hypotheses on the effect of coups and popular mobilization on the likelihood of changes in political institutions. Many studies consider only the effects of coups on transitions to democracy or changes crossing discrete categories. We consider a global sample of all country-years between 1950 and 2019, with institutional changes in all directions, taking into account the timing of the coup and any subsequent changes in political institutions.
Measuring coups, mobilization, and political change
Our main independent variables are coups and popular mobilization. We rely on the Powell and Thyne (2011) dataset, defining coups as “illegal and overt attempts by the military or other elites within the state apparatus to unseat the sitting executive” (p. 252).14 We consider all illegal coup attempts as relevant for our purposes. Coup attempts signal major divisions within the regime and its core supporters, regardless of whether they ultimately succeed. Whereas some data sources only record successful coups, Powell and Thyne (2011) also record failed coup attempts.15
We identify popular mobilization as large-scale non-violent campaigns demanding regime change from the NAVCO 1.3 dataset (Chenoweth and Shay, 2020). NAVCO catalogs non-violent and violent maximalist campaigns mobilizing more than 1,000 participants between 1900 and 2019, demanding regime change, secession, or anti-foreign occupation. As the latter two categories do not focus on political institutional change, we do not include these in our measure of popular mobilization, but add as controls in the analysis. We control for violent campaigns using a dummy for whether a country experienced armed conflict with more than 25 battle deaths in a given year, based on the Uppsala Armed Conflict data (Pettersson and Öberg, 2020).16
Our dependent variable is change on the Polity democracy scale from one year to the other—that is, Polity t − Polity t − 1 (Jaggers and Gurr, 1995). The Polity institutionalized democracy scale ranges from a low of −10 (full autocracy) to a high of 10 (full democracy).17 We are interested in how coups and popular mobilization can promote changes of all sizes toward democracy and autocracy, and do not limit our analysis of transitions that cross some threshold for democracy. By democratization we mean any change toward greater democracy and by autocratization any change toward greater autocracy. The Polity project has a major advantage for relating changes in institutions to events since it provides dates for all changes, while many alternative data sources offer only annual data. Relying on lagged annual data risks missing changes after a coup within the same calendar year, potentially severely underestimating changes after coups. Similarly, coups may be staged in the wake of regime change or institutional reforms, possibly as a response to the change rather than a prior cause. If so, counting co-occurrence within the same year risks misattributing changes to coups. We use the precise dates to identify the temporal order and overcome these potential problems. We compare all cases where coups and Polity changes occur within the same year, and recoding all instances where a coup actually occurs after a change in Polity to the subsequent year.18
Similar ambiguities can also arise with regard to the timing of popular mobilization and Polity changes. In many cases, institutional change occurs during ongoing campaigns, but we could also observe protest emerging as a response to changes in institutions. We apply the same strategy to handle potential problems for campaigns and Polity changes in the same year, using start dates for non-violent campaigns from the NAVCO 2.1 as well as our own coding of campaigns with missing start dates, using newspapers and case-specific literature (Chenoweth and Lewis, 2013). If campaigns emerge after a Polity change, we move the campaign start to the subsequent year to ensure that we do not consider it an apparent prior cause to the change. Seventy campaigns start the same year as a Polity change, but only 14 Polity changes occur prior to campaign start (see Table A-4 in the Supplemental Appendix). Hence, lagging by year would mask a substantial number of cases where Polity changes occur after popular mobilization (56), and our approach ensures against apparent “false positives” where the temporal sequence is reversed.
Capturing effects over time
We use decay functions to account for both immediate and delayed effects of coups and non-violent campaigns. Exponential decay functions are suitable to model persistent effects that weaken over time (Burt, 2000). Our set-up accommodates different types of mobilization and coup sequencing trajectories, including situations where a coup and protests are close in time, but do not take place within the same calendar year. Again, only about 13 percent of all coup attempts see popular mobilization within the same calendar year, but about 40 percent of coup attempts occur within a 4-year window of popular mobilization. We use a decay function defined as: y = 2−t/α, where t is time since the event and α denotes the half-life when the effect is reduced to half of its initial impact.19 We consider a series of half-life specifications for the impacts of coups and mobilization, ranging from 1 to 5 years. We find that a half-time of 2 years appear to fit the data well, and use this for our main specification, but also report results with half-times of 1–5 years in Table A-5 in the Supplemental Appendix. We also consider fixed-window measures, indicating whether there was a coup or non-violent mobilization within the time frame, using windows ranging from 1 to 5 years. Unlike binary windows, where observations are either fully “in” or “’out,” a decay function can capture impact beyond an initial event and reflect a gradually decreasing impact with time.20 The log likelihood values in Table A-5 shows that the models with decay functions outperform the models with windows.
Estimating interaction terms reliably requires a large number of supporting data points, and skeptics may wonder whether our data include sufficient instances of coups and popular mobilization coinciding at the same time or in near proximity to one another. In Table A-1 in the Supplemental Appendix, we show the observed distribution of coups and popular mobilization events. There are 93 instances with a coup and popular mobilization within the current or prior year, which is sufficient to assess conditional effects.
Control variables
Democratization, coups, and popular mobilization are likely to have common correlates that must be considered to isolate the plausible effects. We control for log gross domestic product (GDP) per capita and economic growth, using data from the World Bank (2016). We further include a log of an incumbent’s time in office, based on the Archigos data on political leaders (Goemans et al., 2009), a measure of active armed conflict from the UCDP Armed Conflict Database (Pettersson and Öberg, 2020), and the average Polity level in neighboring countries within 100 km of the outer boundaries (following Hegre et al., 2013). Finally, we include a dummy for the Cold War period, and a dummy identifying countries with two or more coups within the same year. All control variables are lagged by 1 year to avoid including values that may reflect coups or institutional change. In Table A-6, we show the descriptive statistics of the variables included in the models.
Results
We start by comparing changes in the Polity score for cases with and without coups (including the year of the coup as well as the two subsequent years) in Figure 1. The two bars to the left show the share of cases with change (in any direction) for cases with coups (green) and without (dark gray). Change is far more common in observations with coups (25%) than non-coup cases (only 6%). In the second and third bars, we differentiate between change toward more democracy and more autocracy. Democratic and autocratic changes are about equally likely in cases with coups, but for cases without coups, changes toward democracy are about twice as common as changes toward greater autocracy, reflecting the secular trend toward greater democracy over the period. The two rightmost bars for the magnitude of positive and negative changes demonstrate that post-coup changes tend to be larger than changes outside coups, in both directions on the Polity scale. Taken together, this provides initial support for the expectation that coups often trigger political change, and both changes toward democracy and autocracy become more common. We show that the positive impact of coups on any change in Polity score still holds when adding control variables in Table A-7 in the Supplemental Appendix.
We now turn to assessing our propositions under which post-coup changes in specific directions are more or less likely. Table 1 reports a series of country-fixed effects linear regression estimates, where we consider the effect of any type of coup, independent of the final outcome, with t-values in parenthesis. We first consider a sample of all non-democracies, including all countries in the Gleditsch and Ward (1999) list below 7 on the Polity scale. In line with our expectations, we find that a coup without mass mobilization is associated with an increased likelihood of autocratic change, but when mass mobilization is present, democratic change dominates.
| Non-democracies | Non-democracies | Non-democracies | All regimes | All regimes | All regimes | |
|---|---|---|---|---|---|---|
| Coup, decay: α = 2 | −0.200 (−1.49) |
−0.576*** (−4.04) |
−0.487*** (−4.64) |
−0.759*** (−6.61) |
||
| Protest × coup: α = 2 | 1.628*** (4.62) |
0.579* (2.11) |
||||
| Successful coup, decay: α = 2 | −1.005*** (−5.71) |
−1.227*** (−8.50) |
||||
| Protest × successful coup: α = 2 | 1.606*** (3.67) | 0.263 (0.78) | ||||
| Failed coup, decay: α = 2 | −0.0306 (−0.18) |
−0.0353 (−0.25) | ||||
| Protest × failed coup: α = 2 | 1.212* (2.55) |
1.177** (3.13) |
||||
| Protest, decay: α = 2 | 1.080*** (7.52) |
1.089*** (7.62) |
1.131*** (10.29) |
1.102*** (10.10) |
||
| Log GDP per capita, t–1 | 0.180*** (3.60) |
0.0875 (1.75) |
0.0837 (1.68) |
0.0388 (1.20) |
0.0195 (0.61) |
0.0229 (0.72) |
| Economic growth, t–1 | −0.323 (−1.64) |
−0.217 (−1.12) |
−0.210 (−1.08) |
−0.179 (−1.24) |
−0.132 (−0.92) |
−0.111 (−0.78) |
| Log incumbent’s time in office | 0.00796 (1.69) |
0.00809 (1.75) |
0.00690 (1.49) |
0.0129*** (3.51) |
0.0111** (3.06) |
0.00963** (2.66) |
| Ongoing armed conflict | −0.0897 (−0.86) |
−0.114 (−1.11) |
−0.110 (−1.08) |
−0.0384 (−0.49) |
−0.0766 (−0.99) |
−0.0712 (−0.92) |
| Democratic neighborhood | 0.0132 (1.50) |
0.00573 (0.66) |
0.00628 (0.73) |
−0.00864 (−1.35) |
−0.0102 (−1.61) |
−0.0103 (−1.64) |
| Cold War | 0.0118 (0.13) |
0.0899 (1.00) |
0.106 (1.18) |
0.0260 (0.40) |
0.0762 (1.18) |
0.0984 (1.52) |
| Other non-violent campaign, t–1 | 0.0236 (0.07) |
0.153 (0.47) |
0.124 (0.38) |
0.298 (1.11) |
0.348 (1.31) |
0.311 (1.18) |
| More than one coup, t–1 | −1.802*** (−4.74) |
−1.903*** (−5.08) |
−1.832*** (−4.87) |
−1.463*** (−4.49) |
−1.496*** (−4.64) |
−1.404*** (−4.35) |
| Constant | −0.981** (−2.75) |
−0.536 (−1.52) |
−0.499 (−1.42) |
−0.228 (−0.89) |
−0.181 (−0.71) |
−0.209 (−0.83) |
| N | 4,097 | 4,097 | 4,097 | 6,446 | 6,446 | 6,446 |
| Log likelihood | −8,504.8 | −8,432.4 | −8,423.9 | −12,485.8 | −12,399.6 | −12,374.1 |
| BIC | 17,092.8 | 16,964.7 | 16,964.3 | 25,059.3 | 24,904.5 | 24,870.9 |
GDP: gross domestic product; BIC: Bayesian information criterion; OLS: ordinary least squares.
t statistics in parentheses.
*p < 0.05, **p < 0.01, ***p < 0.001.
The first column in Table 1 shows that when considering the effect of coup attempts only, and not mobilization, the estimated coefficient on changes in the Polity score is negative, although far from statistically significant. In the second column, we add popular mobilization and an interactive term for mobilization and coup attempts. We now find a negative and significant coefficient for coup, suggesting that we are more likely to see autocratic changes when coups take place without popular mobilization. We find a positive coefficient for popular mobilization, indicating a positive relationship with changes on the Polity scale, in line with previous research. We also note that the positive coefficient for popular mobilization is much larger than the negative coefficient for coup attempts. When we turn to the interactive or combined effects of coups and popular mobilization, we find a positive coefficient. The net effect now depends on the sum of the relevant coefficients. Since the main coefficient for popular mobilization and the interactive terms with coups are larger than the main coefficient for coups, the net predicted change must be positive when we have coups and popular mobilization, consistent with our claim that changes toward greater democracy are more likely to take place following coups accompanied with popular mobilization. Moreover, the interactive term for popular mobilization and coups are of a magnitude similar to the individual coefficient for popular mobilization, indicating that coups and splits within a regime can boost further the democratizing effects of popular mobilization.
In the third column, we distinguish between successful and failed coup attempts. As can be seen, we now find a larger negative coefficient for successful coups, and a much smaller and not significant negative coefficient for failed coups. This indicates that autocratizing changes in the aftermath of coups in the absence of popular mobilization mainly arise from successful coups. Based on these results, failed coups are by themselves not “helpful” for democratization in the absence of popular mobilization, just “less bad” than successful coups. Again, we find large positive coefficients for mobilization as well as the interactions between mobilization, and failed and successful coups. This suggests that popular mobilization can help curtail the autocratizing effects of coups and improve the prospects for democratization. Our estimates suggest that failed coups have a slightly stronger democratizing effect than successful coups, but we cannot conclude that the difference between the two is statistically significant given the relatively large confidence intervals and limited observed cases.
The implied net substantive effects of coups and popular mobilization depend on several terms as well as the time since events. We plot the marginal predicted impact on changes in the Polity score in Figure 2(a) and (b). For simplicity, we assume that coups and protest occur simultaneously, and examine how the predicted impact is reflected at the event (year 0) and subsequent years, declining with the decay functions specified above. The horizontal axis denotes year after the initial event, while the vertical axes denote either predicted change by year in Figure 2(a) or cumulative predicted change since year 0 in Figure 2(b).21
As can be seen in Figure 2(a), we see a clear predicted decline in the Polity score following a successful coup without protest. The impact is the largest in the year of the events, but remains non-negligible in the following years. For a failed coup absent protest, the negative change is very small at the outset and not significantly different from zero. (For successful coups, the negative predicted values are significantly different from 0 up to the second year.) For the scenarios involving protest, the predicted change in Polity is positive and substantial. We see the largest predicted changes when protest is accompanied by coups, with a higher predicted positive change for failed coups than successful coups, but even successful coups and mobilization have a larger predicted impact than protest not accompanied by a coup. The net predicted impact of a failed coup with protest is almost 80 percent larger than for mobilization not accompanied by a coup.
Since our model set-up is based on country-year data, the predicted values are also annual expected changes. Most countries do not see a series of small changes by each year, and there is a great deal of variation in the timing and magnitude of any changes following coups and popular mobilization. Figure 2(b) provides an alternative way to illustrate the implied cumulative effect by adding the predicted impact by year for the first year for the most likely autocratizing scenario (i.e. a successful coup with no popular mobilization) and the most likely democratizing scenario (i.e. a failed coup accompanied by popular mobilization). This suggests substantively large net predicted effects over a 4-year period, with negative changes of around −2 on the Polity scale for successful coups without mobilization, and positive changes around 6 for a failed coup with popular mobilization. Although it is difficult to account for the specific timing of changes, these results suggest a high likelihood of substantial changes associated with combinations of coups and popular mobilization, in line with our expectations.
In columns 4, 5, and 6, we repeat our analyses on a sample of all country-years. This allows us to consider the impact of coups on political change beyond non-democracies and changes for countries above the ⩾7 threshold on the Polity scale. Previous research has mainly focused on the prospects for democratization after coups in autocracies, but our theory and discussion has also highlighted the ability of popular mobilization to contain the autocratizing effects of coups. The full sample also allows us to also consider coups in democracies. Moreover, we can examine if our results also hold when we include countries ⩾7 but below the maximum Polity score 10 (which may include recent democracies or countries with some less democratic traits). We find generally similar results for our models estimated on the full sample. We find a larger negative coefficient for coup attempts on change in Polity in column 4, potentially reflecting added cases of democratic breakdown after coups. When we add popular mobilization and the interactive term in column 5, we find results similar to the non-democracy sample, although the interactive term is just below the 0.05 significance level for a two-tailed test and only significant in a one-tailed test. When differentiating between coup outcomes in column 6, we find a large and negative coefficient for successful coups. However, the results also suggest that popular mobilization can offset the autocratizing effect. For failed coups, we again find little evidence for a systematic impact on autocratization without popular mobilization, and a net positive impact on change in Polity when failed attempts are accompanied by popular mobilization.
Robustness tests
To examine whether the key findings are sensitive to model specifications, we run a number of robustness tests, and our findings remain robust across these. Skeptics may wonder whether the specific sequence of events matters. We have emphasized the role of expectations, and based on our argument, the implications should be the same if (a) coups occur during protest (or in the recent aftermath of protest) or (b) new protest follow a coup attempt, since in either case incumbents will have greater incentives to try to promise reform. As mentioned previously (within the same calendar year), coups are more likely to take place following recent mobilization than for new mobilization to follow a coup (there are only three cases of popular mobilization emerging after a coup without prior mobilization, see Table A-2 in the supplemental appendix). However, to examine possible differences across sequences, we conduct additional analyses where we remove specific sequences, first taking out all cases where popular mobilization precedes a coup attempt (Table A-8), and second all cases where coup attempts precede the emergence of protest (Table A-9). The results remain substantively similar across these alternative tests.
Table A-5 presents the model for all coups (not separating by outcome) with different specifications of the independent variables, including decay and window functions. The negative effect of coups holds for decay functions up to 4 years and for window specifications up to 2 years. The positive interaction effect between coup and protest holds for all decay functions and windows up to 3 years. In Table A-10, we apply the same set-up distinguishing by coup outcome. The insignificant results of failed coups hold across all specifications, while the positive interaction with protest holds for all decay functions, but only the 1-year window. The negative effect of successful coups holds across all decay functions and for window functions up to 2 years. The positive interaction effect between successful coup and protest holds for all decay functions and windows up to 3 years. Again, the effect of protests holds for all specifications.
Using the V-dem Liberal Democracy Score and the same lagging approach as in the Polity analysis (see Table A-11 in the supplemental Appendix), we find a positive interaction effect in the context of social mobilization for failed coups and successful coups continue to show a negative effect. We also test whether the results are limited to after the Cold War (see Tables A-12 and A-13 in the Supplemental Appendix). We find similar results during and after the Cold War, with an effect of coups conditional on popular mobilization, and the interactive terms when differentiating between outcomes are similar across the samples although not consistently significant in the shorter period with the lower number of cases. There is no evidence that coups generally promote democratization in the Post Cold War period, as some have suggested. The coefficients for coups in general and successful coups are negative and significant. The results for coups independent of outcome are robust to Jackknife resampling, where we re-estimate the model leaving out each individual country from the estimation sample (see Table A-14), although the interaction effects by distinct outcomes are not consistently significant. Finally, we show in Table A-15 that the results hold when dropping all control variables and do not arise simply as an artifact of the particular model specification.
Discussion and conclusion
Previous research on the impact of coups on democratization has reached seemingly contradictory conclusions. Whereas some contend that coups provide a good opportunity for unseating dictators and promote democratic reform, skeptics emphasize how coups are likely to foster authoritarianism and repression. We argue that the consequences of coups are heterogeneous and that the expected effects must be considered relative to other forces that affect the prospects for democratization. We focus on the role of popular mobilization.
We assert that the effect of coups hinges on popular mobilization. Coups are more likely to foster democratization when popular mobilization is present, while autocratic rule is more likely absent mobilization. We have suggested that popular mobilization increases pressure for democratization through a number of mechanisms. Most directly, mobilization increases the ability of civil society actors to make credible demands and sanction governments. Moreover, in the aftermath of a coup, vulnerable leaders are more likely to offer concessions and try to enlist popular support to hold on to power. Besides, competition between rival elites can increase the power of civil society and accordingly the value of democratic institutions as a rational compromise for sharing power.
Our empirical analysis provides support for our arguments. In the absence of popular mobilization, coup attempts increase the likelihood of negative changes on the Polity score, while democratic change is more likely after coups in the context of popular mobilization. We show that the autocratizing effect in the absence of popular mobilization is primarily driven by successful coups, similar to research on increased repression and corruption, and failed coups do not have consistent effects on autocratization. In the context of popular mobilization, democratization is more likely following coups than mobilization alone, and the effects after failed coups appear to be more robust than the ones for successful coups.
We have only considered cases with ongoing or recent popular mobilization. However, our postulated mechanisms arguably pertain to capacity for popular mobilization or plausible threat potential. Some coups may not be preceded by prior mobilization, and plausible motivation and opportunities for mobilization can play a similar role even if collective action may be difficult. Although the likelihood of mobilization cannot be fully known ex ante, some cases clearly have much higher prospects for popular mobilization and protest, and this plausible potential threat should make elite instability more likely to result in democratization.
Our analyses demonstrate the need to consider pressure from above and pressure from below as complimentary and possibly related rather than separate or competing forces. Traditional military coups have become less common, at least outside Africa, but we often observe alternative types of efforts by elites to oust unpopular leaders, including “legal” means such as impeachment proceedings, declarations of no-confidence, or even threats of international intervention. In Gambia, for example, President Jammeh refused to hand over power after unexpectedly losing elections in 2016, citing electoral fraud.22 Jammeh’s refusal to hand over power triggered popular mobilization, and after an intervention by Economic Community of West African States (ECOWAS) forces in January 2017, Jammeh agreed to go into exile. This example underscores the wider applicability of a combined pressure from below and above beyond coups, and how crises and unraveling elite coalitions provide opportunities for democratization.
Acknowledgments
We are grateful for helpful comments from Charles Butcher, Christoph Dworschak, Scott Gates, Belén González, Carl Henrik Knutsen, Gabriel Leon, Håvard Mokleiv Nygård, Lynn Nygård, Mauricio Rivera, Laura Saavedra-Lux, and Tore Wig.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a grant from the Research Council of Norway (275955/F10).
ORCID iDs
Marianne Dahl https://orcid.org/0000-0002-0560-3850
Kristian Skrede Gleditsch https://orcid.org/0000-0003-4149-3211
Footnotes
1. Following Powell and Thyne (2011), we use coups to denote both successful and failed coup attempts.
2. On the Polity scale (ranging from −10 for a full autocracy to 10 for a full democracy), Equatorial Guinea changes from 2 before the coup attempt to −7 after the coup.
3. This also holds in the post-Cold War period.
4. These effects pertain primarily to successful coups and not failed coup attempts (Bennett et al., 2021; Lachapelle, 2020).
5. Svolik (2012) estimates that coups accounted for 68 percent of all non-constitutional exits from office among non-democratic rulers in the period 1946–2008. Non-constitutional removals alone understate the threats to leaders. Many such as Milosević and Honecker have resigned when facing popular mobilization, and rulers can introduce reform in order to avoid losing power (Gleditsch et al., 2022; Gleditsch and Ward, 2006).
6. Chenoweth and Stephan (2011) estimate that non-violent campaigns are more likely to succeed than violent campaigns. In the NAVCO 1.1 data, 59 percent of non-violent campaigns for regime changes have succeeded, compared to only 27 percent of violent campaigns. Gleditsch et al. (2022) find that leaders lose power about twice as often after non-violent campaigns than after violent campaigns.
7. Collier (1999) argues that popular mobilization played an important role in generating elite splits in canonical cases of elite-pacted transitions such as Spain and Peru (see also Wood, 2001: 866).
8. Chenoweth and Stephan (2008, 2011) emphasize how non-violent campaigns have been more likely to succeed when they generate large-scale defection within the security apparatus (see also Nepstad, 2011, 2013). However, they compare only active campaigns and do not consider coups specifically.
9. Miller (2021: 210–215) examines how violent shocks, including coups, mediate the effects of popular mobilization, but does not examine how popular mobilization can modify the effect of coups or violent shocks. We show in the Supplemental Appendix that we see more cases where protest precedes coups than protest following coup attempts, see Tables A-1 (events in different years) and A-2 (events in same year). Many events are close in time and the precise order can be difficult to determine, as expectations may predate recorded dates, but it is clearly not the case coups always precede mobilization.
10. We return later to how elite division and popular mobilization can have reinforcing effects beyond the immediate coup attempt period.
11. Geddes (1999: 122) notes how “it is common for officers who seize power in military coups, for example, to attempt to concentrate power in their own individual hands, to hold plebiscitary elections to legitimate their personal rule, and to create parties to organize their supporters”.
12. Two exceptions are Marinov and Goemans (2014) and Chenoweth and Stephan (2011), who look at change over 5 and 10 years, respectively.
13. Many transition processes are completed over a relatively long time span. For example, although the Solidarity movement helped bring down communist rule in Poland in 1989, the first democratic presidential election did not take place until December 1990, and many data sources do not code a transition to democracy until 1991.
14. Some use the term coup in a very broad sense, encompassing any situation where an incumbent is forced to leave power. For example, the senate’s ousting of President Rousseff in Brazil in 2016 is sometimes referred to as a coup, see https://www.bbc.co.uk/news/world-latin-america-37271465.
15. Most events in their data involve the security apparatus, but palace coups are also included, and some do not have direct military involvement.
16. We use data on armed conflict rather than the more restricted set of violent campaigns NAVCO, limited to events with more than 1,000 deaths. The number of deaths is typically much lower than the number of active participants, and the armed conflict data provide a more comprehensive and comparable list of violent events (Dahl et al., 2021).
17. The Polity IV measure is based on a series of subcomponents, measuring the competitiveness of executive recruitment, openness of executive recruitment, constraints on the chief executive, regulation of participation, and competitiveness of participation.
18. This affects 13 observations, listed in Table A-3 in the Supplementary Appendix. For country-years with more than one Polity change or coup, we include the last event within the calendar year. We control for whether more than one coup occurred during a year. Using a smaller time unit than the year such as months, weeks, or days would risk inflating the apparent N without adding additional information.
19. For example, the half-time of the painkiller ibuprofen is about 2 hours, reflecting a very rapid decay, while uranium-235 has an extremely slow half-time of over 700 million years. Scholars have proposed alternative and more comprehensive approaches to reflect the impact of past events and cumulative effects. Crescenzi and Enterline (2001) propose a model to reflect history in dyadic interactions, adding past conflictual and cooperative events with a decaying influence over time, and Reid et al. (2021) generalize this to a conflict environment index for the risk of civil wars, including the impact of nearby events. Our data are less suited for the former approach since we do not have detailed event data or detailed information on the specific actors. We focus only on the relationships within individual countries here, although we acknowledge that both mobilization and coup attempts may be influenced by nearby events.
20. For an example, with a decay function with a half-life of 2 years, an observation with protests at t and a coup at t + 1 will have a value of 1 at t + 1 for the coup indicator and 0.701 for the protest indicator.
21. The 95 percent confidence intervals in Figure 2(a) are based on draws from the coefficient and variance-covariance estimates (King et al., 2000).
22. “Gambian president Yahya Jammeh rejects election result,” The Guardian, 9 December 2016.
Data availability statement
Data and code to replicate the analyses in this manuscript is available on the Harvard Dataverse. Email: [email protected] & [email protected].
References
Accornero G (2013) Contentious politics and student dissent in the twilight of the Portuguese dictatorship: analysis of a protest cycle. Democratization 20(6): 1036–1055.
Acemoglu D, Robinson JA (2001) A theory of political transitions. American Economic Review 91(4): 938–963.
Bayer M, Bethke FS, Lambach D (2016) The democratic dividend of nonviolent resistance. Journal of Peace Research 53(6): 758–771.
Bennett DL, Bjørnskov C, Gohmann SF (2021) Coups, regime transitions, and institutional consequences. Journal of Comparative Economics 49(2): 627–643.
Bermeo N (2016) On democratic backsliding. Journal of Democracy 27(1): 5–19.
Bove V, Rivera M (2015) Elite co-optation, repression, and coups in autocracies. International Interactions 41(3): 453–479.
Brancati D (2014) Pocketbook protests: explaining the emergence of pro-democracy protests worldwide. Comparative Political Studies 47(11): 1503–1530.
Bueno de, Mesquita B, Siverson RM, Woller G (1992) War and the fate of regimes: a comparative analysis. American Political Science Review 86(3): 638–646.
Burt RS (2000) Decay functions. Social Networks 22(1): 1–28.
Casper BA, Tyson SA (2014) Popular protest and elite coordination in a Coup d’état. Journal of Politics 76(2): 548–564.
Chambers P, Ufen A (2020) Causes, effects, and forms of factionalism in Southeast Asia. Journal of Current Southeast Asian Affairs 39(1): 3–16.
Chenoweth E, Lewis OA (2013) Unpacking nonviolent campaigns. Introducing the NAVCO 2.0 dataset. Journal of Peace Research 50(3): 415–423.
Chenoweth E, Shay CW (2020) NAVCO 1.3 codebook. Available at: https://doi.org/10.7910/DVN/ON9XND/PTMCCV
Chenoweth E, Stephan MJ (2011) Why Civil Resistance Works: The Strategic Logic of Nonviolent Conflict. New York: Colombia University Press.
Collier P (2009) A coup for democracy. The Guardian. Available at: https://www.theguardian.com/commentisfree/2009/jan/15/africa-regimes-ghana-guinea
Collier RB (1999) Paths Towards Democracy: The Working Class and Elites in Western Europe and South America. Cambridge: Cambridge University Press.
Crescenzi MJC, Enterline AJ (2001) Time remembered: a dynamic model of interstate interaction. International Studies Quarterly 45(3): 409–431.
Dahl M (2016) Pressure from below: the effect of civil society mobilization on democratic change. PhD Thesis, Norwegian University of Science and Technology, Trondheim.
Dahl M, Gates S, Gleditsch KS, et al. (2021) Accounting for numbers: group characteristics and the choice of violent and nonviolent tactics. Economics of Peace and Security Journal 16(1): 5–25.
Dahlum S (2022) Joining Forces: Social Coalitions and Democratic Revolutions. Typescript, University of Oslo.
Dahlum S, Knutsen CH, Wig T (2019) Who revolts? Empirically revisiting the social origins of democracy. Journal of Politics 81(4): 1494–1499.
Davenport C, RezaeeDaryakenari B, Wood RM (2022) Tenure through Tyranny? Repression, dissent, and leader removal in Africa and Latin America, 1990–2006. Journal of Global Security Studies 7(1): 1–17.
Debs A (2016) Living by the sword and dying by the sword? Leadership transitions in and out of dictatorships. International Studies Quarterly 60(1): 73–84.
Decalo S (1990) Coups and Army Rule in Africa: Motivations and Constraints. New Haven, CT: Yale University Press.
Derpanopoulos G, Frantz E, Geddes B, et al. (2016) Are coups good for democracy? Research & Politics 3(1): 2053168016630837.
Derpanopoulos G, Frantz E, Geddes B, et al. (2017) Are coups good for democracy? A response to Miller (2016). Research & Politics 4(2): 2053168017707355.
Easton MR, Siverson RM (2018) Leader survival and purges after a failed coup d’état. Journal of Peace Research 55(5): 596–608.
Fernandes T (2015) Rethinking pathways to democracy: civil society in Portugal and Spain, 1960s–2000s. Democratization 22(6): 1074–1104.
Frantz E, Ezrow NM (2011) The Politics of Dictatorship: Institutions and Outcomes in Authoritarian Regimes. Boulder, CO: Lynne Rienner.
Geddes B (1999) What do we know about democratization after twenty years? Annual Review of Political Science 2: 115–144.
Gleditsch KS, Ward MD (1999) Interstate system membership: a revised list of the independent states since 1816. International Interactions 25(4): 393–413.
Gleditsch KS, Ward MD (2006) Diffusion and the international context of democratization. International Organization 60(4): 911–933.
Gleditsch KS, Olar RG, Radean M (2022) Going, going, gone? Varieties of dissent and leader exit. Journal of Peace Research. Epub ahead of print 18 October.
Goemans H, Gleditsch KS, Chiozza G (2009) Introducing Archigos: a dataset of political leader. Journal of Peace Research 46(2): 269–283.
Graham LS (1992) Redefining the Portuguese transition to democracy. In: Higley J, Gunther R (eds) Elites and Democratic Consolidation in Latin America and Southern Europe. Cambridge: Cambridge University Press, pp. 282–299.
Hegre H, Karlsen J, Nygård HM, et al. (2013) Predicting armed conflict, 2010–2050. International Studies Quarterly 57(2): 250–270.
Hegre H, Nygård HM, Ræder RF (2017) Evaluating the scope and intensity of the conflict trap: a dynamic simulation approach. Journal of Peace Research 54(2): 243–261.
Hess D, Martin B (2006) Repression, backfire, and the theory of transformative events. Mobilization 11(2): 249–267.
Hollyer JR, Rosendorff BP, Vreeland JR (2015) Transparency, protest, and autocratic instability. American Political Science Review 109(4): 764–784.
Huntington SP (1991) The Third Wave: Democratization in the Late Twentieth Century. Norman, OK: University of Oklahoma Press.
Jackson RH, Rosberg CG (1982) Personal Rule in Black Africa: Prince, Autocrat, Prophet, Tyrant. Berkeley, CA: University of California Press.
Jaggers K, Gurr TR (1995) Transitions to democracy: tracking democracy’s third wave with the polity III data. Journal of Peace Research 32(4): 469–482.
Johnson J, Thyne CL (2018) Squeaky wheels and troop loyalty: how domestic protests influence Coups d’état, 1951–2005. Journal of Conflict Resolution 62(3): 597–625.
Karatnycky A, Ackerman P (2005) How freedom is won: from civic resistance to durable democracy. International Journal of Not for Profit Law 7: 47.
Kieh GK, Agbese PO (2004) The Military and Politics in Africa: From Engagement to Democratic and Constitutional Control. Aldershot: Ashgate.
Kim NK (2017) Anti-regime uprisings and the emergence of electoral authoritarian-ism. Political Research Quarterly 70(1): 111–126.
Kim NK, Kroeger AM (2019) Conquering and coercing: nonviolent anti-regime protests and the pathways to democracy. Journal of Peace Research 56(5): 650–666.
Kim S (2000) The Politics of Democratization in Korea: The Role of Civil Society. Pittsburgh, PA: University of Pittsburgh Press.
King G, Tomz M, Wittenberg J (2000) Making the most of statistical analyses: improving interpretation and presentation. American Journal of Political Science 44(2): 347–361.
Lachapelle J (2020) No easy way out: the effect of military coups on state repression. Journal of Politics 82(4): 1354–1372.
Linz J, Stepan A (1996) Problems of Democratic Transition and Consolidation. Baltimore, MD: Johns Hopkins University Press.
Londregan JB, Poole KT (1990) Poverty, the coup trap, and the seizure of executive power. World Politics 42(2): 151–183.
Marinov N, Goemans H (2014) Coups and democracy. British Journal of Political Science 44(4): 799–825.
Miller MK (2012) Economic development, violent leader removal, and democratization. American Journal of Political Science 56(4): 1002–1020.
Miller MK (2014) Elections and elite violence on the road to democratization, 1800-2010. Paper Presented at Annual Meeting of the American Political Science Association. Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2451632
Miller MK (2016) Reanalysis: are coups good for democracy? Research & Politics 3(4): 2053168016681908.
Miller MK (2021) Shock to the System. Princeton, NJ: Princeton University Press.
Miller NN (1971) Military Coup in Uganda. American Universities Field Staff 10: 1–18.
Nepstad SE (2011) Nonviolent Revolutions: Civil Resistance in the Late Twentieth Century. New York: Oxford University Press.
Nepstad SE (2013) Mutiny and nonviolence in the Arab Spring: exploring military defections and loyalty in Egypt, Bahrain, and Syria. Journal of Peace Research 50(3): 337–349.
Onwumechili C, Carle D (1998) African Democratization and Military Coups. Westport, CT: Greenwood.
Palacios Cerezales D (2017) Civil resistance and democracy in the Portuguese revolution. Journal of Contemporary History 52(3): 688–709.
Pettersson T, Öberg M (2020) Organized violence, 1989–2019. Journal of Peace Research 57(4): 597–613.
Powell J (2012) Determinants of the attempting and outcome of Coups d’état. Journal of Conflict Resolution 56(6): 1017–1040.
Powell JM, Thyne CL (2011) Global instances of coups from 1950 to 2010: a new dataset. Journal of Peace Research 48(2): 249–259.
Rasler K (1996) Concessions, repression, and political protest in the Iranian revolution. American Sociological Review 61(1): 132–152.
Reid L, Myrick R, Kadera KM, et al. (2021) Conflict environments and civil war onset. Journal of Global Security Studies 6(2): ogz064.
Rivera Celestino M, Gleditsch KS (2013) Fresh carnations or all thorn, no rose? Non-violent campaigns and transitions in autocracies. Journal of Peace Research 50(3): 385–400.
Shen-Bayh F (2018) Strategies of repression: judicial and extrajudicial methods of autocratic survival. World Politics 70(3): 321–357.
Singh N (2014) Seizing Power: The Strategic Logic of Military Coups. Baltimore, MD: Johns Hopkins University Press.
Snyder R (1992) Explaining transitions from neopatrimonial dictatorships. Comparative Politics 24(4): 379–399.
Stephan MJ, Chenoweth E (2008) Why civil resistance works. The strategic logic of nonviolent conflict. International Security 33(1): 7–44.
Svolik MW (2012) The Politics of Authoritarian Rule. Cambridge: Cambridge University Press.
Svolik MW (2015) Which democracies will last? Coups, incumbent takeovers, and the dynamic of democratic consolidation. British Journal of Political Science 45(4): 715–738.
Tansey O (2016) The limits of the “democratic coup” thesis: international politics and post-coup authoritarianism. Journal of Global Security Studies 1(3): 220–234.
Teorell J (2010) Determinants of Democracy: Explaining Regime Change in the World, 1972–2006. Cambridge: Cambridge University Press.
Thyne CL, Powell JM (2014) Coup d’Etat or Coup d’Autocracy? How coups impact democratization, 1950–2008. Foreign Policy Analysis 12(2): 192–213.
Tusalem RF (2010) Determinants of Coup D’état events 1970-90: the role of property rights protection. International Political Science Review 31(3): 346–365.
Ulfelder J (2005) Contentious collective action and the breakdown of authoritarian regimes. International Political Science Review 26(3): 311–334.
Waldner D, Lust E (2018) Unwelcome change: coming to terms with democratic backsliding. Annual Review of Political Science 21: 93–13.
Wig T, Rød EG (2016) Cues to coup plotters: elections as coup triggers in dictatorships. Journal of Conflict Resolution 60(5): 787–812.
World Bank (2016) World Development Indicators 2016. Washington, DC: World Bank.
Wood EJ (2001) An insurgent path to democracy popular mobilization, economic interests, and regime transition in South Africa and El Salvador. Comparative Political Studies 34(8): 862–888.
Yukawa T, Hidaka K, Kushima K, et al. (2022) Coup d’Etat and a democratic signal: the connection between protests and coups after the cold war. Journal of Peace Research 59: 828–842.
Zengin H (2021) Why do some coups lead to democratisation? Democratization 29(5): 1–16.
Biographies
Marianne Dahl is Senior Researcher, Peace Research Institute Oslo (PRIO).
Kristian Skrede Gleditsch is Regius Professor of Political Science, University of Essex and Research Associate, Peace Research Institute Oslo (PRIO).
Supplementary Material
Please find the following supplemental material visualised and available to download via Figshare in the display box below. Where there are more than one item, you can scroll through each tab to see each separate item.
Please note all supplemental material carries the same license as the article it is here associated with
