Salah Ben Hammou
New Lines Magazine recently published a rich first-person narrative by Dr. Faysal Abbas Mohammed, a retired Syrian-Canadian professor, on the experience of Alawite dissidents against the Assad regime in Syria. While narratives of Syrian politics emphasize the Alawite sect’s support for the Assad family, Mohammed provides a more nuanced perspective. Specifically, he draws on his experience in Syria as an Alawite dissident during the reign of President Hafez al-Assad to recount different forms of Alawite opposition against the regime. This includes the author’s own efforts to organize anti-Assad protests and demonstrations after the dictator’s rise in 1970. The author also mentions one striking detail in his narrative - in 1972, Alawite officers attempted to oust Assad in a coup d'etat. These officers were supporters of Assad’s deposed rival, Salah Jadid, but the putsch was crushed and purges of the armed forces followed.
Out of several available datasets documenting coups d’etat around the world, not one collection includes the failed 1972 coup.
Though one might argue that the event’s exclusion is due to its incompatibility with coding criteria, its absence from “candidate lists” suggests that scholars have simply not encountered the event before. This possibility would be unsurprising given that coup researchers near-exclusively rely on Western and popular international media sources during the data collection process. As a result, researchers are neglecting to systematically consult a list of regional and local sources, possibly omitting several coup events. Below, I emphasize this issue by honing in on the treatment of Syria in coup data. I then speak broadly about the perils of overlooking specialized sources for data collection. Put simply, I urge scholars to take more adequate steps towards “decolonizing” coup data through the integration of local sources and regional scholarship.
SYRIA IN COUP DATA
Once known as “the world’s most unstable country”, Syria is well-known to have boasted one of the highest rates of coups d’etat in the world during its post-colonial era. One US ambassador called the country’s experience in the 1960’s “the stability of instability.” The country experienced its first coup in March 1949, led by Colonel Husni al-Zaim, who was later deposed and executed a mere 137 days later by his former comrade Colonel Sami al-Hinnawi. Hinnawi himself was later ejected from power in December 1949 by Colonel Adib al-Shishakli, who would consolidate power again in 1951 and rule Syria until a coup by Baathist-affiliated Druze officers unseated him in 1954. This instability accelerated in the aftermath of Syria’s failed merger with Gammal Abdel Nasser’s Egypt, which ended by a secessionist coup in 1961 and saw several subsequent coup events throughout the decade.
To their credit, large-N coup datasets such as Powell & Thyne’s Global Instances of Coups (P&T hereafter) and the recently released Colpus data project largely showcase much of Syria’s early political turmoil, particularly concerning successful and failed coup attempts. For instance, the datasets do a good job of capturing the onslaught of successful and failed coups throughout the 1960’s. Projects including coup plots and rumors such as the Cline Center (which uses machine learning to data scrape potential events off the web) and the Center for Systemic Peace also appear to capture the general atmosphere of instability.
However, to scholars familiar with the country’s history of civil-military relations and political development, there are glaring omissions in Syria’s data coverage beyond the failed 1972 coup. The most obvious omission once again comes during the regime of Hafez al-Assad - when his brother, Rifaat, attempted to seize power in March 1984. The coup attempt occurred amid Hafez’s failing health and represented perhaps the most serious threat to the autocrat’s rule. Despite the event’s importance, only one dataset released in August 2021 - the Colpus dataset by John Chin, David Carter, and Joseph Wright - includes it. Their codebook indicates that the authors encountered the event while consulting secondary literature to write case narratives about pre-identified cases. Similarly, until recently (by Powell & Thyne), no dataset includes the failed August 1968 coup by officers loyal to Major General Ahmed Suadani. Following P&T’s inclusion of the event, the Colpus project dismissed the event as highly ambiguous given that reports on the event allegedly originated from the right-wing Lebanese newspaper al-Nahar a week later. However, al-Hayat, another Lebanese newspaper, reported the event prior to al-Nahar as did the French newspaper Le Monde. Further, reports emphasized clashes between army units in Aleppo. The Syrian government, represented by Colonel Abdal Karim al-Jundi, would go on to further implicate Suadani in an interview with the Egyptian Press, subsequently jailing the officer for a quarter-century. Further, scholars of Syrian politics such as Hinnebusch (2015), Moubayed (2006), and Khatib (2010) reference the event.
These omissions would be less likely if scholars more seriously engaged with localized sources. It should be noted that a wealth of local resources are available for data collection purposes. Hicham Bou Nassif’s 2020 paper on pre-Ba’ath Syrian coups is an excellent example. Bou Nassif consults several memoirs penned by Syrian civilians and military officers in the 20th century and provides several novel cases unavailable in any large-N project. This includes a thwarted coup plot in 1950 & 1954, another failed attempt in 1955, two plots in 1956, and two failures in 1957. While these events may not meet the conceptual criteria of all datasets - such as the coup plots - these events should still certainly show up in candidate lists. Coup datasets generally present the 1950’s as a decade of fewer coups in Syria but Bou Nassif’s consultation of local sources suggests otherwise.
THE PERILS OF “COLONIZED” COUP DATA
Scholarship focused on other forms of contentious politics - such as protests, riots, and terrorism- continues to recognize reporting bias in their data. The same cannot be said of coup research. That this has failed to garner any serious attention is striking given that nearly every mainstream coup dataset mostly relies on a) Western and international media sources and b) previous coup datasets similarly neglecting local sources and scholarship, as shown in the table below. More specifically, scholars appear to rely on newspapers such as the New York Times and the Washington Post, sources found in LexisNexis and ProQuest as well as encyclopedias such as Keesing’s Records of World Events and Encyclopedia Britannica. In addition, candidate lists also appear to be compiled using other pre-existing large-N datasets such as Singh (2014) and the P&T project. In short, coup datasets have become somewhat of an echo chamber as a result.
Why should we care about reporting bias in coup data, exactly?
First, a reliance on Western sources compromises scholars’ accuracy in data collection and subsequent empirical tests. Coup reporting by media outlets is not a random phenomenon and can likely be skewed towards specific types of events. In particular, there may be a bias to capture successful coups or particularly violent and bloody failed coups given the high-profile nature of the events. More mundane failed coups and coup plots are less likely to receive coverage as a result. In an ongoing project with Jonathan Powell and Bailey Sellers, we utilized regional and local sources combined with academic scholarship on Syria to produce a collection of coup data (successful, failed, and plots) between 1949-1968. Our research yielded 29 different events compared to the efforts of CSP (14) and CCD (17). For more on this work, check out on our ongoing coup narratives on Syria - drawn from a collection of country-specific scholarship and local sources - here.
Although including Syria’s omitted events does not dramatically change its overall “coup-proneness”, the timing of the omissions can have broader implications. As mentioned, the 1950’s - a period which is thought of as the country’s “democratic” era - appears less promising when including several of the omitted plots and failures found by Bou Nassif. Similar misperceptions could be made of Hafez al-Assad’s 30-year reign, which is depicted as generally coup-free both at his regime’s initiation and throughout his tenure. This would suggest that a) Assad did not face instability from Jadid supporters after the seizure of power and b) faced few challenges during his tenure. However, the 1972 coup signals that Assad faced a backlash to his rule early on just as the 1984 coup by his brother Rifaat shows that Hafez faced challenges from within his inner circle later in his tenure. For scholars interested in questions of coups and autocratic stability, using data which omits these events could bias any meaningful statistical test.
Second, and on a normative level, disproportionately relying on Western sources for data collection de-centers the experiences and histories of the countries in question. This perpetuates exclusionary practices in scholarship and places primacy on the objectives, concerns, and biases of non-local, Western actors. This is particularly problematic considering that local voices have historically been excluded and marginalized in the academy. Seriously reckoning with inclusion in the academy extends into the best practices we promote during data collection. Failing to do so results in “colonized” data. It should go without saying that local oral histories and narratives, memoirs, archives, and scholarship should be seriously integrated into any meaningful data collection project.
Ultimately, one might argue that accessibility and limited resources shape the decision to rely on non-local sources. This is a valid counterargument. However, scholars can take steps both in the immediate and long-term future to alleviate concerns of Western source bias and practical limitations. In the immediate short-term, scholars should integrate academic research reliant on local sources as valuable repositories for data and information. The Colpus project, for instance, mentions that they identified 74 new coup events in secondary literature (pg. 5). This is laudable but also suggests that these sources should be front and center to data collection rather than supplementary to general media sources. As a more long-term goal, scholars should seek to make coup data collection more collaborative, promoting interdisciplinary research by working with historians, sociologists, and anthropologists for data collection. These efforts should heavily emphasize trans-regional cooperation for research by prioritizing collaboration with scholars and researchers based in universities abroad. This helps to promote inclusion with historically-marginalized voices in Western academia as well as bolster the academic rigor of coup research.
Jonathan Powell & Mwita Chacha
We’ll start by unequivocally stating coups are problematic events that should not be fomented and should be treated with extreme skepticism. To the degree “good” coups occur, they often result by accident. Though legitimately well-meaning coupists may (occasionally) exist, they face an array of challenges and rarely have the ability to successfully navigate post-coup politics. Even if coups are occasionally well-intentioned, they can still worsen social, political, and economic crises rather than improve them. We can, however, try to learn what might have gone right in the cases with better outcomes, and what went wrong in the bad.
In reference to “good” outcomes, Nic Cheeseman’s recent contribution to The Africa Report points to the rarity of coups leading to democracy. He builds on his earlier essay that identified the negative consequences of coups for civilian rule and constitutional order, this time highlighting the failure of even the “best case scenarios” to produce democratic outcomes. Ultimately, he concludes that in a continent beset by over 100 coups, military interventions in politics have rarely had a “positive and sustained democratic impact.”
Cheeseman rightfully questions the degree to which coups can act as “springboards for democracy.” We build on his comments by pointing to two issues we feel are critical to understanding the issues. First, the high standard adopted in the essay obfuscates many lessons that can be learned from post-coup politics, both in terms of democracy and broader political and social outcomes. Though we fundamentally reject the idea of a “good coup,” using “positive and sustained democratic impact” is an exceptionally high bar that is very difficult to cross. Second, it is impossible to evaluate a coup’s impact without gauging the context in which it occurs and considering the broader prospects for democracy. Needless to say, coups occur in contexts that are hardly conducive to democracy. Addressing these issues, we argue, can help us look beyond the coup itself and look for lessons from history.
A HIGH BENCHMARK FOR SUCCESSFUL DEMOCRATIZATION
Though agreeing with Cheeseman’s general point, focusing solely on outcomes with a “positive and sustained democratic impact” is not an investigation of “good” coups, but instead approaches a veritable unicorn in the study of democratization. Applying this standard to other political phenomena will inevitably lead us to conclude that virtually every indicator fails the democracy test. This is because the likelihood of democratization is exceedingly low to begin with, sustaining democratization over the long term is even lower, and the context in which coups typically occur are especially poor environments for democratic transitions.
For example, Thyne and Powell’s commonly cited study on coups and democratization concluded that an average authoritarian regime has a 1-2% chance of democratization in a given year–outside the context of coups. Joe Wright’s influential study on foreign aid and democratization pointed to similar rates of transition, again, outside the context of coups. And one must keep in mind that these studies are referring to an initial transition, not a “sustained” one, whose probability will be even lower. A recent review conducted by Scott Mainwaring and Fernando Bizzarro, for example, indicated that more than half of third wave (those born between 1974-2012) democracies reverted to authoritarianism, while fewer than one-fifth of these saw major improvements over the long-term. Only Namibia, Cabo Verde, and São Tomé and Príncipe, accounting for no more than 0.3% of the continent’s population, represented Africa in the 23 cases of major advances.
Cheeseman speaks of “frighteningly bad odds” of a “positive and sustained democratic impact” following coups, but this is precisely the nature of authoritarianism more generally. Moving away from authoritarianism is itself a very low-probability affair, and staying away from authoritarianism is an even lower one. In spite of this, most readers will likely be surprised at how common democratic transitions actually are. Jonathan Powell and Mwita Chacha explored post-coup trajectories in an earlier post at the Monkey Cage. They used monthly regime type data from the REIGN project started by Curtis Bell. It is important to clarify that decisions to code a country as a democracy are independent of any consideration of coups and incorporates a range of dynamics beyond the return to civilian rule and holding of an election. Using Powell and Thyne’s coup data, Powell and Chacha found that the post-Cold War era has seen nearly half of coup cases (against authoritarian governments) followed by a democratic transition within three years.
This hardly qualifies as sustained, however, and it is the case that many of these transitions are quickly unraveled. Rather than scoff at a presumed frailty of such regimes, those researching post-coup transitions are instead trying to determine which factors can help them last. In doing so, it is important to understand the environment we are investigating.
THE UNDEMOCRATIC CONTEXT OF COUPS
The low prospects for democratic transitions and survival appear even more dire when considering the environment in which coups occur. Coups, of course, do not occur randomly and researchers have pointed to a number of (unsurprising) factors that are consistently associated with the act, such as poverty, low rates of economic growth, having small and underfunded militaries, weaker ties to the global economy, ongoing domestic political instability, etc. These dynamics are inverse correlates of democratic transitions and democratic failure. Not only is this the case, but in the African context these dynamics seem to be worsening when it comes to coups.
Above we illustrate the mean GDP per Capita (constant 2015 US$) in Sub-Saharan African over time. We illustrate GDP per capita for all states in the region (green) and specific country-years that experienced a coup. It is not exactly a shock to see that wealthier states are more immune to coups. What is perhaps surprising is that the wealth gap between the typical country and the coup-afflicted one has become increasingly lopsided over time. GDP per capita in the 2010s reached 3x that of states that had coups. Further, recent coups have targeted states that are on average poorer than those targeted during the Cold War. While the continent has experienced sustained economic growth in recent decades, coups continue to target the poorest states. Paradoxically, as the world has witnessed a precipitous decline in coups, we have seen an improvement in many of the conditions that could provide a boost toward post-coup transitions, including national wealth.
Cheeseman’s essay refers to “best case scenarios” in the sense that there are a few commonly referenced “good” outcomes. However, there are no “best case scenarios” in which to investigate post-coup democratization prospects since coups are clustered in a subset of countries that are otherwise least likely to see sustained democratic growth. This can help us contextualize Mali’s 1991 coup, which targeted a long-entrenched dictatorship that had recently killed hundreds of protesters who had dared to ask for change, and had mismanaged the economy to the tune of a 2.5% GDP decline the previous year. The case was hardly suggestive of a positive outcome, and combined with a plethora of other challenges, the obvious conclusion was that any experiment with democracy would quickly fail. Mali instead spent the next 20 years as an often-referenced success story. A new military intervention two-plus decades later should not distract us from Mali’s 20 years of democratic rule, especially given the exceptionally challenging conditions from which Mali’s democracy was born.
These challenges extend beyond wealth, of course. Powell and Chacha raised the issue of dominant party legacies after Robert Mugabe’s ouster in Zimbabwe. While the event prompted considerable excitement and occasional sympathies that a transition had effectively already occurred with Mugabe's exit, it quickly became clear that the new regime had little interest in opening up the political system. Powell and Chacha's assessment, illustrated below, identified but a single case of dominant party systems democratizing within three years of a coup.
DEMOCRACY AS ONE OF SEVERAL “POSITIVE” OUTCOMES
Democracy is also but a subset of outcomes that can be described as “positive.” In many cases, particularly during the Cold War, there is little utility for using democratization as a metric as it was on few actors’ radars, was a low priority among both domestic and international actors, and coups almost invariably occurred in contexts in which more immediate challenges prevailed.
For example, Thomas Sankara’s coup-born government does not qualify as having had a “positive and sustained democratic impact.” In fact, it can’t claim a short-term democratic impact either, nor any sustained long-term impact given his brief tenure. He is, however, celebrated for having commenced a range of positive programs in Burkina Faso. Not only are these discounted when fixated solely on the issue of democratization, Sankara likely would not have been able to pursue many of his critical reforms within the context of a democracy, as many were quite unpopular with more conservative elements of society.
In other cases, democracy might be an accidental consequence of coups attempting to promote other positive outcomes. The actions of Portuguese soldiers in its 1974 Carnation Revolution had the immediate goal of ending the Novo Estado Regime’s colonial wars. Though democratization was not a specific goal of the coup, within three years the country had both witnessed its own transition and set off democracy’s “third wave” globally. However, any celebration of the event must be tempered with tragic short-term consequences. The transition period was accompanied by multiple further coup attempts, large scale civil violence, and a complete economic meltdown. Any assessment of “good coups” is destined to be a complicated affair and focusing on a single outcome could keep us from understanding important gains.
As it is, the African context has seen many coups that have been argued to have prevented further political deterioration or social unrest. Just prior to Portugal’s coup, Seyni Kountche’s 1974 ouster of Nigerien president Hamani Diori occurred in the context of worsening personalism and an inability to respond to a major drought that had plagued the Sahel. While the average person struggled during famine, President Diori’s family was accused of hoarding and profiteering from food aid. That there was support for the coup was hardly surprising, and while Kountche made no real effort to democratize, the post-coup administration represented a substantial improvement over what had immediately preceded it.
Christophe Soglo’s first coup in 1963 Dahomey was widely interpreted as having prevented a civil war. As a coup leader, Soglo’s declaration that his move was a “renovation” instead of a revolution is easy to dismiss. Scholars at the time, however, referred to the coup as “caretaker” and “referee,” while later studies with the benefit of hindsight still referred to Soglo’s intervention as a “corrective intervention” or “arbitrator coup” that kept a rapidly deteriorating political crisis from getting worse. Though Soglo would intervene again two years later, this was again a direct response to a rapidly worsening political crisis. While Soglo and the Dahomeyan army proved unable to fix their country’s politics, this must be understood within the context of a broader political context in which civil war was a likely alternative outcome.
LEARN FROM SHORT-TERM AND LOOK BEYOND COUPS
It is easy to dismiss the quick unraveling of a democracy as a sign that a country did not “really” transition to democracy. However, these events often reveal legitimate political changes that are unraveled by competing actors and interests or processes outside the control of the armed forces altogether.
Mauritania’s 2005 coup saw elements of the presidential guard overthrow what they described as President Taya’s “totalitarian” regime. Colonel Ely Ould Mohamed Vall’s junta would navigate a transitional regime toward a new constitutional and a widely-praised election in which soldiers were forbidden to participate. Officers such as Vall were widely reported to have political ambitions, the fractured outcome of the parliamentary poll prompted fears that the new regime would be ungovernable, and there was clear interest among the armed forces in extending the transition period. However, they ultimately allowed the election to play out without their participation or interference. That the army did return to power 16 months later should not prompt us to ignore the short-term impacts of the 2005 coup. We should instead acknowledge what limited successes had occurred and to try to learn from them.
Perhaps the most important challenge to understanding post-coup developments is to move beyond a fixation on the coups themselves, and particularly successful ones. Though some might refer to coups as potential “springboards,” springboards only capture one small part of a multi-staged exercise. Anyone vaulting from a springboard must also execute their maneuver and stick the landing. The outcome already faces a high level of difficulty, but post-coup leaders have the added hazard of various actors who have incentives to sabotage their performance.
More plainly stated, understanding democratic transitions requires far more than focusing on high-profile events such as coups themselves. These events must be interpreted within the contexts they occur, and particular attention must be given to parsing out whether it is the coup, failed or successful, that is driving any political changes or whether the coup is itself a product of broader political or societal events already in motion.
For example, the 1982 failed coup in Kenya may have been followed by Moi’s authoritarian entrenchment. Yet closer inspection reveals that Moi had already been increasing his hold on power prior to the coup attempt. His aim, prior to the failed coup had been reducing the influence of Kenyatta holdovers in his government. Conversely, the failed coup in Zambia against Kenneth Kaunda in 1990 reportedly reinforced Kaunda’s commitment to returning to multiparty politics. However, the coup took place in the midst of ongoing mass discontent and Kaunda’s moves towards political liberalization. Both coup attempts had divergent outcomes influenced by their unique political contexts.
All cases are different, and assessing whether any particular event was “good” requires considering what would have happened in its absence. Cheeseman notes that of all of Africa’s coups, perhaps less than 5% had a “positive and sustained democratic impact.” But what was the likelihood of sustained democratic impacts in the absence of coups? For example, what would Mali’s history of democracy have looked like in the absence of ATT’s March 1991 coup?
It is also important to remember that coups often reflect attempts to prevent further deterioration. While many discussion’s of Mali’s 2012 refer to the disastrous developments that followed it, fewer commentaries acknowledge the coup itself was a direct product of an already deteriorating security situation. To be clear, the coup failed to prevent a further slide and instead contributed rapid territorial losses that amounted to roughly half the country. In other cases, worse outcomes might have been avoided. While not able to “fix” the political system or promote democracy, Soglo’s 1963 coup is credited by many with preventing civil war. Had the coup that brought Sankara to power not been attempted, there almost certainly would have still been a coup undertaken by more conservative elements of the armed forces in August 1983. Sankara taking power may have “failed” to bring demonstrable improvements in terms of democracy, but they action might have prevented a worse outcome. In short, the appearance of no change can sometimes reflect a major achievement.
Welcome to the Arrested Dictatorship blog. Posts on recent events are periodically updated as more information becomes available. It is currently edited by Jonathan Powell and Salah Ben Hammou at the University of Central Florida.
Decolonizing Coup Data, Salah Ben Hammou
Coups and Democracy, J Powell & Mwita Chacha
Coups & Clickbait, J Powell
Iraq 1936, Salah Ben Hammou
Failed Coup...Successful Transition? Salah Ben Hammou & J Powell