In a paper forthcoming at Foreign Policy Analysis, Clayton Thyne and I suggest that attempted coups can prompt authoritarian leaders to open their political systems, potentially leading to democratic transitions. Our work was recently applied to an ill-fated December 30 attempt to unseat Gambian President Yahya Jammeh. In contrast implementing liberal reforms, Jeffrey Smith and Alexander Noyes demonstrate that Jammeh’s actions have been definitively undemocratic.
While I doubt Jammeh's offer to rule for “a billion years” will be realized, I completely agree when Smith & Noyes claim “Gambia’s longtime dictator isn’t going anywhere any time soon.” However, I should clarify that our paper does not investigate the type of act undertaken in Gambia and, if anything, our theory anticipates the status quo (authoritarianism) should prevail. I'll comment on this and then to get to a far more important implication of their post: social science essentially has very little to offer regarding cases like Gambia.
Failed Coups and Democratization
Our investigation of 170 global episodes of authoritarianism shows that democratic transitions are more than twice as likely when a successful or failed coup occurred in the previous 3 years. I'll avoid the technical details of our model since the trend is easily illustrated with descriptive statistics:
Using regime type data from Cheibub and colleagues, this graph looks at countries that were non-democracies, like Gambia, the year prior to the coup. As we move from left to right we see how the percentage of cases that are democracies varies. The solid line shows a sample of countries that experienced a coup in year 0, while the dashed line shows the trend in countries that did not. (a quick data note: we originally used the Polity data for our FPA paper; I use the Cheibub measure here since it covers a wider range of cases).
Countries that experienced a failed coup are well over twice as likely to be a democracy (16.4% versus 6.6%) at the end of the third year, eventually widening to an 11% gap by the end of the fifth year (20.7% versus 9.4%). This trend holds up rather strongly when controlling for other common determinants of democratization.
Coup or no Coup?
Many will colloquially refer to the Gambian “coup attempt,” but scholarly definitions do not consider the December 30 event to be an attempted coup d’état. Luttwak’s classic Coup d’état: A Practical Handbook, for example, required coups to be undertaken by a “segment of the state apparatus” and decades of scholarship generally follows suit. Clay and I are especially clear about this distinction in our dataset. Though news outlets initially reported the Gambian plot was carried out by elements of the military and/or Presidential Guard, subsequent reports clarified these were former members, primarily expatriates living in the United States and Europe.
This conceptual distinction is not merely semantic in this case. Our “perhaps over-optimistic” expectation that failed coups can act “as credible signals that leaders must enact meaningful reforms” was anticipated specifically because a part of the state apparatus attempted to unset the leader. We have no reason to believe actions by regime outsiders, including former members of the military will influence leaders’ decisions via the mechanisms we discuss in our paper. Instead of a “resounding no” to our argument for failed coups, the case would fit our more mundane expectation for non-events (no transition).
This reflects a common tendency to lump different types of anti-regime activity together. Our coup dataset originated in part due to the tendency for them to be conflated with civil war. This has similar theoretical implications. As Hultquist notes, “coups do not require the popular mobilization that largely informs the theory about civil war processes.” However, models testing those theories would often conflate the two so long as the coup attempt was "bloody" enough to reach a sufficient body count for classification as a civil war. James Fearon and David Cunningham recognized the bias this would introduce to duration models early on, while folks like Thyne and Hultquist have recently considered other conflict dynamics.
Lessons for Social Science
The Gambian case, and the discussion by Smith & Noyes more generally, points to a few ways in which scholarship could be seriously improved.
First, failed coups can range from what Kebschull referred to as a “superficial, shotgun attempt by a few, poorly organized officers who fired a few shots in the air” to cases that saw collaboration amongst high-ranking officials across the entire security apparatus. The former case would likely have less of an influence on subsequent decisions than the latter, if they matter at all. Had some of the would-be Gambian putschists hailed from the current ranks of the security apparatus (or if reliable info comes to light showing some actually did), it would probably still have far less of an influence on Jammeh than if his Presidential Guard had fought off hundreds of soldiers led by folks from the military high command. Future studies of the consequences of failed coups would do well to distinguish the degree to which they legitimately threatened the executive’s tenure, how different elements of the security apparatus behaved, or how the public reacted. In short, conditionalities are likely important.
Second, the literature could be vastly improved by considering other post-coup outcomes. As an excellent paper by Nikolay Marinov and Hein Goemans mentions, "While coups have been a staple of twentieth-century politics, their aftermath has eluded systematic scrutiny." Beyond our findings for democratization, the 230+ failed coup attempts since 1950 have produced plenty of “bad aftermaths.” Increased authoritarianism (Daniel Arap Moi’s actions following Kenya’s 1982 failed coup), civil war (e.g., post-2002 Côte d’Ivoire), and one-sided mass killing (Samuel Doe’s reprisals after a failed 1985 coup) are just a few that spring to mind. Understanding the conditions that contribute to any of these diverse outcomes would be a welcome addition to scholarship.
Third, though little scholarly attention has been given to failed coups, even less has been given to events like the December 30 effort against Jammeh. Infamous European-born guns-for-hire like Bob Denard and Mad Mike Hoare have long since left the political scene, but these type of acts have not gone away. Equatorial Guinea’s 2004 “Wonga Coup” saw well-funded mercenaries plotting to overthrow the Obiang government, which was again attacked by armed men from the Niger Delta region five years later. Clay and I do not investigate acts like these and I know of no comprehensive effort to do so. But they are certainly worth study. As it stands, social science essentially offers nothing to systematically explain these events or their consequences.
Finally, studies could also do more to consider “plots” or “rumors.” In contrast to our Foreign Policy Analysis paper, conspiracies that are discovered and quashed prior to execution could signal the strength of the coup-proofing apparatus. Coup “plots” could also be no more than a pretense to purge opponents. Dwyer’s original post on the Gambia case at Jay Ulfelder’s Dart-Throwing Chimp noted “there is speculation as to whether some [previous plots against Jammeh] were real or simply ways to purge members of the military.” Kebschull more directly referred to these types of acts as “deliberately contrived nonsense.”
My best guess is that Smith and Noyes are on the mark when speculating that successive coup "plots" could contribute to authoritarian entrenchment. Though it is inevitably difficult to verify and document plots, they are still regularly reported and—like Turkey’s “Sledgehammer Plot”—could teach us important lessons about a range of political dynamics.