Elite Change without Regime Change: Authoritarian Persistence in Africa and the End of the Cold War

Abstract

Because the end of the Cold War failed to produce widespread democratic transitions, it is often viewed as having had only a superficial effect on Africa’s authoritarian regimes. We show this sentiment to be incorrect. Focusing on the elite coalitions undergirding autocracies, we argue that the end of the Cold War sparked profound changes in the constellation of alliances within regimes. It was an international event whose ripple effects altered the domestic political landscape and thereby enticed elite coalitions to transform and meet the new existential threat they faced. We demonstrate our argument using cabinets as a proxy for elite coalitions, showing that their composition drastically changed at the end of the Cold War. Africa’s authoritarian leaders dismissed many of the core members of their cabinets and increasingly appointed members of opposition parties to cabinet portfolios. Such changes, we argue, represent the dynamic responses that enabled autocracies to persist.

INTRODUCTION

The end of the Cold War marked the beginning of a new era. The competition between the West and the Soviet Union was no more and a new international order was being formed that would produce ripple effects across the globe. As the West and its emphasis on liberal democracy emerged victorious, authoritarian regimes, particularly across Africa, appeared increasingly untenable. Omar Bongo, Gabon’s ruler and one of Africa’s longest serving autocrats, was keenly aware of this when observing the collapsing communist regimes in Eastern Europe, prompting him to say in February 1990 that “[t]he winds from the East are shaking the coconut trees in Africa” (Packham 2004, 209). ¹ Indeed, facing domestic and international pressures for change, most African autocrats adopted multiparty elections and initiated other liberalizing reforms (Bratton and van de Walle 1997; Cheeseman 2015; Lindberg 2006; Riedl 2014).

Rather than ushering in democratization, however, these liberalizing reforms became the veneer behind which many authoritarian leaders and regimes remained firmly in control (Bratton and van de Walle 1997; Diamond 2002; Levitsky and Way 2010; Ottaway 2003; Schedler 2002). This resilience is remarkable considering the existential threats facing so many autocrats at the end of the Cold War. After all, the collapse of the Soviet Union produced what Gunitsky (2017) calls a hegemonic shock—a rare but far-reaching event where the sudden decline of a great power becomes the catalyst for sweeping domestic changes across the world. While the pressures for political reform emanating from this shock were immense, authoritarianism nonetheless prevailed. How, then, did autocracies persist? How did they successfully navigate the hostile political landscape they faced in the wake of the hegemonic shock?

We point to elite coalitions as the principal source of autocratic resilience. Competition between the West and the Soviet Union had provided African leaders with the necessary resources to engender loyalty and punish dissent among the regime’s elite (Schmidt 2013). However, with the Soviet Union in rapid decline and the nature of relationships between the West and Africa changing at the end of the Cold War, existing patterns of domestic alliances in African autocracies were destabilized. From rulers to regime insiders to dissidents in the population, it was clear that a new age was on the horizon where joining the opposition would no longer mean entering political exile. It was a period where the constraints of the past were quickly fading and the status quo was no longer a viable option. Take Kenya’s Mwai Kibaki, for example. Kibaki was a regime stalwart, serving in a variety of prestigious cabinet positions between 1966 and 1991, including Minister of Commerce, Minister of Finance, Vice President, and Minister of Health. However, he seized the opportunity at the end of Cold War and defected from the ruling Kenya African National Union party to challenge incumbent Daniel arap Moi in the 1992 presidential election. In light of these emerging rifts within elite coalitions, rulers had to reevaluate whether their current alliance base was sufficiently strong to ensure their continued political survival.

We argue that this crisis brought on by the end of the Cold War represented an existential threat, which sparked fundamental changes in the composition of elite coalitions, thereby giving way to new alliances. As O’Donnell and Schmitter (1986) argue in their work on transitional moments, the reconfiguration of elite coalitions is a survival mechanism. Incumbents either adjust or risk becoming obsolete and, as we argue below, the formation of new alliances and shedding of old ones was a means to counter the pressures of the end of the Cold War. Deprived of their usual tools, rulers had to pursue alternative strategies and the composition and operation of elite coalitions was an area they could directly affect. The result was a period of transformation where previously calcified elite coalitions suddenly became malleable. Leaders dynamically responded to their changing political landscapes by reconfiguring their alliance bases, which helped them stave off the push for democratization. The resilience of authoritarian regimes was thus achieved by a burst of change in the elite coalitions undergirding those regimes.

We examine the effect of the end of the Cold War on elite coalitions in African autocracies through changes in cabinet composition. Using data from the WhoGov dataset (Nyrup and Bramwell 2020), we analyze cabinet change qualitatively in Gabon under the leadership of Omar Bongo and quantitatively using a sample of 39 African autocracies between 1966 and 2010. Our analyses show that the composition of elite coalitions and the nature of power sharing between leaders and elites fundamentally changed at the end of the Cold War. We find that ministers, including those with substantial experience in the cabinet and those in the most important portfolios, faced a sharp rise in their risk of exiting the cabinet at the end of the Cold War. And, as we illustrate, the extent of these changes cannot be explained by common determinants of cabinet change such as elections, leadership transitions, or failed coup attempts. Moreover, we show that rapid changes to elite coalitions did not persist beyond the end of the Cold War period. Once leaders reconfigured their elite coalitions to meet the new political environment, which included the cooptation of opposition party leaders, they returned to their prior practices of fostering a relatively stable core of ministers to anchor their regimes. Our analyses therefore show that the end of the Cold War was indeed a hegemonic shock, unrivaled in the decades before or since in its scope to effect change among elite coalitions across Africa.

This article makes two primary contributions. First, our findings emphasize the importance of elite coalitions for authoritarian stability. We join scholars such as Brownlee (2007, 10) who argues that the durability of regimes is best understood by looking to “the nerve center of authoritarianism: the ruling organization and the coalition it houses.” African autocrats adjusted to the hegemonic shock at the end of the Cold War by reconfiguring their elite coalitions to meet the challenges of their new political environment. These changes provide additional insights into the process of learning and adaptation that, as Gunitsky (2017) argues, undermined the democratic wave at the end of the Cold War. Similarly, Meng (2020) emphasizes the importance of executive constraints and elite power sharing for authoritarian stability. We add to this research by highlighting the mechanism through which the end of the Cold War shifted the balance of power away from African leaders and toward elites, thereby forcing leaders to make concessions.

Second, our findings provide deeper insights to the literature on political development in Africa after the Cold War. The lack of democratization has led some scholars to discount the effects of the end of the Cold War and instead stress the continuity of politics in Africa (Bleck and van de Walle 2019). Shifting our attention from the regime level to elites, we show that the end of the Cold War led to transformational change in elite coalitions and power-sharing dynamics, even where incumbent autocrats remained in office. In short, the absence of democratization did not signal a continuation of status quo elite politics. Africa’s authoritarian leaders made extensive changes to their inner circle of elites in response to the hegemonic shock caused by the collapse of the Soviet Union.

SHOCKS, ELITE COALITIONS, AND UNCERTAIN TRANSITIONS

In their classic book, O’Donnell and Schmitter (1986) offer insights into the moment when authoritarian regimes teeter at the edge of transition. Whereas earlier scholarship was marked by a long-term view of regime change (Lipset 1959; Moore 1966), O’Donnell and Schmitter (1986, 6) instead focused on a specific time interval “delimited…by the launching of the process of dissolution of an authoritarian regime,” which concludes either in “the installation of some form of democracy, the return to some form of authoritarian rule, or the emergence of a revolutionary alternative.” This approach treats the transitional moment as discrete object amenable to deeper analysis and cross-country comparisons. And, it is this approach that has informed the literature on the end of the Cold War and its effect on African regimes.

With the end of the Cold War marking the beginning of a transitional moment, some observers optimistically assumed that authoritarian regimes across Africa would crumble and give way to democracies. The initial signs seemed to point in this direction as regimes liberalized and introduced multiparty elections. Gradually, however, these expectations proved to be a “democratizing bias” where perceived democratization was really an entrenchment of a different type of autocracy, namely competitive authoritarianism (Levitsky and Way 2010). Scholars have thus stressed continuity as the defining feature of post-Cold War politics across Africa. So much so, in fact, Bleck and van de Walle (2019, 4) posit the transition from Cold War to post-Cold War saw “many of the same men and rather few women remain in positions of power in the region.”

Yet, if authoritarian regimes managed to persist, we argue that it was not because of a lack of change among those in power. Quite the opposite. Authoritarian regimes survived precisely because elite coalitions adjusted to the new post-Cold War landscape. Put succinctly, we posit that elite change is the mechanism linking the end of the Cold War on the one hand and authoritarian persistence on the other. In making this claim, we harken back to O’Donnell and Schmitter’s (1986, 19) observation that transitional moments are driven by “divisions within the authoritarian regime itself.” Although the outcome of transitional moments is highly uncertain, they are invariably marked by splits among the elite. The critical insight O’Donnell and Schmitter (1986) provide is that these splits need not be terminal. They do not foreshadow an imminent collapse of a helpless elite. Instead, elite coalitions reconfigure themselves to survive.

Garretón (1986, 111) spells out the underlying logic at play here. In an attempt to survive, a regime experiencing a transitional moment will make “defensive or reactive adjustments and shuffle its short-term alliance.” In the process, the regime’s “hegemonic nucleus will break up as it turns to partial solutions” where “sectors which previously supported it will fragment and split away.” In other words, elite coalitions are not static entities. Authoritarian persistence does not mean the underlying elite coalition remains intact through the transitional moment. Likewise, regime collapse does not imply an insulated elite too obtuse to confront the problems at hand. Rather, the regime’s nucleus dynamically evolves in an attempt to meet the challenges it faces. Its changing composition during the transitional moment reflects the ongoing search for solutions in an uncertain time. Previously excluded actors who can address a particularly potent threat during this crisis may now be coaxed to join the coalition. Others, in contrast, may be pushed out if their power base erodes and they can no longer aid in the regime’s survival.

This metamorphosis highlights the coalition’s malleability, the characteristic that enables it to absorb the shock of the transitional moment, albeit within limits. The key, as we elaborate below, is to understand the end of the Cold War as a shock the coalition could still absorb. Although the collapse of the Soviet Union was a monumental event, its ripple effects were funneled through elite coalitions who acted as a buffer. As such, it was the coalition’s shifting alliance base that made authoritarian persistence possible. Moving straight to questions of democratization—or the lack thereof—misses this critical link.

Elite Coalitions and the End of the Cold War

Authoritarian regimes contend with threats from both within and without. On the outside, there are the masses who can topple the regime when sufficiently mobilized. On the inside, there is the danger of elite fracture either through coups or defections. From violent suppression of the masses (Nassif 2021) to strategic co-optation of the opposition (Arriola, DeVaro, and Meng 2021), from shuffling (Woldense 2018) to the sharing of spoils to generate loyalty among the elite (Szakonyi 2018), the literature has identified a range of tools that help regimes meet the dual threat to ensure their survival. Yet, when confronted with rare and unprecedented events, even the most robust toolkit will fail to anticipate the newly emerging threats. This is especially the case when facing what Gunitsky (2017, 2–3) calls a hegemonic shock, “moments of sudden rise and decline of great powers” which “act as powerful catalysts for cross-border bursts of domestic reform.” As we argue, the end of the Cold War brought about a vicious dynamic for autocrats. It drastically reduced the number of available tools while simultaneously generating more and more problems.

Since independence, African leaders had leveraged Cold War competition for client states to attain economic, military, and ideological resources (Schmidt 2013; Young 2012). With those resources, African leaders were able to suppress their domestic opposition more effectively and engender loyalty among the elite. According to Clapham (1996, 156), so bountiful was this relationship that the “inexhaustible supply of arms and aid from an all-powerful external patron encouraged rulers to suppose that their own hegemonic ambitions were ultimately unstoppable.” By the mid-1980s, however, Soviet client states were quickly losing access to their source of arms and aid (Dunning 2004; Gunitsky 2017). At the same time, Western countries were quickly changing their approach to foreign assistance, prioritizing democracy and human rights promotion over its previous focus on containing communism (Bearce and Tirone 2010; Bermeo 2011; Dunning 2004; Gunitsky 2017; Kim and Kroeger 2017). This marked a seismic shift not only in the international arena, but also in the domestic political landscape (Gunitsky 2017). With foreign assistance conditioned upon domestic reform and the ideological protection of the Cold War evaporating, the toolkit for survival began to shrink. No longer could the domestic opposition be as easily marginalized nor the elites be enticed with the aid of an international superpower. Indeed, the once favorable international arena that helped autocrats bolster their rule now actively undermined them.

Worse yet, this hegemonic shock occurred at a time when domestic conditions across the continent were “more than ripe for upheaval” (Decalo 1992, 14). As Joseph (1997, 378) notes, “[p]re-1989 Africa had become politically and economically calcified” and experiments with single party rule began to lose their appeal. After more than two decades of failed promises—where military and personal regimes fared no better—the demand for change was mounting (Bratton and van de Walle 1992a; Joseph 1997; Nyong’o 1992; Young 2012). Adding to these woes were the ongoing economic crises. The debt burden saddling African countries throughout the 1970s became all the more acute in the late 1980s when debt payments accounted for significant portions of yearly budgets (Decalo 1992). Structural adjustment programs mandated by the International Monetary Fund to curb bureaucratic bloat and introduce private enterprise only served to further starve regimes of their domestic revenue flows (Cooper 2002). Coupled with the continuing decline in the value of agricultural exports—the principle means of acquiring foreign currency for many countries—incumbent leaders had few resources with which to satisfy their citizens or regime elites (Bratton and van de Walle 1997).

The rapidly changing international environment combined with widespread domestic discontent made the end of the Cold War a focal point for opponents to coordinate in protest against African autocrats. Any attempts to violently suppress the masses would come at great risk not least because of the increasing international pressure for human rights and multiparty elections. For instance, in his speech at the 1990 Franco-African summit, French President François Mitterrand made this clear by stating that France would now deliver aid “more enthusiastically” to countries engaged in democratic reforms (Riding 1990). The result of these emerging pressures was the introduction of a transitional moment, which, though uncertain in its outcome, engendered a predictable response from the affected regimes. Benin’s Mathieu Kérékou was one of the first African leaders to face the effects of the hegemonic shock. Cut off from external aid, Kérékou’s regime was struggling to maintain the basic functions of government when student and teacher protests broke out in 1989. Kérékou acted quickly to save the regime as protests spread by shuffling his cabinet, releasing political prisoners, holding a national conference, and returning to multiparty elections (Bratton and van de Walle 1997). Over the next four years, leaders across the continent found themselves in similar fights for survival. While some leaders, like Kérékou, would not succeed in extending their tenure, the survival response was ubiquitous with leaders forging new alliances and making calculated concessions in an effort to make it through the tidal wave that was the end of the Cold War.

With the accompanying government pronouncements and journalistic coverage, liberalization was, no doubt, the most visible sign of change. Yet the most potent impact of the end of the Cold War was its corrosive effect on the political bonds among domestic elites. Alliances that once anchored regimes weakened as the political terrain started changing. The ability to suppress outsiders and command loyalty among ruling elites was cast into doubt. Leaders had to reassess whether their existing alliances should be maintained, partially reshuffled, or abandoned altogether. To be sure, coalition change was not just initiated by leaders. Elite coalition members had options previously unavailable to them as they could now defect from the incumbent’s party and still remain politically relevant. Similarly, newly empowered opposition leaders could no longer be ignored and became worthy targets of co-optation. With leaders deprived of the fuel that had previously powered their political machinations, they were now compelled to forge new alliances to meet the challenges at hand—ushering in a period of elite transformation.

This deliberate transformation of the elite was critical. It was at once one of the few remaining options left in the toolkit and it had the potential to act as a buffer against the intense but brief pressures brought on by the end of the Cold War. As Gunitsky (2017, 42) argues, the end of the Cold War led to “the proliferation of hybrid regimes—countries that adopted the trappings of democracy to satisfy the immense post-shock pressures for reform, yet learned to infuse them with autocratic rule as the pressures for democracy faded.” The ability to infuse autocratic rule under these hybrid regimes was not a coincidence, but largely attributable to the transformation of elite coalitions. Leaders were able to undermine the brief moment of opposition coordination in the wake of the hegemonic shock by coopting select opponents. Bratton and van de Walle (1992b, 38) highlighted this point when writing that:

leaders tried to sidestep the need for meaningful reform by revolving the ranks of the political elite. They used cabinet reshuffles in an effort to prevent rivals from taking advantage of popular discontent and instigated “party restructuring” campaigns in order to ensure that local officials were correctly transmitting the official party line.

These elite shuffles provide a much needed avenue for change without threatening the authoritarian nature of regimes. Continuity on the regime level (i.e., authoritarian persistence) masks this subtle, but crucial point. Once we take elites into account, we see that regimes survived not due to the absence of change but because of it.

Yet the persistence of authoritarian regimes through elite change came with a cost. Even established autocrats were placed back at the negotiation table and forced to revise existing power-sharing arrangements with their new coalition partners. As Meng (2020, 15) notes, “…Many leaders went from being unconstrained in their rule to accepting formal limits on their authority in constitutions. Presidents went from serving as their own vice president and defense minister to delegating these powers to other elites.” These changes, we argue, were driven by the infusion of new actors and alliances in the elite coalition. After all, the previous power-sharing arrangements were brokered by the old guard who represented the past. And, just as negotiations settled previous disputes, new ones were needed to suit the incoming actors looking to assert themselves within the ruling coalition.

In this respect, the end of the Cold War shifted the balance of power away from the ruler while ensuring authoritarian persistence (Meng 2020). Hampered by depleted resources and a growing opposition, autocrats were at a disadvantage when bargaining with elites. Political liberalization meant that opposing the leader no longer carried the threat of political exile. Elites could now challenge leaders and still expect to return to the halls of power, placing them in a stronger position to demand concessions from the leader. Without the levers of punishment readily available, leaders now had to focus their co-optation efforts on emboldened elites and accommodate them accordingly.

Our argument thus posits a direct link between authoritarian persistence, elite coalition transformation, and the new power-sharing arrangements emerging at the end of the Cold War. The hegemonic shock induced by the end of the Cold War produced strong, but temporally bounded pressures for political liberalization. It opened up a transitional moment where a malleable elite coalition became the main buffer between authoritarian persistence and full-fledged democratization. As leaders faced increasing opposition in a new political environment, they transformed their elite coalitions to co-opt and divide their opposition. This survival strategy thwarted transitions to democracy in many cases, but did lead to new power-sharing arrangements.

In stressing coalition change, our argument stands in contrast to an alternative theory for coping with the end of the Cold War. While the masses may protest, regime insiders usually have a vested interest in seeing the status quo persist (Bueno de Mesquita et al. 2003; Jackson and Rosberg 1982). After all, their political livelihood is tied to the regime and if the latter falters, it can spell the end of their stay at the apex of power. For leaders, surrounding themselves with loyal elites is already a priority during ordinary times and may therefore become the upmost importance when facing a crisis. The incentives for both sides are such that we may expect a hardening of the coalition. Under this argument, it is less, not more change that should be the defining feature of elite coalitions at the end of the Cold War. In the coming sections, we show that this was not the case. We demonstrate empirically that the hegemonic shock led to significant changes in elite coalitions. Indeed, our analysis underscores that the end of the Cold War was unique in altering domestic coalition dynamics across the entire region.

CABINETS, ELITE COALITIONS, AND THE CASE OF GABON

We examine our claims about the effect of the end of the Cold War on elite coalitions in Africa by focusing on cabinets. Cabinet portfolios represent some of the most prestigious and prized political positions short of the chief executive. This is particularly the case in many African countries where legislatures are historically weak and the executive branch plays a large role in policymaking and the distribution of state resources. ² For members of the incumbent’s party and opposition party leaders, being appointed to the cabinet means entering the halls of power. Ministerial appointments come with significant influence over policymaking and implementation and ministers can also use the resources at their disposal to expand their own patronage networks and reward their constituents (Arriola 2009; Barkan and Chege 1989; Zolberg 1969).

The cabinet’s elevated status in the regime makes it the site where elite power-sharing arrangements are most readily observable. For instance, scholars have regularly looked to cabinets to assess claims of favoritism toward particular ethnic groups, regions, or political parties, either in terms of the distribution of cabinet seats or the political power held by particular groups (Ariotti and Golder 2018; Francois, Rainer, and Trebbi 2015; Raleigh and Wigmore-Sheperd 2022; Ricart-Huguet 2021). Similarly, others have emphasized how cabinets are used by leaders to manage threats from elites. Where the threat of coups looms large, Arriola (2009) finds that rulers deliberately inflate the size of the cabinet to hamper coordination among elites. Autocrats may also shuffle their ministers (Kroeger 2020), refuse to appoint a Prime Minister, or reserve for themselves sensitive portfolios like minister of defense to secure their position in power (Meng 2021; Woldense 2018). In short, the cabinet’s composition is a product of the underlying power-sharing dynamics and thus a good way to observe changes to elite coalitions.

Gabon: A Motivating Case

Before moving to the quantitative analyses, we examine cabinet change descriptively using Gabon under the leadership of Omar Bongo as a motivating case. Bongo’s period in office provides an interesting case because he ruled from December 1967 until his death in June 2009, covering nearly the entire period under study here. Moreover, the continuation of Bongo’s tenure after the Cold War presents at least an outward appearance of regime stability. Yet, as we show below, the apparent continuity in Bongo’s regime came only with significant changes to his elite coalition.

Just prior to independence in August 1960, Gabon signed 15 coopération agreements with France related to matters of defense, foreign policy, economics, education, and culture (Gardinier 1982). Such agreements were part of France’s broader Africa strategy for much of the Cold War that sought to maintain the French “field of action” (Chipman 1989, 165). France implicitly guaranteed the protection of Gabon’s leader against domestic threats, which was affirmed by French intervention in 1964 to reinstate President Léon M’ba after he was ousted in a military coup (Nugent 2012). French military protection continued after Omar Bongo assumed the presidency following Léon M’ba’s death in 1967, with Gabon hosting one of the largest contingents of French forces in Africa throughout the Cold War (Chipman 1989). These security guarantees helped ensure continued French domination of Gabonese oil production through the French state-owned company Elf (Yates 2012).

At the end of the Cold War, however, the hegemonic shock helped to produce increasing domestic and international pressure on Bongo for political reform. Opposition to Bongo’s regime initially emerged within the Catholic Church in the early 1980s, which led to the formation of the Mouvement de Redressement Nationale (MORENA) party (Nugent 2012). While Bongo was able to repress MORENA opponents, student protests in early 1990 over poor facilities and instructor shortages quickly escalated into broader anti-government protests that were not as easily repressed (Bratton and van de Walle 1997). Complicating matters more, these protests corresponded with a period of change in French policy toward Africa. French President François Mitterrand began his second term in 1988 seeking to end France’s personalized relationships with African leaders. Particularly, as revolutions spread across Eastern Europe in 1989, Mitterrand was no longer willing to provide unconditional aid to non-democracies that regularly violated the human rights of their citizens (Chafer 1992). As Gardinier (2000, 225) states, Omar Bongo faced “a lack of political support from the Socialist regime in Paris…and even unfriendliness, on occasion.” With French willingness to support the status quo in Gabon waning, Bongo took a more conciliatory approach to his domestic opponents by holding a national conference in March 1990. The new constitution drafted at the national conference put an end to one party rule by Bongo’s Parti Démocratique Gabonais (PDG) and multiparty legislative elections were held in September and October 1990.

Although Bongo was able to maintain control over the national conference and his position as president (Nugent 2012), he made significant changes to his inner circle following the conference. Drawing on the WhoGov dataset (Nyrup and Bramwell 2020), Figure 1 displays these changes by tracking the tenure of individual ministers in Bongo’s cabinet between 1968 and 2008. The red points on the graph indicate that a minister exited the cabinet in a particular year, whereas the gray points indicate minister survival. The upward movement of points across years on the x-axis indicates increases in the tenure of ministers. What is immediately apparent in Figure 1 are the two triangular shapes separated at the 1990 mark. Their similarity in shape reflects Bongo’s approach in managing his cabinet. While Bongo occasionally reshuffled his cabinet before and after 1990, he still retained significant numbers of experienced ministers to maintain continuity within the elite coalition.

Figure 1. Cabinet Change in Gabon, 1968–2008

Yet Bongo’s actions following the national conference in March 1990 represent a sudden and drastic break with the past. Only 7 of 41 ministers from 1989 were retained in 1990. As Gardinier (1994, 28) explains, “Bongo showed a remarkable capacity to adjust to the rapidly changing conditions…Encouraged by the reformist elements with the PDG (including his son Ali), he dropped from the cabinet longtime ministers who opposed changes or showed little real competence.” For instance, Prime Minister Léon Mébiame who had held a variety of ministerial portfolios since 1967 was replaced with technocrat Casimir Oyé-Mba. Bongo’s removal of senior ministers like Mébiame is particularly surprising as they are generally thought to serve as the anchor stabilizing the ruler’s coalition (Francois, Rainer, and Trebbi 2016). The era of the PDG controlling all ministerial portfolios also came to an end. In 1990, members of the MORENA, Association Pour le Socialisme au Gabon, and Union Socialiste Gabonais parties were brought into the cabinet. Bongo would continue to preside over multiparty cabinets for the remainder of his tenure. Thus, between 1989 and 1991, Bongo had almost completely overhauled his cabinet.

Furthermore, just as sudden as Bongo had deviated from his normal management style, he returned back to it post 1991—thus giving rise to the two distinct triangles of Figure 1. Rather than ushering in a perpetual cycle of instability, we see Bongo fostering a new crop of elites over the next decade to replenish the senior ranks he had decimated in 1990. Hence, the end of the Cold War did not so much change the style in which Bongo ruled, but who he ruled with. He may have survived the end of the Cold War, but doing so came with profound change to his inner circle, the scale of which was never repeated at any other time during his reign.

Data and Methods

Were the changes in the cabinet of Omar Bongo at the end of the Cold War an aberration or part of a larger trend across African autocracies? Our argument points to the latter—the end of the Cold War upended elite coalitions across the continent. To test this, we turn to data on cabinet composition from the WhoGov dataset (Nyrup and Bramwell 2020). WhoGov records the composition of cabinets on a yearly basis for 177 countries. This includes the names of cabinet ministers along with their portfolios, gender, and party affiliation. These minister-level data allow us to reconstruct the cabinet in the same way we did for Gabon as displayed in Figure 1. In accordance with the scope of our argument, we limit our focus to cabinets in sub-Saharan African autocracies, which we identify using Geddes, Wright, and Frantz’s (2014) regime-type classification data. ³ The unit of analysis is a country-year where, for a given year, a country is included in our sample if it is coded as authoritarian. This results in a sample of cabinets from 39 African countries under authoritarian rule between 1966 and 2010. ⁴ The dataset includes 8,133 individual ministers and 34,215 minister-year observations. To ensure that our results are not driven by some idiosyncratic feature of our sample, we replicate our analyses using alternative sample specifications, which are displayed in Section A3 of the Supplementary Material. Suffice it to say, our substantive results remain robust to such alternate specifications.

Measuring Coalition Change and Empirical Expectations

Our point of departure is that if the end of the Cold War altered elite coalitions across the continent, we should observe exit patterns similar to those displayed in the early 1990s during Bongo’s reign. To this end, there are two main quantities of interest for our analyses below. The first is the probability of exiting the cabinet for all ministers. In focusing on this quantity, we are highlighting the end of the Cold War’s unique empirical footprint. Although failed coup attempts, elections, and leadership changes all increase the probability of minister exit, on their own, they are country-specific events. They are unlikely to have a direct and immediate spillover effect on the minister exit rates of neighboring countries, let alone the region as a whole.

In contrast, the end of the Cold War stands apart because its effects radiated throughout the African continent. It was not a country-specific event, but a regional one that we expect triggered an increase in the probability of minister exits across African autocracies. That is, if changes in the international system at the end of the Cold War did lead to a transformation of elite coalitions, we should see a sharp increase in the risk of minister exit across our sample that is temporally bounded in the late 1980s and early 1990s. Figure 2 provides a descriptive account of this. Pooling the data by calendar year, the graph displays both the overall percentage of ministers exiting the cabinet on a yearly basis and the 5-year moving average.⁵ As can be seen, the late 1980s coincide with a steep increase in minister exits, which then decline in the mid-1990s. In the coming analysis, we examine more closely whether the spike in minister exit was indeed exceptional during this time period.

Figure 2. Minister Exits across Sub-Saharan Autocracies, 1966–2010

Our second quantity of interest is whether the effect of the end of the Cold War is heterogeneous across different types of ministers. We contend that the end of the Cold War produced deep and meaningful changes within the elite coalitions of African autocracies. However, it may be the case that the most important ministers anchoring the leader’s inner circle made it through the hegemonic shock at the end of the Cold War largely unscathed. Leaders may have provided the impression of coalition change by simply removing the most expendable elites. While we show that this was not the case under the leadership of Omar Bongo in Gabon, other leaders may have only sacrificed ministers of lesser importance while keeping their core alliances intact. We test this by narrowing our focus to the fates of the most important ministers.

We conceptualize minister importance in two ways. First, previous research suggests that a minister’s risk for dismissal increases across their first several years of tenure but then declines once they pass the initial high risk period (Bokobza et al. 2021; Francois, Rainer, and Trebbi 2016; Quiroz Flores 2009). We leverage this pattern by positing that long serving ministers represent core members of the leader’s elite coalition. An increase in exit risk alone is therefore not sufficient, for it may be the case that leaders at the end of the Cold War focused their dismissals on less experienced ministers while retaining their longest serving ministers. This would produce increased turnover, but not necessary signal a fundamental shift in the leader’s alliance base. Mass exits of long serving ministers, on the other hand, do signal deeper changes in the leader’s elite coalitions. Thus, examining patterns of exit risk for ministers with short and long tenures provides one way of determining the extent of elite coalition change at the end of the Cold War.

In addition to seniority, the importance of ministers within the elite coalition may also be linked to the portfolios they hold. Indeed, ministerial portfolios vary widely in the coercive and financial resources they control. Portfolios such as defense, interior, and finance command far greater resources than those like culture, youth, or women’s affairs. The core members of a leader’s coalition may thus be limited to those in portfolios controlling more extensive coercive and financial resources. If this is the case, increases in minister exit risk at the end of the Cold War would only signal deep changes in the elite coalition if they were also experienced by ministers in portfolios with significant coercive and financial resources.

Taken together, we measure coalition change by focusing on both the overall exit rates of ministers and the types of ministers that are removed. Overall exit rates speak to the general volatility within the coalition, whereas the types of ministers that are targeted point to the depth of the coalition change. Our argument implies that we see a spike in both volatility and depth around the end of the Cold War.

Outcome Variable

The outcome variable in our analyses is a binary indicator that equals one in a minister’s last yearly observation in the WhoGov dataset and zero for all other years. For example, if a minister is present in the dataset in 1999 but not in 2000, they are coded as exiting the cabinet in 1999. Ministers that remain in the cabinet in 2010 (the final year in our dataset) and are also present in the full WhoGov dataset in 2011 are right-censored.

Explanatory Variables: Measuring the Timing of the Hegemonic Shock

As our theory suggests, leaders responded to the hegemonic shock at the end of the Cold War by transforming their elite coalitions. However, exactly when this shock manifested itself and when its pressures subsided varies across the countries in our sample. To address the difficulty of measuring the precise timing of the shock to elite coalitions, we use a two-step coding strategy. We start by setting the time window within which we expect the end of the Cold War to have operated. Following other scholars who point to the late 1980s and early 1990s as the critical period (Dunning 2004; Gunitsky 2017), we construct the variable Cold War end (CWE), which equals one in years 1988–1992 and zero in all other years. To ensure that our results are not driven by arbitrary cutoff points, we perform a series of placebo tests where we vary the start and end points. ⁶ The second coding step concerns the years within this time window that are to be included in the analysis. The challenge here is the problem of false positives, which can be illustrated using the example of Omar Bongo’s cabinet management displayed in Figure 1. The end of the Cold War affected Bongo’s cabinet management in 1989 and 1990, whereas years 1988, 1991, and 1992 represent the pre- and post-Cold War status quo with much less turnover in the cabinet. By including these status quo years in our time window, we effectively dilute the effect of CWE.

We approach this problem of false positives by structuring our analysis along a continuum anchored on each end by the least and most favorable specifications. To explain the underlying logic, consider CWE $ {}_{jt} $ , where j is the country index and t is the set of years selected from the time window for analysis. We move between the least and most favorable specifications by changing the set t. For this, let n be the number of years we draw from the time window such that the set t has n elements where $ 5\hskip0.35em \ge \hskip0.35em n\hskip0.35em \ge 1 $ . For the least favorable specification, we set $ n=5 $ . That is, CWE $ {}_{jt} $ contains all the years in the time window and, as such, systematically includes false-positive observations. For the purpose of estimating the effect of CWE, this specification works against our hypothesis by biasing the results toward zero.

In contrast, we gradually eliminate false positives by reducing n. When n is less than 5 (i.e., when we are drawing fewer years than available in the time window), we choose t by selecting the years with the highest turnover. For instance, in the case of Bongo, when $ n=2 $ , we choose the years 1989 and 1990 for t and when $ n=1 $ , t reduces down to just 1989. Stated more generally, for each country j and a given number of n draws, t is the set of years which display the highest exit rates. ⁷ In this context, there will be more false-positive observations in the analysis as n approaches its maximum, thereby yielding the least favorable specification. Conversely, as n moves toward its minimum, the number of false-positives decreases, thus producing the most favorable specification. The benefit of this coding scheme is that it provides us with both conservative and generous estimates, which we report in our results below.

Controls

Our theoretical argument claims that the end of the Cold War had a unique and transformational effect on elite coalitions in African autocracies. Although our design does not allow us to rule out all possible confounders, we consider several variables that have also been shown to influence cabinet change. Some of these control variables raise concerns of posttreatment bias since they too were plausibly influenced by the end of the Cold War. We guard against this in our main analyses by estimating model specifications with and without controls. For models including controls, we ease the presentation of results by dividing the controls into three groups: baseline, tenure, and event controls.

The set of baseline controls includes several variables. First, personalist autocracies have higher rates of cabinet turnover than other autocratic regimes (Kroeger 2020). We adjust for differences in the level of personalization of power using the time-varying measure of personalism from Wright (2021). Second, a leader’s access to natural resource wealth, which is associated with personalism (Fails 2020), may also influence cabinet change. For instance, oil wealth may have insulated leaders from the shock of the end of the Cold War in the same way it helped prevent democratization (Hendrix 2018). As such, we adjust for the natural logarithm of oil and gas rents as measured by Ross and Mahdavi (2015). Third, the risk of exiting the cabinet for individual ministers may be influenced by the overall number of cabinet ministers. Thus, we control for cabinet size. Fourth, we include the natural logarithm of GDP per capita as measured by Gleditsch et al. (2002) to adjust for the impact of economic changes on the composition of elite coalitions. Last, the WhoGov dataset creates a problem of left truncation in minister tenure because it begins collecting cabinet composition data in 1966. For countries that gained their independence before 1966, this means that ministers recorded in 1966 have a tenure of 1 year even though their true tenure is unknown. We address this problem by including the variable left truncated, which equals one for ministers present in the dataset in 1966 from countries that were independent before 1966, and zero for all ministers where tenure is known.

The second group of controls focuses on tenure. It adjust for the tenure of both individual ministers and the leaders they serve. Existing scholarship shows that minister tenure has a non-monotonic effect on exit risk in authoritarian regimes (Quiroz Flores 2009). The risk of exit increases at low levels of minister tenure, but decreases significantly after ministers have survived several years in the cabinet. Scholars have also linked leader tenure to elite purges, although there is disagreement on the direction of this relationship (Sudduth 2017; Svolik 2012). Both minister and leader tenure are included in the model as cubic polynomials to account for any nonlinear duration dependence (Carter and Signorino 2010).

Finally, the event controls include several country-specific events that have been shown to influence the risk of minister exit. ⁸ The first are elections, which are found to be the one of the strongest predictors of cabinet change in autocracies (Kroeger 2020). The election variable equals one in all years where an executive or legislative election took place and zero otherwise. ⁹ Data on legislative and executive elections are taken from the NELDA dataset (Hyde and Marinov 2012). Second, the fate of ministers is strongly tied to the fate of leaders. We address this by including the variable Leader exit, which equals one during a leader’s final year in office and zero otherwise. Leader data are taken from the WhoGov dataset (Nyrup and Bramwell 2020) and were cross-checked with the Archigos dataset (Goemans, Gleditsch, and Chiozza 2009). Finally, scholars have shown that failed coup attempts often result in minister dismissals and reshuffles, particularly in personalist autocracies (Bokobza et al. 2021; Easton and Siverson 2018; Kroeger 2020; Woldense 2022). We therefore include the variable Failed coup, which equals one in years when at least one failed coup attempt occurred and zero otherwise. Data on failed coup attempts are taken from Powell and Thyne (2011).

Empirical Model

We examine the relationship between the end of the Cold War and the probability of minister exit using discrete-time event history models. Specifically, we estimate the probability that minister i from country j who is present in the cabinet in year t exits the cabinet before the cabinet observation at year $ t+1 $ using logistic regression. In our main analyses, we opt for a pooled approach to estimation following the equation below where $ {\mathrm{CWE}}_t $ represents a version of our CWE variable and $ {\mathrm{X}}_{it} $ and $ {\mathrm{Z}}_{jt} $ are vectors of control variables at the minister and country levels, respectively. In each model, we cluster standard errors by leader. While the conditional (“fixed effects”) logit model is sometimes preferred in panel data settings to control for unobserved time-invariant unit-level characteristics, severe limitations on model interpretation make the conditional logit model unattractive as our main source of inference. Nevertheless, we replicate our main pooled analyses using conditional logit models that adjust for time-invariant leader characteristics in the Supplementary Material. The results remain consistent with the findings shown below.

$$ \begin{array}{rll}\mathrm{Pr}({\mathrm{Minister}\ \mathrm{exit}}_{ijt}=1)={\mathrm{logit}}^{-1}(\beta {\mathrm{CWE}}_t+\gamma {X}_{it}+\theta {Z}_{jt}).& & \end{array} $$

RESULTS

We begin by testing the first part of our argument, which expects a significant increase in minister turnover at the end of the Cold War. For this, see Figure 3, which plots the average marginal effects of CWE across several model specifications.¹⁰ Following the procedure discussed above, we vary the size of the CWE window between 5 years (least favorable specification) and 1 year (most favorable specification) across the x-axis. For each CWE window size, we estimate four models. As discussed above, other predictors of minister exit may have also been influenced by the end of the Cold War, introducing the possibility of posttreatment bias when these variables are entered into the model as controls. Thus, our first model includes only CWE as an explanatory variable. The next three models increase in complexity with the second model adding the baseline controls, the third adding the tenure controls, and the fourth adding the event controls.

Figure 3. Average Marginal Effects of Cold War End Window Variables with 95% Confidence Intervals

As can be seen in Figure 3, the cabinets in Africa’s authoritarian regimes experienced significant turnover at the end of the Cold War. The average marginal effect for CWE in each model specification is positive and significant, indicating an increase in minister turnover at the end of the Cold War. In the specification with no controls, our least favorable 5-year CWE window is associated with an 8 percentage point increase in the probability of minister exit. While this increase in minister exit risk may seem relatively modest, it is important to remember that it is our most conservative estimate.¹¹ The least favorable CWE window biases our results toward zero because it systematically includes false positives. Shrinking the window size for CWE prunes these false positives and, as a result, produces larger average marginal effect estimates. For instance, Figure 3 shows that, in the model with no controls, our most favorable 1-year CWE window is associated with a 33 percentage point increase in the probability of minister exit. From our least to most favorable specification, we therefore find evidence for our claim that, relative to other years, elite coalitions changed more rapidly at the end of the Cold War.¹² These findings also hold when limiting the sample to leaders with tenures beginning before and ending after the CWE period (see Section A3.4 of the Supplementary Material).

Despite these findings, it may be the case that the most important ministers anchoring the leader’s inner circle made it through the shock of the end of the Cold War largely unscathed. As discussed, our first approach to examining this question is by distinguishing between more experienced senior ministers and less experienced junior ministers. We determine the cutpoint between the two by identifying the year where a minister’s exit probability falls below the probability of exit at the beginning of a minister’s tenure. Figure 4 displays the results showing a drop in a minister’s sixth year in the cabinet. Accordingly, we code ministers with a tenure less than 6 years as junior ministers and those with a tenure greater than or equal to 6 years as senior ministers.

Figure 4. Individual Minister Exit Probabilities with 95% Confidence Intervals by Minister Tenure

Note: Estimates were obtained using Model 1 of Table A20 in the Supplementary Material with all quantitative covariates set to their median values (with the exception of minister tenure) and all event covariates to zero.

If our findings above are driven primarily by superficial changes, increases in exit risk should be more pronounced for junior ministers. We test this by estimating models of minister exit for each CWE window size, interacting CWE with the senior minister dummy. We then calculate the unconditional first difference in exit probability for the CWE and non-CWE periods for both senior and junior ministers.¹³ Figure 5 shows that changes to elite coalitions were not limited to junior ministers. Their senior counterparts similarly faced increases in their probability of exiting the cabinet. Under our least favorable 5-year coding of CWE, the first difference is 7.1 percentage points for junior ministers and 11.4 percentage points for senior ministers. Under each of our more favorable codings of CWE, the estimated first difference for senior ministers continues to be larger than that for junior ministers, although the difference is only statistically significant when using the most favorable specification. Whereas junior ministers faced an estimated 32.5 percentage point increase in the probability of exit under our most favorable specification, senior ministers faced a 48.3 percentage point increase. Far from remaining unscathed, senior ministers saw their risk of exit grow rapidly at the end of the Cold War. ¹⁴

Figure 5. Difference in Minister Exit Probabilities at CWE = 1 and CWE = 0 for Junior (Tenure $ \le $ 5 years) and Senior Ministers (Tenure $ \ge $ 6 years) with 95% Confidence Intervals

Our second way of examining whether important ministers were impacted by the end of the Cold War focuses on the coercive and financial resources available to ministers. Ministers with extensive coercive and financial resources are often critical to the survival of the leader. Therefore, we compare the fates of ministers in survival-critical portfolios including defense, interior, finance, prisons, police, intelligence, and national security to ministers in other portfolios. If the changes in elite coalitions demonstrated above were only superficial, we would expect ministers in survival-critical portfolios to be relatively less affected by the end of the Cold War than those in other portfolios. For analysis, we adopt the same approach to examine the fates of survival-critical and other ministers as we did for junior and senior ministers above. ¹⁵ The first difference estimates presented in Figure 6 are similar to those in Figure 5. Using our least favorable 5-year CWE window, ministers in other portfolios experienced an estimated 8.2 percentage point increase in exit probability and those in survival-critical portfolios an 8.5 percentage point increase. The estimated first differences for both survival-critical and other ministers increase as the CWE coding becomes more favorable, and the estimates for survival-critical and other ministers at each CWE window size are similar one another. Under our most favorable CWE window, other portfolio ministers face an estimated 38 percentage point increase at the end of the Cold War and survival-critical ministers a 31 percentage point increase. This provides additional evidence that the changes in elite coalitions documented in Figure 3 were not limited to those in the least important or influential positions.

Figure 6. Difference in Minister Exit Probabilities at CWE = 1 and CWE = 0 for Ministers in Survival-Critical and Other Portfolios with 95% Confidence Intervals

Taken together, our analyses support the claim that the end of the Cold War brought deep changes in elite coalitions. Not only did cabinet turnover increase generally, but it did so for both senior ministers and ministers in survival-critical portfolios. What remains untested, however, is our key assumption—namely the unique regional feature of the end of the Cold War. Simply put, our research design posits that, absent the hegemonic shock at the end of the Cold War, we should not see elevated cabinet exit risk clustering around the same time across Africa. Stated in terms of our analysis, the increases in cabinet change we find using our chosen CWE time window are not present in other time windows that do not correspond with the end of the Cold War.

We examine this claim by performing a placebo test, which we construct by coding additional indicator variables for each possible 5-year time window in our dataset. We then estimate a model following the same specification as Model 1 in Table A20 in the Supplementary Material for each of the possible 5-year window variables. The results of the placebo test are presented in Figure 7, which plots the logistic regression coefficients and 95% confidence intervals. The x-axis in Figure 7 indicates the start year of each 5-year time window variable. For example, the coefficient presented for 1980 is estimated using a placebo CWE variable, which equals 1 in years 1980, 1981, 1982, 1983, and 1984. The results from the placebo test confirm that the end of the Cold War produced transformational change in the elite coalitions of African autocracies that is not present during other periods. The only time windows with positive and significant coefficients are the ones whose window starts between 1987 and 1992. This increases our confidence that the impact of the end of the Cold War on elite coalitions in African autocracies is historically unique. At no other period did autocrats across the region collectively make such drastic changes to their elite coalitions. ¹⁶

Figure 7. Placebo Test Examining All 5-Year Time Windows with 95% Confidence Intervals

Beyond the End of the Cold War: Stability amid Change

Our focus thus far has been on the immediate aftermath of the end of the Cold War. As our findings show, autocracies responded to the hegemonic shock that ushered in a new international order by changing their underlying elite coalitions. Looking beyond the immediate aftermath, we see from recent work by Meng (2020) that the more long-term impact of this elite reconfiguration process was a new power-sharing arrangement. Weakened by the end of the Cold War and overseeing rapidly changing elite coalitions, leaders had less bargaining power, which enabled incoming coalition members to extract concessions and further depersonalize regimes. From constraining rulers through term limits to securing important cabinet portfolios like the Ministry of Defense, what followed the shift in elite coalitions was a more equitable power-sharing arrangement, albeit still within authoritarian regimes (Meng 2020).

Furthermore, cabinets during the Cold War were often made up of members of the ruling party, military officers, or nonpartisan technocrats. But, as scholars like Arriola, DeVaro, and Meng (2021) and Bokobza and Nyrup (2021) argue, electoral authoritarian leaders in Africa and around the world have used cabinet appointments as a cooptation tool to fragment the opposition. Figure 8 illustrates this very point. Plotting the mean number of political parties in the cabinet each year across all countries in our sample, we can see that the extensive turnover witnessed in the cabinets of African autocrats coincides with an increase in the number of opposition parties appointed to the cabinet. The end of the Cold War thus marks a stark break with the past. As previous alliances were upended, new and different elite coalitions were being built to accommodate the new political environment.

Figure 8. Mean Number of Parties in the Cabinet in African Autocracies, 1966–2010

Yet, even though elite coalitions transformed in important ways upon transitioning into the post-Cold War era, leaders did revitalize one crucial practice. With their coalitions reshuffled to meet the new political reality, leaders sought to regain the stability they had lost in their most senior ranks—the reliable core of elites who anchor the regime. Rather than endure the risk of a constantly shifting coalition, leaders looked instead to replenish their senior ranks so as to reproduce, once again, a ruling coalition with a stable core. The case of Gabon’s Omar Bongo above provides such an example of a leader returning to their previous cabinet management strategy after the end of the Cold War. Bongo drastically changed the composition of his cabinet, including at the most senior ranks. Yet, after transitioning to a new coalition and withstanding the immediate shock, he promptly returned to grooming a new group of senior elites with whom he could rule.

Bongo’s behavior was not aberration, but part of a more general pattern among Africa’s authoritarian leaders at the time. To show this, we estimate the probability of senior minister survival in the cabinet where we distinguish between three time periods: Cold War, end of the Cold War, and post-Cold War. In making these time distinctions, we are drawing particular attention to the changing dynamics as regimes transitioned from the tumultuous aftermath of the end of the Cold War and settled into the new realities of the post-Cold War era. As can be seen from our results displayed in Figure 9, the period effect is readily apparent. The probability of senior minister survival first declines and, after the dip, recovers to levels similar to observed ones during the Cold War. That is, whereas the end of the Cold War induced the severing of ties with senior ministers, post-Cold War reverted back to the status quo of the Cold War period. It thus appears that once leaders had appointed new groups of elites with top government positions, they returned to their previous methods of elite coalition management.

Figure 9. Probability of Senior Minister Survival during Cold War (Years $ \le $ 1987), Cold War End, and Post-Cold War (Years $ \ge $ 1993) Periods with 95% Confidence Intervals

Note: Probability estimates are derived from Table A64 in the Supplementary Material with all quantitative values set at the median and all events set to zero.

CONCLUSION

Writing on the political developments in the post-Cold War era, Meng (2020, 15) notes that “the most dramatic change that occurred in Africa in the 1990s was not that ordinary citizens could vote in multiparty elections—it was that elites recalibrated the political system amongst themselves to entrench their stability.” Our study offers a deeper look into this elite recalibration processes so critical for authoritarian persistence post-Cold War. Although we agree with the literature that the end of the Cold War failed to usher in democratization, our study shows that the impact of this monumental event was far from superficial. We find that it profoundly changed the constellation of alliances within regimes. Using cabinets as the proxy for power-sharing arrangements among rulers and elites, we captured the transformational changes that occurred in the composition of the regime’s upper echelon. The end of the Cold War sparked an exodus of both senior and powerful ministers and, in cases like Gabon under Bongo, entailed a nearly complete overhaul of the cabinet.

As we argue, the transformation in elite coalitions was a survival response to the seismic shift in the domestic political landscape following the end of the Cold War. The new era posed an existential threat where past alliances no longer carried the same political capital they once did. Though risky, abandoning past coalitions in favor of new ones better positioned regimes to navigate the uncharted terrain lying ahead. Recalibration was thus necessary and it took the form of a morphing elite coalition. The end of the Cold War, then, was an agent of disruption that severed existing alliances, introduced new elites to the coalition, and placed even established autocrats in a weaker bargaining position for the new rounds of negotiations taking place during this period.

Our study also highlights an important distinction between the period we have referred to as the end of the Cold War and the more long-term post-Cold War period. The end of the Cold War was a tumultuous time with much uncertainty. It was a time where even the most senior coalition members were suddenly as expendable as their junior peers. Yet making extensive changes to elite coalitions, particularly at the most senior levels, is rarely a sustainable long-term strategy. Autocrats seek instead to develop a relatively stable core of elites to anchor their regimes. Cold War and post-Cold War elite politics share this tendency toward stable core coalitions. The end of the Cold War, on the other hand, appears to have served as the stepping stone, the critical stage where coalitions morphed to shed the old alliances and make room for the formation of new ones.

Related, our study also shows just how unique the end of the Cold War was in its impact on the continent as a whole. Ordinarily, country-specific factors drive domestic alliance formation where a successful coup or even revolution in one country rarely alters elite coalitions in other countries, let alone the entire continent. The end of the Cold War, in contrast, did exactly that. It left its imprint on the continent as whole and in so doing, accomplished the rare feat of engendering profound changes among domestic elites across different countries. In this respect, our study reinforces Gunitsky’s (2017) designation of the end of the Cold War as a hegemonic shock.

Our paper also suggests an important avenue for future research. While we document coalition change, others can expand on our findings by focusing on the kinds of new alliances that emerged. There is much research on the role of ethnic, regional, and socioeconomic factors in shaping elite alliances across Africa. How these factors came together during this critical period right after the end of the Cold War is unclear, however. While the analyses are only exploratory, in Section A11 of the Supplementary Material, we show that the post-Cold War period is associated with a slight increase in the number of included ethnic groups as measured by the Ethnic Power Relations data (Vogt et al. 2015). We also find increases in the number of parties included in the cabinet and cabinet size in the post-Cold War period. Nevertheless, future work is needed to develop both theories to explain specific changes in cabinet composition and empirical tests to evaluate those theories. Additionally, our findings may have relevance to the rise of China’s involvement in Africa. As China continues to rise, it may produce material and ideological conditions similar to those of the Cold War. Future research should investigate whether such changes have implications for domestic political alliances in Africa.