Foreign Pressure and the Politics of Autocratic Survival

Abel Escribà-Folch and Joseph Wright

https://doi.org/10.1093/acprof:oso/9780198746997.001.0001
Published: 2015 Online ISBN: 9780191809262 Print ISBN: 9780198746997

Search in this book

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

CHAPTER

2 Autocratic Regimes and their Collapse 

Abel Escribà-Folch, Joseph Wright

https://doi.org/10.1093/acprof:oso/9780198746997.003.0002 Pages 28–48

Published: September 2015

Abstract
This chapter introduces the regime classi cation used throughout the book and describes the main
characteristics of personalist, party-based, and military regimes. This classi cation is based on
observable characteristics of the relationship between the autocratic leader, the military, and the
support party (if it exists) to identify the group of elites that have e ective control over policy
decisions and political appointments. The chapter then explains why examining autocratic regime
failure is more theoretically appropriate than looking at individual leader survival: in some regimes the
regular replacement of leaders is an indicator of their organizational strength and stability. The
chapter then de nes regime change and introduces the two types of transitions that occur when
autocratic regimes collapse: democratic transition and transition to a subsequent dictatorship. It is
shown that coerced regime transitions usually result in a transition to a new dictatorship, while non-
forced regime changes tend to lead to a democratic transition.

Keywords: autocratic regimes, autocratic leaders, personalist, dominant party, military, regime change,
democratic transition, autocratic transition
Subject: Comparative Politics
Collection: Oxford Scholarship Online

The uprising in Egypt in spring 2011 that chased President Mubarak from o ce brought a military junta to
power. They oversaw multiparty elections for a constituent assembly in November 2011 and presidential
elections the following year. This delayed transition allowed the military to ascertain the most likely new
party to win power, the Muslim Brotherhood’s (MB) Freedom and Justice Party; and gave them ample
opportunity to bargain with the MB in attempts to obtain protections once the transition to civilian rule
took place. Prior to the constituent assembly elections, the journalist David Kirkpatrick articulated the logic
of a delayed transition from the military’s perspective:
 [T]he military is seeking to slow down a democratic transition until it feels certain that its position
and privileges will remain unassailable even under civilian rule. Some here have advocated o ering
the military special rights including immunity from prosecution in civilian courts, protection from
1
oversight of their operations and budget.

While the election brought President Mohamed Mursi to power, his victory was quite narrow, with the
military-backed candidate, Ahmed Sha k, coming a close second in each round of balloting. Thus the
military junta not only correctly bargained with the eventual winners, but they nearly won the presidency
outright in a multiparty contest. And while the Mursi government did not necessarily rule in close
consultation with the military—and even attempted to circumscribe military power by purging senior

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

o cers—in the end the soldiers took back power by force, executing a coup in July 2013. A year later, the
new military leader General Abdel Fattah al-Sisi dropped all charges against Mubarak and his sons. Thus the
Egyptian military only ceded power after attempting to stack the transition in their favor; and when the
elected president ruled without protecting the military’s interests, they took back power.

These two mechanisms—bargaining with civilian opposition parties and credibly threatening, or in some
p. 29 cases executing, a coup—are common features of military behavior in transitions to democracy. Because
military regimes have the capacity to protect their interests in some fashion, they often successfully bargain
with civilian elites in what becomes a peaceful transition to democracy.

In contrast, when domestic unrest erupted in Libya in the spring of 2011, the Gadda regime failed to
bargain with domestic opponents and was unable to secure a safe exile. Instead, the regime aimed to wipe
out the opposition and fought to the end, with Gadda dead and his eldest son imprisoned. Gadda ’s
regime, a prototypical example of a personalist dictatorship, ended violently; and far from bringing a new
democracy, the collapse of Gadda ’s rule left a Libyan state largely controlled by the militias that lled the
power vacuum after the last of Gadda ’s inner circle ed Tripoli.

Most regime changes in dictatorships are carried out by domestic actors. Therefore, assessing common
foreign policy tools requires paying close attention to the internal dynamics and institutional
characteristics of the target regimes. Only military interventions (but not all of them) directly overthrow
autocratic regimes. Other coercive foreign policy tools only destabilize autocracies when they alter the
incentives or capabilities of domestic actors, be it the regime elites or the opposition. And even when foreign
pressure is successful in ousting a dictatorship, this often does not mean democracy will follow.

This chapter introduces the regime classi cation used throughout this study, detailing the internal politics
2
of party-based, personalist, and military regimes. We then explain why we need to understand the survival
of autocratic regimes and not just their leaders. Because many autocratic regimes survive by regularly
rotating executive leadership, changing leaders does not actually destabilize autocratic rule. Instead
executive turnover is a sign of authoritarian resilience.

We then turn to the question of “what comes next when autocratic regimes collapse?” Regime changes may
yield two political outcomes: either transition to a new democracy or to a new dictatorship. We show how
regime collapse and what comes next vary substantially across autocratic contexts, and demonstrate that,
historically, transitions to new autocracies outnumber democratic transitions. Most importantly, we show
that forced regime transitions usually result in a transition to a new dictatorship, while peaceful regime
changes tend to result in democratization.
p. 30
Autocratic Rule

To assess whether international pressure destabilizes dictatorships, we rst examine the domestic factors
that underpin stability in di erent autocratic contexts. Distinct dictatorships pursue particular strategies to
retain power over domestic threats; and because the relative strength of the leader, the military, and the
support party varies across di erent types of regimes, the fate of their rulers and elite supporters after
losing power di ers as well. The institutional structures of regimes—namely the relationships between the
leader, the military, and the party—can explain these di erences. The strategies dictators pursue to retain
power and their expectations about the chances that they retain post-transition power in uence the

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

regimes’ propensity to fall and the manner in which they collapse. Furthermore, as we discuss in Chapter 3,
these di erences allow us to examine these regimes’ distinct vulnerability to alternative instruments of
statecraft.

Varieties of Autocratic Rule

Dictatorships are as di erent from each other as they are from democracies (Geddes, 1999). These
di erences have important political consequences. We focus on distinct types of autocracies that give us
leverage along the following dimensions: how leaders are selected and replaced; the nature of their support
coalition; the degree of the leaders’ control over policy and leadership selection; the strategies they use to
survive in power; and the likely consequences of leaving o ce.

Scholars of dictatorships have o ered several categorizations of autocratic regimes (Huntington, 1968;
Huntington, 1991a; Bratton and van de Walle, 1997; Chehabi and Linz, 1998; Wintrobe, 1998; Geddes, 1999;
Hadenius and Teorell, 2007; Magaloni, 2008; Cheibub et al., 2010). During the height of the third wave of
democratization in the 1990s, comparative scholars turned to studying these transitions, which in turn
spawned even more regime categorizations focusing on hybrid or transitional autocracies—countries that
had the trappings of democracy, but where leaders and parties did not leave o ce in fair and free elections
or where political participation was still severely restricted (Valenzuela, 1992; Karl, 1995; Zakaria, 1997;
Collier and Levitsky, 1997; Levitsky and Way, 2010). In this study, we use a categorization of dictatorships
generated from the question of “who rules?” (Brooker, 2000). The data set we employ distinguishes among
p. 31 personalist, party-based, and military rule (Geddes et al., 2014a). This approach builds on the
sociological literature of regimes and treats the institutional structure of the regime and the preferences of
leaders as central to the task of distinguishing regime types (Janowitz, 1960; Weber, 1964; Huntington,
1968; Nordlinger, 1977; Linz, 2000).

This approach di ers from recent studies that focus on a speci c attribute of autocratic rule to examine
formal political institutions and autocratic survival (Hadenius and Teorell, 2007; Gandhi, 2008; Svolik,
2012). For example, Gandhi (2008) theorizes about the level of institutionalization (parties and legislatures)
and uses military leaders and monarchies as measures of organizational structures upon which a dictator
can rely for support. Svolik (2012) studies how dictators use institutions to deal with two key threats: that
from the masses and that from the elites. These approaches bene t from thinking about a characteristic of
autocratic rule that varies in its extent rather than relying on typologies. Other recent work picks one feature
of autocratic rule, such as the existence of a long-dominant political party or competitive elections in the
post-1989 period, to de ne the group of countries under study (Brownlee, 2007a; Levitsky and Way, 2010;
Bunce and Wolchik, 2011). Our analysis compares distinct types of autocracies throughout the world since
the mid-1940s (or the year of the country’s independence) to allow comparisons both across dictatorships
with di erent institutional structures and across time as the nature of international pressure changes.

While regime typologies attempt to measure important theoretical concepts, they are nonetheless imperfect
proxies, with well-known limitations. For example, the data do not capture variation in the personalization
 of power over the lifetime of a regime (Hadenius and Teorell, 2007). Communist party rule in China and the
former Soviet Union are categorized as party-based regimes, even though consolidation of personal power
undoubtedly varied over time (Svolik, 2012; Weeks, 2014).

The typology we employ nonetheless provides substantial explanatory power. Many of the ndings in this
book suggest that personalist dictatorships stand apart. For example, economic sanctions only destabilize
personalist dictatorships, though this rarely leads to democratic transition. Similarly, military intervention
can increase the likelihood of regime change, but not necessarily in the direction of a new democracy. And
while critics of human rights prosecutions argue that they may entrench autocratic rule, we only nd
evidence of this in personalist regimes. Our results are not alone in ngering personalist dictatorships as

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

distinct from others. Increasing evidence from comparative politics and international relations studies
concurs (Peceny and Beer, 2003; Weeks, 2008; Wright, 2008; Gurses and Mason, 2010; Wright and Escribà-
Folch, 2012; Way and Weeks, 2014).

p. 32 Although separating personalist dictatorships from others helps explain many important outcomes,
employing a typology does not in itself o er an explanation for why di erent dictatorships are distinct from
one another. Because the regime data are coded using descriptive tendencies to categorize regimes
according to the observed institutional structure of the elite in power, they are of little use in understanding
the origins of these diverse settings (Haber, 2006; Slater, 2010).

Our task is less ambitious. While we take the institutional structure of dictatorships as exogenous to help us
understand how external pressure in uences political change, we acknowledge that domestic political
arrangements are the outcome of strategic interactions among domestic actors and have prior causes rooted
in the historical political economy of the country. We employ observable characteristics of the relationship
between the leader and the military and his support party to circumscribe the institutional rules through
which autocratic elites aggregate power and make policy decisions. The regime typologies are simply groups
of dictatorships that share many of the same characteristics.

For our purposes, these categories serve as measures of constraint to help us understand the preferences of
3
elite actors, while fully acknowledging that the latter are determined in part by the former. In using regime
types as a heuristic for institutions, we understand this term in a Northian sense as a set of formal and
informal constraints (North, 1990). This view of institutions allows us to model the systematic behavior of
individuals under a particular contextual constraint when faced with a foreign-generated shock to the
endowments available for dictators to sustain their rule.

Personalist rule
Personalist rule occurs when power is concentrated in the hands of a single individual leader (Geddes,
2003). Early theories of personalist rule focus on patrimonialism to explain how leaders rule without an
ideological or rational-legal source of legitimacy (Weber, 1964; Roth, 1968; Chehabi and Linz, 1998). For
example, Bratton and van de Walle (1997, 62) emphasize that personalist leaders frequently stand above the
law, and thus face few institutional or legal constraints on their behavior. Geddes (1999) notes that
personalist leaders consolidate power over policy and personnel recruitment, which typically means
p. 33 extricating power from military o cers and party leaders. A de ning feature of personalist regimes,
therefore, lies in whether advancement in regime organizations such as the military or party depends on a
personal connection to the ruler.

Government stability in these regimes does not typically rely on power-sharing agreements made credible
through formal institutions, but is based on personal loyalties bought with immediate rewards and material
bene ts, often distributed through an elite patronage network (Snyder, 1992; Bratton and van de Walle,
1994; Chehabi and Linz, 1998). Political loyalty in these contexts therefore depends on the rewards accrued
 from clientelistic practices and not on credible promises about the future behavior of the dictator or access
to decision-making. For example, in the former Zaire, President Mobutu rewarded supporters with
appointments to state-regulated industries; in the Dominican Republic, Trujillo used access to land to
control elites.

This group of regime insiders comprises a relatively small support coalition which often amounts to a group
of individuals with family, ethnic, friendship, or clan ties to the leader. Jackson and Rosberg (1984, 424)
describe this feature of personal rule as a, “system [that] favours the ruler and his allies and clients: its
essential activity involves gaining access to a personal regime’s patronage or displacing the ruler and
perhaps his regime and installing another.” The coalition of supporters may not be durable in the long-

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

term, especially when the rewards of patronage wear thin. While enclaves of personalist rule can persist
across time, they still nonetheless rest on the ability of a particular leader (and/or his or her family) to
reward loyalty with material bene ts (Hutchcroft, 1991).

Personalist dictators seek control of the military by purging senior o cers who may threaten their rule,
replacing them with soldiers who are then dependent on the dictator for their position. In the former Zaire,
Mobutu swiftly executed a number of high-ranking o cers shortly after coming to power, placing
untrained junior o cials in command “either because of their family attachment and personal loyalty, or
because [he] was too incompetent to pose a threat to Mobutu’s continued rule” (Schatzberg, 1988, 61).

To forestall threats from the military and coup-proof their regimes, these leaders not only reshape the
military command, but frequently rotate o cers to weaken the cohesiveness of the military, often by
creating paramilitary organizations (Quinlivan, 1999; Belkin and Schofer, 2003). Continuing with the
Mobutu example, Schatzberg (1988, 59) notes that “[h]igh-ranking soldiers…are well aware of [his]
frequent rotations of o ceholders to ensure that none can build an autonomous base of power.” Mobutu’s
military was too weak to put down internal insurrection in the 1960s without substantial aid from other
countries, particularly the U.S. and Belgium (Callaghy, 1984, 161). In the mid-1990s, what was left of
p. 34 Mobutu’s military disintegrated when Kabila’s rebel army marched across the former Zaire from the East
(Prunier, 2009). After military o cers ousted Emperor Selassie in Ethiopia in 1974, Colonel Mengistu
proceeded to execute rival o cers, starting with the highest ranking Eritrean o cer (Abate, 1983, 35–6).
Trujillo, with the help of U.S. military advisors, created a modern military in the Dominican Republic but
kept personal control by appointing close family members, related either by blood or marriage, to
leadership positions, even refusing to give weapons to some o cers (Wiarda, 1968, 50–1). These examples
contrast with military regimes where the nominal leader rarely has the power to unilaterally alter the
leadership structure and hierarchy of the military, and risks being replaced if he does (Stepan, 1971, 165–7).

In addition, in personalist regimes, parties are frequently the creation of the leader rather than a pre-
existing political organization, such as an independence movement or labor union. And they typically play
little role in helping elites rotate regime leadership. Both Mobutu’s (MPR) and Trujillo’s (PD) parties
disintegrated once they lost power. Instead of constraining leaders, political institutions such as legislatures
or parties in these regimes may simply help them control elites (Lust-Okar, 2005; Wright, 2008); and rather
than elites rotating the leader, the leader rotates elites to prevent them from accumulating power. In the
Dominican Republic, Trujillo at rst forced only two of twelve Senators and nineteen of thirty-three
Deputies to “resign.” In his second term, the Senate saw twelve resignations for thirteen seats and forty-six
resignations for thirty- ve lower house seats. In his third term, thirty-two Senators (nineteen seats) and
122 Deputies (forty-two seats) “resigned” (Wiarda, 1968, 48).
 Military rule
While personalist dictatorships typically lack an organizational form of rule, power in other dictatorships
resides in a collective organization such as the military or a dominant party. Geddes (2003) codes military
regimes using questions about whether the military hierarchy has been maintained, the role of senior
o cers in policymaking, and the rules for leadership succession. Military regimes are thus “governed…with
the support of the military establishment and some routine mechanism for high level o cers to in uence
policy choice and appointment” (Geddes, 2003, 72). Even if the dictator is a former o cer and coup leader,
as is the case in many personalist regimes, the dictatorship is not considered a military regime unless there
is clear evidence that the leader shares decision-making power with other o cers. Hence, a military regime

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

entails rule by an o cer constrained by other o cers, not rule by a military strongman (Geddes et al.,
2014b).

p. 35 Military regimes stand apart from other autocracies because o cers may value the corporate interests of
the military, such as an ample military budget and control over the military hierarchy, more than staying in
power. Nordlinger (1977, 51–3) argues that the military’s corporate interests, especially its preference for
4
autonomy, are due in part to the professionalization of military o cers. Perhaps the most vital corporate
interest is the unity of the military itself, which has implications for their survival in power. Geddes (1999,
140) suggests that “because most o cers value the unity and capacity of the military institution more than
they value holding o ce, they cling less tightly to power than do o ce holders in other forms of
authoritarianism.”

The preference for military unity and against factionalism makes military regimes particularly vulnerable to
elite divisions (O’Donnell and Schmitter, 1986). As a result, elites in these regimes are more likely to bargain
with civilian elites to negotiate a transition to democracy, especially when the military has a credible
guarantee that their corporate interests will be protected (Dix, 1982). One guarantee stems from the
military’s capacity to stage a coup should the new authorities encroach on their interests (Sutter, 1995).
This makes possible extracting institutional arrangements to protect military elites. For example, the
Chilean Constitution of 1980, written by the military regime of General Pinochet, guaranteed ample budgets
for the military and allowed Pinochet to remain as commander of the military, ensuring civilians would not
be able to directly interfere in military appointments (Valenzuela, 1992).

The preference for internal military unity also has implications for how these regimes rule. To prevent
internal con icts, leaders develop power-sharing agreements to rotate leadership and prevent any
individual from consolidating power. Military juntas are therefore frequently composed of leaders from
di erent military branches to keep any one branch from dominating. For example, prior to the Argentine
coup in 1976, o cers reached a deal among all three service branches to ensure sharing of power and
preserve unity within the armed forces (Fontana, 1987, 45–8). In Chile, though Pinochet remained the
president throughout, “a partial separation of powers and the adoption of decision by unanimity set
fundamental constraints preventing any Junta member from dominating” (Barros, 2002, 8).
p. 36 Party rule
The nal autocratic category is dominant party regimes. In these dictatorships, the regime party and not the
leader “controls the career paths of o cials, organizes the distribution of bene ts to supporters, and
mobilizes citizens to vote and show support” (Geddes, 1999, 124). Magaloni (2006) points out that, if these
regimes can dominate politics (and elections), they present themselves as “the only game in town.” For
ambitious politicians, winning power means being part of the party, since the party provides opportunities
for career advancement. Further, party-based co-optation often conditions the bene ts of senior
membership on costly service during early stages of a member’s career, giving junior cadres a stake in
regime survival (Svolik, 2012). Additionally, party organizations can provide members with a durable

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

framework wherein to resolve di erences, bargain and advance their in uence (Brownlee, 2007a). As a
result, dominant party systems are able to generate and to maintain a cohesive and loyal leadership cadre.

Party dictatorships typically have large support coalitions, built and sustained by institutions that provide
information about societal demands that mobilize support for the regime. Institutionalization of the regime
confers credibility to its policy concessions and power-sharing agreements, thus helping it to mobilize both
citizen and elite cooperation (Gandhi, 2008; Svolik, 2012). In an institutionalized forum, information and
demands can be revealed, and factions can reach compromises within the regime. Credible power-sharing,
however, can only succeed when the dictator faces a credible threat of being removed by regime insiders
(Magaloni, 2008; Svolik, 2012; Boix and Svolik, 2013). In post-Mao China, for example, political reforms
helped prevent the concentration of power in one individual by setting term limits for the regime leader,
establishing retirement-age provisions for party leaders, and ensuring that collective decision-making
bodies, such as the Standing Committee of the Politburo, meet regularly.

Parties also operate as patronage distribution networks that mobilize mass support. A dominant party
regime’s grip on power may hinge on its capacity to monopolize and politicize public resources (Greene,
2010). To this end, these regimes distribute perks and privileges by diverting funds from state-owned
enterprises, appointing supporters to public positions in the administration and state corporations, and by
delivering targeted public goods to speci c constituencies. Chhibber (1996, 130), for example, notes that
state intervention in the Algerian economy by the dominant party (FLN) provided the regime “with a
formidable instrument for drawing support from a wide range of economic interests.”

p. 37 While access to public resources for rent-distribution may be essential for dominant parties to survive, their
support coalitions may nonetheless be quite deep. Magaloni (2006), for example, nds that older voters
were more likely to support the dominant party (PRI) in Mexico during the 1990s, which suggests that older
voters, having experienced decades of good economic performance under PRI rule, were more likely to have
enduring loyalties towards the dominant party despite economic decline in the last decade of PRI rule.

The capacity to e ectively deliver patronage rents and credibly promise policymaking in uence helps party
regimes mobilize public support and co-opt rivals and potential opposition leaders. In Egypt, for example,
the rise of a young new business elite in the 1990s posed a threat to Hosni Mubarak’s ruling National
Democratic Party (NDP). This new group at rst sought to create its own party to compete against the NDP.
Instead of forming a new rival party, however, this new group, led by the president’s son, was bought o
with promises of leadership positions within the NDP (Brownlee, 2007a, 133–4).

Finally, the party organization also serves as an instrument of control. The regime extends power through
local party o cials and other territorial branches to ensure control of civil society organizations and
mobilize support. Broad sectors of the population are integrated into the regime through “administered
mass organizations,” or “formal organizations structured and managed by the state’s ruling apparatus to
shape mass social action for the purpose of implementing public policy” (Kazsa, 1995, 218). These
organizations help extend party control into society as they are used to channel material rewards, to
 organize support, o ces and honors, and a self-directed local administration. In the end, these regimes
“typically do not need to rely on outcome-changing fraud or bone-crushing repression to maintain their
rule” (Greene, 2010, 723).

Figure 2.1 shows the number of countries under the rule of di erent types of dictatorship from 1946 to 2010
(Geddes et al., 2014a). The number of autocracies rose in the 1950s and 1960s during de-colonization,
particularly in Africa and Asia. The trend in the number of dictatorships reverses with the start of the third
wave of democratization in the mid-1970s, falling from its peak in 1977 to less than sixty in 2010.

Figure 2.1.

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

Autocratic regime types across time. Smoothed trend line: three-year moving average.

Data source: Geddes et al. (2014a).

The most common type of dictatorship are party regimes; the trend in the number of these regimes follows
the overall trend, rising from 1950 through the early-1970s, then declining in the late 1980s with the fall of
Communism in Eastern Europe. The rise of military rule, particularly in Latin America, begins in the early
1960s. Yet, while the third wave of democratization has seen a marked decline in military regimes, the
p. 38 number of personalist dictatorships has increased. This trend can be explained in part by the fact that
military regimes in the post-Cold War era tend to be very short-lived, while personalist regimes last longer.
Further, many of the post-Soviet dictatorships are personalist, where leaders from the Soviet era
consolidated power. In 2010, personalist regimes were almost as common as party-based ones, with
military dictatorships numbering the fewest.
 Autocratic Regimesʼ Stability

The analysis in this book considers the survival of autocratic regimes. We want to know if a particular
foreign policy tool in uences the survival of regimes and, if so, what comes next after regime collapse:
democracy or a subsequent autocracy. In some cases, deposing the leader will cause the regime to fall as well
because the latter cannot survive without the former. In other cases, however, policies that are linked to a
particular leader leaving o ce may not uproot the regime and will cause little damage to the power of the
p. 39 elite who support the regime. The next section explains how we de ne autocratic regimes and introduces
the two categories of transitions that occur when they collapse: democratic and autocratic transitions. The

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

following section shows how di erent types of autocracies are likely to collapse and demonstrates that this
varies by autocratic regime type. It also shows that the mode in which regimes collapse is highly correlated
with the political outcome of a transition.

Regimes and their Leaders

In 1979 long-standing dictatorships in Iran and Nicaragua collapsed as revolutionaries forced hated
dictators from power. These events entailed much more than simply replacing one dictator with another.
Not only did the ruling families leave power, ee the country, and face the threat of assassination, but
landed business elites and families that supported the Somozas in Nicaragua were left powerless, while in
Iran a secular, foreign-backed monarchy was replaced by a clerical regime.

In both cases, the new dictatorship was, at least at rst, less autocratic than its predecessor. For example, in
Iran’s rst election after the Shah’s ouster, the country selected a secularist president who was also the
Commander-in-Chief of the military. The initial Sandinista government in Nicaragua included opposition
members, such as Violeta Chamorro, on the ruling Council of National Reconstruction and allowed token
opponents to be seated in the Council of State (Williams, 1994, 177). But neither the Sandinista government
nor the clerical regime in Iran were democracies. Revolution in these countries ousted dictatorships and
replaced them with new ones.

These political transitions are fundamentally di erent than democratic transitions because even though
dictatorship collapsed, the new regime was not a democracy. Further, these transitions from one autocracy
to another entailed much more political change than many constitutional (and even unconstitutional)
leadership changes because the incumbent elites lost power and were replaced by an entirely new group of
elites. In contrast, many irregular leadership changes—such as the replacement of one member of the
Argentine military junta with another in the early 1980s or the assassination of Egyptian President Anwar
Sadat in 1981—simply reshu e the leadership atop a dictatorship without substantially altering the group
which holds power.

While many studies examine how domestic and international factors in uence leadership survival, we argue
that this approach is not suitable for understanding how dictatorships remain in power. Because many
p. 40 autocracies rotate leadership among a group of elites, we cannot assume that ousting a particular leader
will necessarily destabilize the regime. That is, we cannot always equate regime and leadership survival in
autocratic contexts.

Autocratic regimes are de ned as a set of formal and informal rules for choosing leaders and policies
(Geddes et al., 2014a). Informal rules are included because autocracies often hide the rules that shape and
constrain political choices, even though dictatorship routinely coexists with many of the formal institutions
seen in democracies. The informal rule central to distinguishing one autocratic regime from another is the
rule that identi es the group from which leaders can be chosen and determines who in uences leadership
choice and policy. To remain in power, regime leaders must retain the support of members of this group, but
 leaders also have substantial ability to in uence the membership of the group, especially after initial
leadership selection.

In some cases the concepts of regime and leader are nearly identical. For example, in many personalist
regimes, we can fruitfully think of the leader as the regime: when one falls, the group of elite surrounding
the leader also tends to lose power. Distinguishing between leaders and regimes in these countries provides
little analytic advantage. For example, Ferdinand Marcos, Gadda , and Mobutu were the sole leaders in their
countries during their regimes. While this is not true of all personalist leaders because sometimes they die
and are replaced by family members, or name a family member as president without really giving up power,
5
these cases are relatively rare.

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

In other dictatorships, however, distinguishing between leaders and the regime itself can aid our
understanding of how groups of elites who support the regime behave. The distinction between leader and
regime is perhaps most important in regimes where elites regularly and peacefully rotate leadership. This
can be done through regularized term limits in regimes with national elections (e.g., Mexico, post-Nyerere
Tanzania), and even in dictatorships without direct executive ballots (e.g., post-Mao China). This type of
regularized leadership turnover is conceptually distinct from regime failure. In fact, regularized leadership
change in these dictatorships is part of the institutionalization of these regimes that aids their longevity
(Magaloni, 2006; Ezrow and Frantz, 2011b). Once the regime is fully institutionalized, there is little
opportunity for leaders to consolidate power in their own hands, and leaders face little prospect of losing
o ce before the end of their term.

p. 41 Consequently, distinguishing between the end of a leader’s tenure in power and regime failure is
particularly salient in countries with dominant parties because leadership turnover is actually a
consequence of a prior factor (institutionalization) which causes both frequent turnover of leaders and the
relative longevity of the regime itself. Thus leadership turnover does not represent instability in these
regimes but is a feature of autocratic survival—that is, the opposite of regime failure.

The Mexican case perhaps best illustrates the distinction between regime and leader. Data on leader survival
typically codes each of the transfers of power between PRI presidents as separate leadership failures. This
means that the transition from Salinas to Zedillo (1994) and the transition from Zedillo to Fox (2000) are
both coded as regular, peaceful, and constitutional leadership transitions. The rst of these leader exits
entailed the continued survival of PRI rule while the latter marked the end of the autocratic period and a
transition to democracy.

Some military regimes can also peacefully rotate leadership. In Argentina, for example, the military
attempted to institutionalize leadership succession, but largely failed in the execution, mixing consensus
alternation of leaders with coup threats and actual coups. Prior to the 1976 coup, senior o cers reached a
deal to share power among the various service branches to ensure military unity after seizing power
(Fontana, 1987, 45–6). This agreement stipulated that the acting president should rst retire from the
military before assuming power. In practice, however, to reach an agreement on whom to choose as the rst
president, o cers agreed that General Videla should remain in the military while still serving as president.
While the pre-coup agreement suggested a “rotating presidency,” the rst transfer of power did not take
place until 1981 amid general crisis environment, at which point the military elite agreed on a consensus
candidate, General Viola. He lasted less than a year and was peacefully replaced with an interim president,
General Galtieri, though the military command had threatened a coup throughout Viola’s brief tenure
(Fontana, 1987, 129). Galtieri was then ousted in a coup following Argentina’s disastrous defeat in the
Malvinas War (1982), but the military regime did not end until Bignone handed power back to civilians in
1983.
 6
Figure 2.2 plots the failure rates of regimes and leaders for each regime category. The failure rate is the
p. 42 number of failures (leader or regime) divided by the number of years at risk of failure. We do not count
natural death or ousters due to foreign invasion.

Figure 2.2.

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

Leader and regime failure rates. All leader failure rates exclude natural death and foreign invasion. Years: 1946–2004 (Goemans et
al.), 1946–2002 (Svolik), 1946–2010 (Geddes et al.).

Data sources: Goemans et al. (2009); Svolik (2012); Geddes et al. (2014a).

All measures indicate that autocratic instability is highest among military regimes. Leader failure occurs in
roughly 20 percent of regime-years. This number drops to 17 percent when we exclude leader exits that
occur at the end of a term limit. The regime itself fails in 13 percent of regime-years. Using regimes instead
of leaders as the unit of analysis thus drops the baseline failure risk by roughly one-third.

Personalist dictators are less likely to fail. Their failure rate is 8 percent, dropping only slightly to 7.5 when
we exclude term-limit failures, and to just under 7 percent when we consider regimes. Whether personalist
leaders or regimes are the unit of analysis, the failure rate does not vary substantially.

Party regimes illustrate the di erence between leaders and regimes most clearly. Leaders in these regimes
fail at a rate of 7 percent—a similar rate to that of personalist leaders. However, if we exclude term-limit
exits, the failure rate drops to just over 5 percent. The regime failure rate, however, is less than 3 percent.
The di erence in failure rates for party leaders and regimes is the starkest, falling by almost two-thirds.
p. 43 That is, the failure rate of leaders in these autocracies is nearly three times the regime failure rate. As we
noted earlier, this di erence is due largely to the regular manner in which party regimes rotate leadership.
Institutionalizing a regime to the point where it can peacefully rotate leaders can help these regimes stay in
power (Ezrow and Frantz, 2011b). If we simply model leader survival in these dictatorships, we would be
treating a mechanism for preserving autocratic power (i.e. leadership turnover) as a marker of instability
when in fact it is often just the opposite.

These patterns demonstrate that regime and leader failure are di erent concepts, and that these di erences
vary systemically by the type of autocracy. When elites have a mechanism for regularly rotating leaders, this
may be a sign of stability. The di erence between leaders and regimes shows up in party dictatorships and
to a lesser extent in military regimes because they are organized through an institution that allows elites to
have a say in leadership selection. Autocratic leaders and their exits from power are easier to identify than
regimes and their failures, and thus make for more transparent units of analysis. But focusing on leaders
 obscures one of the most important features of autocratic politics—namely that some of the most durable
dictatorships survive precisely because they rotate rulers.

Autocratic Regime Collapse

In this section we examine the various threats dictators face. Although recent theories of regime transition
7
focus on popular revolution as the key threat to autocratic rule, dictators are removed from power in a wide
variety ways, with most autocratic leaders overthrown by regime insiders and not by a “revolution from
below” (Svolik, 2009, 2012).

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

As noted earlier, while democratization is often equated with the demise of autocratic rule, it is certainly not
the only outcome that can transpire after autocratic regime collapse. Coups, assassinations, and rebellions
feature prominently in the demise of dictatorships, but not all autocratic exits need be violent. Some
dictators leave power peacefully, for example when they lose an election or when the military negotiates a
peaceful transfer of power to civilians. How dictators lose power and the type of regime likely to replace
them varies considerably across di erent types of regimes. Moreover, the manner in which regimes collapse
and whether the next regime is democratic are highly correlated.

p. 44 Since 1946, roughly half of autocratic regime collapses result in a transition to democracy. The other half of
regime collapses entails one dictatorship replacing another, as occurred in the Iranian and Nicaraguan
8
revolutions in 1979 or in Zaire in 1997. In all these cases, a new (and in some cases revolutionary)
autocratic regime replaced a prior dictatorship—the Shah’s regime in Iran, the Somoza dictatorship in
Nicaragua, and Mobutu’s one-man rule in Zaire.

The left panel of Figure 2.3 shows that the propensity for democracy to follow regime collapse varies
considerably across autocracies. Following the collapse of military regimes, roughly two-thirds of the time
a new democracy emerges. The opposite pattern occurs in personalist regimes: roughly two-thirds end in
transition to a new autocracy. In dominant party regimes, just over half end in a new autocracy. These
patterns give us initial insight into what we might expect to occur if foreign pressure destabilizes a
dictatorship: ousting a personalist dictatorship may not be a reliable strategy for promoting democracy, but
pushing military regimes to relinquish power may be.

Figure 2.3.

Democratic and autocratic regime transitions. Years: 1946–2010.

Source: Geddes et al. (2014a).

 To link foreign pressure to the distinct trajectories that occur after autocratic regime collapse, the right
p. 45 panel examines whether the manner in which a regime collapses is correlated with what comes next after
regime failure. We divide regime collapse events into two categories: forced regime changes and peaceful
ones. The former include foreign invasions, coups, uprisings, and ouster by insurgents in a civil war;
9
peaceful events include elections and rule changes made by regime elites. More than 80 percent of the
autocratic regimes that collapsed as a result of some forceful action were followed by a subsequent
dictatorship, while almost three-quarters of regime collapse events that were not caused by a forcible action
were followed by a new democracy. Finally, when we put the insights from these graphs together, we nd
that, unsurprisingly, regime collapse in personalist dictatorships (85 percent) is most likely to be forced
while in military regimes (43 percent) it is most likely to be peaceful. Party regimes (63 percent) fall in

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

between.

These descriptive tendencies suggest that theories of democratization which posit violent strategies—such
as coups, rebellions, revolutions, and insurgencies—against a dictatorship as the key vehicle of democracy
are at best predicting a relatively small share of democratization episodes, though the unrealized threat of
10
violent may play a role. Further, these patterns indicate that peaceful regime change is the most likely
route to democratization. This suggests that external pressure that foments violent or forced regime change
in dictatorships is unlikely to yield democracy on a consistent basis. Forced transitions usually result from
factional con icts, ethnic rivalries, and protracted group con icts in highly exclusionary regimes where
outsiders have no option of peacefully accessing power.

The data on regime change in dictatorships illustrate why the institutional di erences among regimes
matter for how they collapse and what comes next. Three stylized facts emerge. First, military regimes are
the most unstable; but they are also the most likely to end peacefully and to be followed by a democracy.
Whether this pattern emerges from the preferences of leaders in professionalized militaries, as Geddes
(1999) argues, or from their greater capacity for violence, as Debs (2010) suggests, it has implications for
our analysis because this pattern highlights both how regimes’ survival strategies and the likely post-exit
p. 46 fates of their leaders in uence outcomes. Geddes’ explanation for military withdrawal assumes leaders
in these regimes prefer protecting military unity or other corporate military interests to retaining executive
power. Increasing the costs of remaining in power, or alternatively raising the bene ts of leaving power, can
sway military regimes to hand power to civilians in a new democracy. Debs’ argument focuses on a strategic
decision of military leaders to cede power to democrats because losing power to another dictator is
potentially very costly, as he may try to eliminate military leaders.

Both of these explanations rest on the institutionalization of the regime in the organization of the military
and on the post-exit consequences of losing power. While the professional organization of the military may
engender particular preferences for unity (Nordlinger, 1977), that these regimes tend to end in peaceful,
bargained transitions is largely explained by their capacity to protect their interests after a regime
transition (Huntington, 1991b; Geddes, 1999). The high capacity for organized violence in these regimes
stems from the fact that military organizations can more easily overcome collective action problems to
credibly threaten subsequent rulers and are thus likely to cede power to a democratic leader who is less
likely to eliminate them (Debs, 2010). In short, the military as an organization—whether through the
preferences it breeds or the collective action capacity it enables—can explain why these dictatorships are
not only short-lived, but also the most likely to peacefully transition to democracy.

Second, while personalist dictatorships are more stable than military rule, the former are more likely to lose
power in a coerced transition in which a subsequent dictatorship comes to power. Because rulers in these
regimes principally employ private goods patronage and repression of rivals to survive in o ce, they often
lack institutionalized procedures for dealing with challengers. Consequently, they are less capable of
peacefully selecting new leaders from among the regime elite. This exclusionary logic leaves alienated elites
and regime outsiders without access to rents and other political or economic opportunities. Challengers are
 therefore more likely to resort to violence, such as a coup or rebellion, to oust the incumbent (Snyder, 1998;
Roessler, 2011). As Bratton and van de Walle (1994, 465) argue, “[i]ncumbent and opposition leaders are
usually so polarized as a result of winner-take-all power struggles that there is slim possibility that
moderate factions from either side can negotiate an agreement. Instead transitions unfold along a path of
escalating confrontations until one side or other loses de nitively.” Because they often weaken
organizations such as parties and the military, personalist rulers have less capacity to retain leverage after
regime change. Therefore, negotiated transitions are less likely and violent ousters more likely. Indeed,
p. 47 leaders in these regimes “usually tr[y] to remain in power as long as they can” (Huntington, 1991a, 588).
Or as Bratton and van de Walle (1997, 86) put it, elite supporters in these dictatorships, “face the prospect of
losing all visible means of support in a political transition,” and thus “have little option but to cling to the

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

regime, to sink or swim with it.”

Finally, dominant party regimes are the most stable because they mobilize support from a large coalition
through both institutional co-optation and broad patronage. As a result, these regimes often collapse when
long-term structural processes undermine their capacity to retain support by reducing citizens’ dependence
on state provision of goods and the party’s monopoly over state resources (Magaloni, 2006; McMann, 2006;
Greene, 2007). On the other hand, party-based regimes fall between personalist regimes and military
regimes in their propensity to transition to democracy. “Like o cers, single-party cadres can expect life as
they know it to continue after liberalization or even regime change” (Geddes, 1999, 141). If the former
regime party has a broad and deep support coalition, they can compete in and sometimes win post-
transition elections. Electoral power gives them leverage to protect elites’ interests and prevent bad post-
transition outcomes. For this reason, dominant party dictatorships also frequently end in negotiated
transitions.

Conclusion

This chapter presents the classi cation of authoritarian regimes we use throughout this book and examines
their main characteristics. We distinguish between personalist, party-based, and military regimes. Such
classi cation is based on observable characteristics of the existing relationship between the leader, the
military, and the support party to identify which of them has e ective control over policy decisions and
political appointments.

We also make the case for why we focus on autocratic regimes rather than leaders when analyzing
authoritarian stability: in institutionalized regimes, especially dominant party regimes, the regular
replacement of leaders is an indicator of their organizational strength and stability. In contrast, in
personalistic regimes, the fall of the leader usually entails the fall of the whole regime as well. Our measure
of regime collapse marks a fundamental change in the group of elite who rules and selects the leader.

Finally, we emphasize one key distinction often neglected by scholarly work on regime change: when
authoritarian regimes collapse, the establishment of a democracy does not automatically follow. In fact, one
of two outcomes may transpire, either a transition to a new democracy or a transition to a new dictatorship.
p. 48 Further, how regimes fall and what comes next vary substantially across autocratic contexts: personalist
regimes tend to be forcibly overthrown and be followed by a new autocracy, while military regimes are the
most likely to end peacefully and be replaced by a democratic regime. Hence, the data reveal that there
exists a strong correlation between the type of collapse and the outcome of transitions. In particular, we
show that forced transitions usually result in a new dictatorship, while non-forced regime changes tend to
lead to a democratic transition. This is relevant for the study of foreign pressure because some foreign
policy tools, by increasing the likelihood of a forced regime change, may not promote democracy.
In the next chapter we develop a theoretical model of regime transitions that incorporates these stylized
facts and accounts for both types of political outcomes. We also discuss two elements in which regimes
di er that are of particular importance to understanding regime transitions and the in uence of foreign
pressure: the strategies dictators use to retain power, and the post-transition fate of their leaders and elites.

Notes
1 David Kirkpatrick, “Egyptʼs Military Expands Power, Raising Alarms,” The New York Times, 14 October 2011.
2 We exclude monarchies from the analysis because these dictatorships tend to be highly stable and we therefore have less

Downloaded from https://academic.oup.com/book/8804/chapter/154964950 by Universitat Pompeu Fabra user on 13 April 2024

room to examine how foreign pressure influences their survival. In fact, there is only one monarchy, in Nepal, that
transitioned to democracy in the post-war period. See Menaldo (2012) on the durability of monarchies in the Middle East
and North Africa.
3 See Roeder (1993) for an analysis of how institutions operate to constrain elite behavior during transitions.
4 As Bellin (2004, 145) points out, the institutionalization of the coercive apparatus in an autocratic regime “is rule-
governed, predictable, and meritocratic. It has established paths of career advancement and recruitment; promotion is
based on performance, not politics; there is a clear delineation between the public and private that forbids predatory
behavior vis-à-vis society; and discipline is maintained through the inculcation of a service ethic and strict enforcement of
a merit-based hierarchy.” These qualities build on the Weberian concept of institutionalization that contrasts with
patrimonialism.
5 A er François Duvalier died in Haiti, his son replaced him as regime leader. Other personalist rulers change the o icial
head of state even while maintaining power behind the scenes. For example, Rafael Trujillo named his brother President
of the Dominican Republic in 1952 and Anastasio Somoza García installed his uncle, Víctor Manuel Román, as puppet
president of Nicaragua a er the 1947 coup.
6 The leader data is from Svolik (2012) because this source records whether the leader exited o ice when an executive term
limit expired. We recoded the data to account for leaders who leave o ice a er the end of the regime, as defined in the
regime data. For example, we account for the fact that the regime data codes the end of military rule in Chile in 1989, while
the leader data codes Pinochet stepping down in 1990.
7 See, for example, Boix (2003) and Acemoglu and Robinson (2006).
8 Subsequent regimes coded as “ceasing to exist” or “failed state” are classified as transition to subsequent dictatorship,
except East Germany (1989) and the Soviet Union (1990) which are treated as transitions to democracy. We exclude regime
failures that end in occupation by a foreign military (e.g. Iraq 2003).
9 While we use the term forced regime change to categorize the use of force during the regime collapse event, a separate
variable in the data from Geddes et al. (2014a) codes the number of deaths that occur during the regime collapse event.
Thus some events which we include under forced regime change do not entail deaths, such as bloodless coups.
10 Formal theories of democratization in Boix (2003) and Acemoglu and Robinson (2006) rely on the credible threat of
revolution against a dictator, not necessarily a realized revolution, to yield predictions about the likelihood of
democratization. Our model of regime change in the next chapter incorporates this possibility as well.