View metadata, citation and similar papers at core.ac.

uk brought to you by CORE

Institute for International Integration Studies provided by Research Papers in Economics

IIIS Discussion Paper

No.209 / March 2007

Presidential Succession and Democratic Transitions

Alexander Baturo

Political Science Department, Trinity College Dublin

 IIIS Discussion Paper No. 209

Presidential Succession and Democratic Transitions

Alexander Baturo,
Political Science Department, Trinity College Dublin

Disclaimer
Any opinions expressed here are those of the author(s) and not those of the IIIS.
All works posted here are owned and copyrighted by the author(s).
Papers may only be downloaded for personal use only.
 Presidential Succession and Democratic
Transitions
Alexander Baturo
Lecturer
Trinity College Dublin
Political Science Department
2-3 College Green, Dublin 2, Ireland
Email: [email protected]
Phone: +353-1896-3483

February 25, 2007

Abstract

Why might presidential succession in partly- and non-democratic regimes render

the probability of democratic transition more likely? Many presidential regimes in
developing world are highly personalist and their stability depends on the strength
of their rulers. Transitions are often initiated and driven by elite splits, and the
process of presidential succession triggers these splits and uncertainty along the chain
of command. Building upon previous work on liberalizing elections (Howard and
Roessler 2006), I find that presidential designated successors lose elections more often
than the long-standing incumbents, which increases the probability of democratic
change, since the former compete against the pro-democratic opposition in a recent,
1990-2004 period. I also find that the presence of hegemonic parties mitigates these
effects.

I would like to thank Marc M. Howard and Philip Roessler for generously sharing their data on
liberalizing electoral outcomes with me. I also wish to thank Ken Benoit, Jos Elkink, Julia Gray, Slava
Mikhailov and Matthew Wall for their insightful comments.
Introduction

After nearly two decades in office, on 7 January 2001 President Rawlings of Ghana stepped

down, passing power to the pro-democratic opposition after his designated successor and

Vice-President Atta Mills lost a second round of the December 2000 presidential election,

despite the support of the ruling party, military, media and the outgoing president himself.

One of the most important elements of this contest was that the charismatic incumbent

Rawlings who won two elections in the past did not run due to term limits (Gyimah-Boadi

2001, 107). Indeed, in the 2000 elections ‘a great wave of excitement swept the country

with President Rawlings no longer on the ballot paper, and the scent of change in the air.’1

The hopes of the Ghanian electorate were not unfounded, as Rawlings’s successor indeed

lost, and, following the election, Ghana’s Freedom House ranking was changed from ‘Partly

Free’ to ‘Free’. Does presidential succession in partly- and non-democratic regimes render

the probability of democratic transition more likely and why?

Throughout history, the failure to achieve smooth transitions of power brought havoc

to the governments. The need to transfer power has been an important component of

promoting the rise of new political patterns (Burling 1974, 2). Political stability, long-

term economic policy and growth, international alliances, and, as this paper argues, the

very success of transitions to democracy are affected by political decisions that rulers take

during succession and how they depart from power.

In this paper I focus on one important aspect of transition in presidential regimes: the

effects of presidential succession on democratic change. I investigate whether successors are

more likely to lose elections than the long-standing incumbents and whether it affects the
1
IRIN. 2004, November 26. ‘Kufuor Likely to Be Re-elected in Ghana’.

2
probability of democratic transitions. When presidents the world over present themselves

for re-election, they are more likely to win. But when incumbents step down, very often

they designate their successors in their stead. They can select from the ranks of their own

party (Mexico under PRI), their relatives (Azerbaijan 2003, Syria 2000), public officials or

people they share business interests with (Ukraine 2004). But they often do so at the peril

of their regime.

I argue that, when presidents appoint successors in nondemocratic or partly democratic

regimes, the uncertainty in the ruling coalition over whether rents will continue to be

distributed as well as over whether designated successors are strong enough to deter or

punish potential challengers often causes elite splits. This, in turn, makes an opposition

victory more likely. Transitions to democracy almost always result from splits within

authoritarian regimes, as many researchers have noted (Di Palma 1990; Karl 1990; Kitschelt

1986; Meyer 2004; Przeworski 1991; Stepan 1986). I test this argument empirically, focusing

on 90 elections in partly- and non-democratic presidential regimes, from 1990 to 2004. I

find that longstanding incumbents obtain more votes and are less likely to lose elections

than their designated successors, and that democratic transitions occur more often when

incumbents step down and successors run in elections in their stead. I also find that the

presence of hegemonic parties somewhat reduces the handicaps of successors.

The findings of this paper suggest that defeats at so-called ‘stunning elections’ are

driven not only by the opposition’s mobilization, but also by the absence of incumbent

that reduces the costs of the collective action. This follows the work of Geddes (1999) and

Huntington (1991), who have argued that the absence of a leader due to the incapacity

or death of a personalist ruler often precipitates elite splits and eases collective action

problems within the opposition. The broader implication is that democracy practitioners

3
should focus their attention not only on free and fair elections, but also on the executive

tenure restrictions and the observance thereof.

This paper proceeds as follows. First, I elaborate on the process of presidential succes-

sion and why it affects democratic transitions in presidential regimes in developing world.

To put it simply, I expect incumbents to win, and designated successors to lose. To put it

simply, I expect incumbents to win, and designated successors to lose. Next I discuss how

I build upon the data on liberalizing electoral outcomes (Howard and Roessler 2006) and

present the results of logit and OLS estimations and discuss the impact of incumbency on

the probability of democratic change. I then focus on the outlier cases in which incumbents

ran and lost or successors ran and won in order to illuminate the causal mechanism and

see whether theory holds under adverse conditions.

Succession, Incumbency and Democratic Transitions

The most dangerous time in office for the presidents is the first one or two years in office

(Bienen and Van der Walle 1991). Earlier studies on political survival found that leaders,

once they survive the initial stage in office, have an incumbency advantage over their

challengers (Bueno de Mesquita et al. 2003; Bienen and van der Walle 1991). Likewise,

Cheibub and Przeworski (1999) found that among presidents who faced elections without

impending term limits only 27 per cent were defeated, and the rest either won them or

did not run. It seems that incumbents are more difficult to defeat in elections than other

electoral candidates. Incumbent reelection goes beyond the personal concerns of presidents

and their challengers, however, and has larger implications for the fate of the country’s

political system.

4
 Democracy is the only form of government that is designed to arrange for an orderly

succession of leaders. In contrast, when a dictator dies, very often ‘war of all against all’

follows and the strongest contender emerges as a new dictator. The difference between suc-

cession in dictatorships as described by the public choice literature (e.g., Kurrild-Klitgaard

2000; Tullock 1987) and succession in modern presidential regimes lies in the fact that the

latter is legitimated by elections. Often described as electoral authoritarian regimes (Lev-

itsky and Way 2002; Ottaway 2003; Schedler 2002), these, unlike democratic regimes, in

which ex ante electoral uncertainty is one of the inherent features of democracy (Przeworski

and Limongi 1997), are designed to prevent electoral defeats through fraud or intimidation

of the opponents (Schedler 2002, 37). When incumbents step down for whatever reason,

elections are supposed to facilitate an orderly succession by conferring legitimacy on the

new ruler. These regimes oscillate between democracy and dictatorship, and can gravi-

tate to either, or remain in between. The outcomes of succession impact on their future

democratic development.

The issue of succession is the most severe problem for dictatorships throughout recorded

history and the hereditary succession is the most stable form of succession for dictatorships

(Egorov and Sonin 2005; Kurrild-Klitgaard 2000; Tullock 1987). A dictator cannot appoint

his successor in advance without undermining his own authority. If it were, everyone would

coordinate around the latter and then the first position would be endangered (Burling

1974).

In the majority of cases, the inherent uncertainty of succession triggered infighting and

political instability. When Kenyan President Daniel arap Moi decided not to participate

in the 2002 elections and obey term limits, he designated Uhuru Kenyatta as his successor.

The absence of incumbent running made the president himself and the regime insiders

5
less motivated to spend large amounts of money on the electoral campaign and led to

splits within the governing party, KANU, among those who wanted to choose their own

presidential candidate and those standing by the incumbent’s choice (Howard and Roessler

2006; Throup 2003). Presidential succession in Zambia was even messier, with two former

vice presidents — Godfrey Miyanda and Genenarl Christon Tembo — running against

another former vice president, Mwanawasa, who was selected as the governing party’s

presidential candidate for the 2001 elections (Burnell 2001). Similarly, despite appointing

his Prime Minister as his successor in Ukraine in 2004, President Leonid Kuchma failed to

transfer his enormous powers to the former and remained in firm control until his departure.

The united and strong opposition increased the costs of suppression, while regime elites

became split between supporters of the designated successor and those that hesitated up

until the last moment. This ultimately led to a democratic breakthrough in Ukraine.

Thus, I put forward the following hypothesis:

• H1 : Successors are more likely to lose elections than incumbents. The corollary of this
hypothesis is that I expect successors to gain lower number of votes than incumbents,
all things being equal.

In the absence of strong institutions in personalist presidential regimes, rewards and

the access to rents depend on the access to a president, in the words of Robert Bates (2001,

72), “the fountain of privilege.” If some elite groups are not confident in their continuing

access to rents under a new ruler, or unsure of a new ruler’s ability to provide rents, they

are likely to designate an alternative candidate instead, or even place bets on several likely

winners.

However, the very uncertainty of elections creates focal point for societal coordination

6
and enables the horizontal communication that is needed for the collective action (Barzel

2002). In turn, the latter often produces the so-called ’stunning elections’ phenomenon,

in which a suddenly mobilized opposition inflicts a surprising defeat on a non-democratic

regime (Huntington 1991, 175-80; Markoff 1996, 113-4; Thompson and Kuntz 2004).

The absence of a long-standing ruler compounds uncertainty and increases the prob-

ability of democratic change through the following process. Firstly, rulers are at their

weakest in the initial period (Bienen and Van der Walle 1991), when they still have to

prove their strength, neutralize challengers, accumulate resources for their coalitions and

gain experience in office. Designated successors are about to live through this dangerous

period and hence are more vulnerable than longer-serving rulers. Secondly, in the absence

of strong institutions in personalist presidential regimes (later on I also check whether

hegemonic parties ease the costs of succession), rewards and the access to rents depend on

the access to a leader (Bates 2001; Geddes 1999). If some elite groups are not confident in

their continuing access to rents under the new ruler, or unsure of a new ruler’s ability to

provide rents, they are likely to designate an alternative candidate instead. Thirdly, am-

bitious politicians could feel they are capable of defeating a weaker candidate and gaining

office for themselves, something that they would not dare under the old ruler. For exam-

ple, politicians that failed to obtain party presidential nomination represented very serious

threat to hegemonic party regime in Mexico (Magaloni 2006, 258). In turn, elite splits will

compound the electoral uncertainty and increase chances for regime’s defeat. Finally, elite

splits and succession could affect the chances for democratic change indirectly, via easing

the collective action and breaking the apathy of a population through the novelty effects

of new candidates.

Why would this process necessarily lead to a democratic transition? The incumbents or

7
their designated successors, as representatives of the status quo (stability of partly- or non-

democratic regime), run against the opposition that, in turn, challenges this status quo.

Most of the time, and especially after the end of the Cold War this challenge came from the

groups espousing democratic values rather than socialist, nationalist or pro-independence

creeds, like they did in earlier periods (McFaul 2002; Przeworski in Munck and Snyder

2007).2

If designated successors are more likely to lose, then the absence of incumbents can

be seen as the cause of democratic change during elections. However, if rulers step down

because they simply cannot run, then their absence is the result of increasing regime

democratization rather than its cause. While some presidents do not face restrictions on

the number of terms they can serve (Syria, Egypt), the majority face term limits and have

to step down. However, many term-bound presidents extend their terms and proceed to win

elections: for example, in Uganda in 2005, Belarus in 2004, Gabon in 2003 or Togo in 2002.

These rulers did not designate successors, elite splits were absent and democratic change

did not occur. If these rulers had stepped down and initiated succession, the likelihood of

transition would have gone up only because of their absence, as argued above.

The absence of incumbents, however, is also a manifestation of the overall regime liber-

alization. Presidential succession causes elite splits but it also, as described by an observer

of the Ghanaian politics in the beginning of this paper, changes the expectations about the

likely electoral outcomes and the overall political environment. On average, the Freedom

House civil liberties index in a two-year period before an “incumbent” election is 5.1 and
2
McFaul (2002) proposed that the origins of democracy as the ideology of opposition could be under-
stood if one took into account the balance of ideologies in the international system at the time (1980-90s).
Likewise, Adam Przeworski contemplated that after the disastrous experience with the military dictator-
ships in Latin America, the opposition (the left) came to realize the importance of democracy as the value
in itself (Przeworski cited in Munck and Snyder, 2007 forthcoming, 24). Similar argument was advanced
by Castaneda (1993).

8
with “successor” it is 4.2, where higher values represent less open environment. If ruler

steps down, he or she does not subvert the constitutional process by extending the term,

which also contributes to overall liberalization, for which I account in the empirical section

of this paper. It is also in the interests of the departing ruler to leave office for a more

democratic setting where his interests as a private citizen are not at the mercy of the all-

powerful ruler to be. In this sense, while the absence of incumbent is both the cause of

the electoral breakthrough and the manifestation of the overall democratization, it is the

multifaceted process of presidential succession that causes democratic change directly and

indirectly.

This leads us to a further hypothesis:

• H2 : The fact of successors running in elections (executive turnover) increases the

probability of democratic transitions. Since successors are more likely to lose than
incumbents, and since they are most likely to lose to the opposition groups espousing
democratic ideas, the fact whether an incumbent runs or not affects the probability
of successful democratic transition.

What conditions might allow succession to take place in a more mannered fashion? In

the first half of the 20th century, only a few rulers managed their orderly succession in

developing world (Herz 1952). However, succession does not always have to be unruly.

Sometimes it proceeds smoothly as a successor consolidates his or her power and becomes

a new incumbent (e.g., Russia 2000, Syria 2000, Azerbaijan 2003). The paradigmatic case

of a long-standing regime with an orderly presidential succession is Mexico. Indeed, while

Mexico has remained non-democratic for most of the 20th century, its presidents were

replaced every six years in Mexico. Castaneda (2000) described how Mexican presidents

chose their successors in a process called the dedazo (the finger tap): the former picked

9
the ruler to be at his will, so that the incumbent was the only ‘voter’ deciding on the next

president.

If the presidents possessed such broad powers over succession, why did not they remain

incumbents and dedazo themselves instead? The 1917 constitution banned presidential re-

election, and when president Alvaro Obregon (1920-24) avoided this ban, winning a second

term, he was subsequently assassinated in 1928. Following this, the Mexican elites estab-

lished a new umbrella party, PRI, as a collective agreement to prevent one single individual

from becoming dictator (Hall 1990). The assassination of Obregon created a powerful focal

point for coordination against possible reelection attempts in the future (Magaloni 2006,

8). The all-powerful party, combined with the specific institutional agreement, placed in-

surmountable constraints on the executive should the latter decide to prolong his rule,

lowered the stakes of losing office, and, by making regime elites into stakeholders of the

status quo, largely prevented the splits during succession.

Brownlee (2004) focused on regime stability and breakdown as functions of elite unity

and defections, which, in turn, depended on whether ruling political parties were capable

of mediating inter-elite conflicts. Magaloni (2006) and Smith (2005) explored the determi-

nants of stability of hegemonic party regimes and found that hegemonic parties mediated

elite conflicts and perpetuated these regimes beyond the lives of individual rulers.

We can state these propositions in our last hypothesis:

• H3 : Hegemonic parties reduce probability of democratic transition during presidential

succession.

To sum it up, elite splits are almost always precipitate transition to democracy. Pres-

10
idential succession often generates these splits, as elites are less likely to split under a

long-standing ruler than under designated successor. The fact whether incumbent or his

successor runs in elections affects the probability of democratic change: successors are

more likely to lose than incumbents and they do so to the pro-democratic opposition. I

also expect that the presence of hegemonic parties mitigates these effects.

Domestic and International Determinants of Transitions to Democ-

racy

I define incumbents as presidents that have been in office for at least one year prior to elec-

tions. Successors are presidential candidates from the same party as presidents, relatives,

or all those designated as successors and/or campaigned for by the departing presidents.

In all but three elections analyzed in this paper I was able to identify designated succes-

sors. The exceptions are cases in which incumbent presidents were ousted in coups or

were forced to resign and thus neither participated in elections nor designated successors.

Prime Ministers, Vice Presidents or other officials of the incumbent regime that run for

the presidency are defined as successors.

In order to investigate the impact of incumbency on the probability of democratic

transitions, and to control for other determinants, I build upon a model from a recent paper

by Howard and Roessler (2006) that investigates the occurrences of so-called liberalizing

electoral outcomes in competitive authoritarian regimes.3 Their paper takes into account

both the actor-driven and structural parameters that are likely to influence democratic

transitions. Crucially, one of the parameters in their model is the incumbent turnover —
3
I would like to thank Marc M. Howard for generously sharing these data.

11
whether an incumbent is present at elections or not.

Their sample contains 50 elections held in parliamentary and presidential electoral

autocracies in the period of 1990-02. Howard and Roessler operationalize competitive

autocracies by excluding those countries that received Freedom House (henceforth FH )’s

score of 2 or lower (democracies) — indeed, democracies do not need to democratize —

on the one hand, and those with the worst FH ’s score of 7 (‘closed’ autocracies), and also

those where the winning party or candidate received over 70 per cent of the vote, on the

other hand (ibid, 368). That is, the authors focus on competitive, and exclude hegemonic

authoritarian regimes. They also exclude foundational elections. Some of the elections

occur in the same country twice or even three times. The dependent variable is liberalizing

electoral outcome (increases in the levels of democracy scores in election year: at least by

1 on FH score and by 3 on Polity score).

I build upon this dataset in the following manner. Firstly, I exclude elections in non-

presidential regimes (e.g., Albania, Singapore and Malaysia). Secondly, to the 35 presiden-

tial elections held in 1990-2002 I add 55 more elections from 1990-2004. The modified data

include 90 elections in all presidential regimes, even those with incumbents gaining more

than 70 per cent of votes. The sample now includes hegemonic authoritarian regimes, such

as Egypt, Uzbekistan or Tajikistan. The reason behind the inclusion is twofold. Firstly,

the opposition might boycott certain elections, which would inflate executive vote shares

but does not necessarily render such regimes uncompetitive. For example, in 1996 elec-

tions in Zimbabwe the incumbent Mugabe gained 92.8 per cent after two main contenders

withdrew from the race in protest. Later, in the 2002 election, Mugabe gained 56.2 per

cent, even though the FH democracy scores deteriorated from 5 in 1996 to 6 in 2002.

Secondly, the 70 per cent cut point is rather arbitrary. Kyrgyzstan was not included in

12
Howard and Roessler’s (2006) study, yet President Akayev, who gained 74.5 per cent in

2000 elections (above the 70 per cent threshold), was subsequently ousted after the 2005

elections in a splendid example of a liberalizing electoral outcome. The presidential succes-

sion of 2003 in Azerbaijan nearly resulted in another liberalizing electoral outcome despite

its high executive vote margins.

These broader criteria, however, leads to the inclusion of several non-competitive plebiscite

“elections” (Syria, Egypt), where the chances for democratic breakthrough are minuscule.

I additionally estimate the sample of elections below 75 per cent margins separately. Also, I

match the executive turnover with leaders at the time of elections and identify incumbents

and their designated successors. There are only three elections with both incumbents and

their designated successors absent (following military coups). I examine every election and

report the name of the incumbent and successor (Gleditsch, Goemans and Chiozza 2006),

as well as their electoral performance. Variable definitions and data sources are described

in appendix.

Thus, I re-estimate and re-specify the Howard and Roessler (2006) model, focusing

on succession in presidential regimes, and add hegemonic party and international democ-

racy diffusion variables. I also estimate an OLS regression, with votes cast for incum-

bent/successor as the dependent variable and using a number of new parameters that are

likely to affect the vote.

The liberalizing electoral outcome is a function of whether a successor or an incum-

bent runs (Successor ), opposition unity (Opposition coalition), the average number of

anti-governmental demonstrations in year prior to election and election year (Opposition

mobilization), economic factors (Growth), international and global factors (FDI flows and

13
Foreign Aid per capita, both averaged as above), Democracy as the regime’s average FH

civil liberties score for the two years prior to the election; Prior Liberalization: whether

regimes experienced changes on democracy score 5 years prior to elections. I refer the

reader to Howard and Roessler (2006) article for a more detailed discussion (ibid, 370-

74). Their major explanatory variable is the Unity of the Opposition. Thus, the model

includes actor-centered parameters, as well as economic and political structural factors and

is well-suited to test the impact of presidential succession on democratic transition while

controlling for other parameters.

The diffusion of ideas and policies across borders is a topic that has been of concern

to social sciences in the past several decades and has recently been tackled by political

methodologists.4 In a recent paper on diffusion of democracy, Gleditsch and Ward (2006)

employ a battery of parameters to measure diffusion, such as the proportion of democracies

locally and globally, as well as democratic transitions locally.

To account for the demonstration effects of ‘stunning elections’, I add two new vari-

ables: Regional Liberalization and Regional Transitions. I count the number of liberalizing

outcomes in the region, 3 years prior to election.5 . Clearly, the opposition’s efforts to

mobilize and win elections in one country were triggered and mobilized by the success-

ful efforts in other countries; the opposition activists traveled and shared their methods

and practices, international donor programs were designed according to the successfully

implemented ones, etc.6 (Carothers 2003; McFaul 2005).

[FIGURE 1 IS ABOUT HERE]

4
Beck et al. 2006; Cederman and Gleditsch 2004; Darmofal 2006; Franzese and Hays 2006; Gleditsch
and Ward 2000, 2006; O’Loughlin et al. 1998.
5
Regional groups are West Africa, Middle Africa, South Africa, East Africa, Middle East and North
Africa, CEE/NIS, Central America, South America, and Asia
6
Washington Post. 2003, November 25. Tbilisi’s “Revolution of Roses” Mentored by Serbian Activists:
Foes of Milosević Trained Georgians, (by Peter Baker) Page A22.

14
 Figure One is a two-by-two matrix that plots the occurrence of democratic transitions

depending on whether incumbent or successor is running in elections. Four elections in

quadrants are included for illustrative purposes. In 67 elections, or 74 per cent of cases,

incumbents run in elections, and in 23 elections (26 per cent) incumbents do not run for

one reason or another. As can be seen, when an incumbent runs, a liberalizing electoral

outcome occurs in only 12 per cent of elections (8 cases) in presidential regimes (1990-

2004). In contrast, democratic breakthrough occurs in 52 per cent of elections (12 cases)

in which an incumbent does not run.

Comparing vote shares, we can see that even if incumbents lose, they still gain 7 per

cent more than losing successors, and when the former win, they perform better by 11

per cent, than winning successors. Overall, incumbents obtain 68.6 (18) v. successors’

49.9 (21) per cent of popular vote in the first and/or only round. This difference could

be attributed to the fact that successors are weaker and thus gain fewer votes: because

they cannot commit electoral fraud on a sufficiently large scale; or they lack resources to

rally and buy their supporters or to provide goods for their winning coalition; or because

of unusually high international pressure.

While the size of fraud is difficult to estimate, the Database of Political Institutions,

DPI (Beck et al. 2001) includes a variable coded as 1 if allegations of fraud were reported.

On average, incumbents win 71.5 v. 66.1 per cent in fraudulent v. ‘clean’ elections, while

successors gain 49 in ‘clean’ and only 43.8 per cent in fraudulent elections. While we

should treat these data with caution, it could also indicate that successors are less adept

at manipulating the ballot when there is uncertainty along the chain of command.

15
Model Estimation and Discussion

Models 1 – 4 are specified as logit regressions, with a binary dependent variable denoting

democratic change or not. Model 5 is an OLS regression with per cent of votes cast for

the executive (incumbent or his successor) in first and or only round (Beck et al. 2001) as

the dependent variable. Models 1 – 3 are estimated on the same estimation sample. All

four models are specified with robust (cluster-adjusted) standard errors. Unit of analysis

is election.

The first, ‘base’ model follows the specification in the original Howard and Roessler

(2006) article, extending the sample to 90 elections in presidential regimes, with robust

(cluster-adjusted) standard errors. Like in their estimation, the unity of the opposition,

incumbent turnover and mobilization have statistically significant effects. In fact, in all

three logit models (1 – 3), when Successor runs instead of incumbent, the probability

of democratic change increases. Likewise, when the opposition is united, the probability

increases in all three specifications. Economic growth exerts a negative influence on demo-

cratic transitions: economic performance helps to sustain regimes in power the world over

(Bueno de Mesquita et al. 2003, 302-9). Likewise, low levels of democracy reduce the

chances for change — no doubt, democratic breakthrough was much nigher in Ukraine of

2004, with democracy levels of 4 (on a seven-point scale, with the 7 being the worst score),

than in neighboring Belarus with its score of 6.

[TABLE 1 IS ABOUT HERE]

Models Two and Three are estimated on the basis of the same sample of 90 elections in

presidential regimes (1990-2004). In Model Two I test the impact of the hegemonic party

on electoral outcomes. Clearly, the presence of the hegemonic party reduces the chances

16
for democratic change. The effects of Successor remain significant, however, whether the

former runs in a hegemonic-party regime or not. In Model Three I add two ‘diffusion’ pa-

rameters, indicating overall regional levels of democracy and instances of democratic tran-

sitions in the region. Clearly, demonstration effects do seem to be at work: the Regional

Transitions variable is positively significant, indicating that successful ‘stunning’ elections

in the region affect the probability of similar elections nearby. Interestingly enough, pre-

ceding transitions in the region seem to cancel the effects of preceding mobilization in the

country — the effects of this variable become insignificant. The overall degree of Regional

Liberalization does not affect democratic change, however. I perform the likelihood-ratio

test on these three models and the results indicate that the prediction is improved with

each new specification. Model 4 is estimated on a more ‘democratic sample’ – elections in

which winners obtain less than 75 per cent. As can be seen, the latter estimation produces

results very similar to the more inclusive models.

Finally, I employ OLS regression (Model Five) in order to test whether successors are

expected to obtain fewer votes. In this model I include all variables from Model Two

apart from Prior Liberalization. If binary models are capable of predicting whether regime

elites win or lose, the OLS model with a similar specification should predict votes cast

for the incumbents/successors. If the latter obtain less than 50 per cent in majoritarian

elections, they lose. I expect successors to obtain fewer votes, the unity of opposition

should decrease votes cast for regime representative, growth should improve vote shares,

and lower democracy scores should be associated with the larger shares.

I also included several variables that are expected to influence votes cast: Political

Constraints (Henisz 2002), Fraud (Beck et al. 2005) and vote cast for the executive in

previous elections (Previous Vote). Political constraints make it more difficult to commit

17
fraud and amass large margins. They also account for the overall institutionalization (yet

they also are correlated with Democracy). Reported fraud indicates that votes margins

are probably inflated. Votes cast in the past election are included for two reasons: to

control for the typical executive vote in a country as an additional control for the degrees

of democracy, and also because regime elites coordinate election returns on the basis of

past performance so as not to look weaker than in the past. The ratio of votes cast to

votes cast in previous election is 1.03; for incumbents it is 1.1 and for successors it is 0.85:

overall, the executives tend to gain more in each consecutive elections, while successors

gain less than the outgoing rulers. Simpser (2005) and Magaloni (2006) explain possible

reasons behind amassing large margins of victory: to discourage opposition supporters

from voting, establish and signal a long-term dominance and co-opt the opposition. Large

margins can also result from the competition between regional operators as to who could

deliver the largest vote, etc.7

Results of Model 5 indicate that Previous Vote affects the proportion of votes cast for

incumbent at present, Democracy is also highly significant: more dictatorial rulers do gain

more votes. Economic Growth is also conducive for gaining more votes in presidencies

throughout the world. Likewise, the fact of Successor running decreases the vote share

received by the representative of the regime.

[FIGURE 2 IS ABOUT HERE]

Figure 2 visualizes results. I plot the predicted values of democratic transition during

elections (liberalizing electoral outcome, LEO) and Democracy (Freedom House score, 2
7
In 1999 elections in Kazakhstan, regional governors responsible for rallying the voters to polling stations
and ensuring required turnout, apparently competed against each other on who would be able to ‘deliver’
the highest vote for the president. As a result, Nazarbaev gained unprecedented 91.15 per cent of vote
cast. One journalist rather appropriately called him ‘Nursultan the Ninety-Two Percent.’

18
years prior to elections, averaged), weighted by votes cast for incumbent/successor. Points

filled with colour represent observations where democratic transitions actually occur. The

graph shows that the probability of transition is lowest for big-vote-winning incumbents,

they also tend to be more authoritarian. In both categories of incumbents and successors,

there is a slight evidence of a mild negative relationship between levels of Democracy and

probability. Evidently, the model predicts that incumbents will succeed.

Finally, holding all independent variables at their means, I calculate first differences

to assess the influence of changes in parameters of interest on the probability of demo-

cratic breakthrough (Model 3) and calculate expected values for votes cast for the regime

representatives (Model 5).8 Figure 3 plots the predicted changes in probabilities and val-

ues. The lines represent 95 per cent confidence intervals, and all included parameters are

statistically significant.

[FIGURE 3 IS ABOUT HERE]

The overall probability that the dependent variable takes on the value of 1 is 8 percent

(s.e. = 4) holding all predictors at their means. I find that all things being equal, the

probability of transition increases by 34.8 per cent (s.e. = 15) when the incumbent does

not run. Interestingly enough, when a successor runs in the hegemonic party regime, the

former still increases the chances for change, but with less magnitude: by 13.3 per cent.

Also, the shift from the closed authoritarian regime to electoral democracy (from 7

to 2.5 on Democracy scale in 2 years prior to election) results in the drop of probability

of transition by 24 per cent. This is a rather extreme counterfactual scenario, as such

significant improvement can hardly occur. Note, however, that Benin democratized very
8
See King et al. 2000; Tomz et al. 2003.

19
rapidly: from FH = 7 in 1989 to 2.5 in 1991. There is no surprise that during this

opening the incumbent president Kerekou suffered defeat at elections. Likewise, successful

democratic breakthrough in the region gives the boost for democratization in this region

by 5.5 per cent, but 3 successful ‘stunning’ elections in a row increase it by 31 per cent.

Indeed, electoral revolutions in Slovakia 1998, Croatia 1999, Serbia 2000, Georgia 2003,

Ukraine 2004 and Kyrgyzstan in 2005 was inspired by successes of their neighbors.9 All

things being equal, incumbency gives additional 8.3 per cent of votes in a first round.

Comparing successors’ performance in hegemonic and non-hegemonic party regimes, the

former are expected to gain 3 per cent more than the latter, however.

Incumbents, Successors and Democratic Transitions

In the theoretical section of this paper, I proposed the hypotheses that successors are

more likely to lose elections than incumbents and this fact increases the probability of

democratic transitions. The results of my statistical analyses support these hypotheses:

successors gain a smaller proportion of votes than incumbents, and their participation in

election increases the chances of democratic transition. Indeed, when an incumbent runs,

breakthrough occurs in only 12 per cent of such elections (8) and does not occur in 88 per

cent (59). When a successor runs, a liberalizing outcome occurs in 52 per cent (12) and

does not occur in 48 per cent (11).

The hypotheses stipulated in the beginning of this paper predicted that the incumbents

should win and successors lose. Since we live in a probabilistic world, the quadrants

representing these outcomes in Figure 1 are not empty. In an ideal world, every time an
9
E.g., Eurasian Daily Monitor. 2004, October 13. “Ukrainian Authorities Target Student Youth
Election-Monitoring Groups” 1 (104).

20
incumbent runs, there is no democratic change, and every time a successor runs, there is a

change. In other words, the upper left and the lower right quadrants of this figure should

be empty. In order to test under what conditions theory does not hold, in this section I

focus on outliers: elections when incumbents run and lose, and successors run and win.

Incumbents Run and Lose

Incumbents run and lose only in 12 per cent of cases in the sample (8 out of 67). A

closer look at these elections could reveal what it takes to defeat a standing incumbent in

elections in the developing world. I discard three cases in which presidents either did not

participate in elections or did not lose, even though democratic change occurred (see note to

Figure 1). Altogether, five incumbent presidents lost: the incumbent president Ion Iliescu

of Romania lost in 1996 (only to retain office back in 2000); President Noriega of Nicaragua

lost reelection in 1990 (the only Latin American ‘reelection loser’ according to Cheibub and

Przeworski 1999) - only to return to office in 2006; in Benin the multiparty elections were

introduced after a long one-party rule and incumbent Kerekou quite unexpectedly failed to

win the first round of 2001 presidential elections and lost in the second round (to return to

office in 1996) (Decalo 1997); Slobodan Milošević of Yugoslavia lost in 2000 in a democratic

revolution (‘Serbian October’), and President Abdou Diouf of Senegal lost to Abdoulaye

Wade in 2000.

The defeats of Milošević in Serbia and of Diouf in Senegal indicate that long-standing

incumbents are extremely difficult to unseat. In the latter case, Diouf has won 4 reelections

as a president with an unlimited mandate since 1981, yet each time with smaller margins.

Finally, in 2000 he lost. ‘Many analysts were predicting that Diouf would prevail again,

with Wade being confirmed as the eternal front-runner among the challengers’ (Ottaway

21
2003, 101). In February of 2000 Diouf was able to win the first round of the presidential

election, yet he was unable to garner the necessary majority of the popular vote required

to prevent a run-off election. It is very likely that if incumbents fail to win a first round,

they fail to win the second, as their failure mobilizes their opposition and signals their

weakness – the opposition became united for a second round only after Diouf failed to win

in a first round. The former minister Niasse joined the second-runner Wade in exchange

for the promise of a prime ministerial post. Through a series of political arrangements in

the opposition camp and following a series of protests, the opposition candidate Abdoulaye

Wade was finally able to defeat President Diouf in a second round.

While these cases indicate that incumbents are not immune to losing, these are also

examples of the extraordinary efforts that are necessary to unseat the standing presidents:

all five cases can be counted as ‘watershed’ or ‘stunning’ elections (Huntington 1991). It is

easier to defeat successors. One can also view transitions in these elections as the result of

the accumulation of random hazards (Przeworski and Limongi 1997) – the long-standing

rulers could have failed because they encountered more problems and made more errors

than usual — in other 59 elections, incumbents prevail.

Incumbents Do Not Run, Successors Win

There are 11 elections recorded as successors running and winning (the upper left

quadrant). I discard three cases in which either there were no successors or they lost, even

if not to the pro-democratic opposition (see note to Figure 1).

There are eight elections (out of 23, or 35 per cent), which successors won. In 2 cases

the designated successors had a very early start and consolidated their power base in

advance and managed to prevent possible elite splits during succession. In Russia an aging

22
and highly unpopular President Yeltsin appointed Prime Minister Putin as his designated

successor in 1999 well in advance of the forthcoming 2000 poll. Then Yeltsin retired three

months prior to the 2000 elections so that Putin became the de facto president for the

crucial pre-election period, and was able to consolidate his power base. In Azerbaijan in

2003, terminally ill incumbent president, Heidar Aliev, began the electoral campaign but

then at the last moment withdrew in favor of his son, Ilham. Having a designated successor

prevented elite splits and uncertainty.10 Like in Russia, Aliev junior had an unusually

early start and had been a de facto incumbent president prior to the vote. Together with

Azerbaijan in 2003, Syria in 2000 represents another case of dynastic succession, in which

Bashar Al-Assad succeeded his father, endorsed by a plebiscite. Elite were assured that

the status quo would remain not only by the family ties between the rulers, but also by

the presence of the hegemonic party.

Likewise, in five more elections presidential succession was facilitated by a strong hege-

monic party (Mexico 1994, Malawi 2004, Djibouti 1999, Mozambique 2004 and the Zambia

2001). I elaborated on Mexican exceptionalism and the stringent adherence to the no re-

election rule in the theoretical section. In Mexico in 1994 president Salinas honored term

limits and stepped down, like all Mexican presidents had done for 60 years prior to him.

The election of Zedillo ensued, even though he was not the expected successor. In Djibouti,

82-year-old president Gouled stepped down in favor of his nephew, Gulleh, who promptly

became the new ruler. The hegemonic People’s Rally for Progress party ensured a smooth

succession. Likewise, in Malawi, having failed to extend his term, the outgoing president

Muluzi appointed Bingu wa Mutharika, Minister of Economic Planning and Development

and his opponent in 1999 presidential elections, as his successor for 2004 elections, which
10
There were reports that the President Aliev was actually dead while running for the Presidency.
Al-Ahram International Edition. 2003, August 7-13. “Like Father, Like Son”. No. 650.

23
the latter won.11 (VonDoepp 2005). In Mozambique president Chissano appointed Ar-

mando Guebuzo as his successor and the candidate from the ruling, Frelimo party, and

stepped down, following the victory of Guebuzzo.

In Zambian elections of 2001 the outgoing President Chiluba did not support Mwanawasa,

a former vice president, who was selected as the governing party, MMD’s presidential can-

didate for 2001 elections. The splits within the ruling elite were so severe that two more

former vice presidents, Godfrey Miyanda and General Christon Tembo, ran against each

other and another former vice president, Mwanawasa (Burnell 2002). The designated suc-

cessor was endorsed by the governing party, however, and prevailed.

To sum up, out of eight cases of successors running and ‘winning’ (no democratic

change), six successors won elections with the help of the hegemonic parties and two

were appointed well in advance, so that they could consolidate their early incumbency

advantage. In fact, two were sons of the incumbent presidents. In all eight cases elite splits

did not occur, successors remained in control, displaying all the characteristics of standing

incumbents, and proceeded to win the elections. The opposition failed to coordinate the

collective action and to mobilize for the ‘stunning elections.’ These exceptions suggest that

under certain circumstances, successors do not have to lose and succession can proceed in

an orderly manner. The process of presidential succession seems to be mitigated by the

presence of strong hegemonic parties: institutions ensure the continuation of careers and

policies of regime elites despite the uncertainty of executive turnover (Brownlee 2004).

Strong, hegemonic parties are able to engineer the ex post elite contracting, and the early

delegation of power that signals that successors are in control and de facto incumbents,

facilitates presidential succession.

11
Chronicle. 2003, July 7. “Malawi: Dictatorship Defeated!”

24
Conclusions and Future Research

Clearly, the absence of incumbents is important for democratic transitions: it increases

the probability of a successful democratic breakthrough by 35 per cent during elections

in presidential regimes. Controlling for other factors, when incumbent presidents are ab-

sent, successors are more likely to lose elections. Thus, they obtain 8 per cent less than

the longer-standing rulers irrespective of the degree of authoritarianism. Successors in

hegemonic-party regimes tend to perform better, however. Because since 1990 partly- and

non-democratic regimes are challenged by the pro-democratic opposition, rather than by

the pro-Soviet, nationalist, fascist or other anti-system forces of the earlier periods, defeat

at elections is very likely to lead to democratization.

The findings in this paper indicate that we should not neglect the dynamics of succession

in presidential regimes when we try to understand and predict democratic transitions.

Using statistical analyses and a careful examination of all the outlier observations, I found

that the stipulated hypotheses hold: presidential succession does increase the chances for

democratic breakthrough in the developing world. The process of democratic transition

thrives on uncertainty and elite splits. Presidential succession often triggers these splits

and uncertainty, because successors are relatively weak and have yet to consolidate their

incumbency advantage. As the experience of battling with Milošević in 2000 and Abdu

Diouf in 2000 suggests, defeating a standing incumbent is possible, but extremely difficult.

Incumbent turnover occurs, however, only if incumbents do not run for one reason or

another. In the sample, 23 elections record executive turnover. Out of these 23 cases, 3

presidents were ousted in coups or forced to resign and could not run, 4 died in office, 4

stepped down even though they could stand, and 12 presidents stepped down because of

25
term limits. In this paper, I treated executive turnover as an independent variable, but

it would be interesting to explore in the future what makes some presidents obey their

constitutions and step down when required, and others to attempt and extend their rule.

Out of those 12 in the sample that honored term limits, two presidents clearly flirted with

the idea of a third term, but in the end stepped down: Chiluba of Zambia and Muluzi of

Malawi.12 In a number of elections turnover did not occur because presidents scrapped

restrictions: for example, in Gabon in 2003 or Togo in 2002. Clearly, what determines

presidential decision-making during succession is something to explore in the future, both

theoretically and empirically.

Term limits can be observed because of a very powerful precedent and the agreement

among elites, like the example of single six-year terms in Mexico illustrates. Recent litera-

ture argues that institutions matter in dictatorships (Gandhi and Przeworski 2006; Brown-

lee 2004): they can provide guarantees to the groups whose support is often necessary for a

ruler to stay in power, serve as a contract between members of the winning coalition, coopt

potential challengers, etc. Also, rulers could introduce and obey term limits in the process

of self-imposed liberalization and signalling their ‘type’ to the international community.

The succession behaviour and its outcomes have direct effects on the prospects of democ-

racy in many developing countries. To measure the affinity between incumbents and their

successors and to map this affinity to the outcomes of succession is an interesting project

to study in the future. Likewise, the presence of hegemonic parties seems to reduce the

uncertainty and costs of succession in presidential regimes. There are other possible factors

influencing how succession interacts with transition. Magaloni stipulates that when presi-

dential elections take place concurrently with legislative elections, stakes can be lower for
12
Times of Zambia (Lusaka). 1998, October 13. “Will Chiluba Desire Another Third Term.” 13 October
1998; Chronicle. 2003, July 7. “Malawi: Dictatorship Defeated!”; VonDoepp (2005).

26
incumbent elites because they can lose presidential office, but still retain legislative seats

(Magaloni 2006, 231-233). Indeed, I find that 60 per cent of democratizing elections in

this sample are concurrent (legislative and presidential), while only 40 per cent of regular

elections are. Further studies should focus on the interaction of incentives and institu-

tions during succession and study which institutions and elite pacts could ease the ex post

contracting in order to ensure transition to and consolidation of democracy in developing

world.

27
 References

Acemoglu, Daron, and James A. Robinson. 2006. Economic Origins of Dictatorship and Democ-
racy. Cambridge; New York: Cambridge University Press.

Banks, Arthur. 2005. Cross-National Time-Series Data Archive, 1815-2003. Binghamton, New
York: Arthur S. Banks, CD-ROMs.

Barzel, Yoram. 2002. A Theory of the State: Economic Rights, Legal Rights, and the Scope of the
State. Cambridge: Cambridge University Press.

Bates, Robert H. 2001. Prosperity and Violence: the Political Economy of Development. New
York, W. W. Norton.

Beck, N., Gleditsch, K. S. and Beardsley, K. C. 2006. “Space is more than Geography: Using
Spatial Econometrics in the Study of Political Economy.” International Studies Quarterly 50(1):
27-44.

Beck, Thorsten, George Clarke, Alberto Groff, Philip Keefer, and Patrick Walsh, 2001. “New
Tools in Comparative Political Economy: The Database of Political Institutions.” World Bank
Economic Review. 15(1): 165-176 (version DPI 2004).

Bienen, Henry, and Nicolas Van de Walle. 1991. Of Time and Power: Leadership Duration in the
Modern World. Stanford, Calif.: Stanford University Press.

Brownlee, Jason. 2004. “Ruling Parties and Durable Authoritarianism”, Paper presented to the
Annual Meeting of the American Political Science Association, Chicago, September 2-5, 2004.

Bueno de Mesquita, Bruce, Alastair Smith, Randolph Siverson, James Morrow. 2003. The Logic
of Political Survival. Cambridge, Mass.: MIT Press.

Burkhart R.E. and Michael S. Lewis-Beck. 1994. “Comparative Democracy-the Economic Devel-
opment Thesis.” American Political Science Review (88): 903-910.

Burling, Robbins. 1974. The Passage of Power: Studies in Political Succession. New York:
Academic Press.

Burnell, Peter. 2002. “Zambia’s 2001 Elections: the Tyranny of Small Decisions, ‘’Non-decisions”
and “Not Decisions.” Third World Quarterly 23(6).

Carothers, Thomas. 2003. “Why Dictators Aren’t Dominoes.” Foreign Policy.

Castaneda, Jorge. 1993. Utopia Unarmed: The Latin American Left After the Cold War. New
York: Knopf.

Castaneda, Jorge. 2000. Perpetuating Power: How Mexican Presidents Were Chosen. New York:
The New York Press.

Cederman, Lars-Erik and Kristian Gleditsch. 2004. “Conquest and Regime Change: an Evolu-
tionary Model of the Spread of Democracy and Peace.” International Studies Quarterly 48(3):
7296-7303.

Cheibub, Jose and Adam Przeworski. 1999. ‘’Accountability for Economic Outcomes”. In Democ-

28
 racy, Accountability, and Representation, edited by Adam Przeworski, Susan Stokes, Bernard
Manin. Cambridge; New York: Cambridge University Press.

Colomer, Josep. 2000. Strategic Transitions: Game Theory and Democratization. Baltimore, MD:
Johns Hopkins University Press.

Darmofal, D. 2006. “Spatial Econometrics and Political Science.” Working paper, University of
South Carolina.

Decalo, Samuel. 1997. ‘Benin: First of the New Democracies.’ In John F. Clarke and David E.
Gardiner. (Eds.) Political Reform in Francophone Africa. Boulder, CO: Westview Press. 4161.

Deutsch, Karl. 1961. “Social Mobilization and Political Development.” American Political Science
Review 55 (3): 493-510.

Di Palma, Guiseppe. 1990. To Craft Democracies: An Essay on Democratic Transitions. Berkeley:

University of California Press.

Diamond, Larry J. 1993. “Economic Development and Democracy Reconsidered.” American

Behavioral Scientist (35).

Egorov, George, Konstantin Sonin. 2005. “The Killing Game: Reputation and Knowledge in
Non-Democratic Succession.” mimeo.

Epstein, David, Robert Bates, Jack Goldstone, Sharyn O’Halloran and Ida Kristensen. 2006.
“Democratic Transitions”. American Journal of Political Science 50 (3): 551-569.

Franzese, Robert and Jude Hays. 2006. “Strategic Interaction among EU Governments in Ac-
tive Labor Market Policy-making: Subsidiarity and Policy Coordination under the European
Employment Strategy.” European Union Politics 7(2): 167-189.

Franzese, Robert J. 2002. Macroeconomic Policies of Developed Democracies. Cambridge, UK;

New York: Cambridge University Press.

Gandhi, Jennifer and Adam Przeworski. 2006. “Cooperation, Cooptation, and Rebellion under
Dictatorships.” Economics and Politics 18(1): 1-26.

Gleditsch, Kristian and Michael Ward. 2006. “Diffusion and the International Context of Democ-
ratization.” International Organization 60: 911-933.

Gleditsch, Kristian, Hein Goemans and Giacomo Chiozza. 2006. Archigos: A Data Base on
Leaders 1875 - 2004.

Gleditsch, Kristian and Michael Ward. 2000. “Peace and War in Time and Space: the Role of
Democratization.” International Studies Quarterly 43: 1-29.

Gyimah-Boadi, E. 2001. “A Peaceful Turnover in Ghana.” Journal of Democracy (12) 2: 103-117.

Hall, Linda. 1981. Alvaro Obregon: Power and Revolution in Mexico, 1911-1920. College Station:
Texas A&M University Press.

Henisz, Witold. 2002. “The Institutional Environment for Infrastructure Investment.” Industrial
and Corporate Change 11(2) (POLCON 2005 Release).

29
Herz, John H. 1952 “The Problem of Successorship in Dictatorial Regimes; A Study in Comparative
Law and Institutions.” Journal of Politics 14 (1). 19 - 40.

Howard, Marc, Philip Roessler. 2006. “Liberalizing Electoral Outcomes in Competitive Authori-
tarian Regimes”. American Journal of Political Science 50 (2): 365-381.

Huntington, Samuel. 1991. The Third Wave: Democratization in the Late Twentieth Century.
Norman: University of Oklahoma Press.

Karklins, Rasma and Roger Petersen. 1993. “Decision Calculus of Protesters and Regimes -
Eastern-Europe 1989.” Journal of Politics 55(3): 588-614.

Karl, Terry. 1990. “Dilemmas of Democratization in Latin America”. Comparative Politics 23(9).

King, Gary, Michael Tomz, and Jason Wittenberg. 2000. “Making the Most of Statistical Analyses:
Improving Interpretation and Presentation.” American Journal of Political Science 44, no. 2
(April): 347-61.

Kitschelt, Herbert P. 1986. “Political Opportunity Structures and Political Protest – Antinuclear
Movements in 4 Democracies.” British Journal of Political Science 16: 57-85.

Kuran, Timur. 1991. ‘’Now Out of Never: the Element of Surprise in the East European Revolution
of 1989.” World Politics 44(1): 7-48.

Kurrild-Klitgaard, Peter. 2000. “The Constitutional Economics of Autocratic Succession.” Public

Choice 103 : 63-84.

Lerner, D. 1958. The Passing of Traditional Society. NY: Free Press Glencoe.

Levitsky, Steven and Lucan Way. 2002. “The Rise of Competitive Authoritarianism”. Journal of
Democracy 13 (2): 51-65.

Lipset, S. M. 1959. “Some Social Requisites of Democracy.” American Political Science Review
53: 69-105.

Magaloni, Beatriz. 2006. Voting for Autocracy: Hegemonic Party Survival and Its Demise in
Mexico. Cambridge: Cambridge University Press.

McAdam, Doug, Sidney G. Tarrow, and Charles Tilly. 2001. Dynamics of Contention. Cambridge;
New York: Cambridge University Press.

McFaul, Michael. 2005. “Transitions from Postcommunism.” Journal of Democracy 16 (3).

McFaul, Michael. 2002. “The Fourth Wave of Democracy and Dictatorship: Noncooperative
Transitions in the Postcommunist World”. World Politics 54: 212-44.

Meyer, David S. 2004. “Protest and Political Opportunities.” Annual Review of Sociology 30:
125-145.

Muller, Edwin N. 1995. “Economic Determinants of Democracy.” American Sociological Review

60, 966-982.

Munck, Gerardo L. and Richard Snyder. 2007 forthcoming. Passion, Craft and Method in Com-
parative Politics. Baltimore: The Johns Hopkins University Press.

30
O’Loughlin, J., M. Ward, C. Lofdahl, J. Cohen, D. Brown, D. Reilly, K. Gleditsch and M. Shin.
1998. “The Diffusion of Democracy, 1946-1994.”Annals of the Association of American Geogra-
phers 88 (4).

Ottaway, M. 2003. Democracy Challenged: the Rise of Semi-authoritarianism. Washington, D.C.,

Carnegie Endowment for International Peace.

Przeworski, Adam. 1991. Democracy and the Market: Political and Economic Reforms in Eastern
Europe and Latin America. Cambridge; New York: Cambridge University Press.

Przeworski, Adam, Jose Cheibub, Fernando Limongi. 2000. Democracy and Development: Political
Institutions and Well-being in the World, 1950-1990. Cambridge: Cambridge University Press.

Przeworski, A., Fernando Limongi 1997. ”Modernization: Theories and Facts.” World Politics 49:
155-84.

Rueschemeyer, Dietrich, Evelyn H. Stephens and John D. Stephens. 1992. Capitalist Development
and Democracy. Chicago, IL: University of Chicago Press.

Schedler, A. 2002. ‘’The Menu of Manipulation.” Journal of Democracy 13 (2):36-50.

Shugart, Matthew and John Carey. 1992. Presidents and Assemblies: Constitutional Design and
Electoral Dynamics. Cambridge; New York: Cambridge University Press.

Simpser, Alberto. 2005. “Making Votes Not Count: Strategic Incentives for Electoral Corruption”.
PhD dissertation, Stanford University.

Smith, Benjamin. 2005. “Life of the Party: the Origins of Regime Breakdown and Persistence
under Single-Party Rule.” World Politics 57: 42151.

Stepan, Alfred. 1986. “Paths Towards Redemocratization: Theoretical and Comparative Consid-
erations”. In O’Donnell, Schmitter and Whitehead (eds.). Transitions from Authoritarian Rule:
Tentative Conclusions About Uncertain Outcomes. Baltimore: Johns Hopkins University Press.

Thompson, Mark and Philipp Kuntz. 2004. “Stolen Elections: The Case of the Serbian October.”
Journal of Democracy 15 (4):159-172.

Throup, David. 2003. “The Kenya General Election: December 27, 2002.” CSIS: Africa Notes,
No. 14. Washington: Center for Strategic and International Studies.

Tomz, Michael, Jason Wittenberg, and Gary King. 2003. CLARIFY : Software for Interpreting
and Presenting Statistical Results. Version 2.1. Stanford University, University of Wisconsin,
and Harvard University.

Tsebelis, George .2002. Veto Players: How Political Institutions Work. New York: Russell Sage
and Princeton University Press.

Tullock, Gordon. 1987. Autocracy. Boston: Kluwer Academic.

VonDoepp, P. 2005. “The Problem of Judicial Control in Africa’s Neopatrimonial Democracies:

Malawi and Zambia”. Political Science Quarterly 120 (2): 275-301.

31
Variable Definitions and Data Sources

Democratic Change (Liberalizing Electoral Outcome) 1 if MPOLITY≥3 and MFH≥1

(democratization) in election year, 55 new presidential elections are added to H&R data.

Vote share Per cent of votes in the 1st/only round, cast for incumbent/successor, DPI.

Successor 0 if incumbent, 1 if successor (incumbent does not run, or president served <
1 year prior to elections). Every effort was made to identify the designated successor in
cases in which incumbents do no run. Table 1 in Appendix lists all incumbent presidents
and their successors.

Opposition Coalition 1 if Opposition united, 0 otherwise. The author followed the

coding by H&R, 55 new obs. are added.

Opposition Mobilization The average number of anti-governmental demonstrations, a

year preceding elections and election year, added from Banks.

Growth GDP growth per capita, per cent, averaged 2 years prior to elections, WDI.

FDI FDI as per cent of GDP, averaged 2 years prior to elections, WDI.

Foreign Aid Aid per capita, averaged 2 years prior to elections, WDI.

Democracy Averaged FH civil liberties score, 2 years prior to elections, FH.

Prior Liberalization0 if F Ht−5 >= F Ht−1 , 1 otherwise, FH.

Hegemonic Party 1 if regime is hegemonic party autocracy at the time of the election,
as coded in Magaloni (2006, 36–41): (1) regularized multiparty competition, (2) the chief
executive and legislature are elected, (3) the incumbent held office for more than 20 y., (4)
the ruling party never lost elections. I cross-check these regimes and add those coded as
the single-party regimes in Smith (2005, 424).

Regional Transitions Number of transitions (DV = 1) in the region (West Africa,

Middle Africa, South Africa, East Africa, Middle East and North Africa, CEE/NIS, Central
America, South America, and Asia), 3 years prior to an election.

Regional Liberalization Averaged FH score in the region, year preceding election and
election year.

Previous Vote Vote share cast for the winner of the previous elections, DPI.

Political Constraints PolconIII : the feasibility of policy change on the [0, 1] scale, where
0 is the unlimited authority, Henisz (2002).

Electoral Fraud Fraud reported in elections, yes (1) or no (0), DPI.

Data sources: H&R: Howard and Roessler (2006); FH: Freedom House (various years); DPI: Database of
Political Institutions (Beck et al. 2001); WDI: World Development Indicators (World Bank); GFD: Global
Financial Data (GFD database); Banks: Cross-National Time-Series Data Archive (Banks 2005).

32
 Figure 1: Incumbency and Democratic Transitions

11 elections* 12 elections
(48 per cent) (52 per cent)

SUCCESSOR
Malawi 2004 Ghana 2001

INCUMBENT Uganda 2001 Senegal 2000

59 elections 8 elections**
(88 per cent) (12 per cent)

NO LIBERALIZING YES
ELECTORAL OUTCOME
fi

*Even though the executive turnover and no transition were observed, in 3 cases successors either lost or
there were no successors. In Guinea-Bissau following the coup neither incumbent Veiera nor his successor
participated in elections. In Guatemala of 1990 designated successor Cabrera Hidalgo did not even get
past the first round. In Iranian elections of 1997 president Rafsanjani stepped down and his conservative
‘successor’ lost to the ‘reformist’ candidate Khatami. If we discard these 3 cases, the proportion goes
down to 35 per cent (8 elections) **In Peru 2000 the incumbent Fujimori ran for a third term, and,
proper speaking, won. He was subsequently ousted in a series of post-election protests. ‘Rose revolution’
in Georgia in 2003 followed the parliamentary elections in which incumbent did not run. In Ghana in
1996 the opposition made advances in the legislative elections, yet the incumbent president Rawlings won
presidential contest that same year, so this case is not an incumbent’s defeat as such. If we leave out these
3 cases, the proportion changes to 8 per cent (5 elections).

33
 Table 1: Democratic Transitions During Elections in Presidential Regimes

Logit Regression (1 – 4) OLS (5)

Model 1 Model 2 Model 3 Model 4 Model 5
Base with with Party Excluding Vote Share
Model Party & Diffusion votes>.75 (OLS)
Successor 2.377*** 3.080*** 2.919*** 2.888*** -8.190**
(.599) (.920) (.766) (.586) (1.636)
Opp. Coalition 1.983** 2.257** 1.719** 1.948*** -3.908
(.870) (.962) (.723) (.705) (2.884)
Opp. Mobilization .462*** .365*** .162 .223 -1.336
(.066) (.048) (.163) (.343) (1.055)
Growth -.049* -.075** -.103** -.123* .419**
(.030) (.025) (.032) (.048) (.155)
Foreign Aid .001 .003 .008 .001 -.021
(.007) (.006) (.009) (.006) (.032)
FDI -.102 -.138 -.118 -.039 -.012
(.117) (.184) (.161) (.074) (.203)
Democracy (FH) -.702** -.873** -.670** -.397* 6.341**
(.356) (.349) (.293) (.233) (1.610)
Prior Liberalization .447 .239 .530 .700 –
(.713) (.742) (.824) (.825)
Hegemonic Party – -2.044** -2.138** -2.375** 2.066
(.951) (.934) (.895) (2.270)
Regional Transitions – – .928** .893** –
(.386) (.438)
Regional Liberalization – – -.749 -.489 –
(.714) (.765)
Previous Vote – – – – .376**
(.075)
Political Constraints – – – – -12.224
(11.493)
Electoral Fraud – – – – 1.906
(2.863)
Intercept -.073 1.014 3.109 1.126 11.661
(1.395) (1.266) (3.276) (3.541) (11.037)
N 90 90 90 60 90
Log-Likelihood -29.195 -26.495 -24.003 -21.703 F = 22.71
Pseudo R2 .39 .44 .50 .43 R2 = .62

Models 1 - 4 estimate the probability of democratic transition during elections, 1990-2004, logit regression,
with robust (cluster-adjusted) standard errors. Model 5 estimates the vote share cast for incumbent or
successor in first/only round, OLS regression, with robust (cluster-adjusted) standard errors. *Significant
at .1 level, **significant at .05 level, *** significant at .01 level. Robust (cluster-adjusted) standard
errors in parentheses. Adjustments are made for within-region correlation, regions are specified as follows:
West Africa, Middle Africa, East Africa, South Africa, Middle East and North Africa, NIS/CEE, Central
America, South America, Asia.

34
 Figure 2: Predicted Democratic Transitions, Incumbents and Successors

incumbent successor
1

.75
Predicted Probability

.5

.25

1 2 3 4 5 6 7 1 2 3 4 5 6 7
Democracy
Predicted values Predicted LEOs
Graphs by incumbturnover

Plot of predicted values of democratic transition during elections and democracy (Freedom House averaged fi

score, 2 years prior to elections), weighted by votes cast for incumbent/successor in first/only round, based
on the results of Model 3. Points filled with colour represent observations where democratic transitions
actually occur. If they are located above .5 probability line, they are deemed to be predicted correctly.
Larger circles represent larger vote shares, in [23, 99] interval. The graph shows that the probability
of transition is lowest for big-vote-winning authoritarian incumbents. The model predicts 13 transitions
correctly, and fails to predict 7. Overall, the model predict 88 per cent of observations correctly (p> .5
and transition; p< .5 and no transition).

35
 Figure 3: Simulated Effects on the Dependent Variable

Predicted Changes in Probabilities

Regional Transitions: 1 to 3
Changes in Parameters

One Regional Transition

Democracy (FH): 7 to 2.5

Opp. Coalition

Successor in Heg. Party

Successor
0 .1 .2 .3 .4 .5 .6 .7
Changes in Predicted Probability of Democratic Transition

Predicted Changes in Expected Values

Previous Vote: 50 to 75 pc
Changes in Parameters

Democracy (FH): 3.5 to 5

Incumbency in Heg. Party

Incumbency
0 .05 .1 .15 .2 .25 .3
Changes in Expected Value of Vote Shares

fi

Simulated effects represent the change in the predicted probability of the dependent variable using Model
3 (predicted probabilities of democratic transition) and Model 5 (expected values of votes cast for presi-
dent/successor in first/only round), given changes in particular predictors. For example, the fact of suc-
cessor running in elections (rather than incumbent) increases predicted probability of democratic change
by 34.8 per cent. However, when incumbent runs in the hegemonic party regime, it increases the proba-
bility by only 13 per cent. Likewise, incumbent is expected to obtain 8 per cent more than successor, but
successors hegemonic party system gain 3 more than successors in regular regimes.

36
 Appendix

Table 1: Incumbents, Successors and Transitions

Transition Incumb. Election Leader Vote Share Prev. Leader

transition incumbent Benin 1991 Kerekou 27.2
no incumbent Burkina Faso 1998 Campaore 87.5
transition successor Cote d’Ivoire1 2000 Guei 32.7 Bedie
no incumbent Gambia 1996 Jammeh 55.76
no incumbent Gambia 2001 Jammeh 52.6
transition incumbent Ghana 1996 Rawlings 58.3
transition successor Ghana 2000 Atta-Mills 44.8 Rawlings
no incumbent Guinea 1998 Conte 56.1
no incumbent Guinea 2003 Conte 95.6
no successor Guinea-Bissau 19992 Kumba Yala 56.11 Vieira
no incumbent Mauritania 1997 Taya 90.1
no incumbent Mauritania 2003 Taya 66.7
no incumbent Niger 2004 Tandja 40.7
no incumbent Nigeria 2003 Obasanjo 61.8
no incumbent Senegal 1993 Diouf 73
transition incumbent Senegal 2000 Diouf 41.6
no incumbent Sierra Leone 2002 Kabbah 70.1
no incumbent Togo 1998 Eyadema 52.1
no incumbent Togo 2003 Eyadema 57.8
no incumbent CAR 1999 Patasse 51.6
no incumbent Cameroon 1997 Biya 92.6
no incumbent Cameroon 2004 Biya 75.2
no incumbent Chad 2001 Deby 67.3
no incumbent Congo 2002 Sassou-Nguesso 89.4
no incumbent Eq. Guinea 1996 Obiang 99
no incumbent Eq. Guinea 2002 Obiang 99
no incumbent Gabon 1998 Bongo 51.18
no successor Djibouti 1999 Guelleh 72.02 Gouled
no incumbent Kenya 1997 D. arap Moi 40.6
transition successor Kenya 2002 Kenyatta 30.6 Moi
no incumbent Malawi 1999 Muluzi 52.4
no successor Malawi 2004 Mutharika 35.9 Muluzi
no incumbent Mozambique 1999 Chissano 52.3
no successor Mozambique 2004 Guebuza 65 Chissano
no incumbent Rwanda 2003 Kagame 95
no incumbent Uganda 1996 Museveni 75.5
no incumbent Uganda 2001 Museveni 74.2
no successor Zambia 20013 Mwanawasa 72.5 Chiluba
no incumbent Zimbabwe 1990 Mugabe 83.5
no incumbent Zimbabwe 1996 Mugabe 92.7
no incumbent Zimbabwe 2002 Mugabe 56.2
no incumbent Algeria 2004 Bouteflika 85
no incumbent Egypt 19994 Mubarak 99
no incumbent Iran 1993 Rafsanjani 94.5
Continued on next page
1
 Table 1: Incumbents, Successors and Transitions (continued)

Transition Incumb. Election Leader Vote Share Prev. Leader

no successor Iran 1997 Khatami 63.3 Rafsanjani
no incumbent Sudan 2000 Ahmad al-Bashir 86.5
no incumbent Syria 19994 Assad 99.9
no successor Syria 20004 B. Assad 99
no incumbent Tunisia 1999 Ben Ali 99
no incumbent Tunisia 2004 Ben Ali 94.5
no incumbent Armenia 1996 Ter-Petrosian 83
transition successor Armenia 1998 Kocharian 52 Ter-Petrosyan
no incumbent Armenia 2003 Kocharyan 48.3
no incumbent Azerbaijan 1993 H. Aliev 99
no incumbent Azerbaijan 1998 H. Aliev 77.6
no successor Azerbaijan 2003 H. Aliev 76.8 Aliev
no incumbent Belarus 2001 Lukashenko 75.6
no incumbent Belarus 20045 Lukashenko 79.4
no incumbent Croatia 1997 Tudjman 57
transition successor Croatia 2000 Granic 26.7 Tudjman
no incumbent Georgia 2000 Shevarnadze 40.1
transition incumbent Georgia 20036 Shevarnadze 42.2
no incumbent Kazakhstan 1999 Nazarbaev 81
no incumbent Kyrgyzstan 1995 Akayev 71.6
no incumbent Kyrgyzstan 2000 Akaev 74.5
no incumbent Romania 1992 Iliescu 85.1
transition incumbent Romania 1996 Iliescu 47.34
no incumbent Russia 1996 Yeltsin 35.79
no successor Russia 2000 Putin 50.6 Yeltsin
no incumbent Russia 2004 Putin 71.1
no incumbent Tajikistan 1999 Rakhmonov 97
no incumbent Ukraine 1999 Kuchma 38
transition successor Ukraine 2004 Yanukovich 41.4 Kuchma
no incumbent Uzbekistan 2000 Karimov 91.9
no incumbent Yugoslavia 1996 Milosevic 57.46
transition incumbent Yugoslavia 2000 Milosevic 37.8
transition successor Dominican Rep. 19967 Reynado 36 Balaguer
no successor Guatemala 1990 Hidalgo 38.65 Cerezo
transition successor Guatemala 19958 Arzu 42.6 Carpio
no successor Mexico 1994 Zedillo 50.39 Salinas
transition successor Mexico 2000 Labastido 36.1 Zedillo
transition incumbent Nicaragua 1990 Daniel Ortega 45
no incumbent Peru 1995 Fujimori 44.52
transition incumbent Peru 20009 Fujimori 49.8
transition successor Peru 20019 Paniagua 49.87 Fujimori
no incumbent Venezuela 200410 Chavez 59
transition successor Sri Lanka 1994 Dissanayake 35.9 Premadasa
no incumbent Sri Lanka 1999 Kumaratunga 62.2
no incumbent Taiwan 1996 Lee Teng-hui 54
transition successor Taiwan 200011 Lian Chan 23.1 Lee Teng-hui
Continued on next page

2
 Table 1: Incumbents, Successors and Transitions (continued)

Transition Incumb. Election Leader Vote Share Prev. Leader

1
President Bedie (1993-99) was overthrown in a coup in 1999. Military junta encouraged Guéï to run in
elections as its candidate but he lost. 2 Military coup of 1999 ousted president Veiera. 3 Although the party
of President Chiluba - MMD - elected Mwanawasa as its presidential candidate for 2001 elections, Chiluba
did not endorse Mwanawasa. 4 Plebiscites (to endorse the candidate). 5 2004 Parliamentary elections and
a third term referendum. 6 2003 Parliamentary elections, following which the incumbent was ousted. Vote
share is reported for the incumbent parties. 7 President Balaguer and his party supported Reynado in
the first round of 1996 elections. 8 Arzu defeated Carpio, who was an interim president (1993-95) after
President Elias and his Vice-President were forced to resign after constitutional violations. 9 Fujimori
was the winner of 2000 elections, even though democratic transition ensued after corruption allegations.
Fujimori did not have a successor in 2001 elections. 10 2004 Recall referendum. 11 The outgoing president
Lee failed to ensure the victory of his successor. It appears that he helped the opposition candidate to
win instead. In 2000 Freedom House rated Taiwan as democracy, yet this election is included as it was the
first alternation in power in Taiwan.

3
Institute for International Integration Studies
The Sutherland Centre, Trinity College Dublin, Dublin 2, Ireland