Professional Documents
Culture Documents
Archigos A Database On Political Leaders
Archigos A Database On Political Leaders
1
We are grateful to Bruce Bueno de Mesquita, Lindsay Heger, Tanisha Fazhal, Kimuli Kasara, Brett
Ashely Leeds, Nicolay Marinov, Michael Ross, Idean Salehyan, and Branislav Slantchev for comments.
The Archigos data are available at http://mail.rochester.edu/∼hgoemans/data.htm. The National Science
Foundation declined to support the Archigos project. Gleditsch acknowledges the support of the Centre
for the Study of Civil War. Goemans was supported by a PEPR grant from the Wallis Institute at the Uni-
versity of Rochester. Email: hgoemans@mail.rochester.edu, ksg@essex.ac.uk, chiozza@berkeley.edu
Abstract
Scholars in Political Science and associated disciplines have for a long time theorized about the
role of leaders. A fundamental hurdle in empirical research on the role of leaders has been the
limited availability of data and information about individual leaders. We provide a new data set
on all leaders from 186 countries between 1875 and 2004 entitled Archigos. We illustrate the
utility of these data by demonstrating how variation in leader attributes not provided by existing
data sources are powerful predictors of other variables of interest. We show that leaders who
enter power in an irregular manner are also much likely to leave power irregularly and have a
higher risk of experiencing civil war. Leaders that exit irregularly are much more frequently
punished after leaving office. Crises interactions differ depending on whether leaders face each
other for the first time or have prior experiences. Irregular leader changes help identify political
change in autocracies not apparent from data that consider whether institutions are democratic
or not. Finally, transitions to democracy in the third wave are more likely to fail in instances
where autocratic rulers were punished after leaving office.
1 Introduction
To analyze politics, scholars in Comparative Politics and History have long emphasized the role
of individual leaders, a perspective that more recently also has gained currency in mainstream
International Relations research. This approach has the advantage of focusing specifically on
decision makers, their personal incentives, as well as how their incentives and constrains are
shaped by their political environment, allowing new theories to be built on a solid methodolog-
ically individualist basis. Moreover, a common focus on leaders in Comparative Politics and
one field into research in the other field, and so bring both fields in closer contact.
A major stumbling block for such a unifying approach has been the lack of data on leaders
for empirical research. Pioneering efforts by Blondel (1987), Bienen and van de Walle (1991)
and Bueno de Mesquita and Siverson (1995) steadily compiled an expanding data set of leaders.
However, these initial data sources suffered from some measurement problems—such as when
two or more leaders were coded to rule a country at the same time and significant gaps in the
sequence of leaders in power—and contained very little information on the individual leaders
across fields, and open up new avenues for research, this article introduces a new data set entitled
Archigos (of the Greek term for ruler o αρχιγ óς), which contains information on leaders from
186 countries between 1875 and 2004. Below, we first briefly survey the history and some of the
merits of a focus on leaders and then illustrate how a series of important empirical questions can
be addressed with the new Archigos data. These questions by no means exhaust the potential
1
applications of Archigos.
We show how Archigos can throw new light on the causes and consequences of political
instability in various forms, on how, when and why leaders lose office and the associated con-
sequences of losing office and on the international interactions between leaders. To briefly list
the illustrative patterns we identify, leaders who enter power in an irregular manner are more
likely to also leave power in an irregular manner, and leaders who exit irregularly are much
more frequently punished after they lose office. Leaders who enter power irregularly also face
a higher risk of civil war. We show that whereas the normative force of a prior crisis settlement
carries over into crisis involving leaders who have interacted previously, this does not apply
in crises where leaders face each other for the first time. Compared to the well-known Polity
data which largely focuses on whether institutions are democratic, Archigos’s data on irregular
leader changes makes it possible to identify political transitions and institutional changes in a
novel way. In particular, this data makes it possible to also identify major political change in au-
tocracies. Finally, transitions to democracy in the third wave are more likely to fail in instances
where autocratic rulers were punished after leaving office. We end with a brief conclusion that
suggests other potentially fruitful avenues of research that can be explored with Archigos.
Four decades ago, the study of leaders figured prominently in the field of International Rela-
tions. In the 1960s and 1970s, following the pathbreaking work of Snyder, Bruck and Sapin
lar, leaders, largely from an organizational and psychological perspective. Although since the
2
1950s a significant number of scholars emphasized the role of the international system, it was
the publication of Waltz’s Theory of International Politics (1979) that led to an overriding and
field-wide emphasis on the international system as the primary unit of analysis. This focus on
the system was in turn superseded by a focus on states and dyadic relations between states in the
1990s, at least partly the result of the work on the democratic peace phenomenon. It deserves
note that this shift was significantly facilitated by a wealth of new sources of country specific
data, such as the Polity data on institutional characteristics. We believe that the focus of the
field has now come full circle as researchers increasingly examine international political behav-
ior from the perspective of leaders (Bueno de Mesquita and Siverson, 1995; Bueno de Mesquita
et al., 1999, 2003; Goemans, 2000b; Gelpi and Grieco, 2001a; Chiozza and Choi, 2003; Chiozza
and Goemans, 2003, 2004; Marinov, 2005; Horowitz, McDermott and Stam, 2005).
Scholars in Comparative Politics have often examined domestic politics by a focus on cab-
(Dodd, 1984; King et al., 1990; Warwick, 1994; Grofman and Van Roozendaal, 1997; Dier-
meier and Stevenson, 1999; Leblang and Bernhard, 2000). However, since the pathbreaking
work of Bunce (1981), Blondel (1987), Ames (1987) and Bienen and van de Walle (1991), re-
searchers in comparative politics begun to systematically focus on the specific leaders who set
policy (Betts and Huntington, 1985/86; Cheibub and Przeworski, 1999; Przeworski et al., 2000;
Stokes, 2001).
A generation of scholars since Downs (1957) have adopted the simplifying assumption that
leaders choose policies to stay in office. Leaders’ policy choices then depend on the anticipated
effect of these policies on their tenure, and leaders will pick policies that maximize their time
in office. Building on this assumption, scholars have argued that policies such as decisions
3
to initiate or continue international conflict (Levy, 1989; David, 1991; Fearon, 1994; Downs
and Rocke, 1994; Leeds and Davis, 1997; Schultz, 2001b; Bueno de Mesquita et al., 2003;
Colaresi, 2004; Mansfield and Snyder, 2005; Horowitz, McDermott and Stam, 2005; Lai and
Slater, 2006), impose or comply with international sanctions (Marinov, 2005), promote eco-
nomic development (Bates, 1981; Wintrobe, 1998; Przeworski et al., 2000; Gelpi and Grieco,
2001b; Jones and Olken, 2005) and political reform (Bueno de Mesquita and Siverson, 1997)
or time elections (Warwick, 1994; Smith, 2003; Kayser, 2005) are fundamentally driven by the
leader’s desire to maximize his or her tenure in office. The assumption that leaders choose poli-
cies to stay in office has proven enormously influential and fruitful in comparative politics and
Archigos makes it possible for the first time to directly test these arguments. Scholars can
examine if a certain policy choice affects the tenure of leaders. A more ambitious approach
would take into consideration the endogeneity of a policy choice. Scholars can use Archigos
to generate an instrument for the latent risk of losing office and employ this instrument in their
regressions on policy choice.1 Chiozza and Goemans (2003), for example, use a two-stage
probit estimator to explore the reciprocal relationship between the probability of losing office
Although existing research on individual leaders has generated new insights, it has largely
been limited to the tenure of individual leaders and when they enter and leave office. Scholars
rarely tried to establish indicators of other potentially important forms of variation among lead-
1 The leader’s age would seem a variable exogenous to most policy choices, although
Horowitz, McDermott and Stam (2005) point to how testosteron levels in men decline with
with age and that older leaders therefore should be less likely to go to war.
4
ers such as the manner in which they acquire or lose political office or their political experience.
As we will show in this paper, the new information in Archigos on how leaders enter and leave
political office as well as their post-tenure fate provides new sources of variation that can help
This new focus on the incentives of individual leaders in both fields sits well with the current
call to integrate theoretical and empirical models (EITM) and the methodological individualism
of rational choice theory. Moreover, a focus on individual leaders and their incentives helps
us increase explanatory variation in empirical work. Even though it is a central tenet of the
logic of comparisons that one cannot explain variation with a constant (Przeworski and Teune,
1970), a great deal of research in International Relations rely on indicators that are dangerously
close to time-invariant (Bennett and Stam, 2004). Typical explanatory variables such as regime
type, great power status, and contiguity change too little—if at all—to be able to account for
dynamic phenomena such as conflict initiation (Chiozza and Goemans, 2003), the imposition of
sanctions, (Marinov, 2005) or international financial instability (Bernhard and Leblang, 2002).
A focus on leader characteristics introduces more variation as the median leader stays slightly
less than two years in office, and leaders vary significantly in their political and personal char-
Finally, new comprehensive data on leaders and their characteristics hold the promise of
providing new answers to questions that could not be answered with existing data sources. For
example, whereas traditional data on the degree of democracy tell us little about political change
within autocracies, our data make it possible to develop new indicators for regime transitions
in authoritarian regimes by recording not only when leaders alternate but how one leader loses
office and how his or her successor enters office. Moreover, the Archigos data also make it
5
possible to explore and discover new patterns. For example, Archigos makes it possible to
explore why some leaders “come back” and regain office after having lost it, and how the
importance of political experience in the form of previous spells in office affects their behavior.
In this section we provide a brief overview of the Archigos data. Archigos is a data base of
political leaders. It specifically identifies the effective leader of each independent state, as coded
by Gleditsch and Ward (1999), between 1875 and 2004. By effective leader, we mean the person
that de facto exercised power in a country. We relied upon primary and secondary sources, as
well as our knowledge of the particulars of each state, to inform our coding decisions. 2
In most cases, identifying effective rulers is relatively clear and uncontroversial, but in other
cases the coding may be more contentious. Many countries, for example, have multiple heads
of states. In some cases, the formal head of a state may primarily be a ceremonial position—
position such as a prime minister. In other cases, a strongman (or woman) may formally step
down but remain the effective leader behind the scenes, as in the case of Rafael Trujillo in the
Dominican Republic, who did not formally occupy the presidency between 16 August 1938 and
We generally followed a simple coding rule. In Parliamentary regimes, the Prime Minister
2 Principal sources include Lentz (1994, 1999), Keesing’s, http://www.rulers.org
6
is coded as the leader, in Presidential systems, the President. In regimes that combine elements
of both parliamentary and presidential systems—i.e., Finland, France and Portugal—we code
the president as the leader since in these regimes presidents typically control foreign policy. In
communist states we generally code the Chairman of the Party as the effective ruler. The exten-
sive case description file available online specifies the full rationale, reasoning and references
Archigos identifies not just when but also how the leader gained and lost power. Leaders
can enter in 1) a regular manner, according to the prevailing rules, provisions, conventions and
norms of the country, 2) an irregular manner, and 3) through direct imposition by another state.
Note that support of a particular choice of leader or attempts to indirectly influence leader selec-
tion by a foreign state do not qualify as imposition by another state. In cases where a previous
over or an interim leader is appointed by the legislature as prescribed by rules and procedures
Leaders can lose office in 1) a regular manner, according to the prevailing rules, provisions,
conventions and norms of the country, 2) an irregular manner, 3) through direct removal by
another state, and 4) as a result of a natural death, under which we include illness or suicide.
Examples of a regular loss of office include voluntary retirement, term limits and defeat in
elections. A loss of office is considered irregular when the leader was removed in contravention
of explicit rules and established conventions. Most irregular removals from office are done by
domestic forces and are overwhelmingly the result of the threat or use of force as exemplified in
coups, (popular) revolts and assassinations. Assassinations may or may not have a clear political
motivation; we prefer to make no judgments about the “real” intention behind assassinations. In
7
a very few cases, whether a leader dies of natural causes or is assassinated may be disputed. We
clarify our judgments in the case descriptions when potentially controversial. As in the case of
entry into office, to qualify as a removal by another state requires the direct removal of a leader,
as in the case of a successful invasion. Cases where another country is perceived or known to
have orchestrated the removal of a leader through a coup carried out by domestic forces (for
example, Allende in Chile or Mossadeq in Iran) are not coded as foreign removal, but simply as
Archigos records the leader’s post-tenure fate in the period up to one year after the leader
lost power. This period is chosen to exclude as much as possible that we measure punishment
due to a leader’s behavior after he or she lost office rather retribution directly due to his or
her behavior while in office.3 Archigos records one of four types of post-exit fates: when a
an embassy legally is considered as belonging to a foreign state, we code cases where leaders
take refuge in the embassy of another state as exile. We code house arrest as imprisonment.
Archigos records only the highest level of punishment (there are many cases where a relatively
brief house arrest was followed by exile). For leaders who died while in office, we code their
post-tenure fate as missing, as we do for leaders who died less than six months after they left
office.
In addition, Archigos records several personal characteristics of the leader such as date of
or killed more than one year after he or she lost office, but we make no claim to provide a
comprehensive coding of the leader’s fate beyond one year after he or she lost office.
8
4 Exploring the Utility of Archigos
In this section, we provide a brief summary of some of the information in the Archigos data to
demonstrate some of its contributions and potential. Our review will necessarily be selective,
and primarily intended to demonstrate how variation in observed leader characteristics in the
data gives rise to dramatic differences in interesting outcomes such as the expected manner of
losing office, the probability of punishment after leading office, international conflict behavior
and regime transitions. These are all new empirical patterns that could not be examined us-
ing existing data sources. Although the descriptive patterns we show are striking and highly
origins of these patterns here. Nevertheless, we believe that these examples illustrate how the
data may be used, and provide pointers to some of the research questions that the Archigos data
As we have previously noted, much of the existing work on leaders in international relations
simply records a given leader X in power over some time interval. Typically, the data does
not consider any attributes of the leaders themselves. However, the information in Archigos
strongly suggests that leader-specific characteristics—such as the manner in which leaders at-
tain office, or the number of their previous spells in office—has a strong influence on subsequent
9
4.1.1 Leader Entry and Political Experience
Table 1 presents the observed distribution of the number of previous spells in office by the
leader’s manner of entry. For ease of exposition, we collapse the variable with information
on the number of previous spells in office into a simple dummy variable. The cell entries
in parentheses in the main part of the table display the relative frequency distribution within
each row, i.e., the conditional frequency of the number of previous spells in office given the
particular manner of entry. The figures in parentheses for the margin totals indicate the relative
frequency of the values of the individual variables in percentages. The table suggests an answer
to the question of how the manner of entry relates to previous political leadership experience.
Specifically, Table 1 shows that leaders who enter irregularly tend to be newcomers. Whereas
roughly 19% of all leaders have served in office at least once before, this percentage drops to
irregular means, most returning leaders gain office through regular means. So-called “come
back” leaders—who return to office after having lost office—also differ from novices in their
international conflict behavior. Additional exploratory research shows they are significantly less
Table 2 examines whether the political experience gained from serving previous spells in
office affects how leaders lose office. We see that while overall roughly 68% of all leaders
10
lose office in a regular manner, this percentage increases to roughly 76% for leaders who have
served in office before. It may be that leaders return to office in countries with particular types
of regimes, such as stable democracies, with a high leadership turnover and deeply entrenched
norms of regular leadership succession. Alternatively, leadership experience may allow the
leader to rely on entrenched political networks of support. ‘Come back’ leaders thus differ from
Research on coups has largely focused on country-specific variables such as economic develop-
ment (Londregan and Poole, 1990; Belkin and Schofer, 2003, 2005). Archigos makes it possible
to examine whether and which leader-specific factors might play a role in the irregular removal
from office of leaders. For example, Archigos makes it possible to trace whether and how the
manner of entry influences the likely form of exit.4 Table 3 shows the observed distribution of
the manner of exit from office by the manner of entry. As before, the cell entries in parentheses
display the relative frequency distribution within each row. As can be seen, although regular
exits are by far the most common form of exit—i.e., 68.48 % of all the cases in Archigos—the
distribution of exit modes look very different among leaders that have experienced irregular
4 Archigos distinguishes between two particular subcategories of natural deaths: cases where
a leader retires due to ill health, and cases where a leader lost office as a result of suicide. Given
the rarity of such cases, Table 3 keeps all these sub-categories with exit by natural death.
11
entries. Indeed, for leaders who gain power by irregular means, the modal loss of power is also
irregular—i.e., 47.64% of all exits, or over 50% if we include the cases where a leader is de-
posed by a foreign state. Stated differently, the data suggest that the likelihood that a leader will
lose power by irregular means is over three times greater if that leader came to power through
irregular means. This clearly demonstrates that if we wish to assess the prospects that a leader
will be deposed or lose power through irregular means such as a coup, we should take into
Archigos makes it possible to examine the causes and consequences of political instability for
both the leader and domestic politics more broadly conceived. We next examine how the man-
ner of entry into office affects civil war. Much of the literature on civil war and insurgency has
argued that the risk of civil war should vary with political instability or regime effectiveness.
In stable democratic regimes, the risk of civil war should be low, since open political systems
afford many opportunities for groups to further their aims by political means. Effective autoc-
racies should be able to enact sufficient repression to make the cost of insurgency so high as to
deter violent dissent. The risk of civil war should thus be highest in politically unstable regimes
without fully democratic political institutions, since these provide neither meaningful avenues
for nonviolent political participation, nor effective political control or repression to deter dissent
(Hegre et al., 2001; Fearon and Laitin, 2003; Muller and Weede, 1990).
12
These arguments have typically been tested using the Polity scale of a country’s degree of
democracy, and many studies have found evidence that the risk of civil war seems highest for
countries with values in the middle of this scale. The use of Polity to test arguments about
institutions and civil war has recently been criticized by Vreeland (2005), who argues that these
findings follow by construction, since two of the subcomponents in the Polity scale that are
robustly associated with civil war—the Competitiveness of Participation (PARCOMP) and the
country experiences civil war. None of the other subcomponents of the Polity index appear to
The manner of entry data in Archigos provide an alternative approach to explore the link be-
tween political instability and the risk of civil war. An effective autocracy that exercises control
over leader selection should not see irregular leader entries. Nor should established democracies
with regular elections see irregular leader changes. Recall we previously showed that leaders
who enter irregularly are much more likely to be deposed in an irregular manner, at least until
they have had a long time to settle in office. This in turn means that irregular entry is a good
indicator of future irregular change and instability. As such, the irregular entry information in
Archigos provides an alternative indicator of political instability not plagued by the potential
tautological nature of the Polity data in civil war studies. Following the logic of the arguments
relating civil war to instability reviewed above, we would expect that civil wars are more com-
mon under leaders that have entered power irregularly than those that have entered in a regular
manner. Table 4 compares regular and irregular leader entries in Archigos with information on
civil wars from Gleditsch (2004).5 As can be seen, civil wars are almost twice as common in
5 We exclude all cases where states participate in civil wars in other states, since these reflect
13
cases where leaders entered power irregularly, of whom 22.83% experience civil wars, com-
pared to cases where leaders have entered power regularly, of whom only 12.64% experience
civil conflict. This suggests that the arguments relating political instability or opportunities to
conflict may be correct, even if the Polity based tests may have relied on flawed indicators.
Additional exploratory analysis has shown that a leader’s manner of entry has a broad and
significant impact on several important policies, ranging from international conflict, to eco-
nomic growth, to political reform. Moreover, in each of these cases, preliminary analyses sug-
gest that the impact of an irregular manner of entry, while strong initially, dissipates over time. 6
The new information on how leaders lose office in Archigos not only allows us to better study
which leaders are irregularly removed from office, but can also throw new light on several other
Just as the leader’s manner of exit depends on his (or her) manner of entry, the chances that a
leader will be punished after leaving office differ dramatically by the manner of exit. Table 5
decisions to intervene rather than challenges to a leader’s authority and control.
6 In other words, we suggest that researchers include not just manner of entry, but also man-
14
demonstrates that although the majority of leaders do not suffer any punishment in the year after
leaving power—i.e., in 74.9% of all cases—the chances of post-exit punishment are very high
for leaders who lose office in irregular ways. In particular, only less than 20% are able to avoid
post-tenure punishment altogether, while almost half of all leaders who lose office irregularly
are quickly forced into exile (e.g., 42.71%). This finding suggests that if we wish to understand
the incentives of individual leaders, we may need to consider the consequences they are likely
to face if they lose office. In other words, leaders may care about how their policies affect not
To elaborate, consider the so-called theory of gambling for resurrection, which argues that
leaders become more likely to continue seemingly inefficient conflict when they face a high risk
of losing office (Downs and Rocke, 1994). The risk of losing office now features prominently in
the explanation of conflict (Bueno de Mesquita and Siverson, 1995; Bueno de Mesquita et al.,
1999, 2003; Goemans, 2000b; Siverson, 1996). Notwithstanding many decades of research
on diversionary war, empirical support for the proposition that leaders resort to conflict when
facing problems at the domestic scene has been decidedly inconclusive (Leeds and Davis, 1997;
Levy, 1989; Gelpi, 1997b; Miller, 1999; Mansfield and Snyder, 1995). We believe that one
reason for the lack of strong empirical findings can be found in the assumption common to
almost all empirical studies that losing office is the worst that can happen to leaders. With
the exception of Goemans (2000b,a), almost all studies ignore the consideration that leaders
might be punished after leaving office. Most European and North American heads of state
15
can indeed look forward to a comfortable retirement in the wake of a foreign policy failure.
C.f., for example, the Dutch Prime Minister Wim Kok, who resigned due to the parliamentary
discussion of Dutch responsibility in the Srebrenica massacre in Bosnia, but subsequently went
on to hold important and very well-paid positions on the boards of several large companies and
in the European Union. By contrast, an autocratic ruler like Saddam Hussein faces a very high
likelihood of post-tenure punishment. Taking into account the probability and expected costs
of punishment can make it rational for a leader to initiate force or continue to stand firm in
crises, in ways that may seem very risky gambles to scholars who consider only the probability
of losing office.
The new information in Archigos can also help suggest new research questions and puzzles that
have not yet received much attention. For example, what is the fate of authoritarian leaders
when, finally, their dictatorship collapses, and what does their fate presage for the stability of a
new democracy? Huntington (1991, 231) suggested that for careful democratic leaders who are
trying to balance the trade-off between justice and freedom, “the least unsatisfactory course may
well be: do not prosecute, do not punish, do not forgive, and, above all, do not forget.” Although
Huntington’s (1991) claim is provocative and controversial, it has remained largely unexplored.
Kaminski et al. (2006, 298) note that the lack of systematic research on this topic might reflect
the belief that each transition is the outcome of a unique experience. The Archigos data can
contribute to the identification of more systematic patterns and help assess the implications of
the punishment of authoritarian leaders for the longevity of a democratic transition and quality
of democratic institutions.
16
In Table 6, we take a preliminary stab at illustrating the relationship between the fate of lead-
ers and the stability of democracy. From 1974, the conventional starting year of the third wave
as we show in Table 6, the non-democratic leaders managed to avoid any personal punishment,
and only a handful were sent to exile or prison (17.9%). Figure 1 presents the estimates of
the Kaplan-Meier survival probabilities for the third wave democracies. It shows a marginally
higher failure propensity for transitions in which the authoritarian leaders suffered some form
of punishment.
Consideration of the manner of losing office and the leader’s post-exit punishment opens
up other questions. Consider, for example, the very high frequency of post-tenure punish-
7 Following Jagger and Gurr (1995, 474), we count as transitions to democracy cases where
the democracy minus autocracy scale in the Polity data crosses the threshold of +7 (see also
Atkinson, 2006, 517–518). We recognize that the Polity data identify some potentially contro-
versial transitions. For example, the transfer of power Yeltsin to Putin is considered a transition
to democracy, as the Polity score for Russia increases from +4 to +7 on 25 March 2000, re-
contrast, the Freedom House data codes Russia as becoming increasingly authoritarian from
1998 and on. However, the potential controversies here stem from the Polity data, rather than
Archigos, and for the purposes of this example we simply use the transitions identified by Polity
as given.
17
ment in many states. In Haiti, no fewer than 64% of previous leaders have suffered post-exit
punishments. Between Presidents Nord, who was removed in December 1908, and Velbrun-
Guillaume, who was removed in July 1915, Haitian presidents were, successively, exiled, ex-
iled, bombed and blown up, imprisoned, exiled, executed, exiled, and, particularly gruesome,
dragged from the French legation by an angry mob and “impaled on the iron fence surrounding
the legation and torn to pieces”(Lentz, 1999, 219). These findings raise the obvious question
who would want to take such a risky job as being the president of Haiti? Indeed, Velbrun-
Guillaume clearly knew the risks, since his father, Teresias Simon Sam, who had also been
president, had been forced into exile. One possible answer may be that even if the risk of death
as President is high in states such as Haiti, the opportunity costs of not being in power may
be even higher. A contender not in power risks persecution from ruling leaders, and the risk
of post-tenure punishment is not necessarily worse, considering the alternatives. Even though
holding office is risky, it also offers the possibility of accumulating vast amounts of personal
wealth, and the prospects of such personal gain might make assuming the presidency suffi-
ciently attractive, even if a safe life after the loss of office cannot be guaranteed. Although
more theoretical work is needed to understand the incentive structures of contenders in risky
and unstable states, Archigos provides an empirical basis for research along these lines.
Archigos makes it possible to identify political instability other than changes in a country’s
overall degree of democracy. Almost all work on institutional features in international re-
lations and cross-national research over the last two decades has focused on the distinction
18
tively well-defined category and different definitions by and large classify the same states as
not. As such, a number of very different types of political systems are often lumped together as
autocracies such as the Soviet Union, fascist regimes such as Nazi Germany, and kleptocracies
such Mobutu’s Zaire, which have little in common apart from not being democratic. Moreover,
a great deal of instability and changes between quite different regimes would not appear as
Consider the revolutions in Cuba and Iran, in 1959 and 1979 respectively, which entailed
fundamental political changes. Since both states remain “non-democracies” before and after
their revolutions, however, neither would qualify as undergoing a regime change by common
Some researchers have tried to identify variation within autocratic regimes and changes be-
tween different autocratic regimes by looking at the structure of their institutions (i.e., whether
they have a personalistic government with a single ruler, or a rule by a dominant party?) or
identifying large changes in policy. However, these approaches to identify regime changes tend
to rely on post hoc classifications of whether we see large or influential changes or not. 9 Ob-
8 In the Polity data, for example, Iran is assigned a Polity score (institutionalized democracy–
autocracy) of -10 before the 1979 revolution and a score of -6 after 1981. Likewise, Cuba’s
associated with changes in ruling coalitions. Many leaders that have pursued centralized eco-
nomic planning have later enacted privatization and market reforms when opportune, as seen
19
viously, not all leader changes can be considered a change in the ruling coalition and regime
changes. Papa Doc’s transfer of power to his son Baby Doc can hardly be characterized as a
change of regime. But Archigos allow identifying changes in ruling coalitions or regimes in
autocracies by examining whether leader entry and exit occur in a prescribed (as in the case of
transfers to a designated successor) or an irregular manner (as in the case of the Iranian revolu-
tion). More specifically, irregular regime transitions can be identified in cases where leaders are
removed irregularly relative to the prevailing rules and practices of a state and the new leader
enters irregularly. We add the clause on irregular leader entry, to exclude cases where leaders
are forced to resign or removed irregularly, but where a designated vice-president then assumes
Table 7 compares the population of years with and without irregular transitions, measured
as irregular exit followed by irregular entry within a window of twelve months, to years in
which we observe transitions in the Polity data, in terms of changes that move countries to and
from the threshold for democracy. As can be seen, we have almost 50% more cases of irregular
transitions (277) than we have cases of transitions between democracy and autocracy (190).
Moreover, there is relatively little overlap between the cases of irregular regime transitions as
defined by leader entry and exit and transitions that to and from democracy in the Polity data.
This strongly reinforces our argument that a great deal of significant political change is simply
in Vietnam in the 1990. Moreover, rulers sometimes find it helpful to set up ruling parties and
other institutions. For example, President Calles of Mexico founded the party later known as
the Institutional Revolutionary Party (PRI) to stop the violent struggle for power among fac-
tions in the wake of the revolution. His influence in the party then allowed him to dominate the
20
not reflected in the Polity data. Gleditsch and Choung (2006) conduct an empirical analysis
of autocratic regime crises and the likelihood of transitions to democracy or new autocracies.
Similarly, studies of political instability based on Archigos should also be helpful for studies
of the effects of political instability on growth and conflict (Feng, 1997; Mansfield and Snyder,
1995, 2005).
Archigos makes it possible to probe beneath the country- or country-dyad-specific patterns that
leader-dyads introduces additional explanatory variation into the interaction between countries
and thereby can help throw new light on some previously unexplained patterns. For example,
Archigos makes it possible to disaggregate the study of conflict by examining how individ-
ual leaders may acquire reputations in deterrence interactions or in the face of agent-specific
Huth (1997, 43) suggests that reputations pertain to leaders more than to countries:
The importance of reputations may well fade as the international strategic environ-
ment changes over time and as new leaders assume power within defender states
of defenders in this approach, the first requirement will be to construct a data base
21
Archigos helps fill this gap.
To take heed of Huth’s (1997) suggestion, we have identified the leaders who were involved
in the 122 repetitive international crises compiled by Gelpi (1997a). That data illustrates how
challengers’ choices to escalate, back down or seek a compromise solution are mediated by two
factors: the defenders’ strategies, on the one hand, and the presence or absence of a previous
settlement agreement. Gelpi’s (1997a) findings show that normative factors affect crisis bar-
gaining behavior. But is there more to learn from those crisis interactions if we switch focus to
In Table 8, we observe that the largest proportion of repetitive crises involves the same
challenger squaring off with the same defender (42.62%), and nearly 78% of those deterrence
encounters involved at least one leader who had been involved in a previous interaction. Bar-
gaining choices also reflect the history of previous interactions. For example, one of the major
findings in Gelpi’s (1997a) study is that the presence of a legitimate settlement would induce
the challengers to comply or seek a compromise solution to the crisis. A leader-based analysis
qualifies this finding by showing that this pattern only pertains to crises that are re-initiated by
the same leader who had been involved in the previous interaction. In Table 9 we observe that
new challengers entirely disregard the presence of a previous settlement when they determine
their final response in a crisis. The normative force of a crisis settlement, in other words, does
not carry over to new challengers, who would likely attempt to change the status-quo out of
22
5 Conclusion
The Archigos data constitute the most extensive systematic data set on political leaders. Unlike
other alternatives, it provides a great deal of additional information about leaders, including
how they acquire and lose office, as well as what happen to leaders after they leave office. We
have shown that this information is associated with striking differences in political behavior and
the likelihood of particular events. To recap, leaders who enter power in an irregular manner
are also much likely to leave power irregularly and have a higher risk of experiencing civil war,
whereas leaders that exit irregularly are much more frequently punished after leaving office.
Irregular leader changes help identify political change in autocracies not apparent from data that
solely consider whether institutions are democratic. Transitions to democracy in the third wave
have failed more often in instances where autocratic rulers were punished after leaving office.
We have shown that a shift from country-dyads to leader-dyads makes it possible to uncover
new patterns and offer a finer-grained explanation of international crisis behavior. Specifically,
we showed that crisis interactions differ depending on whether leaders have faced each other
previously. These are all new empirical findings that could not be explored in existing data
sources. Although selective, our overview shows how the more comprehensive information on
leaders and their characteristics in Archigos bear considerable promise of providing answers to
new and old research questions, and open up new avenues for research on individual leaders as
decision-makers.
23
References
Ames, Barry. 1987. Political Survival: Politicians and Public Policy in Latin America. Berke-
Bates, Robert. 1981. Markets and States in Tropical Africa: The Political Basis of Agricultural
Belkin, Aron and Evan Schofer. 2003. “Toward a Structural Understanding of Coup Risk.”
Belkin, Aron and Evan Schofer. 2005. “Coup Risk, Counterbalancing, and International Con-
Bennett, D. Scott and Allan C. Stam. 2004. The Behavioral Origins of War. Ann Arbor:
Bernhard, William and David Leblang. 2002. “Democratic Processes, Political Risk, and For-
Betts, Richard K. and Samuel P. Huntington. 1985/86. “Dead Dictators and Rioting Mobs: Does
10(3):112–146.
Bienen, Henry and Nicholas van de Walle. 1991. Of Time and Power: Leadership Duration in
24
Blondel, Jean. 1987. Political Leadership: toward a general analysis. London: Sage Publica-
tions.
Bueno de Mesquita, Bruce, Alastair Smith, Randolph M. Siverson and James D. Morrow. 2003.
Bueno de Mesquita, Bruce, James D. Morrow, Randolph M. Siverson and Alastair Smith. 1999.
“An Institutional Explanation of the Democratic Peace.” American Political Science Review
93(4):791–807.
Bueno de Mesquita, Bruce and Randolph M. Siverson. 1995. “War and the Survival of Political
Bueno de Mesquita, Bruce and Randolph M. Siverson. 1997. “Nasty or Nice? Political Sys-
tems, Endogenous Norms and the Treatment of Adversaries.” Journal of Conflict Resolution
41(2):175–199.
Bunce, Valerie. 1981. Do New Leaders Make a Difference? Executive Succession and Public
Policy under Capitalism and Socialism. Princeton, N.J.: Princeton University Press.
Cheibub, José Antonio and Adam Przeworski. 1999. Democracy, Elections, and Accountabil-
ity for Economic Outcomes. In Democracy, Accountability, and Representation, ed. Adam
Przeworski, Susan C. Stokes and Bernard Manin. Cambridge: Cambridge University Press
pp. 222–249.
Chiozza, Giacomo and Ajin Choi. 2003. “Guess who Did What: Political Leaders and the
25
Management of Territorial Disputes, 1950-1990.” Journal of Conflict Resolution 47(3):251–
78.
Chiozza, Giacomo and H. E. Goemans. 2003. “Peace through Insecurity: Tenure and Interna-
Colaresi, Michael. 2004. “Aftershocks: Postwar Leadership Survival, Rivalry, and Regime
David, Steven R. 1991. Choosing Sides: Alignment and Realignment in the Third World. Bal-
de Marchi, Scott. 2005. Computational and Mathematical Modeling in the Social Sciences.
Diermeier, Daniel and Randy T. Stevenson. 1999. “Cabinet Survival and Competing Risks.”
Dodd, Lawrence C. 1984. “The Study of Cabinet Durability: Introduction and Commentary.”
Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harpers and Collins.
Downs, George and David M. Rocke. 1994. “Conflict, Agency and Gambling for Resurrec-
tion: The Principal–Agent Problem Goes to War.” American Journal of Political Science
38(2):362–380.
26
Fearon, James D. 1994. “Domestic Political Audiences and the Escalation of International
Fearon, James D. and David D. Laitin. 2003. “Ethnicity, Insurgency, and Civil War.” American
Feng, Yi. 1997. “Democracy, Political Stability and Economic Growth.” British Journal of
Gelpi, Christopher. 1997a. “Crime and Punishment: The Role of Norms in Crisis Bargaining.”
Gelpi, Christopher. 1997b. “Democratic Diversions: Government Structure and the External-
Gelpi, Christopher and Joseph M. Grieco. 2001a. “Attracting Trouble: Democracy, Leader-
ship Tenure, and the Targeting of Militarized Challenges.” Journal of Conflict Resolution
45(6):794–817.
Gelpi, Christopher and Joseph M. Grieco. 2001b. “Democracy, Interdependence, and the Lib-
eral Peace.” Paper prepared for delivery at the 2001 Annual Meeting of the American Political
Gleditsch, Kristian S. and Michael D. Ward. 1999. “A Revised List of Independent States since
Gleditsch, Kristian Skrede. 2004. “A Revised List of Wars Between and Within Independent
27
Gleditsch, Kristian Skrede and Jinhee Lee Choung. 2006. “Autocratic Transitions and Democ-
Goemans, H. E. 2000a. “Fighting for Survival, The Fate of Leaders and the Duration of War.”
Goemans, H. E. 2000b. War and Punishment: The Causes of War Termination and the First
Grofman, Bernard and Peter Van Roozendaal. 1997. “Modelling Cabinet Durability and Termi-
Hegre, Håvard, Tanja Ellingsen, Scott Gates and Nils Petter Gleditsch. 2001. “Toward a Demo-
cratic Civil Peace? Democracy, Political Change, and Civil War, 1816-1992.” American
Horowitz, M., R. McDermott and Allan C. Stam. 2005. “Leader Age, Regime Type, and Violent
Huntington, Samuel P. 1991. The Third Wave: Democratization in the Late Twentieth Century.
Jagger, Keith and Ted Robert Gurr. 1995. “Tracking Democracy’s Third Wave with the Polity
Jones, Benjamin F. and Benjamin A. Olken. 2005. “Do Leaders Matter? National Leadership
and Growth since World War II.” Quarterly Journal of Economics 120(3):835–864.
28
Kaminski, Marek M., Monika Nalepa, and Barry O’Neill. 2006. “Normative and Strategic
Kayser, Mark A. 2005. “Who Surfs, Who Manipulates? The Determinants of Opportunis-
tic Election Timing and Electorally Motivated Economic Intervention.” American Political
King, Gary, James E. Alt, Nancy E. Burns and Michael Laver. 1990. “A Unified Model of
34(3):846–871.
Lai, Brian and Dan Slater. 2006. “Institutions of the Offensive: Domestic Sources of Dispute
Leblang, David and William Bernhard. 2000. “The Politics of Speculative Attacks in Industrial
Leeds, Brett Ashley and David R. Davis. 1997. “Domestic Political Vulnerability and Interna-
Lentz, Harris M. 1994. Heads of States and Governments: A Worldwide Encyclopedia of Over
Lentz, Harris M. 1999. Encyclopedia of Heads of States and Governments, 1900 through 1945.
Levy, Jack S. 1989. The Diversionary Theory of War: A Critique. In Handbook of War Studies,
ed. Manus I. Midlarsky. Ann Arbor, MI: University of Michigan Press pp. 259–188.
29
Londregan, John B. and Keith T. Poole. 1990. “Poverty, The Coup Trap, and the Seizure of
Mansfield, Edward D. and Jack Snyder. 1995. “Democratization and the Danger of War.” Inter-
Mansfield, Edward D. and Jack Snyder. 2005. Electing to Fight: Why Emerging Democracies
Marinov, Nikolay. 2005. “Do Economic Sanctions Destabilize Country Leaders?” American
McGillivray, Fiona and Alastair Smith. 2000. “Trust and Cooperation through Agent Specific
Miller, Ross A. 1999. “Regime Type, Strategic Interaction, and the Diversionary Use of Force.”
Muller, Edward N. and Erich Weede. 1990. “Cross-National Variation in Political Violence: A
Przeworski, Adam and Henry Teune. 1970. The Logic of Comparative Social Inquiry. New
York: Wiley.
Przeworski, Adam, Michael E. Alvarez, José Antonio Cheibub and Fernando Limongi. 2000.
Democracy and Development: Political Institutions and Well-Being in the World, 1950-1990.
30
Schultz, Kenneth A. 2001b. Democracy and Coercive Diplomacy. Cambridge: Cambridge
University Press.
Siverson, Randolph M. 1996. “Thinking about Puzzles in the Study of International War.”
Smith, Alastair. 2003. “Election Timing in Majoritarian Parliaments.” British Journal of Poltical
Science 33(3):397–418.
Snyder, Richard Carlton, H. W. Bruck and Burton Sapin, eds. 1962. Foreign Policy Decision-
Making; An Approach to the Study of International Politics. New York: Free Press of Glen-
coe.
Stokes, Susan C. 2001. Mandates and Democracy; Neoliberalism by Surprise in Latin America.
Vreeland, James Raymond. 2005. “A Problem with Polity: Unpacking Anocracy.” Typescript,
Waltz, Kenneth N. 1979. Theory of International Politics. New York: Random House.
Warwick, Paul V. 1994. Government Survival in Parliamentary Democracies. New York, NY:
Wintrobe, Ronald. 1998. The Political Economy of Dictatorship. New York: Cambridge Uni-
versity Press.
31
Figures and Tables
0 5 10 15 20 25 30
Years
Punished?
No Yes
1
Table 1: Previous number of spells in office by manner of entry
2
Table 2: Type of exit by number of previous spells in office
3
Table 3: Manner of exit by manner of entry
4
Table 4: Civil war by manner of entry
5
Table 5: Post-tenure fate by manner of entry
6
Table 6: Democratic Transitions and the Fate of Leadersa
8
Table 8: Leaders in Repetitive International Crises, 1929–1979
Defender Challenger
Different Same Sum
Different 27 (22.13) 24 (19.67) 51 (41.80)
Same 19 (15.57) 52 (42.62) 71 (58.20)
Sum 46 (37.70) 76 (62.30) 122 (100%)
9
Table 9: Challenger Resolve and Previous Settlement
Same Different
Challenger Challenger
Settlement Settlement
No Yes Sum No Yes Sum
Compliance 9 11 20 5 11 16
(24.32) (28.21) (26.32) (27.78) (39.29) (34.78)
Compromise 2 12 14 3 5 8
(5.41) (30.77) (18.42) (16.67) (17.86) (17.39)
Intransigence 26 16 42 10 12 22
(70.27) (41.03) (55.26) (55.56) (42.86) (47.83)
Sum 37 39 76 18 28 46
(100) (100) (100) (100) (100) (100)
χ2 9.68 p=.008 .80 p=.672
10