When governments “rebel”: How internal power struggles can make

When governments “rebel”:
How internal power struggles can make
governments reignite civil war
Joakim Kreutz
Department of Peace and Conflict Research, Uppsala Universitet1
This is work in progress and comments are appreciated. Please contact the author for correct citation and most
recent version – [email protected]
Abstract
This article presents a new line of inquiry into the issue of civil war recurrence, focusing on specific motivations for
governments to reignite Civil War. The article argues that when governments are vulnerable to internal power struggles,
they may decide to reignite a dormant civil war to obtain the loyalty of key allies, in particular the armed forces. This
proposition is tested on all cases of civil war recurrence 1970-2002 and is consistently supported even when controlling for
a plethora of potential confounding factors. In extending the analysis, the study finds that even though the vulnerability to
intra-government power struggles is correlated to government-initiated civil war recurrence, it does not increase the risk of
rebel-initiated or tit-for-tat resumption of civil wars.
I wish to thank Mats Hammarström, Magnus Öberg, Kristine Eck, Hanne Fjelde, Halvard Buhaug, Daniel Strandow, Margareta
Sollenberg, David R. Davis and participants at the International Studies Association Annual Convention 2010 for helpful comments
and suggestions on earlier drafts of this paper.
1
As night fell on 7 August 2008, Georgian forces launched an artillery barrage on the rebel-controlled territory of
South Ossetia followed by a ground incursion the following day. The offensive reignited the third round of civil
war between these belligerents in less than two decades of Georgian independence. Why is it that some
countries, like Georgia, repeatedly fail to resolve intrastate disputes while others can overcome decades of
destructive conflict without recurrence? A growing literature has identified explanatory factors relating to the
conflict issues, the previous outcome of fighting, and the citizens’ quality-of-life as important for why renewed
rebellion is more or less likely (see Mukherjee 2006, Walter 2004.) However, a common assumption in this
literature is that the decision to resume warfare always initiates from the rebel side; a claim that is not consistent
with empirical evidence. Drawing on detailed case histories, recurrence of civil war can be identified through
three different pathways. The first consist of moves by former rebels to reorganize and ambush government
positions and troops. The second is offensive actions from both former belligerents, escalating from verbal
provocations to violent interaction, so-called tit-for-tat resumption of fighting. The third path, which is the focus
for this paper, are government attacks on rebel camps or into territory controlled by former rebels.
In previous research on civil war recurrence, scant attention has been given to the role of the
2
government. Scholars’ have noted that governments may spoil ongoing peace processes (Höglund and Stedman
2006), or implement policies that produce grievances and opportunities for rebels to remobilize (Collier,
Hoeffler and Söderbom 2008; Quinn, Mason and Gurses 2007), but there has been little investigation into the
issue of why governments choose to reignite the warfare. This oversight follows a simple logic: it should always
be in a government’s interest to avoid civil war because of how a conflict destabilizes a country. Governance and
economic life becomes more difficult, resources needs to be diverted to maintain the war effort, and the
constituency may become increasingly critical of the regime. Despite this, empirics indicate that every year on
average at least one civil war resume through government offensives against former adversaries.
This paper examines when governments decide to reignite a dormant civil war. I contend that
there is one particular situation when it can be rational for government leaders to choose this strategy; when it is
ravaged by intra-government power struggles. I argue that when governments are vulnerable to such internal
infighting, factions seek support from key actors outside the government which creates incentives for leaders to
resume a civil war. The logic for this is that individuals or intra-party factions that want to challenge the
incumbent can not openly announce their ambitions but need to covertly assemble a power base. The de facto
ascent to power is controlled by either, or a combination of, the public opinion, party and business elites, and
commanders of the armed forces, and thus the incumbent and challengers will compete for their support. I
argue that in this high-stakes competition for the support of these external key actors, the proposition of an
armed assault on former rebels can be a useful strategy since it can provide spoils, a rally-round-the-flag-effect,
and lead to the strategic deployment of non-trustworthy troops away from the capital.
The diversionary war literature has theorized why governments may decide to start – but not resume - conflicts, but has primarily
focused on interstate disputes or a sub-set of intrastate ethnic competition (de Figueiredo and Weingast 1999, Lai and Slater 2006,
Lake and Rothchild 1996, Miller 1995, Powell 2009, Tir and Jasinski 2008, Weeks 2009.)
2
The aim of this study is thus to answer the question, why do governments decide to reignite civil war? This
project adds to the scholarly understanding of civil war recurrence in several ways. First, it moves beyond the
prevailing assumption that only rebels restart civil wars and draws attention to the need for a more nuanced view
of the actors involved. Second, it identifies intra-elite competition as a potential source of aggressive behavior
from a state against opponents that have few or no connections to these elites. Third, this paper presents unique
new data that identify different pathways to civil war recurrence and employs this data for an empirical
investigation into the behavior of governments in post-conflict settings. Spanning the 1970-2002 period, this
analysis finds support for the proposed relationship between a country’s vulnerability to intra-government
competition and decisions of governments to reignite civil war. This result holds even when the empirical
investigation is expanded to include possible factors that may confound the findings; these include conflict
specific and institutional features that could provide alternative incentives for why governments may decide to
reignite civil war. Finally, these findings also have potential implications to the larger literature on diversionary
theory and conflict onset. Whereas existing theorizing almost exclusively have focused on democratic leaders
and/or ethnic competition, my results indicate that intra-elite competition increases the risk that governments
regardless of regime type will “pick a fight” against any domestic opposition. This challenges conceptions of
ethnic competition as different from other settings and suggest that government strategy may be more
determined by opportunities for aggressive policies than incentive structures.
Rationality and civil war recurrence
Existing scholarship on civil war recurrence generally departs from a rational-choice perspective that suggests
that the start, end, or resumption of conflicts is determined by belligerents view on whether there is more to
gain from fighting than from sustaining the peace (Mason and Fett 1996; Quinn, Mason and Gurses 2007).
However, the empirical investigation of this cost-benefit analysis has almost exclusively focused on the rebel
side. This is based in the logic that the government consistently wants to avoid civil war recurrence, while the
former rebel side has a more instrumental approach towards the use of violence. Apart from the inherent
human costs of fighting and its’ detrimental effect on economic development and societal structures, a civil war
portrays the government as weak since it indicates poor legitimacy, an inability to repress the opposition, or
both, which in the long run will further weaken the government vis-à-vis competitors. This sensitivity to the
onset or recurrence of civil war can be used strategically by non-government actors to force concessions simply
through threats or limited use of violence (Boyle 2009; Hultman 2009; Pape 2003.) This does not suggest that
the government consistently is portrayed as without influence in a post-conflict setting. In order to avoid a
resumption of a conflict, the government can impose democratic rule, ensure economic development, or offer
concessions in the form of autonomy or power-sharing that removes the incentives for rebels to leave their
civilian lives (Collier, Hoeffler and Söderbom 2008; Hartzell 2009; Licklider 2002; Walter 2004.) At the same
time, the government can also increase the costs for the opposition by limiting the opportunities for rebels to
remobilize (Fuhrmann and Tir 2009; Quinn, Mason and Gurses 2007).
For governments, the experience of civil war thus incurs several costs while there are few, if any,
obvious benefits. Since the primary goal for a government is to maintain its hold on power, a decision to reignite
civil war is generally not rational. When faced with opposition, it is preferable to employ strategies such as
offering concessions, to increase the legitimacy of the regime, or repression, to defeat the opposition while it
remains weak (Enterline and Gleditsch 2000; Gandhi and Przeworski 2006; Lacina 2009). However, such
strategies are mainly useful if the opposition that needs to be appeased or repressed is clearly defined and
emancipates from outside the ruling coalition itself. I argue that during intra-government power struggles, it is
difficult to use such low-cost strategies. In such situations, it may be rational for a government faction to initiate
a civil war as a means of ensuring the loyalty of key constituents and thereby defeating intra-government
competitors.
The outcome of an intra-government power struggle depends on which faction can credibly
project that it is most capable of ruling the country. For simplicity, I distinguish between so-called incumbent
and challenger factions in the discussion below. Whereas the issue of intra-government power struggles
primarily has been investigated in authoritarian settings (Gandhi and Przeworski 2007; Svolik 2009; Tullock
1987; Weeks 2009), the incentive structures for the phenomenon should be the same regardless of regime type
(see Balán 2009; Belloni and Beller 1976.) At the heart of the argument is the assumption that political leaders
consistently want to maximize their power and influence over government policy. Thus, there will always be
incentives for challengers to try to usurp their superiors and for the incumbents to try to prevent this (Tullock
1987). Whereas the ascent to power in authoritarian settings most often consist of a coup d’etat, intragovernment challengers in a democracy are more likely to focus on party congresses or similar established
processes. For example, challengers within the Indian Congress Party unsuccessfully tried in 1995 to force Prime
Minister Narasimha Rao to convene a meeting with the All-India Congress Committee for a vote on his position
as leader of the party (Andersen 1996.)
The characteristics of intra-government challengers make it more difficult for the incumbent to
offer policy concessions or distribute spoils without weakening his or her position. First, the challengers with the
most realistic chance of success will already be in positions close to the incumbent. As the incumbent is aware
that a challenge is most likely to emancipate from the upper echelons of the bureaucracy, he or she will have
incentives to rely on a small coalition whose loyalty easily can be assured by perks and privileges or monitored
by the security apparatus (Bueno de Mesquita et al. 2003.) The members of this coalition will also have been
promoted by the incumbent under the explicit assumption that they are loyal and will not try to challenge his or
her rule. The downside is that small coalitions are more vulnerable for so-called palace coups as the leader have
little support in society (Lai and Slater 2006). Second, the challengers must conceal their ambition from the
incumbent until the plan is executed. If the incumbent learns about plans to replace him or her, the challengers
will be arrested or deposed beforehand. This creates a dilemma for the challengers, as they need to
simultaneously seem weak in the eyes of the incumbent in order to not be seen as a threat while convincing the
necessary supporters that they are stronger and more capable than the incumbent as a ruler. Third, challengers
have incentives to vilify other challengers and misrepresent themselves as being loyal to the incumbent, even if
their true intentions are aimed at becoming a challenger. If an individual challenger learns that other challengers
are plotting against the incumbent, he or she can gain rewards from the incumbent by revealing the plot.
However, the incumbent does not know whether the person revealing this information is doing so as a sign of
loyalty or simply an attempt to remove other competing factions. Indeed, the incumbent can not even know
whether the revealed plot actually existed, or if the accusations are part of an attempt to remove his or her
loyalists. As a consequence, the key facet of intra-government power struggles is that challengers conceal their
plans from the incumbent and other members of the ruling coalition while trying to assure necessary support
from key actors external to the government to present themselves as a better alternative than the incumbent
ruler.
I argue that this intra-governmental competition for support from key actors creates incentives for
government members to reignite a civil war. Even though the incentives for internal power struggles always
exist, the government is most vulnerable for these during and immediately after an attempted reshuffle. When
an incumbent is challenged, he or she has nothing to lose by gambling on a risky strategy to potentially remain in
power. Similarly, if a challenger manages to usurp the incumbent, it is possible that a counter-attempt is
launched or that other challengers see an opportunity to take power before the new regime is formalized. As the
outcome of the intra-government struggle effectively is determined by which faction is most successful in
creating alliances with key supporters, the costs of reigniting civil war will be outweighed by the gains of political
survival for the individual leader. The resumption of a civil war may be an attractive prospect for incumbents
and aspiring challengers as this can serve as both a direct and indirect path to rally support to his or her side in
the intra-government power struggle. During conflict, it is possible for political leaders to offer spoils in the
form of promotions and loot to key individuals in the military, as well as increased resources and autonomy to
the military as an institution (Pion-Berlin 1992.)
In addition to providing potential spoils to supporters, the resumption of conflict may also
increase the public popularity of an individual thus making his or her removal more costly and less likely. It has
been suggested that politicians can emphasize potential threats to citizens to create a so-called “rally around the
flag” effect (Downs and Rocke 1994; Smith 1996). In a post-conflict society, the conflict history provides a
precedent that confirms the credibility of the threat while government forces can be assumed to be militarily
superior to the – partly demobilized – rebels. Even though this tactic most often is available to incumbents, it
can also be used by intra-governmental challengers. A challenger or incumbent who claims that it is necessary to
pre-emptively strike against such threats will seem strong and capable, and may be able to rally significant
supporters to his or her side. For the military, supporting this faction will not be controversial since it
corresponds with its’ mandate and self-image of being the protector of the state. Following accusations of
corruption in late 2007, Georgian President Mikheil Saakashvili responded with repression and by denouncing
his critics as an attempted “Russian-backed coup” (ICG 2007.) Continuing the anti-Russian rhetoric, he then
quickly called and won Presidential elections while Georgian troops launched attacks on South Ossetian rebel
territory along the Russian border which eventually led to the resignation of the civil war.
Finally, sometimes the aim for any of the competing government factions is not to secure the
support from the military but rather to ensure that the armed forces remain excluded from the power struggle. If
the incumbent or challengers primarily have gathered a strong position for example within a party, they will
prefer that the power struggle is settled through a congress or a steering committee meeting without
involvement of the armed forces. Reigniting a civil war in such a situation will provide an opportunity to deploy
troops at a distance from the capital, as well as increase the rivalries between branches to make the military less
coherent and thereby a less important ally for the upcoming power struggle. A decision to reignite civil war can
obviously be motivated by a combination of these reasons.
In sum, I content that it can be rational for a government to decide to reignite a civil war is when
this can help a faction survive a intra-government power struggle. This does not imply that governments always
will choose this strategy when faced with such a situation. If that was the case, governments would reignite civil
war more often than what is observed. Indeed most governments try all other possible measures that are less
costly, such as policy concessions, distribution of spoils, and repression to prevent or win intra-government
power struggles. The decision to reignite a civil war will only be attractive when the stakes are high, such as
when the government factions are unsure about who will be supported by the military, when defeat in the power
struggle is costly, and when other options are unlikely to have an effect. Thus, the theoretical argument
presented here leads to one main empirical prediction:
H1: The more vulnerable a government is to intra-governmental power-struggles, the more likely it is to reignite civil war.
Conflict and Institutional Specifications
Although there is a rich and growing literature focusing on intra-governmental features and conflict propensity,
no previous study has explored the phenomenon of government reigniting civil war. However, in order to
explore the relationship between sensitivity to intra-governmental power struggles and a government’s decision
to reignite civil war, I have reviewed related scholarly work and identified factors that may confound the
findings of the theoretical argument presented here. These factors can be divided into two sets of specifications.
I first discuss conflict specific factors and thereafter institutional features that could be confounding factors.
Conflict-specific factors
In existing work on civil war recurrence, it has been acknowledged that some conflict issues may be more
difficult to settle than others (DeRouen and Bercovitch 2008). In particular it is argued, that conflicts where the
parties consider the stakes to be high or where the issues are fundamentally irreconcilable are more difficult to
settle and more likely to resume (Crocker, Hampson, and Aall 2005). One aspect that would be of particular
interest for the government to control is a territory rich in natural resources. In a country where the economy is
dependent on the extraction of natural resources, other sectors of the economy are often underdeveloped,
increasing the relative value of the resource-rich region. In addition, extraction of rents from these resources
may be crucial for an individual in power to distribute rewards to his or her loyalists (Le Billon 2004).
H2a: Governments are more likely to reignite civil war if the conflict concerns territories rich in natural resources.
Walter (2006) presents an extension of the argument relating to the role of resources. She argues that even
territories lacking resources are sensitive for the government because it is concerned with the reputation costs of
conceding to any separatist demands. If the government allows a separatist movement to secede or flourish in a
non-strategic area, this will encourage separatist movements in other, more important, areas. Thus, it is rational
to react harshly towards all separatist moves regardless of the value of the territory at stake.
H2b: Governments are more likely to reignite civil war if the conflict concerns territory.
The final conflict-specific feature that can increase the likelihood that governments decide to reignite civil wars
concerns how the earlier conflict ended. Within the civil war recurrence literature, findings of whether victories
or peace agreements lead to most post-conflict stability are inconclusive (Fearon and Laitin 2008; Licklider 1995;
Kreutz 2010). There is, however, a general agreement that the outcome of the previous conflict is the most
powerful factor to signal future intentions of a warring side (Hoddie and Hartzell 2005). In situations when the
previous fighting ended without a decisive outcome such as a clear victory or the signing of a peace agreement,
there will be more concern that the opponent is preparing for a new offensive. Furthermore, a decisive outcome
will also provide obstacles for a government faction to motivate the need for an armed assault on the former
rebels. Following a victory, former rebels should not pose a relevant threat and after a peace agreement, the
rebels will be able to signal their commitment to peace.
H2b: Governments are more likely to reignite civil war in the absence of a decisive settlement of the preceding conflict..
Institutional features
Scholarship on institutional features and war-propensity has primarily focused on international war (Bueno de
Mesquita et al. 1999; Downs and Rocke 1994; Lai and Slater 2006, Mansfield and Snyder 2002; Miller 1995;
Weeks 2009) or domestic competition specified along ethnic lines (De Figueriredo and Weingast 1999; Fearon
and Laitin 1996; Lake and Rothchild 1996; Tir and Jasinski 2008). Thus, the arguments presented in this section
are taken from studies concerned with different phenomenons than civil war recurrence. Drawing on the logic
presented in this literature, I have identified factors that potentially could influence governments’ decision to
reignite civil war.
There are two potential ways for attaining government power in democratic states: through intraparty/state apparatus promotion and by winning elections whilst only the former is applicable in authoritarian
systems. This would suggest that the stakes are higher during intra-governmental power struggles in
authoritarian systems since the consequences of losing carry greater costs for the individual, most likely resulting
in death, imprisonment, or forced exile (Goemans 2008; Tullock 1987). Furthermore, since authoritarian rulers’
constituency consists of business and bureaucratic elites rather than the entire population, they should have
fewer qualms about launching military attacks domestically (Miller 1995; Weeks 2009).
H3a: Authoritarian governments are more likely to reignite civil war.
Recent research has focused on disaggregating the study of authoritarian regimes, and argued that there are
important institutional differences among this category. In particular, it has been suggested that regimes
grounded in a secure institutionalized setting will be less vulnerable to intra-government power struggles (Fjelde
2009.) Lai and Slater (2006) argue that there is a pronounced difference between regimes that rests primarily on
party or military institutions. In a one-party state, the steps for advancing to the top of the party/state
bureaucracy are dependent on coherence to party discipline, and thereby decreasing the risk for overt power
struggles. Furthermore, in many instances the advancement is solely based on longetivity, thereby making
attempts to usurp superiors fruitless. In military regimes, or civilian regimes dependent on the military, there is a
lack of institutional capacity which manifests itself in little or no ability for challengers to credibly follow a stepby-step rise to the power rather than attempt to immediately usurp their superiors (Lai and Slater 2006).
H3b: One-party governments are less likely to reignite civil war.
H3c: Military governments are more likely to reignite civil war.
Another line of reasoning focuses on the ability of a government to use repression against political challengers.
The argument is that whereas authoritarian rulers can use unlimited repression against opposition there is no
need for them to launch a full-scale war effort (Enterline and Gleditsch 2000.) On the other hand, democratic
governments rarely have the opportunity to target the opposition unless emergency powers are declared. Thus, it
may be rational for democratic governments to decide to reignite a civil war for the dual purposes of creating a
rally-around-the-flag effect while also opening a window of opportunity for repression of both the former rebels
and other political opposition.
H3d: Democratic governments are more likely to reignite civil war.
Drawing on theories of diversionary war, it has also been suggested that there may be an important distinction
to be made between stable and transitional democracies. The early stages of democratization unleash an intense
competition for power between social groups and an interest, while the institutions expected to regulate this
competition remains under development. Mansfield and Snyder (2002) argue that the transition to democracy
consist of two phases, the first where the old autocratic structures are removed creating a so-called mixed
regime, and the second leading to the establishment of a system of free political competition where the
governments are held accountable by the population. During the first phase, the argument suggests, existing
elites are concerned about losing power, and will be tempted to pursue nationalist, populist policies to maintain
their position (Mansfield and Snyder 2002). Such nationalism can lead to the launching of international war or
targeting domestic enemies, in particular if the former recently have been involved in attempts to secede or
overthrow the state.
H3e: Governments in countries with incomplete democratic transitions are more likely to reignite civil war.
Finally, in the relatively limited scholarship on domestic diversionary war, potential threats are often identified
along ethnic lines (Lake and Rothchild 1996). By using or playing up ethnic cleavages, leaders can draw on
perceived historical injustices to give credibility to their argumentation while introducing an element of fear and
uncertainty among the audience (Brown 1996). This fear, in combination with the ascriptive nature of ethnicity,
also increases the likelihood that moderate members of the ruling ethnic group will be receptive to claims made
by the leader. This is because even if an individual in general is opposed to aggressive action, he or she will fear
that a future loss in the conflict will be followed by purges targeting the losing ethnic group, regardless of
whether these individuals supported the government policy or not (de Figueiredo and Weingast 1999). In effect,
the use of this type of strategy is more likely to be successful if ethnic characteristics can be drawn upon.3
H3f: Ethnically politicized governments are more likely to reignite civil war.
Data and Methods 4
Operational definition of the dependent variable
Data on armed conflict is taken from the UCDP-PRIO armed conflict dataset version 4-2009. An intrastate
armed conflict is defined by UCDP (Uppsala Conflict Data Program) as a contested incompatibility over
government, territory or both between the government of a state and a nongovernmental party that results in at
least 25 battle-related deaths in one calendar year (Gleditsch et al. 2002; Harbom and Wallensteen 2009). The
main advantages with this dataset are the ability to study multiple conflicts in the same country; and the low
fatality threshold for conflict activity, which is ensures that the information on recurrence is measured when
3 Another possible explanation for why ethnically dominated governments may be more likely to pursue offensive strategies is offered
by Heger and Salehyan (2007). They argue that ethnic-based governments proxy “smaller ruling coalitions” that are less sensitive to the
costs of civil war than governments that depend on the support of the entire population.
4 Replication dataset, data notes, and robustness tests including alternative measures of the variables presented below will be made
available in a web appendix.
fighting actually begins. To identify cases of recurring conflict, a dataset was constructed using the information
provided by the UCDP-PRIO dataset covering all post-conflict years since the ending of a previous colonial
war, or civil war.5 The termination of a conflict is identified when an active conflict year (t) is followed by an
inactive conflict year (t+1) or if fighting in a conflict against a specific opponent ends in year t, and resumes
against a different opponent in year t+1. Thus, it is possible for the government to stop fighting one of their
opponents in December in year t and launch an offensive against a different rebel group in January year t+1. All
exact dates for the termination and recurrence of fighting are taken from the UCDP Conflict Termination
Dataset (Kreutz 2010).6 Because of restrictions of available time series data on independent variables, this study
focuses on the time period 1970-2002.
For each recurrence, I then collected specific event-based information on the exact circumstances
for how the fighting resumed, disaggregating between government, rebel, and tit-for-tat characteristics. Information
was sought using local media reported in Factiva database, individual case studies, Keesing’s, and reports from
NGOs and IGOs.
As provocations of different kinds can be expected to be common in post-conflict
situations, and the use of violence is politically sensitive, a conservative approach to interpreting the information
was appropriate. Therefore, the variable was coded as (a) clear government attacks, (b) tit-for-tat recurrence, or
(c) clear rebel attacks.7 For this paper, the dependent variable is a dichotomous variable indicating government
attacks or no recurrence (observations of tit-for-tat and rebel attacks are dropped). Table 1 shows descriptive
statistics for the recurrence variable.
[Table 1 about here]
Independent variables
For the main independent variable, I use Belkin and Schofer’s (2003) coup risk index to identify governments that
are vulnerable to intra-government power struggles. This measure is particularly attractive since it focuses on the
structural factors that create vulnerability for a government without taking into account countermeasures
adopted to prevent an overthrow (Belkin and Schofer 2005). As I argue that there always will be incentives for
challengers to initiate intra-government power struggles, it is specifically the cases when the government is
vulnerable that produce the rationale for a decision to reignite civil war. The coup risk index consists of three
components: the strength of civil society, the legitimacy of the regime, and the previous history of coups in a
country. The strength of civil society is measured by the level of membership in international nongovernmental
Civil war, or intrastate conflict, consists of the two categories intrastate armed conflict and internationalized intrastate armed conflict
from the UCDP-PRIO dataset. Colonial wars are included as the legacy of those have produced potential “threats” that can be
exploited in the form of domestic recruits fighting on behalf of the former colonial administration.
6 To control for robustness, the analysis was also performed on data specifications that excluded colonial conflicts and/or conflicts
with longer duration of peace, with similar results (see web-appendix).
7 The tit-for-tat category includes cases where both sides traded accusations about who actually started the fighting and some cases
when the instigator was unclear.
5
organizations. The legitimacy of the regime is constructed using measures of the competitiveness and degree of
regulation in the political system from the Polity dataset. Coup history is a dichotomous indicator of whether a
successful coup has been executed in the decade preceding each observation. The full Belkin and Schofer index
covers the time period 1960-2000, and range from -4.55 to 7.68 with a mean value of -0.03 (Belkin and Schofer
2003: 609). For the sub-set of post-conflict countries in this study, the corresponding range is -4.55 to 5.43 with
a mean value of 0.21. The variable is lagged one year across all models.
Potential conflict-specific confounding factors are drawn from the UCDP-PRIO Armed conflict
dataset or additions to these data. The Uppsala Conflict Data Program (UCDP) differentiates between conflicts
where the issue primarily concerns government, territory, or both. Conflicts that either exclusively concerns
territory or both territory and government are coded as territorial conflicts. Data on whether the previous conflict
ended without a decisive settlement is taken from the UCDP Conflict Termination Dataset (Kreutz 2010). This
dataset codes conflicts terminating through one of four outcomes: victory, peace agreement, ceasefire or
“other”. The latter indicates an end to conflict activity by a decision by either party to withdraw from hostilities
without an accompanying defeat or negotiated settlement. This “other” category is used as an indicator of the
absence of decisive settlement in the preceding conflict phase. To investigate whether a conflict takes place in a
territory that is rich in resources, I employ data from Buhaug, Gates and Lujala (2009). Their measures for
whether there is oil, gas, or gemstone mining in a conflict zone are colluded into a dummy variable indicating
resource rich territory.
Data on institutional features is primarily drawn from Hadenius and Teorell (2007). I employ their
coding of military regimes, one-party regimes, as well as their composite measure of authoritarian rule. Recent debates
on measurement of different regime type variables have identified potential problems in studies concerned with
how regime type correlate with political violence (Cheibub, Gandhi and Vreeland 2009). In particular, it is
possible that using the composite Polity IV scale on regime types may overestimate the propensity of war in
anocracies. However, much previous literature on the effect of incomplete democratization has used this measure.
Thus, in order to make my findings comparable with existing literature and to put my proposed argument to
maximum scrutiny, I employ the same measure. In line with Mansfield and Snyder (2002), I therefore include a
dummy variable for the first five years after a country has transformed from being a strong autocracy to an
anocracy (but not a strong democracy). Strong autocracies are classified as -7 or lower, and anocracy between -6
and 6 in the Polity IV data (Marshall and Jaggers 2002). Finally, the variable used to indicate an ethnic government
consists of a dummy variable indicating that a government is either monopolized or dominated by a single
ethnic group according to Wimmer, Cederman and Min (2009).
Control variables
In addition to the confounding factors specifically identified for testing the effect of intra-government power
struggles on government decisions to reignite civil war, I also include some variables to control for possible
spurious effects as they can affect both the independent and dependent variables. The first of these is a lagged
measure of GDP per capita, taken from Gleditsch (2002). Poor economic policies are often the trigger for
governmental crises and coup attempts (Londregan and Poole 1990) while it also increases the risk for civil war
recurrence (Quinn, Mason and Gurses 2007). Even though the specific effect of ethnic government is tested
separately, the ethnic composition of a country can influence the stability of government as well as the risk of
recurrence. In countries where ethnicity has become politicized, it is easier for political opposition as well as
potential rebels to mobilize supporters (Taylor 1988; Eck 2009), which increases the likelihood of both political
crisis and the decision of governments to launch preventive attacks. To control for this eventuality, I include a
measure of the ethnic fractionalization in a country, taken from Wimmer, Cederman and Min (2009). This
variable is an index of linguistic fragmentation in a country based on politically relevant ethnic groups. During
the Cold War, it can be suspected that the key ally for factions involved in intra-government power struggles was
external to the state, as support from the superpowers or their regional proxies arguably was the most important
resource. As the pattern of civil war termination and recurrence also have changed since the end of the
superpower rivalry (Fortna 2004), a dummy variable indicating Cold War years is included. I also include a
measure of the costs of preceding conflict episode as this could influence both the vulnerability of the government
and its’ willingness to use force again. The variable is created by dividing the number of battle deaths from the
preceding conflict with the duration of fighting. The measure of battle deaths is taken from Lacina and
Gleditsch (2006), while conflict duration measured in months comes from Kreutz (2010).
Statistical technique
The dependent variable for the empirical investigation is a dichotomous variable indicating whether a conflict
recur thorough government attacks. However, the numbers of observations of conflict recurrence are only a
fraction of the observations of peace. King and Zeng (2001) show that in such situations, a normal logit model
systematically underestimates the probability of rare events and report biased estimates. To correct for this bias,
I employ the RELOGIT procedure developed by King and Zeng (2001). In addition, to control for time
dependence between observations, I include the time since preceding conflict ended, its squared, and its cubed
term as suggested by Carter and Signorino (2007). All models are estimated using robust standard errors and
clustered on county.
Findings
The central hypothesis concerns the relationship between a government’s vulnerability to internal power
struggles and the risk that the government will decide to reignite civil war. Table 2 presents the results of the
statistical analysis. Models 1-5 include possible confounding factors identified in the civil war recurrence
literature, while models 6-11 include alternative explanations drawn from research on regime type and
institutional features. Model 12 include all potentially confounding variables.8
[Table 2 about here]
The results in table 2 serve to confirm the main hypothesis in this paper and are robust for the inclusion of a
multitude of potentially influential competing factors. The analysis suggests that the coefficients and statistical
significance of this relationship is quite powerful through all model specifications.9 Governments that are
vulnerable to intra-government power struggles are more likely to reignite civil war. The table also indicates that
explanations taken from literature on civil war recurrence or institution-specific incentives for aggressive
government behavior have none or only limited effects on the propensity of governments to reignite civil war.
Focusing on models 4 and 5 that indicate that instances where the previous conflict ended without
a decisive settlement are more at risk for government-reignited civil war. This suggests that the ability of former
rebels to send signals of their true intentions influences the opportunity for the government to employ the tactic
of relaunching civil war. The results then confirm the arguments made by de Figereido and Weingast (1999) on
the ability of governments to use diversionary strategies against ethnic competitors in a domestic setting.
According to this view, the signaling of intentions is not primarily important only towards the other belligerent
but also to the constituency of that party. For a government faction to successfully use the resignation of civil
war as a means to ensure support from key elites in the military or the general public, a necessary condition is
that the claims’ regarding hostile intentions of former rebels are confirmed, or at least not negated, by the
actions of said rebels. As previously discussed, this is less likely to succeed when the previous conflict ended
with an outright victory for the present government or if the two parties regularly interact in a peace process.
Taking a closer look at the temporal effect on the likelihood of government-ignited civil war recurrence further
indicates how opportunities for governments to use this strategy can be limited.
[Figure I about here]
As the variable for authority include both military and one-party states (and is based on the same data source), it is excluded from the
aggregated models. By including the authority dummy instead of the military and one-party variables, the effect of internal power
struggle vulnerability becomes more statistically significant, whilst all other results remain similar as in the models presented.
9 The statistical significance is slightly weaker in model 8, when controlling for military governments. This probably is an effect of
multicollinearity between the variables. If the power struggle variable is removed, military regimes are positively correlated and
statistically significant with government ignited civil war recurrence. This correlation between the two independent variables is in line
with existing research that argues that military regimes are more vulnerable to intra-elite dissent than other regimes (Belkin and Schofer
2005; Fjelde 2009; Lai and Slater 2006).
8
Figure I show the predicted probability that the government reignites civil war across the duration of peace.10
Shortly after the end of the preceding conflict, there remains much uncertainty about the intentions of the
former belligerents; whether they are committed to non-violence or planning to launch new attacks. In this
situation, it is possible for government leaders to exploit this uncertainty and propagate for attacks on the
former opponents to secure the loyalty of the armed forces in a state. As time passes without a recurrence of
fighting, the military and the public in the country overcomes their fear of the former enemy and thereby
become less receptive to leaders claiming there remains a threat. Thus, reigniting civil war becomes a less
attractive option for factions involved in intra-government power struggles, and governments will be less likely
to use this strategy. Figure I indicate that if a country remains peaceful for at least 10 years, there remains only a
very small risk that governments will decide to reignite civil war.
Of the control variable, the dummy for cold war is negatively related to the decisions of
governments to reignite civil war, and consistent through all models. This finding does not in itself contradict
the argument made regarding how intra-government power struggles create incentives for governments to
resume a civil war. During the cold war, the outcome of intra-government power struggles were more likely to
be determined by which faction had the best connections with a superpower ally or its’ regional proxy. Thus, it
was more fruitful and less costly to aim one’s resources on improving relations with Washington or Moscow
than towards the domestic armed forces, as the latter primarily dependent on superpower allocation of resources
anyway.
Extending the analysis
Having identified the connection between governments vulnerable to internal power struggles and the decision
to reignite civil war, I also explore whether other pathways to civil war recurrence also are more likely in these
cases. Since the proposed argument contends that the effect of intra-government competition specifically
increases the risk of government-ignited civil war recurrence, I expect that the main independent variable will have
no effect on rebel-ignited or tit-for-tat recurrence of fighting. Figure II show the changes in relative risk for the
three different pathways to civil war recurrence when moving between every tenth percentiles of the values for
the main independent variable.11
[Figure II about here]
Figure II show that whether governments are vulnerable for internal power struggles have a pronounced effect
on the risk for government-ignited civil war recurrence, but not for rebel-initiated or tit-for-tat resumption of
10
11
Figure I is calculated on the basis of Model 12 in Table 2, i.e. including all potentially confounding factors.
Figure II is calculated on the models presented in table 3.
fighting. The lines depict the relative risk change for each path to civil war recurrence. The first, for example,
shows that when moving from the 10th to the 20th percentile on the coup risk index scale, the relative risk of
government attacks increases with 19%, the relative risk of rebel attacks increases with 4%, and the relative risk
of tit-for-tat recurrence decreases with 1%. As theorized, the decision of a government to reignite civil war will
mainly be used as a last resort, when the stakes are high and there may be little time or availability of alternative
strategies. Even though the risk for government attacks is increasing through the observations, it is also
accelerated in substantially vulnerable regimes. When moving from the 60th to the 70th percentile, the relative risk
increases with 39 percent, and between the 70th and the 80th percentile, the increase is 56 percent. The aggregate
change of moving from the 10th to the 90th percentile increases the risk for government attacks with 215%, with
each of the calculated steps statistically significant at 95% confidence level. In contrast, neither rebel-initiated
recurrence nor the tit-for-tat pathway to civil war resumption is statistically significant which suggest that there is
no relationship between the vulnerability to intra-government power struggles and these types of civil war
recurrence.
These results provide further strength to the proposed argument that even though government
may be forced to react to rebel attacks or provocations at any time, they will only deliberately reengage in civil
war when structural vulnerability is present. The finding that governments are willing to reignite civil wars
because of factors that are endogenous to the preceding conflict challenges previous findings from existing civil
war recurrence research. However, an extension of this relationship is that it should be expected that other
factors would be influential for the decisions of rebel to reignite civil war or for the tit-for-tat recurrence. Table
3 show the results from the statistical analysis on different pathways to civil war recurrence.12
[Table 3 about here]
The findings in table 3 should be considered tentative but may provide potential suggestions for further
research. Several prominent studies argue that the post-conflict relationship between former belligerents can be
seen as a bargaining game where the ability to signal peaceful intentions is the key to sustaining the peace. This
view seems supported with regards to conflicts that resume through tit-for-tat hostile activities. In the absence
of a decisive settlement, the parties will be uncertain about each others future behavior and minor provocations
may escalate to a resumption of armed activity.
Civil wars reignited by rebel attacks are different and the findings are somewhat contradictory. For
a rebel group to be able to resume a civil war, it must either have been able to keep its armed forces intact
during the post-conflict hiatus or be efficient in remobilizing fighters. Countries that are ethnically diverse are
more likely to see a resumption of civil war by the rebel side, lending weight to arguments that ethnicity may be
12
The result does not change if all types of recurrence are explored simultaneously in a multinomial logit model.
a useful instrument for rebel mobilization (Eck 2009). The observation that rebels are less likely to resume a
conflict in an authoritarian state can also be a consequence of opportunities for them to remobilize.
Authoritarian governments are more likely to repress political opposition in its’ early stages which will create
obstacles for former rebels. Because of the preceding conflict episode, the remnants of the rebel group will be
tracked by the security forces and have few opportunities to establish the necessary organization to launch
attacks. However, it could be expected that the few opportunities to remobilize would be in the periphery of the
state and in particular concern groups that easily could blend in with civilians (Fuhrmann and Tir 2009). It is
therefore surprising that territorial conflicts are less likely than government conflicts to be reignited by the
rebels. It is possible that the influence of territorial issues on resumption include more aspects than the ability
for rebels to remobilize. It is possible that government sensitivity to territorial demands, as suggested by Walter
(2006), confer that when such ongoing remobilization is discovered, conflict will resume through attacks from
either side.
Conclusion
This paper argue that governments that are vulnerable to internal power struggles have incentives to launch
attacks that reignites civil war. As such power struggles emancipates from within the ruling coalition, it becomes
difficult for the incumbent as well as the challenger to employ less costly alternatives as both are competing for
support form key allies, in particular the armed forces. Thus, reigniting a civil war becomes an attractive option
despite the costs inherent in such a decision. Using unique data, this proposition was tested against all postconflict countries 1970-2002 whereby the proposed hypothesis was supported. As governments became more
vulnerable to internal power struggles, so did the risk that a civil war would restart through government attacks.
The results were robust even when controlling for a plethora of possible confounding factors. In contrast, the
vulnerability of a government did not increase the risk for other pathways that could reignite civil war, such as
rebel attacks or through tit-for-tat dynamics. An important insight from these findings is that studies on civil war
recurrence should acknowledge that there are multiple explanations for why some countries experiences
repeated conflict episodes. Furthermore, the paper also contributed to the growing literature on how different
institutional features influence the propensity for violence through developing a logic for why governments may
choose violent strategies for reasons that are not linked to the targets of this violence. On the evidence
presented here, this is certainly a feature in post-conflict countries which adds to the challenges that a country
faces when trying to overcome a devastating civil war and build a stable peace.
References
ANDERSON, WALTER. 1996. “India in 1995: Year of the Long Campaign.” Asian Survey 36(2): 165-178.
BALÁN, MANUEL. 2009. “Competition by denunciation: The political dynamics of corruption scandals in
Argentina and Chile.”
BELKIN, AARON AND EVAN SCHOFER. 2003. “Toward a structural understanding of coup risk.” Journal
of Conflict Resolution 47(5): 594-620.
BELKIN, AARON AND EVAN SCHOFER. 2005. “Coup risk, counterbalancing, and international conflict.”
Security Studies 14(1): 140-177.
BELLONI, FRANK P., AND DENNIS C. BELLER. 1976. “The study of party factions as competitive
political organizations.” The Western Political Quarterly 29(4): 531-549.
BOYLE, MICHAEL J. 2009. “Explaining Strategic Violence after Wars.” Studies in Conflict and Terrorism 32(_):
209-236.
BROWN, MICHAEL E. 1996. “The causes and regional dimensions of internal conflict.” In The international
dimensions of internal conflict, ed. Michael E. Brown. Cambridge, MA: MIT Press.
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, ALASTAIR SMITH, RANDOLPH M. SIVERSON, AND JAMES D.
MORROW. 2003. The logic of political survival. Cambridge, MA, and London: MIT Press.
BUHAUG, HALVARD, SCOTT GATES, AND PÄIVI LUJALA. 2009. “Geography, rebel capability, and the
duration of civil conflict.” Journal of Conflict Resolution 53(4): 544-569.
CARTER, DAVID B., AND CURTIS S. SIGNORINO. 2007. “Back to the future: Modeling time dependence
in binary data.”
CHEIBUB, JOSÉ ANTONIO, JENNIFER GANDHI, AND JAMES RAYMOND VREELAND. 2009.
“Democracy and dictatorship revisited.” Public Choice (forthcoming)
COLLIER, PAUL, ANKE HOEFFLER AND MÅNS SÖDERBOM. 2008. “Post-conflict risks.” Journal of
Peace Research 45(4): 461-478.
CROCKER, CHESTER A., FEN OSLER HAMPSON, AND PAMELA R. AALL. 2005. Grasping the nettle:
Analyzing cases of intractable conflict. Washington D.C.: United Institute of Peace Press.
DE FIGUEIREDO, RUI J. P., AND BARRY R. WEINGAST. 1999. “The rationality of fear: Political
opportunism and ethnic conflict.” In Civil wars, insecurity and intervention, ed. Barbara Walter and Jack Snyder. New
York, NY: Columbia University Press.
DEROUEN, KARL R., AND JACOB BERCOVITCH. 2008. “Enduring internal rivalries: A new framework
for the study of civil war.” Journal of Peace Research 45(1): 55-74.
DOWNS, GEORGE W., AND DAVID M. ROCKE. 1994. “Conflict, agency, and gambling for resurrection:
The principal-agent problem goes to war.” American Journal of Political Science 38(2): 362-380.
ECK, KRISTINE. 2009. “From armed conflict to war: Ethnic mobilization and conflict intensification.”
International Studies Quarterly 53(2): 369-388.
ENTERLINE, ANDREW J., AND KRISTIAN S. GLEDITSCH. 2000. “Threats, opportunity, and force:
Repression and diversion of domestic pressure, 1948-1982.” International Interactions 26(1): 21-53.
FEARON, JAMES D., AND DAVID D. LAITIN. 1996. “Explaining interethnic cooperation.” American
Political Science Review 90(4): 715-735.
FEARON, JAMES D., AND DAVID D. LAITIN. 2008. “Civil war termination.”
FJELDE, HANNE. 2010, forthcoming. “Generals, Dictators, and Kings: Authoritarian Regimes and Civil
Conflict, 1973-2004.” Conflict Management and Peace Science.
FORTNA, VIRGINIA PAGE. 2004. “Does peacekeeping keep peace? International intervention and the
duration of peace after civil war.” International Studies Quarterly 48(_): 269-292.
FUHRMANN, MATTHEW AND JAROSLAV TIR. 2009. “Territorial dimensions of enduring internal
rivalries.” Conflict Management and Peace Science 26(4): 307-329.
GANDHI, JENNIFER AND ADAM PRZEWORSKI. 2006. “Cooperation, cooptation and rebellion under
dictatorships.” Economics & Politics 18(1).
GANDHI, JENNIFER AND ADAM PRZEWORSKI. 2007. “Authoritarian institutions and the survival of
autocrats.” Comparative Political Studies 40(11): 1279-1301.
GLEDITSCH, KRISTIAN S. 2002. ”Expanded trade and GDP data.” Journal of Conflict Resolution 46(5): 712-724.
GLEDITSCH, NILS PETTER; PETER WALLENSTEEN, MIKAEL ERIKSSON, MARGARETA
SOLLENBERG AND HÅVARD STRAND. 2002. ”Armed conflict 1946-2001: A new dataset.” Journal of Peace
Research 39(5): 615-637.
GOEMANS, H.E. 2008. “Which way out? The manner and consequences of losing office.“ Journal of Conflict
Resolution 52(6): 771-794.
HADENIUS, AXEL AND JAN TEORELL. 2007. “Pathways from authoritarianism.” Journal of Democracy 18(1):
143-156.
HARBOM, LOTTA AND PETER WALLENSTEEN. 2009. “Armed conflicts, 1946-2008.” Journal of Peace
Research 46(4): 577-587.
HARTZELL, CAROLINE A. 2009. “Settling civil wars: Armed opponents’ fates and the duration of peace.”
Conflict Management and Peace Science 26(4): 347-365.
HEGER, LINDSAY AND IDEAN SALEHYAN. 2007. “Ruthless rulers: coalition size and the severity of civil
conflict.” International Studies Quarterly 51(2): 385-403.
HODDIE, MATTHEW AND CAROLINE A. HARTZELL. “Signals of reconciliation: Institution-building
and the resolution of civil wars.” International Studies Review 7 (1): 21-40.
HULTMAN, LISA. 2009. “The Power to Hurt in Civil War: The Strategic Aim of RENAMO Violence.” Journal
of Southern African Studies 35(4): 821-834.
HÖGLUND, KRISTINE AND I. WILLIAM ZARTMAN. 2006. “Violence by the State: Official Spoilers and
Their Allies.” In Violence and Reconstruction ed. John Darby. Notre Dame, IN: Notre Dame University Press.
KING, GARY AND LANGCHE ZENG. 2001. “Logistic regression in rare events data.” Political Analysis 9(2):
137-163.
KREUTZ, JOAKIM. 2010. “How and when armed conflicts end: Introducing the UCDP conflict termination
dataset.” Journal of Peace Research 47(2): 243-250.
ICG (INTERNATIONAL CRISIS GROUP). 2007. “Georgia: Sliding towards Authoritarianism?” Europe
Report No 189. Available at www.crisisgroup.org. Accessed 10 March, 2010.
LACINA, BETHANY. 2009. “Interest groups and internal security: A theory of the strategic origin of political
violence.”
LACINA, BETHANY AND NILS PETTER GLEDITSCH. 2005. “Monitoring trends in global combat: A
new dataset of battle deaths.” European Journal of Population 21(2-3): 145-166.
LAI, BRIAN, AND DAN SLATER. 2006. “Institutions of the offensive: Domestic sources of dispute initiation
in authoritarian regimes, 1950-1992.” American Journal of Political Science 50(1): 113-126.
LAKE, DAVID A. AND DONALD ROTHCHILD. 1996. “Containing fear: The origins and management of
ethnic conflict.” International security 21(2): 41-75.
LE BILLON, PHILIPPE. 2004. “The geopolitical economy of ‘resource wars.’” Geopolitics 9(1): 1-28.
LICKLIDER, ROY. 1995. “The consequences of negotiated settlements in civil war, 1945-1993.” American
Political Science Review 89(3): 681-690.
LICKLIDER, ROY. 2002. “False Hopes? Postsettlement Democracy and the Resumption of Civil War.” Paper
presented at the 2002 Annual Meeting of the American Political Science Association, August 29-September 1,
2002.
LONDREGAN, JOHN B. AND KEITH T. POOLE. 1990. “Poverty, the coup trap, and the seizure of
executive power.” World Politics 42(2): 151-183.
MANSFIELD, EDWARD D., AND JACK SNYDER. 2002. “Democratic transitions, institutional strength,
and war.” International Organization 56(2): 297-337.
MARSHALL, MONTY AND KEITH JAGGERS. 2002. “Polity IV project.” University of Maryland.
MASON, T. DAVID AND PATRICK J. FETT. 1996. “How civil wars end: A rational choice approach.”
Journal of Conflict Resolution 40(4): 546-568.
MILLER, ROSS A. 1995. “Domestic structures and the diversionary use of force.” American Journal of Political
Science 39(3): 760-785.
MUKHERJEE, BUMBA. 2006. “Why political power-sharing agreements lead to enduring peaceful resolution
of some civil wars, but not others?” International Studies Quarterly 50(2): 479-504.
QUINN, J. MICHAEL, T. DAVID MASON AND MEGMET GURSES. 2007. “Sustaining the peace:
Determinants of civil war recurrence.” International Interactions 33: 167-193.
PAPE, ROBERT A. 2003. “The strategic logic of suicide terrorism.” American Political Science Review 97(3): 343361.
PION-BERLIN, DAVID. 1992. “Military autonomy and emerging democracies in South Amerca.” Comparative
Politics 25(1): 83-102.
POWELL, JONATHAN. 2009. “Regime Vulnerability and the Diversionary Threat of Force.”
SMITH, ALASTAIR. 1996. “Diversionary Foreign Policy in Democratic Systems.” International Studies Quarterly
40(1): 133-153.
SVOLIK, MILAN W. 2009. “Contracting on violence: Authoritarian repression and military intervention in
politics.”
TAYLOR, MICHAEL. 1988. “Rationality and revolutionary collective action.” In Rationality and revolution, ed.
Michael Taylor. New York: Cambridge University Press.
TIR, JAROSLAV AND MICHAEL JASINSKI. 2008. “Domestic-level diversionary theory of war: Targeting
ethnic minorities.” Journal of Conflict Resolution 52(5): 641-664.
TULLOCK, GORDON. 1987. Autocracy. Dordrecht: Martinus Nijhoff Publishers.
WALTER, BARBARA F. 2004. “Does conflict beget conflict? Explaining recurring civil war.” Journal of Peace
Research 41(3): 371-388.
WALTER, BARBARA F. 2006. ”Building reputation: Why governments fight some separatists but not others.”
American Journal of Political Science 50(2): 313-330.
WEEKS, JESSICA L. 2009. “Rulers, risk, and restraint: Domestic politics and war in authoritarian regimes.”
WIMMER, ANDREAS, LARS-ERIK CEDERMAN AND BRIAN MIN. 2009. “Ethnic politics and armed
conflict: A configurational analysis.” American Sociological Review 74(2): 316-337.
Appendix: Cases of government-reignited Civil War, 1970-2002
Algeria 1991
Angola 1996 (Cabinda)
Burma/Myanmar 1990 (Mon), 1994 (Arakan), 1995 (Karen), 1996 (Karenni, Mon), 1997 (Karen)
Cambodia 1978
Chad 1987, 1991
Congo-Brazzaville 1997, 2002
Croatia 1995 (Krajina)
Eritrea 1999
Ethiopia 1980 (Oromiya), 1999 (Somali)
India 1995 (Manipur)
Iran 1979 (Kurdistan), 1991, 1993 (Kurdistan), 1996 (Kurdistan)
Iraq 1985 (Kurdistan)
Niger 1997
Rwanda 1997
South Yemen 1986
Thailand 1974
Table 1. Descriptive statistics of civil war recurrence (19702002)
Recurrence side
N
Preceding peace duration (months)
Min
Max
Mean
Median
Government offensive
27
0.07
356.50
69.43
30.92
Rebel offensive
87
0.33
417.05
85.38
39.16
Tit-for-tat offensive
51
0.07
231.03
50.44
24.51
Table 2. Determinants of goverment-ignited civil war recurrence
1
2
3
4
5
6
7
8
9
10
11
12
.27 (.13)**
.29 (.14)**
.31 (.15)**
.31 (.13)**
.36 (.13)***
.32 (.13)**
.28 (.14)**
.24 (.13)*
.27 (.14)**
.28 (.13)**
.27 (.14)**
.36 (.16)**
Resource-rich territory
-
.40 (.55)
-
-
.30 (.56)
-
-
-
-
-
-
.50 (.63)
Conflict over territory
-
-
.42 (.40)
-
.28 (.40)
-
-
-
-
-
-
.41 (.42)
Not decisive outcome
-
-
-
.76 (.37)**
.81 (.43)*
-
-
-
-
-
-
.71 (.42)*
Authoritarian gov.
-
-
-
-
-
-.62 (.72)
-
-
-
-
-
-
One-party gov.
-
-
-
-
-
-
.45 (.59)
-
-
-
.46 (.60)
.04 (.74)
Military gov.
-
-
-
-
-
-
-
.40 (.43)
-
-
.46 (.44)
.21 (.46)
Inc. democratization
-
-
-
-
-
-
-
-
.70 (.57)
-
.63 (.58)
.71 (.56)
Ethnic government
-
-
-
-
-
-
-
-
-
-.21 (.42)
-.30 (.39)
-.55 (.43)
GDP per capita
-.00 (.00)
-00 (.00)
-00 (.00)
-.00 (.00)
.00 (.00)
-.00 (.00)
-.00 (.00)
-.00 (.00)
-.00 (.00)
-.00 (.00)
.00 (.00)
-.00 (.00)
ELF
-.15 (.85)
.52 (.82)
.39 (.83)
-.35 (.84)
.33 (.88)
-.06 (.85)
-.15 (.84)
-.19 (.87)
-.24 (.81)
-.24 (.80)
-.40 (.77)
-.18 (.86)
-.85 (.45)*
-.89 (.43)**
-.94 (.46)**
-.81 (.43)*
-.89 (.43)**
-.91 (.46)*
-.82 (.49)*
-.91 (.44)**
-.93 (.51)*
-.74 (.46)
-.00 (.03)
-.01 (.03)
.00 (.01)
.00 (.02)
-.00 (.02)
-.01 (.02)
-.00 (.02)
-.01 (.02)
-.01 (.02)
-.01 (.02)
.01 (.02)
(.02)***
-05 (.02)**
-.05 (.02)**
-.05 (.02)***
-.05 (.02)**
-.05 (.02)***
-.05 (.02)***
-.05 (.02)***
-.06 (.02)***
-.06 (.02)***
(.02)***
-.05 (.02)**
Peace duration2
.00 (.00)**
.00 (.00)*
.00 (.00)*
.00 (.00)**
.00 (.00)
.00 (.00)**
.00 (.00)**
.00 (.00)**
.00 (.00)**
.00 (.00)**
.00 (.00)**
.00 (.00)
Peace duration3
-.00 (.00)**
-.00 (.00)
-.00 (.00)
-.00 (.00)*
-.00 (.00)
-.00 (.00)*
-.00 (.00)**
-.00 (.00)**
-.00 (.00)**
-.00 (.00)**
-.00 (.00)*
-.00 (.00)
Intra-gov. Power struggle
Cold war
Costs of prev. Conflict
-.91 (.44)** -1.06 (.45)**
-.01 (.02)
-.05
Peace duration
Constant
N
-.05
-1.98
-2.72
-2.66
-2.50
-3.45
-1.59
-2.01
-2.12
-.2.05
-1.81
-2.4
-23.14
(.64)***
(.65)***
(.74)***
(.81)***
(.95)***
(.57)***
(.62)***
(.67)***
(.63)***
(.66)***
(.62)***
(.81)***
2453
2221
2308
2453
2221
2338
2347
2347
2450
2453
2344
2130
Dependent variable is recurrence of civil war initiated by the government actor: robust standard errors in parenthesis; *, **, *** significant at .1, .05, .01 levels. Estimates
computed using Stata 9.
Table 3. Different pathways to civil war recurrence
13
14
15
Gov recur
Reb recur
Tit-for-tat
.43 (.15)***
.10 (.10)
-.01 (.11)
Territory conflict
.31 (.39)
-.51 (.26)*
.28 (.28)
Not decisive outcome
.77 (37)**
.10 (.27)
1.11(.42)***
Authoritarian government
-.83 (.73)
-.45 (.25)*
-.33 (.44)
GDP per capita
-.00 (.00)
-00 (.00)
-00 (.00)
ELF
.30 (.83)
1.33 (.58)**
.19 (.64)
-.76 (.43)*
-.13 (.28)
.15 (.29)
.00 (.02)
.00 (.00)
.00 (.01)
-.05 (.02)**
-03 (.01)***
-.07 (.02)***
Peace duration2
.00 (.00)
.00 (.00)**
.00 (.00)***
Peace duration3
-.00 (.00)
-.00 (.00)
-.00 (.00)**
-2.59 (.85)***
-2.16 (.53)***
-2.14 (.95)**
2209
2209
2209
Intra-gov. Power struggle
Cold war
Costs of prev. Conflict
Peace duration
Const
N
Dependent variable is recurrence of civil war initiated by the government actor/rebel actor/tit-for-tat:
robust standard errors in parenthesis; *, **, *** significant at .1, .05, .01 levels. Estimates computed
using Stata 9.
25
Figure I. Risk of government-ignited civil war recurrence over
time
(bars indicate 95% confidence)
1
0,8
Predicted probability
0,6
0,4
0,2
0
1
2
3
4
5
6
7
8
9
10
Peace years
26