Why Bad Governance Leads to Repeat Civil War

Abstract

Most contemporary civil wars are now recurrences of earlier civil wars. In contrast to classic theories of grievance and opportunity, this article advances a theory of civil war recurrence that highlights the critical role political and legal institutions play in constraining elites in post–civil war states. Such constraints serve as a check on executive power, help incumbent elites credibly commit to political reform, and create a situation where rebels need not maintain militias as a supplementary mechanism to hold political elites in line. All of this reduces the odds of repeat civil war. Using a statistical analysis of post-conflict years, this article demonstrates that strong political institutions are not only significantly and negatively related to repeat civil war but are the primary determinants of whether countries get caught in the conflict trap.

Keywords

civil wars conflict resolution conflict internal armed conflict bargaining international security

The problem of civil war is now almost exclusively a problem of repeat civil war. Fifty years ago, most civil wars were first-time civil wars—new wars pursued by rebels who had previously not challenged the state.¹ But by the 2000s, 90 percent of all civil wars were repeat civil wars—old wars restarted by the same rebels after a period of peace.² In fact, every civil war that has started since 2003 (with the exception of the recent war in Libya) has been a continuation of a previous civil war.

The fact that most civil wars today are repeat wars is striking for two reasons. First, civil wars are not recurring in all countries and all regions. El Salvador, for example, experienced a civil war that ended in 1992 and has never restarted. Instead, repeat civil wars are predominantly concentrated in sub-Saharan Africa and the Middle East (UCDP/PRIO ACD v-4, 2009). Second, countries experiencing multiple civil wars are not experiencing different wars with different sets of rebels. Most are experiencing the same war with the same rebel group or set of rebel groups over time. The multiple bouts of civil war in Lebanon, for example, have consistently been fought between the Christian/Maronite right and the Muslim/Druze left. This suggests that the main challenge surrounding civil war is no longer how to prevent new wars from starting but how to permanently end the ones that have already broken out.

I am not the first to discover that civil wars repeat themselves. Collier and Sambanis (2002) noticed this pattern and called it the conflict trap. Since then scholars have offered two chief explanations for it. The first argues that countries become trapped in cycles of violence because fighting exacerbates the underlying economic and political conditions that motivated rebels to fight in the first place (Walter 2004; Hegre and Sambanis 2006; Quinn, Mason, and Gurses 2007; Justino 2009). Civil war tends to make poor countries poorer and weak states weaker, encouraging additional wars over time. The second argues that certain types of civil war are more prone to repeat violence than others (Licklider 1995; Quinn, Mason, and Gurses 2007; DeRouen and Bercovitch 2008; Kreutz 2010; Toft 2010; Call 2012). Combatants who are unable to defeat each other, who are engaged in long and intense battles, and who fight over stakes that are perceived to be particularly valuable have been viewed as being especially susceptible to repeat civil war.

This article argues that a key factor is missing from existing accounts: the role of political and legal institutions in preventing the repeated outbreak of violence. Civil wars are much more likely to repeat themselves in countries where government elites are unaccountable to the public, where the public does not participate in political life, and where information is not transparent.³ Greater institutionalization serves four purposes. First, these features serve as a check on executive power creating governments that are more likely to serve the interests of a wider population. This creates fewer motives for rebels to return to war. Second, they create multiple nonviolent avenues to influence government policy, making renewed violence less essential as a means to promote change. Third, strong political and legal institutions help incumbent elites credibly commit to the political terms of a peace settlement, making lasting bargains more likely. Finally, checks on executive power create a situation where rebels need not maintain militias and the threat of violence to hold political elites in line.⁴ The result is a situation where combatants have an easier time resolving their underlying differences and supporting a permanent halt to fighting.

What follows is divided into four sections. The first section lays out the main argument regarding the importance of predictable, accountable governance in reducing repeat civil war. Political leaders in very weakly institutionalized environments will have the greatest difficulty committing to and working with former combatants and will, therefore, be most susceptible to recurring violence. The second section presents three additional sets of factors that likely contribute to repeat civil war. These focus on the underlying grievances rebels might have, their ability to sustain a rebellion, and bargaining problems that stand in the way of long-term settlement. The third section introduces the data set of all repeat civil wars between 1945 and 2009 and explains how different hypotheses were tested. The fourth section presents the findings: civil wars were significantly more likely to repeat themselves in countries with few legal and political constraints on government. They were also more likely to repeat themselves in countries that had shown no marked improvement in accountability or transparency once violence broke out in the first place. These institutional controls proved more important than any other factor—including per capita income—in limiting repeat civil war. The fifth section discusses what this means for scholars trying to understand persistent violence and for policy makers struggling with how to resolve long-standing disputes in some of the world’s most fragile states.

The Argument: Why Accountability Matters

The biggest puzzle surrounding repeat civil wars is why combatants continue to fight despite years of conflict and an inability to defeat each other. Why would the government of Myanmar, for example, continue to battle Shan insurgents after more than fifty years of intermittent war, and why would the Shan persist in fighting? To explain repeat civil war, therefore, one must do more than identify the source of ongoing grievances; one must explain why these grievances are not being resolved despite years of rising costs.

Governments and rebels engaged in a civil war have two ways to end violence and produce long-term peace. They can defeat their opponent and eliminate its ability to fight, or they can reach a mutually agreeable compromise with their opponent that resolves their differences and removes the need to battle. Combatants are likely to prefer the first option since overwhelming one’s opponent militarily has the benefit of delivering political benefits now and into the future and potentially the benefit of generating a lasting peace (Fearon 1998). President Mahinda Rajapaksa of Sri Lanka, for example, won election in 2005 in part on his promise to defeat the Tamil Tigers rather than compromise. Within four years, his government had crushed the rebels, creating a widely celebrated and seemingly lasting peace. Similarly, Mao Zedong’s military victory over the incumbent Kuomintang party has brought decades of relative peace and stability to China. Given the choice, therefore, combatants engaged in a civil war are likely to prefer to conquer their opponent rather than compromise.

Combatants in repeat civil wars, however, tend not to have this luxury. Almost by definition, repeat wars represent those conflicts where combatants have been unable to decisively defeat each other or suppress additional dissent. Recurring civil wars, therefore, represent the subset of cases where something short of conquest—namely, compromise and reform—may be the only way for combatants to permanently end the war.

The attractiveness of any political settlement is likely to depend on whether the two sides believe that the proposed reforms will actually be implemented and adhered to over time. Opponents who have recently fought a civil war, however, are unlikely to trust one another. This is likely to be especially true if the two sides have a history of repeatedly fighting, reneging on ceasefires, and signing and then snubbing peace agreements. The Angolan government and the National Union for the Total Independence of Angola (UNITA) rebels, for example, signed numerous peace agreements during their multi-decade civil war that were never implemented. Almost all of them failed to bring peace.

This is where institutional accountability comes in. Combatants’ ability to negotiate their way out of war will depend on their ability to hold each other accountable to the terms of an agreement over time. Rebels understand that promises to reform are less likely to be implemented by governments with limited accountability than those more answerable to the public. Incumbent elites who are answerable to few groups in society and face few institutional restraints make bad negotiating partners. The more accountable the government is to a wide range of people, the easier it will be to credibly commit to share power and reform, and the fewer incentives groups will have to return to violence.

What institutional features are most likely to hold governments accountable and lead to less repeat war? Three features stand out as potentially critical. First, legal checks on the actions of the executive could restrain a leader’s ability to unilaterally renege on concessions and make a negotiated settlement based on reform more likely. Government leaders who operate under a written constitution enforced by an independent judicial system will make more attractive negotiating partners since the rule of law will serve as a check on the ability of the government to renege on its promises, especially after peacekeepers leave. Second, political accountability in the form of active and engaged voters and competing political parties may be able to help regulate the behavior of governments through the threat of political removal.⁵ Thus, even if incumbents resist reform, nonviolent channels would exist to influence behavior. Finally, a free media can serve as a third check on incumbent elites by providing information to the public about any bad behavior by the government. In short, rebels that face government leaders constrained by an independent judiciary, an empowered public, and an open and objective media are more likely to lay down their weapons and are less likely to use violence to keep the government in line.

When Doesn’t Accountability Work?

At least three circumstances exist, however, under which accountability is likely to be irrelevant or insufficient to steer chronic rivals toward a lasting peace. First, the theory presented above assumes that governments and rebels will, at some point in a conflict, prefer settlement to continued war. In reality, one or both sides may have no interest in compromise at any point during a war. Government elites, for example, will have few incentives to build strong institutions if doing so would permanently remove them from power. Saddam Hussein had little reason to negotiate with the majority Shi’a population since demographics would have left him vulnerable to retribution for years of repressive rule. Rebel leaders, on the other hand, may have no interest in settlement if war itself is profitable. UNITA rebels in Angola used violence as a tool to gain and maintain control over territory rich with diamonds.

Second, a majority of citizens may have no interest in compromise, especially if the opposition is a minority group demanding significant concessions. The long-standing stop-and-go war between the Philippine government and various Muslim factions in Mindanao is an example of this. The Philippine government has fairly high levels of accountability, participation, and transparency and yet has experienced multiple bouts of civil war, making it an outlier in the list of recurring wars. In this case, a combination of institutional structure (a majoritarian electoral system) and demographics (a large Catholic majority) has created a situation where the legislature has made it impossible for any president to make real concessions to Muslims. Even if a president were willing to sign such a peace settlement, it is unlikely that he or she would be able to fully implement the deal over time.

Third, if an outside patron has a strong interest in the conflict and would prefer renewed war to peaceful compromise, local institutions may be largely irrelevant. For much of the Cold War, the United States and the Soviet Union conducted proxy battles that were motivated by global dominance rather than local issues. All of this suggests that political accountability is a necessary but far from sufficient condition to permanently end repeat civil wars.

Shouldn’t Accountability Matter at Every Stage of Civil Conflict?

The theory to this point has focused on conflicts that experience repeated cycles of violence and peace—a set of civil wars that now dominates the civil war arena. Still, it is important for substantive and methodological reasons to consider how accountability may affect the initial outbreak of civil war. If political accountability makes it easier for two sides to resolve a dispute, then more accountable governments should be less apt to experience civil war in the first place. An emerging body of literature suggests this is true. In an extensive study of all countries since 1946, Fearon (2011) found that the quality of a country’s governance or institutions strongly predicted its risk of civil war outbreak. The shoddier the governance and the weaker the institutions, the more likely a country was to experience civil war.

This raises the following issue: If accountability played a decisive role in the initial outbreak of civil war, it is possible that institutions have little ongoing influence in repeat civil wars. Countries that experience civil wars are already those countries that suffer from bad governance. There are, however, three reasons to expect political accountability to have an independent effect on repeat civil war. First, accountability could become more influential in the outbreak of repeat civil war as other factors instrumental to the initial outbreak of violence become less relevant. Factors that vary little over time, for example, such as a country’s population, its geographic terrain, and its level of ethnic diversity (all of which have been found to be significantly related to the initial outbreak of civil war) are less likely to be strongly tied to repeat civil war. The same is true for informational asymmetries; once the initial civil war breaks out, combatants should be able to collect important information about each other’s capabilities and resolve, making repeat war unnecessary.

Second, accountability may become more relevant as it becomes increasingly obvious that other means of permanently ending a conflict—such as a decisive military victory or a negotiated settlement enforced by a third party—are unavailable.

Third, accountability is far from a determinative mechanism in the early stages of conflict. For a range of reasons already outlined, some conflicts in countries with strong institutions will still go to war. There will be plenty of cases of repeat civil war occurring in politically accountable countries.⁶ In these countries, if conditions on the ground change or if the resolve of the combatants change, then political accountability could begin to play a role in ending the cycle of violence.

Fourth and perhaps most important, institutions can and do change over the course of these conflicts.⁷ Government leaders engaged in a lengthy civil war with an opponent they cannot defeat have incentives to reform. The government of El Salvador, for example, eventually offered Farabundo Marti National Liberation Front (FMLN) rebels significant concessions partly in response to pressures from Salvadoran business interests to end the war (Wood 2007). As we will see, governments that instituted real political reforms either during a civil war or immediately after peace broke out were significantly more likely to experience a lasting peace.

One Caveat

This article suggests that political and legal checks can help governments credibly commit to reform and eliminate the need for rebels to return to violence. Still, the mechanism by which good governance works to prevent renewed violence is likely to be multifold. Elites that are accountable to the public are liable to govern better or at least in the interests of a wider audience. Grievances, therefore, are likely to be less severe, creating fewer incentives to challenge the state. Leaders operating in countries where power is already divided may also have an easier time making concessions since the cost of political reform is likely to be less threatening to incumbents. Bargaining in these cases may be more seamless. This article cannot conclusively address which of the many mechanisms is most important, but each of the logics points to the same conclusion: civil wars are less likely to recur in countries where strong mechanisms for government accountability exist.

A clear set of predictions, therefore, can be put forward for testing. Civil wars that are fought against governments with limited accountability should be more likely to repeat themselves than civil wars in countries with highly accountable governments. This should be true whether constraints on government behavior come from a strong rule of law, popular participation, or any other factors that hold a government answerable to the larger population. It should also be true regardless of the mechanism that is causing rebels to return to war in countries with weak institutions. The more accountable government leaders are to multiple actors in society, the fewer incentives opponents will have to return to violence and the fewer repeat wars should occur.

Other Plausible Explanations

Civil wars can start (and restart) for many different reasons beyond government institutions. In what follows, I discuss a range of factors that have been found to be influential in other studies of civil war onset and recurrence. These will serve as controls for the analysis that follows.

Repeat Civil War as a Result of Particularly Deep Grievances

The most obvious explanation for repeat civil war has to do with the grievances that motivate groups to rebel in the first place. It is reasonable to assume that rebels will keep fighting as long as their grievances are unresolved and as long as their motive for rebelling remains. Countries with low gross domestic product (GDP) per capita, poor public health, or other features related to low levels of human development almost certainly create groups dissatisfied with the status quo and desirous of change (Murshed 2002).

Civil wars may also intensify ethnic grievances or group rivalries, making renewed war more likely. Once violence breaks out, ethnic identities may be reinforced in ways that make cooperation between groups more difficult and future war more likely.

Repeat Civil War as a Result of Enhanced Opportunity

Countries may experience recurring civil war for another reason. Organizing a rebellion may simply be easier in countries where the opportunity costs for fighting are low. Three features of a state could affect this calculation. The first is poverty.⁸ The poorer a country, the easier it is for rebel leaders to recruit soldiers and maintain an army (Collier and Hoeffler 2004; Collier and Sambanis 2002; Gates 2002). The second is state capacity. The weaker the state, the less able it is to effectively police all parts of the country and the easier it is for aggrieved groups to survive and continue fighting (Fearon and Laitin 2003; Collier, Hoeffler, and Soderbom 2009). Finally, geography may be important for repeat war as well. Rebels who operate in countries with rough or mountainous terrain or have other territory from which to operate are believed to have an easier time evading government repression (Fearon and Laitin 2003; Salehyan 2009).

A related factor is the profitability of violence. Some countries may experience repeated bouts of violence because war itself is profitable to pursue (Buhaug and Gates 2002; Collier and Hoeffler 2004; Fearon 2004; Ross 2004; Humphreys 2005; Lujala, Gleditsch, and Gilmore 2005). This would be the case if war enabled rebels to control lucrative pieces of land, as UNITA did in Angola, from which they could extract valuable raw materials or rents.

Repeat Civil War as a Bargaining Problem

Repeat civil wars, however, could also occur due to a host of bargaining problems. It is possible that stop-and-go wars occur in countries where reliable information on rebel strength is difficult to collect due to topography, shifting alliances, or the sheer size of a country (Fearon 2004; Cunningham 2006). It is also possible that wars recur because periods of violence have not been sufficiently long to reveal full information about the relative strength of the combatants (Doyle and Sambanis 2000; Hartzell, Hoddie, and Rothchild 2001; Regan 2002; Dubey 2004; Fearon 2004; Fortna 2004).

It is also possible that civil wars are repeating themselves because combatants are unwilling or unable to divide the stakes (Toft 2003). Wars fought over particularly valuable territory or over issues that are perceived to be too important to compromise could be those prone to repetition.

Finally, it is possible that civil wars repeat themselves because governments are unable to credibly commit to demobilization, disarmament, and reintegration.⁹ Hartzell and Hoddie (2003) have found that combatants were also significantly more likely to sign a peace agreement if the treaty included specific power-sharing guarantees (Hartzell and Hoddie 2003). Wars that repeat themselves, therefore, could be those in which peacekeepers were not available to help with demobilization, or power-sharing agreements were not included in a settlement (but see DeRouen, Lea, and Wallensteen 2009).

Data and Empirical Analysis

All of these factors almost certainly play a role in the outbreak and continuation of violence. What is missing from these accounts, however, is the role institutional constraints play in decisions by rebels to relinquish violence in favor of peace.

In order to determine which factors influence the outbreak of repeat war and which do not, I utilized conflict data from the Uppsala Conflict Data Program/Peace Research Institute Oslo (UCDP/PRIO) Armed Conflict Database (ACD) Version 4.0 between 1945 and 2009.¹⁰ The ACD records annual data for all intra-state conflicts that result in at least twenty-five battle-related deaths a year.¹¹ I create a time-series cross-sectional data set with each year in each post-conflict peace spell as a separate observation.

The ACD has a rough estimate of the annual battle deaths for each conflict. It does not, however, provide any scheme for identifying when exactly civil wars begin and end and, therefore, does not define an episode of civil war. Thus, an important first step was to define and measure episodes of peace and the onset of repeat civil war. For the main analysis in this article, a conflict had to produce at least a thousand battle deaths in one year to be included as an episode of civil war. I employed this relatively high threshold of deaths to avoid focusing on relatively minor conflicts. An episode of civil war was not coded as having ended until it experienced two years without 1,000 battle deaths and at least one year with fewer than 25 battle deaths. I used this coding rule in order to ensure that ongoing wars with short lapses in fighting or small fluctuations in deaths were not included as multiple separate episodes of conflict. This rather lengthy period of relative peace is critical if we want to eliminate cases where war is essentially ongoing but less severe.¹²

Onset of repeat civil war thus had three conditions. It had to follow two years of relative peace. The new conflict had to surpass 1,000 battle deaths. And most important, it had to be fought between the same combatants over the same issue.¹³ By these criteria, there were a total of seventy-seven periods of peace and twenty-four episodes of repeated civil war in the data set (see the Online Appendix for a list of these cases).^14,15

The main analysis that follows employs a Cox proportional hazards model with peace years (the years between the end of the previous civil war and the start of the next one) as the basic unit of analysis and the onset of renewed war as the end of the peace period. Each year of peace is coded as zero and the first year of renewed war is coded as one. Estimates using a parametric model, the Weibull distribution, or those employing a simple cross-sectional time-series logit model are not substantially different than the results presented here (see the Online Appendix).

Operationalizing Political and Legal Constraints

The main focus here is on specific political institutions and the effect they may have on recurring war. Three areas are presumed to be critical for permanently resolving violent disputes and discouraging additional attacks. These are (1) legal accountability, especially to the rule of law; (2) political accountability in the form of popular participation in government; and (3) transparency in the form of a free press.

The general strategy to measure these political constraints is to focus initially on concrete and objective measures (i.e., the percentage of the population that votes). These types of measures have two advantages. First, they can be calculated with relative accuracy. Second, they eliminate any subjective assessments by coders about how accountable or unaccountable is a government, and thus reduce the chance that the rating is endogenous to civil war. The second strategy is to repeat the analysis with a range of different, more conventional measures derived from different data sets and organizations. If a number of different indicators all suggest that political constraints are influential despite the fact that they differ markedly by their source, by their objective or subjective coding, and by the years and countries for which they are coded, then our confidence in the role of constraints increases.

I first assess legal accountability by focusing on a set of indicators that come from the Institute and Election Project (IAEP) produced by Binghamton University (Regan, Frank, and Clark 2009). The primary variable here, Written Constitution, is a dummy variable indicating whether a formal legal constitution is in place. Fortunately, there are several alternate measures of rule of law that are available from more conventional sources. The first is the Rule of Law measure produced by the World Governance Indicators (WGI). The WGI are expert-based ratings compiled by the World Bank (2012) that annually evaluates six dimensions of governance around the globe.¹⁶ According to the WGI, rule of law assesses the quality of the courts and the enforcement of contracts and property rights, factors that could be key to a government’s ability to credibly commit to deals. The main results for the rule of law are also reconfirmed using a somewhat different Civil Liberties indicator produced by Freedom House (2009). According to Freedom House, the civil liberties index assesses the extent to which the government allows for freedom of expression and belief, associational and organizational rights, rule of law, and personal autonomy without interference from the state.

To measure political accountability, the main model includes one basic indicator of public participation. The Participation variable, produced by Vanhanen (2000) and the International Peace Research Institute Oslo (PRIO), is a single count of the percentage of the population voting in the most recent parliamentary and/or presidential elections (Vanhanen 2000; Gates et al 2006). Given the objective nature of this measure, it is preferred to other measures of political accountability. However, to test the robustness of political accountability in shaping civil war outcomes, I reanalyze the data using a series of different indicators from Freedom House (2009), the World Bank (2012), Polity IV, and Cheibub, Gandhi, and Vreeland (2010) that are detailed later in this article.

Measures of a free press, unfortunately, are much more limited. Freedom House (2009) produces a measure, Freedom of the Press, which is designed to give an overall assessment of media independence. Few viable alternatives exist.

To see if more generic measures of democracy can explain the initiation of war in these preexisting conflicts, I also test three basic measures of Democracy and Autocracy—Polity IV, the Country and Policy Institutional Assessment (CPIA) produced by the World Bank (2012), and Cheibub, Gandhi, and Vreeland’s (2010) alternative measure of democracy.¹⁷

One word of caution should be expressed. With a relatively small N—only seventy-seven peace spells when peace is defined most strictly—the many independent variables put substantial strain on the statistical models tested subsequently. Although the findings should be read with some care, it should also be noted that the data include the entire universe of cases of repeat civil war rather than a representative sample.

Controls

It is possible that conflicts are more likely to repeat themselves simply because grievances and opportunity are particularly high. A number of measures were included, therefore, to control for this. Three measures were incorporated into the analysis to take grievances into account: (1) Income—measured as log of per capita GDP in 2005 US dollars,¹⁸ (2) Ethnic Fractionalization—a measure of the ethnic diversity of the country based on estimates of ethnic group populations, and (3) Religious Fractionalization—a similar measure of religious diversity.¹⁹

The basic model also includes four core measures of opportunity highlighted by Fearon and Laitin (2003) and Collier and Hoeffler (2004) in their past research on civil war.²⁰ Political Instability—presumed to be a measure of state weakness—is a dummy variable that indicates whether there was any positive or negative change in the Polity 2 score in the previous country year. Mountainous Terrain is the log of the percentage of mountainous terrain in the country (plus one) as judged by geographer A. G. Gerard. Population Size is simply the log of the country population. The assumption here is that the greater the number of citizens in a country, the more difficult it is to monitor and control their behavior. Finally, Non-Contiguous Land Mass is a dummy variable that indicates whether the country contains a significant noncontiguous landmass that could be used as a staging ground for rebellion.²¹

The analysis also incorporates a number of different elements associated with bargaining failures. Specifically, I include data on key characteristics associated with the preceding war including whether the war was fought over territory or some other goal, the intensity of the previous war (defined as deaths per year—logged) and in alternate tests, the duration of the previous war (logged), whether the previous war ended with an outright military victory, and whether the two sets of combatants signed a comprehensive peace agreement. Data on previous duration and battle deaths and the outcome of the previous war are derived from the UCDP Conflict Termination Data Set V.2010-1, the UCDP/PRIO Armed Conflict Data Set V4.0, and the UCDP/PRIO Battle Deaths Data Set V3.0 (see Gleditsch et al. 2002; Lacina, Gleditsch, and Russett 2006). In alternate tests, I also incorporate their data on the degree to which comprehensive peace agreements included different aspects of power sharing (Hartzell and Hoddie 2003), as well as data on the number and type of UN peacekeepers on the ground in any given country from the UN Peacemaker website.²²

What Makes Countries More Likely to Suffer from Repeated Civil War?

Table 1 focuses on the three sets of factors that dominate the literature: grievances, opportunity, and bargaining problems. The first model in the table illustrates the relationship between core measures of these three factors and repeat civil war.

Table 1.

The Determinants of Peace Duration—Grievances, Opportunity, and Bargaining. Cox Proportional Hazards Model.

	Coefficient (SE)
Grievance
Ethnic fractionalization	−1.40 (1.59)
Religious fractionalization	1.80 (1.58)
GDP	−.39 (.31)
Opportunity
Population size	−.07 (.28)
Mountainous terrain	.01 (.24)
Political instability	−.67 (.77)
Noncontiguous land mass	.37 (.89)
Bargaining
Previous war intensity	.31 (.21)
Territorial conflict	−1.21 (.67)*
Clear victory	−.41 (.58)
Comprehensive peace agreement	−1.42 (.78)*
Peace keepers in country	.72 (.92)
Log likelihood	−69.96
Observations	1,043
Spells	75
Failures	21

Note: GDP = gross domestic product. SE = standard error.

*p < .10. **p < .05. ***p < .01.

Grievance

The results in Table 1 offer little support for grievance as a primary explanation of repeat civil war.²³ None of the standard measures of grievance is significantly related to repeated episodes of violence even though they were found to be significantly related to civil war onset (see Fearon and Laitin 2003). Repeat civil war is not tied to ethnic division or religious fractionalization in this test.²⁴ Nor is it linked to per capita GDP.²⁵ However, alternate tests do hint at a possible role for grievances. Although Table 1 does not illustrate it, when other types of measures of economic stress such as life expectancy, infant mortality, or the poverty gap are added to the model, they are all nearly significantly related to repeat conflict.²⁶ In each case, when economic conditions are poor, the likelihood of renewed war increases.²⁷

Opportunity

Table 1 shows little support for the role of opportunity in repeat civil war, at least as it is currently measured. None of the core measures of opportunity included in the model is significant. Again, this is quite different from what existing studies of civil war onset have found. A range of different tests reveal few links between alternate measures of opportunity and renewed conflict. Neither the overall rate of unemployment nor the unemployment rate of young adults in the country is significantly related to repeat civil war when added to the model in Table 1.²⁸ The presence of large oil reserves, a resource that Fearon and Laitin (2003) have demonstrated is potentially important for civil war onset, is not significant when added here. Similarly, when Ross’s (2011) more comprehensive data on oil and gas reserves or Lujala, Gleditsch, and Gilmore’s (2005) figures for diamond deposits were tested, no link between profitable resources and repeat conflict could be found. Additionally, when a measure of the percentage of the population that is urban is added to the model, it is also not significant. Finally, higher school enrollment, by any measure available, is unrelated to the reemergence of civil war.

The findings regarding government military strength are less clear. A larger armed force, greater arms imports, and more police are all not demonstrably linked to civil war renewal when added to model 1 in Table 1. But the proportion of all government expenditures that are directed toward the military is significantly and negatively tied to the recurrence of civil war. Countries that spend relatively more on their militaries are significantly less apt to experience repeat war—a relationship that suggests that governments may be able to deter rebel groups from returning to violence if they are able to repress them. It may also suggest that stronger governments are better able to defeat their opponent, making it impossible for them to restart the war. Nonetheless, this finding should be read with care, as military expenditures are available only for a small fraction of the cases.

Although the preponderance of findings suggests that neither opportunity nor grievance is the primary factor driving groups to return to war, both are likely to be critical factors driving the original conflict. The fact that neither plays an empirically evident role in causing combatants to return to war may be because grievance and opportunity are both relatively high across all of the cases. We cannot and should not, therefore, rule out grievance and opportunity as important factors in the larger process leading up to repeat civil war.

Bargaining Problems

The findings reported in Table 1 are more sanguine about the relevance of information problems, divisibility problems, and commitment problems. The empirical relationships are neither strong nor, as we will see in subsequent tables, particularly consistent across different models. However, there are faint signs that a range of bargaining problems may influence whether combatants return to war.

First, there is some indication that credible commitments can help shape the course of peace and the likelihood of renewed conflict. Conflicts that end in comprehensive peace agreements have one-quarter of the risk of restarting in any given year than do conflicts that do have an agreement.²⁹ By resolving the underlying differences driving the war, these agreements, if implemented, remove the motive to fight. This logic gets further support from alternate tests in which the comprehensive peace treaty measure is replaced with Hartzell and Hoddie’s (2003) main power-sharing variable. In this test, power sharing is negatively and significantly related to repeat civil war. When the two parties are able to sign an agreement that gives both sides real power in the government, it appears to reduce the likelihood of returning to war.

Other factors touted as helping to address credible commitment problems do less well. Third-party enforcement may help combatants temporarily end one civil war, but it appears unable to consistently prevent groups from returning to war. When measured by a simple dummy variable indicating the presence of UN Peacekeepers on the ground, this type of intervention is not significantly tied to repeat civil war. Alternate tests also reveal no significant relationship between the number and type of UN personnel, or total expenditures on peacekeeping, and civil war renewal. When alternate codings of peacekeeping were substituted into the model in Table 1, none was found to be significant.³⁰

There are at least three reasons why peacekeeping may have no statistically significant effect on repeat civil war in this analysis. First, the analysis here only notes whether UN peacekeepers existed on the ground not whether they had been deployed to prevent a particular conflict from restarting. Thus, they could be in the country for an entirely different conflict or mission. Second, the absence of a relationship could be the result of endogeneity. Third-party intervention is often offered to a select set of particularly poor and vulnerable countries.³¹ Finally, peacekeepers rarely remain in a country permanently. Peacekeepers, therefore, may be critical to convince combatants to sign and implement a peace settlement but may play no role in the maintenance of peace once they leave (Walter 2002). Given these possibilities, firm conclusions about the value of peacekeeping are impossible to draw from these data.

The results of indivisibility are also somewhat unclear. The one measure of indivisibility included in Table 1—territorial conflict—is significant but not in the expected direction. Wars that were fought over territory have three-tenths the risk of returning to war than non-territorial wars. Alternate tests detailed later in this article also suggest that secessionist war have a lower probability of reigniting.

The one aspect of bargaining problems that gets no apparent support in Table 1 is information. There is little sign that wars that provided combatants with clear information about their relative capabilities resulted in a more lasting peace. Conflicts that ended with a decisive victory by either the government or the rebels were no less likely to return to war.³² Similarly, more intense wars appeared to follow no different path than less intense wars. In alternate tests, I also found no clear link between the duration of the previous war and the likelihood of a repeat civil war. It may be that information matters little in repeat civil wars because these are wars in which the combatants are well known to each other.

Government Accountability

What about government accountability? Are particular institutional features an important missing element in our understanding of repeat civil war, as argued previously?

I begin by examining the link between two broad and traditionally used measures of governance from Polity IV (the level of democracy and the level of autocracy) and renewed conflict. As expected and as other studies have shown, neither democracy nor autocracy is significantly tied to civil war renewal.³³ Less democratic countries are not significantly more likely to return to war. Moreover, alternate tests reveal that the CPIA—a broader World Bank (2012) measure of governance that incorporates both formal institutions and economic policy—and Vreeland’s alternative fixed Polity measure of democracy are also unrelated to repeat war. The absence of any connection between broad measures of regime type and repeat war suggests that if governance matters, it must be related to more specific institutional features that affect government behavior.

The theory outlined earlier suggests that three kinds of constraints—political accountability, legal constraints, and information transparency—may reduce the need for rebels to resort to violence. The next three columns of Table 2 assess each constraint directly.³⁴

Table 2.

Government Accountability and Peace Duration, Cox Proportional Hazards Models.

	Level of democracy	Political accountability	Legal constraints	Information transparency	Freedom House index
Government accountability
Autocracy	−.07 (.20)	—	—	—	—
Democracy	−.01 (.16)	—	—	—	—
Public participation	—	−4.69 (2.44)*	—	—	—
Rule of law (constitution)	—	—	−2.19 (1.07)**	—	—
Free press	—	—	—	.02 (.02)	—
Rule of law + political rights	—	—	—	—	−2.36 (1.18)*
Grievance
Ethnic fractionalization	−1.52 (1.71)	1.40 (2.69)	.54 (1.94)	−3.71 (3.74)	−.97 (1.61)
Religious fractionalization	1.63 (1.61)	−1.03 (2.70)	3.26 (2.27)	2.87 (4.13)	1.64 (1.55)
GDP	−.27 (.37)	.27 (.53)	.29 (.44)	−.71 (.51)	−.07 (.35)
Opportunity
Population size	−.06 (.31)	−.94 (.51)*	−.40 (.35)	.40 (.49)	−.13 (.28)
Mountainous terrain	−.09 (.27)	−.16 (.33)	−.18 (.31)	−.47 (.47)	−.10 (.24)
Political instability	−.63 (.83)	−.32 (.89)	−.64 (.84)	—	−.79 (.77)
Noncontiguous land	.12 (.91)	1.83 (.126)	1.02 (.97)	.67 (1.55)	.69 (.89)
Bargaining
Previous war intensity	.34 (.23)	.08 (.34)	.31 (.26)	.76 (.73)	.27 (.21)
Territorial conflict	−.95 (.69)	.34 (.89)	−.65 (.78)	−.44 (1.05)	−1.02 (.64)
Clear victory	−.36 (.60)	.08 (.84)	.24 (.65)	—	−.46 (.61)
Peace agreement	−.99 (.80)	−.14 (1.34)	−1.93 (1.25)	−1.78 (1.16)	−.89 (.81)
UN peace keepers	−.39 (1.19)	.08 (.34)	1.00 (1.36)	1.45 (1.11)	.80 (.95)
Log likelihood	−60.35	−30.55	−49.64	−22.86	−67.05
N	882	651	904	677	998

Note: GDP = gross domestic product. Values are coefficients and standard errors in parentheses.

*p < .10. **p < .05. ***p < .01.

The results suggest that at least two of these constraints affect the likelihood of repeat civil war.³⁵ Likelihood ratio tests also confirm both of these relationships.³⁶ First, as column 2 demonstrates, widespread political participation appears to help prevent renewed violence. Table 3, which presents the hazard ratios for all five regression models in Table 2, shows that the effect is dramatic. The hazard ratio is .01, which means that the risk of repeat conflict in countries with the highest levels of public participation is 1/100th the level of risk in countries where few members of the public are involved.

Table 3.

Illustrating the Effects of Government Accountability on the Risk of Renewed War.

	Values are hazard ratios (Z scores) derived from Table 2 regression models
Model 1
Autocracy	.93 (−.37)
Democracy	.99 (−.06)
Model 2
Public participation	.01 (−1.92)*
Model 3
Rule of law (written constitution)	.11 (2.04)**
Model 4
Free press	1.02 (1.12)
Model 5
Rule of law + political rights	.09 (−1.99)**

*p < .10. **p < .05. ***p < .01.

Political accountability appears to have an equally dramatic effect on civil war renewal. As shown in column 3 of Table 2, there is a strong relationship between the presence of a written constitution and the likelihood of repeat civil war. Table 3 further illustrates that countries whose laws are enshrined in a constitution have about one-tenth the risk of returning to civil war in any given year as compared to countries without a written constitution.

The one dimension of internal constraints that gets no support in Table 2 is a free press. At least as measured here, greater press freedom is unrelated to the probability of returning to war.

The two measures that matter in Table 2—public participation and the presence of a written constitution—were chosen because they are concrete, straightforward measures that are less likely to be endogenous to any upcoming war and less likely to be influenced by coders’ biases or knowledge of impending events. Still, they are not necessarily the most conventional or comprehensive measures of these two dimensions. To address this concern and to test the robustness of these findings, I repeated the analysis in Table 2 with other available measures of internal governmental constraints (see Table S1 of the Online Appendix).

By almost any measure, political accountability is linked to the renewal of civil war. When I substituted into the model Freedom House’s (2009) broader Political Rights measure, it displayed a significant and robust relationship with repeat civil war. In alternate tests, a number of Polity measures (those designed to measure the ability of the public to openly compete for executive office) tended to be the most highly correlated with civil war outcomes. Similarly, Cheibub, Gandhi, and Vreeland (2010) have several variables that measure whether the executive and legislature are closed to public competition. All are significantly related to repeat civil war outcomes when substituted in the models. This pattern suggests that the ability to compete against and politically oust incumbent elites may be a critical factor in ensuring that peace will last.³⁷ Finally, the Voice and Accountability measure produced by the World Bank (2012) as part of its WGI series was also weakly related to the onset of a repeat civil war. In all cases, peace was more likely to last the easier it was for voters to hold the executive accountable and constrain elite actions.

Similarly, alternate tests confirm the role the rule of law plays in shaping the course of the peace. Both of the available alternate measures of the rule of law are significantly and substantially correlated with repeat civil war. Freedom House’s (2009) broad civil liberties measure is closely linked to civil war renewal when substituted in Table 2. The rule of law measure from WGI and the World Bank (2012) also has a strong relationship with renewed war. In every case, greater adherence to the rule of law is associated with a lower probability of repeat civil war.³⁸

The fact that a wide variety of indicators of governance and accountability are all significant despite the fact that they differ markedly by their source, their subjective or objective coding, and by the years and countries for which they are coded adds to our confidence in the role played by political features that limit the power of the executive. Thus, it is hard to imagine civil war repeating itself in either Aceh or in East Timor now that Indonesia has significantly democratized, improving on all measures of governance since 1999. It is also clear why the Democratic Republic of Congo (DRC), Uganda, and Myanmar continue to experience repeat challenges to the state.

One concern with these results, however, is that it is difficult to establish which political features are most influential. Since measures for the different constraints on government come from different data sets and since each of the different data sets has considerable missing data that do not overlap, two dimensions cannot readily be included in the same model. Moreover, although there are clear theoretical distinctions between the two robust dimensions of internal constraints, empirically the story is more mixed. The two significant measures of constraints in Table 2 (public participation and a written constitution) are only minimally associated with overall measures of democracy and can thus be considered to be distinct from a generic democracy score.³⁹ But the two different measures of constraints are themselves at least somewhat correlated with each other (r = .35). Governments that are subject to one type of constraint tend to be checked by other types of restrictions. Thus, while there is evidence that constraints on government are tied to fewer repeat civil wars, untangling exactly which constraints are most effective is a more difficult endeavor.

The goal, however, is less to determine which constraint matters than it is to highlight the importance of internal constraints as a whole. This in itself is a contribution since political accountability has been largely absent from research on civil wars to date. With this in mind, I present one more general test of political constraints in Table 2. The last column of the Table assesses the relevance of Freedom House’s (2009) Politics and Freedom Index. The Politics and Freedom Index combines the Freedom House subcategories—political rights and civil liberties—into a single composite measure that can be construed as a rough index of the degree to which a government is constrained by its laws and its people. As the last column of Table 2 demonstrates, the risk of repeat war is much higher in countries with few political and legal constraints than it is in countries where the government is sharply constrained by the public and the courts.⁴⁰ The risk, as Table 3 further illustrates, is, in fact, ten times higher when governments are unaccountable.

Another way to assess the role of governance and institutional constraints is to look at change in governance across time. Countries that are able to break out of the cycle of repeat war should do so at least in part because governance has improved over time, while countries that fall back into war should do so at least in part because government has become less open and less credible.

Although it is difficult to rigorously assess changes in governance over time, a look at institutional change between the first measured peace year of a peace spell and the last measured peace year of that spell reveals a stark pattern. Judged by the Freedom House (2009) Politics and Freedom Index, among those conflicts that did not return to war, most tended to occur in countries where governance improved over time. Governance improved in 60.7 percent of peace spells that did not end in renewed war. By contrast, conflicts that reverted back to war generally occurred in countries where governance declined or was stagnant. Governance declined or did not improve in 74.6 percent of the peace spells that ended in conflict.⁴¹ Government attempts to improve governance and create stronger institutional constraints appear to be an important step in ending the cycle of repeat civil war.

Robustness checks

To further test the robustness of these results, I performed a series of additional analyses on different sets with different definition of peace and renewed war. First, given potential concerns about collinearity and other problems, I repeated the analysis in Table 2 with a number of paired-down regression models that included only the key governance measures. Governance measures that were significant in multivariate models remained significant in these bivariate tests (Table 4).

Table 4.

Government Accountability and Peace Duration, Cox Proportional Hazards Models—Reduced Models.

	Level of democracy	Political accountability	Legal constraints	Information transparency	Freedom House index
Government accountability
Autocracy	.03 (.17)	—	—	—	—
Democracy	.05 (.15)	—	—	—	—
Public participation	—	−.03 (.01)*	—	—	—
Rule of law (written constitution)	—	—	−1.23 (.58)**	—	—
Free press	—	—	—	−.00 (.01)	—
Rule of law + political rights	—	—	—	—	−2.34 (.94)**
Log likelihood	−76.56	−40.07	−66.92	−42.34	−67.05
N	1,002	743	936	738	998

Note: Values are coefficients and standard errors in parentheses.

*p < .10. **p < .05. ***p < .01.

Also, to help address issues related to omitted variable bias, I replicated these stripped down regression models with country fixed effects. The results did not alter the main governance story (Table 5).

Table 5.

Government Accountability and Renewed War—Reduced Models-Xtreg with Fixed Effects.

	Level of democracy	Political accountability	Legal constraints	Information transparency	Freedom House index
Government accountability
Autocracy	−.0056 (.0043)	—	—	—	—
Democracy	−.0057 (.0040)	—	—	—	—
Public participation	—	−.0013 (.0003)***	—	—	—
Rule of law (written constitution)	—	—	−.136 (.031)***	—	—
Free press	—	—	—	.0007 (.0005)	—
Rule of law + political rights	—	—	—	—	−.041 (.029)
N	1,075	744	975	764	1,146

Note: Values are coefficients and standard errors in parentheses.

*p < .10. **p < .05. ***p < .01.

In the second set of tests (reported in Table 6), I altered the definition of civil war episodes. Given that the analysis in Tables 1 and 2 rests on only twenty-one cases of repeat civil war, there is some concern that the results may be idiosyncratic. To test this possibility, I switched to a more lenient definition of peace—two consecutive years with fewer than 1,000 battle deaths—and reran the analysis. The results are displayed in Table 6. Even with significantly more cases of repeat civil war, the results are nearly identical.⁴² All else equal, countries with few political and legal constraints are over five times more likely to relapse into war than are countries with full political rights and the rule of law.

Table 6.

Government Accountability and Peace Duration—Alternate Definition of Peace Cox Proportional Hazards Model.

	Coefficient (SE)
Government accountability
Rule of law and political rights	−1.79 (.91)**
Grievance
Ethnic fractionalization	.85 (1.11)
Religious fractionalization	1.46 (1.16)
GDP	−.17 (.24)
Opportunity
Population size	.04 (.22)
Mountainous terrain	−.03 (.19)
Political instability	−.02 (.48)
Noncontiguous land mass	.78 (.58)
Bargaining
Previous war intensity	.06 (.16)
Territorial conflict	−.39 (.44)
Clear victory	−.34 (.53)
Comprehensive peace agreement	−.30 (.66)
Peace keepers in country	−.05 (.80)
Log likelihood	−122.8
N	1,226

Note: GDP = gross domestic product. SE = standard error.

*p < .10. **p < .05. ***p < .01.

A much more difficult robustness test is to employ an entirely different definition of what a conflict represents. To do that, I take data directly from Kreutz’s (2010) analysis of intrastate conflicts. In that analysis, a civil war is redefined as any conflict that leads to twenty-five battle deaths in a calendar year and a conflict episode continues as long as annual battle deaths exceed twenty-five. Peace begins in the first year in which there are fewer than twenty-five battle-related deaths.

Radically changing the definition of a civil war also radically alters the peace periods, which in turn requires significant redefinition of the characteristics of the previous conflict, the groups involved, and other aspects of the empirical model. Thus, in Table 7, I utilize Kreutz’s (2010) coding for data on the previous war including its goals, intensity, duration, and how it ends. The analysis in Table 7, therefore, represents a different set of cases, different measures, and a different methodology.⁴³ The one constant is the inclusion of the Freedom House (2009) Political Rights and Freedom index. The results remain consistent. All else equal, countries with few political rights are more than twice as likely to experience a repeat civil war than countries with expansive political rights.

Table 7.

Government Accountability and Repeat Civil War—War Defined as More than Twenty-five Annual Deaths.

	Coefficient (SE)
Government accountability
Rule of law and political rights	−1.13 (.48)**
Grievance
Ethnic fractionalization	.34 (.50)
Religious fractionalization	.21 (.65)
GDP	−.23 (.18)
Opportunity
Population size	.09 (.10)
Mountainous terrain	.01 (.10)
Noncontiguous land mass	.54 (.29)*
Bargaining
Previous war intensity	−.01 (.01)
Territorial conflict	−.29 (.18)
Clear victory	−.61 (.33)*
Comprehensive peace agreement	−.10 (.46)
Peace keepers in country	−.28 (.37)
Years of peace	−.06 (.02)**
Constant	−2.44 (2.54)
Pseudo R²	.08
N	1,814

Note: GDP = gross domestic product. SE = standard error.

*p < .10. **p < .05. ***p < .01.

As before, other aspects of bargaining appear to be relevant to civil war recurrence. Conflicts that end with a decisive victory appear to be significantly less likely to reemerge than those that end with a negotiated settlement. This mirrors the findings of Licklider (1995) and Toft (2010), but given that it is not consistent across different methods used for defining peace and repeat civil war, readers should not place too much confidence in this finding.⁴⁴

I also performed a range of additional empirical tests on each of these different versions of the data set. First, to address concerns about the large number and potentially nonrandom nature of the missing observations, I used Stata’s MI command series to impute missing data and to rerun the analysis with multiple imputed data set. I tested a range of different imputations models. All led to nearly identical conclusions.⁴⁵

In light of potential unobserved heterogeneity across countries or across conflicts, I tested a series of shared frailty models. The results of these models suggest that unmeasured risk factors are not a problem within the data set. They also mirror the original findings, namely that governance strongly predicts repeat civil war outcomes. All of these additional tests are reported in Online Appendix⁴⁶ (see Tables S5 and S6).

Conclusion

What distinguishes civil wars that repeat themselves from those that do not? The ones that repeat themselves are more likely to occur in the most weakly institutionalized settings. Conflicts that occur in countries with more open political environments and more constraints on their executives are much less susceptible to repeat bouts of violence. This helps to explain why repeat civil wars are predominantly located in the Middle East and sub-Saharan Africa—the two most weakly institutionalized regions of the world. It is not democracy or poverty per se, but very specific accountability mechanisms that make countries more or less likely to experience repeat war.

A focus on governance fundamentally affects the policy prescriptions that follow. To date, organizations such as the World Bank (2012) have invested heavily in economic development believing this was the best way to reduce repeat civil war. Raise the average income of the population and increase economic growth, it was believed, and the risk of violence would decline. These reforms are still likely to be essential for preventing civil war in the first place. However, by focusing on various legal and political accountability mechanisms, this research leads to a complementary and potentially more important “fix” for countries that are already in the grip of repeat civil war. If the international community wants to end the seemingly endless waves of violence in countries such as Somalia, Uganda, and the DRC, it will need to focus on containing executive power and creating governments that are accountable to their populations more so than increasing GDP.

Establishing greater institutional constraints is not an easy task, and there are obvious barriers to this kind of policy prescription. Unconstrained leaders are unlikely to give up power easily. But leaders facing a tough and tenacious rebel group—a group that they cannot effectively put down—may be exactly those leaders with incentives to do so. If the international community can show these leaders that by agreeing to constraints they are more likely to avoid repeat war, it may be able to convince some of them to reform. Moreover, despite the difficult nature of the task, the data indicate that considerable institutional change is already occurring. Many governments do open up, and when they do, the likelihood of returning to war greatly diminishes.

Before the international community moves forward, however, critical questions need to be answered. Governance clearly matters. What is less clear are what aspects of good governance are most important in discouraging repeat civil war. What is also unclear is the exact mechanism by which governance influences outcomes. All of the measures of governance highlighted here make it easier for citizens to check the bad behavior of their leaders. They also make it easier for incumbent elites to credibly commit to implement promised reforms over time. But they also likely lead to fewer grievances and fewer opportunities to recruit soldiers willing to fight. Thus, a carefully designed laboratory experiment, a survey experiment in the field, and/or careful qualitative analysis that traces the causal process over time across several of these cases is needed to determine exactly how and why these constraints matter.

Despite these ongoing questions, scholars and policy makers should learn two things from this study. First, civil wars continue to be a problem because an increasingly small number of countries (Chad, Myanmar, the Philippines, Somalia, the DRC, and Afghanistan) cannot permanently resolve the conflicts they already have. Second, they should be aware that although many different economic factors affect the original start of violence, it is political factors that strongly affect whether peace emerges or wars restart. If the international community wants to reduce the incidence of civil war around the world, it must address the institutional weaknesses that make it so attractive and so necessary for rebels to return to war. In the end, the ongoing problem of repeat civil war appears to be more a problem of weakly institutionalized governments than it is a problem of persistent poverty, economic development, or the presence or absence of short-term peacekeepers.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Portions of this research were funded by the World Bank for their 2011 World Development Report on “Conflict, Security, and Development.

Supplementary Material

The online appendices and data supplements are available at .

Notes

References

Boix

Carles

Svolik

Milan

. 2013. “The Foundations of Limited Authoritarian Government: Institutions and Power-sharing in Dictatorships.” The Journal of Politics 75 (2): 300–16.

Buhaug

Halvard

Gates

Scott

. 2002. “The Geography of Civil War.” Journal of Peace Research 39 (4): 417–33.

Call

Charles T.

2012. Why Peace Fails: The Causes and Prevention of Civil War Recurrence. Washington, DC: Georgetown University Press.

Cheibub

José Antonio

Gandhi

Jennifer

Vreeland

James Raymond

. 2010. “Democracy and Dictatorship Revisited.” Public Choice 143 (1-2): 67–101.

Cingranelli

David L.

Richards

David L.

. 2013. Accessed May, 2013. “CIRI Human Rights Data Project.” Human Rights Quarterly 32 (2). http://www.humanrightsdata.org/documentation.asp. Accessed May, 2013.

Collier

Paul

Hoeffler

Anke

. 2004. “Greed and Grievance in Civil War.” Oxford Economic Papers 56 (4): 563–95.

Collier

Paul

Hoeffler

Anke

Soderbom

Mans

. 2009. “Beyond Greed and Grievance: Feasibility and Civil War.” Oxford Economic Papers 61 (1): 1–27.

Collier

Paul

Sambanis

Nicholas

. 2002. “Understanding Civil War: A New Agenda.” Journal of Conflict Resolution 46 (1): 3–12.

Cunningham

David

. 2006. “Veto Players and Civil War Duration.” American Journal of Political Science 50 (4): 875–92.

10.

DeRouen

Karl R.

Bercovitch

Jacob

. 2008. “Enduring Internal Rivalries: A New Framework for the Study of Civil War.” Journal of Peace Research 45 (1): 43–62.

11.

DeRouen

Karl R.

Lea

Jenna

Wallensteen

Peter

. 2009. “The Duration of Civil War Peace Agreements.” Conflict Management and Peace Science 26 (4): 367–87.

12.

Doyle

Michael

Sambanis

Nicholas

. 2000. “International Peacebuilding: A Theoretical and Quantitative Analysis.” American Political Science Review 94 (4): 779–801.

13.

Doyle

Michael

Sambanis

Nicholas

. 2006. Making War and Building Peace. Princeton, NJ: Princeton University Press.

14.

Dubey

Amitabh

. 2004. Domestic Institutions and the Duration of Civil War Settlements. Unpublished paper. New York: Columbia University.

15.

Fearon

James D.

1998. “Bargaining, Enforcement and International Cooperation.” International Organization 52 (2): 269–305.

16.

Fearon

James D.

2004. “Why Do Some Civil Wars Last So Much Longer than Others?” Journal of Peace Research 41 (3): 275–301.

17.

Fearon

James D.

2011. “Governance and Civil War Onset.” World Development Report 2011. Background Paper. Washington, DC: World Bank.

18.

Fearon

James

Laitin

David

. 2003. “Ethnicity, Insurgency and Civil War.” American Political Science Review 97 (1): 75–86.

19.

Fortna

Virginia Page

. 2004. Peace Time: Cease-fire Agreements and the Durability of Peace. Princeton, NJ: Princeton University Press.

20.

Fortna

Virginia Page

. 2008. Does Peacekeeping Work? Shaping Belligerents’ Choices after Civil War. Princeton, NJ: Princeton University Press.

21.

Freedom House. 2009. Country Reports. New York: Freedom House. http://www.freedomhouse.org/report/freedom-world/freedom-world-2009.

22.

Gates

Scott

. 2002. “Recruitment and Allegiance: The Microfoundations of Rebellion.” Journal of Conflict Resolution 46 (1): 111–30.

23.

Gates

Scott

Hegre

Håvard

Jones

Mark

Strand

Håvard

. 2006. “Institutional Inconsistency and Political Instability: The Duration of Polities.” American Journal of Political Science 50 (4): 893–908.

24.

Gilligan

Stedman

. 2003. “Where Do the Peacekeepers Go?” International Studies Review 5 (4): 37–54.

25.

Gleditsch

Nils

Strand

Havard

Eriksson

Mikael

Sollenberg

Margareta

Wallensteen

Peter

. 2002. “Armed Conflict 1946–2001: A New Dataset.” Journal of Peace Research 39 (5): 615–37.

26.

Grant

Ruth W.

Keohane

Robert O.

. 2005. “Accountability and Abuses of Power in World Politics.” American Political Science Review 99:29–43.

27.

Hartzell

Caroline

Hoddie

Matthew

. 2003. “Institutionalizing Peace: Power Sharing and Post Civil War Conflict Management.” American Journal of Political Science 2:318–32.

28.

Hartzell

Caroline

Hoddie

Matthew

Rothchild

Donald

. 2001. “Stabilizing the Peace after Civil War: An Investigation of Some Key Variables.” International Organization 55:183–208.

29.

Hegre

Håvard

Sambanis

Nicolas

. 2006. “Sensitivity Analysis of Empirical Results on Civil War Onset.” Journal of Conflict Resolution 50 (4): 508–35.

30.

Heston

Alan

Summers

Robert

Aten

Bettina

. 2009. Penn World Table Version 6.3. Philadelphia: Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania.

31.

Humphreys

Macartan

. 2005. “Natural Resources, Conflict, and Conflict Resolution: Uncovering the Mechanisms,” Journal of Conflict Resolution 49 (4): 508–37.

32.

Justino

Patricia

. 2009. “Poverty and Violent Conflict: A Micro-level Perspective on the Causes and Duration of Warfare.” Journal of Peace Research 46 (3): 315–33.

33.

Kaufmann

Daniel

Kraay

Aart

Mastruzzi

Massimo

. 2009. “Governance Matters VIII: Aggregate and Individual Governance Indicators 1996-2008.” World Bank Policy Research Working Paper 4978, World Bank, Washington, DC.

34.

Kreutz

Joakim

. 2010. “How and When Armed Conflicts End: Introducing the UCDP Conflict Termination Dataset.” Journal of Peace Research 47:243–50.

35.

Lacina

Bethany

Gleditsch

Nils Petter

. 2005. “Monitoring Trends in Global Combat: A New Dataset of Battle Deaths.” European Journal of Population 21 (2-3): 145–66.

36.

Lacina

Bethany

Gleditsch

Nils Petter

Russett

Bruce

. 2006. “The Declining Risk of Death in Battle.” International Studies Quarterly 50 (3): 673–80.

37.

Licklider

Roy A.

1995. “The Consequences of Negotiated Settlement in Civil Wars, 1945–93.” American Political Science Review 89 (3): 681–90.

38.

Lujala

Paivi

Gleditsch

Nils Petter

Gilmore

Elisabeth

. 2005. “A Diamond Curse? Civil War and a Lootable Resource.” Journal of Conflict Resolution 49 (4): 538–62.

39.

Marshall

Monty G.

Jaggers

Keith

Gurr

Ted Robert

. 2011. Accessed May, 2013. Political Regime Characteristics and Transitions, 1800-2010 (Polity IV). Center for Systemic Peace. http://www.systemicpeace.org/polity/polity4.htm.

40.

Murshed

S. Mansoob

. 2002. “Conflict, Civil War and Underdevelopment: An Introduction.” Journal of Peace Research 39 (4): 387–93.

41.

Østby

Grudun

. 2008. “Polarization, Horizontal Inequalities and Violent Civil Conflict.” Journal of Peace Research 45 (2): 143–62.

42.

Political Terror Scale. 2013. Accessed May, 2013. http://www.politicalterrorscale.org/.

43.

Quinn

J. Michael

Mason

David

Gurses

Mehmet

. 2007. “Sustaining the Peace: Determinants of Civil War Recurrence.” International Interactions 33 (2): 167–93.

44.

Regan

Patrick. M.

2002. “Third Party Interventions and the Duration of Intrastate Conflicts.” Journal of Conflict Resolution 46 (1): 55–73.

45.

Regan

Patrick M.

Frank

Richard W.

Clark

David H.

. 2009. “New Datasets on Political Institutions and Elections, 1972-2005,” Conflict Management and Peace Science 26 (3): 286–303.

46.

Ross

Michael

. 2004. “What Do We Know about Natural Resources and Civil War?” Journal of Peace Research 41:337–56.

47.

Ross

Michael

. 2011. The Oil Curse: How Petroleum Wealth Shapes the Development of Nations. Princeton, NJ: Princeton University Press.

48.

Salehyan

Idean

. 2009. Rebels Without Borders: Transnational insurgencies in World Politics. Ithaca, NY: Cornell University Press.

49.

Toft

Monica D.

2003. The Geography of Ethnic Violence: Identity, Interests, and the Indivisibility of Territory. Princeton, NJ: Princeton University Press.

50.

Toft

Monica D.

2010. Securing the Peace: The Durable Settlement of Civil Wars. Princeton, NJ: Princeton University Press.

51.

Vanhanen

Tatu

. 2000. “A New Dataset for Measuring Democracy, 1810–1998.” Journal of Peace Research 37 (2): 251–65.

52.

Walter

Barbara F.

1997. “The Critical Barrier to Civil War Settlement.” International Organization 51 (3): 335–364.

53.

Walter

Barbara F.

2002. Committing to Peace: The Successful Settlement of Civil Wars. Princeton, NJ: Princeton University Press.

54.

Walter

Barbara F.

2004. “Does Conflict Beget Conflict? Explaining Recurring Civil War.” Journal of Peace Research 41 (3): 371–88.

55.

Wood

Elisabeth Jean

. 2007. Forging Democracy from below: Insurgent Transitions in South Africa and El Salvador. Cambridge: Cambridge University Press.

56.

World Bank. 2012. Accessed May, 2013. AidData. World Bank. http://www.aiddata.org/content/index/data-search.

57.

World Bank. 2013. Accessed May, 2013. CPIA Indicators. World Bank. http://data.worldbank.org/indicator/IQ.CPA.STRC.XQ.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.39 MB

0.00 MB