An open access publication of the American Academy of Arts & Sciences
Fall 2017

Civil War & the Current International System

James D. Fearon
View PDF

This essay sketches an explanation for the global spread of civil war up to the early 1990s and the partial recession since then, arguing that some of the decline is likely due to policy responses by major powers working principally through the United Nations. Unfortunately, the spread of civil war and state collapse to the Middle East and North Africa region in the last fifteen years has posed one set of problems that the current policy repertoire cannot address well – for several reasons, conflicts in this region are resistant to “treatment” by international peacekeeping operations – and has highlighted a second, deeper problem whose effects are gradually worsening and for which there does not appear to be any good solution within the constraints of the present UN system. That is, for many civil war–torn or “postconflict” countries, third parties do not know how to help locals build a self-governing, self-financing state within UN-recognized borders or, in some cases, any borders.

JAMES D. FEARON, a Fellow of the American Academy since 2002, is the Theodore and Frances Geballe Professor in the School of Humanities and Sciences, Senior Fellow at the Freeman Spogli Institute for International Studies, and Professor in the Department of Political Science at Stanford University. He has written for such journals as Foreign Affairs, American Political Science Review, and The Quarterly Journal of Economics.

This essay provides an overview of the problem of civil war in the post-1945 international system. I first describe global patterns and trends over the whole period, and next sketch an explanation for the spread of civil war up to the early 1990s and the partial recession since then. There is reasonable evidence that United Nations and major-power policy responses since the end of the Cold War have contributed to the global decline in civil war since the early 1990s. However, the spread of civil war and state collapse to the Middle East and North Africa (MENA) region in the last fifteen years has posed one set of problems that the current policy repertoire cannot address well, and has highlighted a second, deeper problem whose effects are gradually worsening and for which there does not appear to be any good solution within the constraints of the present UN system.

The first problem is that compared with conflicts in Asia, sub-Saharan Africa, and Latin America, civil war and state collapse in the MENA region more directly affect the major powers, and possibly international peace and security more broadly. Third-party peacekeeping operations and a panoply of associated aid programs have been deployed to “treat” civil war – torn countries elsewhere, with a measure of success. In most cases, however, it will be impossible to apply this treatment model in the MENA region due to higher costs and other obstacles related to nationalism, the transnational jihadi movement, and the intensity of conflict among the region’s biggest powers.

The second problem is that third-party efforts to build effective, self-sustaining states in countries where states have collapsed due to civil war, misrule, or invasion have mainly been failures. This is painfully evident in the U.S. experiences in Afghanistan and Iraq. To some degree, it appears to generalize to the experience of postconflict peacekeeping operations and foreign aid efforts in a number of low-income countries outside of the MENA region. Third parties do not know how to bring about the construction of self-governing states that can support themselves financially within UN-approved boundaries.

A common misconception is that the contemporary prevalence of civil war is largely a post–Cold War phenomenon. Figure 1 shows that the number of civil wars in progress each year increased steadily throughout the Cold War, already reaching levels in the 1980s greater than at present. There was a rapid increase around the time of the end of the Soviet Union, a spike that contributed to the perception that widespread civil war was a new, post–Cold War international problem. But after reaching a high point of forty-eight ongoing wars in 1992, the prevalence of civil war has actually declined quite a bit, leveling out over the last fifteen years between the high twenties and low thirties.

Figure 1
Civil Wars by Year, 1945–2014

Figure 1

Source: Updated version of the civil war list described in James D. Fearon and David D. Laitin, “Ethnicity, Insurgency, and Civil War,” American Political Science Review 97 (1) (February 2003): 75–90. Available at

The UN state system expanded a great deal over this whole period, but we see basically the same trends if we consider the share of independent countries with civil wars (the dotted line and right axis in Figure 1, calculated omitting microstates that had populations smaller than half a million in the year 2000). It is also clear from these data that “prevalence” is the right word. Major civil conflict has affected roughly one in six nonmicrostates each year since 2000 and almost one in five today; at the peak in 1992, it was nearly one in three.1

Figure 2’s panels break down the trends by region. These mirror the global pattern for the two most conflict-prone regions, Asia and sub-Saharan Africa, and also for Latin America. The most striking exception is the MENA region, which roughly mirrors the other high-conflict regions until around 2003, but has seen a large increase from three wars in 2002 to twelve ongoing wars in 2014.2 All other regions had major declines in civil conflict after the early 1990s.

Figure 2
Civil Wars by Year and Region, 1945–2014

Figure 2

Source: Author’s coding, available at

From the steady increase after 1945 shown in Figure 1, one might suppose that civil wars were breaking out more frequently over time. This is not so. Civil wars have begun over the whole period at a rate of about 2.2 new conflicts per year on average, with at best a very slight trend downward.3 The reason for the impressive increase in prevalence up to the early 1990s is that the rate at which civil wars have ended has been consistently lower, averaging 1.77 per year. Suppose that each morning you pour a random amount of water into a tank and then remove a different random amount of water in the afternoon, with the average amount going in greater than the average amount coming out. The tank will gradually fill up. This same sort of dynamic is behind the gradual increase and the contemporary prevalence of civil war in the post-1945 international system.

A related implication is that the average duration of civil wars in progress has increased over time. The international system has been accumulating long-running conflicts. Figure 3 shows that the average duration of civil wars in progress is currently greater than twenty years, reflecting some very long-running, intractable conflicts in Afghanistan, Myanmar, the Philippines, India, Turkey, and Somalia, among others. Even median durations of wars in progress have climbed to remarkably high levels: it was nineteen years in 2010 and fourteen years in 2014 (the recent fall mainly reflecting the entry of a number of new conflicts in the wake of the Arab Spring).

Figure 3
Accumulation of Long-Running Conflicts, 1945–2014

Figure 3

Source: Author’s coding, available at

Three final observations concern types of civil conflicts. The proportion of civil wars in which rebels have aimed to capture the central government, as opposed to winning greater autonomy or regional secession, has been fairly stable since the 1960s, varying without clear trend between 50 and 60 percent. The proportion in which the combatants have been organized primarily along ethnic rather than ideological lines has increased somewhat over the whole period since World War II, from around 60 percent in the early years to around 70 or 75 percent since the end of the Cold War.4 A much more striking change has been the remarkable increase in the share of conflicts that involve avowedly jihadist rebel groups, from around 5 percent in 1990 to more than 40 percent in 2014 (see Figure 4).

Figure 4
Growth in Wars with a Significant Jihadi Presence, 1945–2014

Figure 4

Source: Author’s coding, available at

The spread and prevalence of civil war in the post-1945 UN system is related to the persistent gap between the rates at which civil wars have broken out and ended. But why have civil wars been easier to start than to end? This section sketches a two-part answer. First, decolonization produced an international system in which most states are former colonies with weak state structures and good conditions for guerrilla warfare or competing local militias. Second, these forms of armed conflict can be highly robust, so that civil wars are hard to end militarily. And they are also hard to end politically because stable power-sharing agreements between armed groups are extremely difficult to arrange within states.

On June 26, 1945, when the UN Charter was signed, there were sixty-four independent states, fifty of which joined that day. As a result of successive waves of decolonization and the breakups of the Soviet Union and Yugoslavia, the UN system has tripled to 193 member states at present. We have an international system composed of many relatively small and administratively, financially, and politically weak states. The median UN member today has a population of about 8.1 million, a bit smaller than New Jersey’s and more than one million fewer than that of the Chicago metropolitan area. Considering the 167 nonmicrostates, the median country has a population of 10.7 million; (approximate) examples include Somalia, Bolivia, and Haiti. Half of all UN member states are former colonies that gained independence since 1960, and more than two-thirds gained independence after 1945.

The colonial powers built state apparatuses in their colonies primarily to facilitate cash crop and natural resource extraction via a capital city, a few roads, and a port where possible. Administration often barely extended to rural peripheries. With the backstop of imperial militaries removed by decolonization, the option to try to use force to capture political control either at the center of a new state or in a region became more attractive for ambitious or abused would-be rebel groups. Postindependence leaders have – most of the time successfully – used state revenues and offices to buy supporting coalitions, reducing the risk of coup attempts and rebellions. But positive shocks to the relative strength of potential rebels versus a central government sometimes occur. These shocks create windows of opportunity to try to seize power or at least get an armed organization over a threshold of military viability against what are often chronically weak government forces.5

For example, the collapse of the Gaddafi regime in Libya in 2011 led to a flow of arms and fighters to northern Mali, providing a positive shock to insurgent capabilities that, in combination with postcoup weakness of the government in Bamako, made for civil war onset. In Iraq, the U.S. invasion and destruction of Saddam Hussein’s Baathist regime created a power vacuum and motivating principle for multiple armed groups to form and seek local or, looking to the longer run, national control. In Syria, the mass demonstrations sparked by the Arab Spring created a window of opportunity for the formation of armed rebel groups, spurred on by the aggressive repression of an Assad regime that saw no prospects for stable and safe power-sharing with a moderate opposition.

Once an armed rebel group gets over the threshold of military viability in a developing country with good conditions for insurgency, civil war can be extremely difficult to end. Civil wars end either by military victory or with a power-sharing agreement. The latter may take the form of greater regional autonomy provisions in the case of autonomy-seeking rebel groups, or the sharing of political and military positions by explicit agreement, or an electoral process in the case of wars fought over a central government.

In civil wars fought over a central government, stable power-sharing deals are hard to reach and implement in the absence of long-term, credible third-party commitments to enforce them.6 Each side has good reason to fear that the other would try to grab full control any chance it got and then use the full power of state forces against an effectively disarmed and exposed losing side. For example, the heart of the problem in the Syrian war has been that Assad and his supporters realistically fear that diluting their control of the Syrian military in any power-sharing deal would create an unacceptable risk of genocide against them: even relatively moderate Sunni opposition figures cannot credibly commit that greater opposition power would not unintentionally head in the direction of control by more extreme factions. Likewise, if opposition forces were to agree to a deal with Assad that gave them no real hold in the state’s military, Assad could not credibly commit not to use the military to punish and secure himself against future trouble from current opposition forces.

Power-sharing deals as means to end autonomy-seeking civil wars are more feasible because powers can be divided between territorially distinct central and regional institutions. Even so, central government fears that regional rebels would escalate autonomy demands from their stronger position and institutional base can make autonomy-seeking civil wars difficult to end via negotiated settlement.

These considerations help to explain a depressing regularity: A large majority of center-seeking civil wars since 1945, and about half of the autonomy-seeking conflicts, have ended by military victory rather than with significant negotiated power-sharing deals.7 Further, military victories, the alternative to power-sharing deals, are usually hard to come by when the mode of fighting is either guerrilla warfare or conflict among urban and semiurban militias in the context of largely collapsed central governments. Some of the strongest and most competent militaries in the world have struggled with guerrilla conflicts without much success. It is not surprising that less well-financed militaries with much worse command-and-control problems would struggle even more and cause even more killing of noncombatants, which can in turn help insurgents with their recruitment efforts.

How have other states and nonstate actors responded to the spread of civil war and the concomitant weakening of formal state structures? There was hardly any collective response until the end of the Cold War “unfroze” the UN Security Council. In the 1990s, the Security Council rapidly assumed the role of the main international institution for coordinating major power and international community responses to the newly discovered – or newly actionable – problem of civil war.

Figure 5 plots the number of UN peacekeeping operations (PKOs) in the field each year. It shows a rapid increase from an average of less than four per year before 1989 – the year of the Namibian UNTAG, which began an era of cooperation among the five permanent members of the Security Council on PKOs – to an apparent steady state of around seventeen missions per year since 1993. Most of the PKOs before 1989 were deployed to facilitate cease-fires or other agreements ending interstate wars, whereas since then, almost all PKO mandates have addressed peacekeeping or “peacemaking” in civil war–torn countries.

Figure 5
Civil Wars and un pkos by Year, 1945–2014

Figure 5

Source: Author’s coding, available at

Peacekeeping operations can be understood as a central part of an “international regime” that has developed since around 1990 to address the problem of civil war in the UN system.8 Their central logic is to try to make power-sharing arrangements – usually including postconflict elections – more feasible by providing third-party monitoring and enforcement capability to address credible commitment problems, like those outlined in the last section. Peacekeeping forces have deployed to oversee and monitor disarmament processes, to help implement postconflict elections, and often implicitly to provide security guarantees for new governments and former combatants. In some cases, peacekeeping operations began as or morphed into military operations against rebel groups, on behalf of a flimsy peace agreement or an extremely weak formal state (for example in Cambodia, Bosnia, Sierra Leone, and Mali).

The international regime for civil war goes well beyond PKOs, however. They are supported and supplemented by the work and money of a host of intergovernmental, regional, and nongovernmental organizations, aid agencies, donor conferences, and election monitoring and human rights organizations – all with programming and intervention theories developed for civil war–torn and “postconflict” countries. International norm entrepreneurs have also been active and somewhat successful in this area, as illustrated by the UN General Assembly’s vote to accept the responsibility to protect doctrine in 2005, and the development of a system of international criminal tribunals and courts focused on human rights abuses and crimes committed mainly in or around civil wars.

The PKO-based international regime for the “treatment” of civil wars has been roundly criticized for (what are argued to be) a number of high-profile and disastrous failures. Notably, in Somalia, Bosnia, Rwanda, and Eastern Congo there have been massacres, even genocide, under the noses of inadequately manned or mandated PKO troops. PKO personnel have moreover repeatedly engaged in sexual exploitation and abuse of locals and, in Haiti, caused a deadly cholera epidemic.9

At the same time, there is a strong case that, overall, the “PKO-plus” treatment has done a great amount of good for relatively small cost. Although they get much less media attention, quite a few missions are plausibly judged as largely or even highly successful. A number of studies have found that even though PKO missions on average go to relatively hard cases for maintaining postconflict peace, PKO treatment is associated with significantly longer peace duration after conflict.10 While it is difficult to be sure, it is plausible that a nontrivial amount of the post-1992 decline in civil war seen in Figure 1 is due to the UN system’s response through PKOs and related interventions.11 A remarkable 41 percent of the civil wars that have ended since 1991 (twenty-one out of fifty-one) have had UN PKOs. This does not mean that the PKO (and associated postconflict aid regime) caused or secured a durable peace in each case. But the evidence from comparisons of similar “treated” and untreated cases suggests that PKOs probably lower conflict recurrence and may increase the feasibility of peace deals that would be less likely without the third-party monitoring and enforcement instruments of the broader regime.12

Obviously, though, all is not well. Far from it, and the problems are deeper and more varied than can be gauged simply by charting the number and magnitude of ongoing civil wars. In this section, I briefly characterize two issues. One is an intractable problem that has become increasingly evident over time. The second is a relatively new cluster of problems associated with the spread of civil war and state collapse to the Middle East and North Africa region indicated in Figures 2 and 4.

First, while the PKO-plus regime has had some success at fostering peace agreements and making them more durable, third-party efforts to build effective, self-sustaining states in countries where states collapsed due to civil war, misrule, or invasion have mainly been failures. This is most clearly illustrated by the U.S. attempts at third-party state-building in Iraq and Afghanistan. In both, U.S. or U.S.-backed invasions destroyed the existing regimes and structures of government, such as they were. In Iraq, the United States attempted to install a democracy that would share power between predominantly Shia, Sunni, and Kurdish parties. Elections, coalition politics, and foreign influences gave control of the top offices to politicians from the majority Shia sect, who feared that genuine power-sharing with Sunnis (for example, in army leadership and the incorporation of Sunnis who had fought against Al Qaeda in Iraq) would open the door to coups or other types of subversion. The Shia-led government excluded Sunni politicians and rank and file to a degree that favored ISIS’s successful conquest of Mosul and much of Western Iraq by the end of 2014. In effect, the Shia governments have preferred exclusion, peripheral Sunni insurgency, and reliance on Iranian-allied militias to the more risky course of power-sharing at the center.

Despite years of training by the United States and many billions of dollars invested, the formal Iraqi army performed terribly after the U.S. withdrawal, completely disintegrating in the face of the ISIS attack on Mosul in June 2014 and losing Ramadi, Falluja, Tikrit, Hit, and other cities to relatively small numbers of ISIS fighters. In Afghanistan, the United States and NATO have tried to build capable army and police forces for even longer – fifteen years – again with disappointing results. Continued U.S. military support appears necessary just to maintain a costly stalemate with the Taliban. Without this support, it is likely that either the government in Kabul would fall or Afghanistan would return to the Taliban-versus-northern-armed-groups civil war of the mid-1990s. Politically, the United States has provided third-party backing for a power-sharing arrangement between competing factions (President Ashraf Ghani and “Chief Executive” Abdullah Abdullah), but the government has been largely dysfunctional.13 The formal, UN-member Afghan state would be unable to survive financially without massive foreign backing: between 70 and more than 90 percent of government revenue comes from foreign aid.14

The present Afghan state is, in effect, a ward of “the international community.” To varying degrees, this is true of what may be an increasing number of UN member states. One rough indicator is the increasing duration of peacekeeping operations. For UN PKOs addressing civil wars, the average duration increased from two years for operations in the field as of 1991 to eleven years for operations in the field as of 2014. In other words, PKOs tend to “hang around,” unable to leave without unacceptable risk of returning to, or worsening of, armed conflict. Another rough indicator is dependence on foreign aid, measured by comparing total aid receipts to total central government expenditure. On average, from 2004 to 2014, for at least one in five UN member states, aid receipts equaled at least half of all government expenditures (whether we consider all states or only nonmicrostates). Looking only at the countries in the World Bank’s “low-income” category for 2014, median aid dependence was a remarkable 86 percent. This suggests that in at least half of these low-income countries, more than half of all (intended) spending on nonmilitary public goods has come from taxpayers in OECD countries.15 Not surprisingly, many of the most aid-dependent countries are either postconflict or mired in conflict. For example, states at or near the top of the list include Liberia, Afghanistan, Somalia, Sierra Leone, the Central African Republic, Cambodia, the Democratic Republic of the Congo, Rwanda, Burundi, Nicaragua, Mozambique, Chad, and Mali.

Higher-income UN members can continue to pay to maintain the semblance of statehood according to UN norms in low-income and civil war–torn countries. Hopefully, in some cases, these subsidies will no longer be needed because state-building will eventually occur. But in other cases, it may be that the construction of capable and effective state institutions can only be carried out by locals – third parties simply do not know how or cannot do it – in processes that will be bloody, slow, and will not necessarily produce functioning states that operate exactly within current UN-recognized borders. Recall that this was the case historically for state-building in most of today’s major powers.16

In sum, while there is a good argument that the PKO-plus regime has been a moderately effective and relatively low-cost means of addressing the spread of civil war in the UN system, the regime has no good answer to the long-term question of how third parties can reliably foster the building of capable, not-awful states in civil war and postconflict settings.

The second major problem stems from the spread of civil war and state collapse into the Middle East and North Africa over the last fifteen years. These are regions in which internal conflict has particularly large negative externalities for the major powers, but also where the PKO-plus treatment regime is difficult and often impossible to apply.

Although the roots are deeper, the rise of civil war and state collapse in the MENA region began in earnest after 9/11, with the U.S. invasions of Afghanistan and Iraq sparking civil (and anti-invader) wars in both countries (see Figure 2). In Yemen, war between the Houthis in the north and the government in Sanaa broke out in 2004, the same year that Pakistan saw one insurgency restart in Baluchistan and another begin in the form of the Pakistani Taliban. The conflicts have continued, escalating in Yemen with the addition of a war in the south involving the local Al Qaeda branch and southern separatists. Following uprisings in the Arab Spring, Libya and Syria collapsed into major wars while in Egypt, a lower-level insurgency developed in the Sinai.

In contrast to civil wars in Africa and the mainly relatively small separatist conflicts in Asian countries, civil war and state collapse in the MENA region has much larger bad consequences for European states and, arguably, for “international peace and security” (the Security Council’s formal charge). Exhibit A is the Syrian war and the rise of the Islamic State in eastern Syria and western Iraq. The massive refugee disaster raises risks of contagion of civil war and state collapse to other states in the region, and has played into the growing pressures on European democratic politics and norms. The war has also led to dangerous escalations of the Saudi-Iranian cold war and U.S.-Russian conflict, along with Kurdish-Turkish and Sunni-Shia conflicts in the region. Elsewhere, anarchy in Libya poses international problems due to refugee flows, while the war in Afghanistan reflects in part and certainly engages the volatile and dangerous conflict between Pakistan and India. The program of some Islamic fundamentalists involved in these conflicts involve terrorist attacks outside the region, and there is no doubt that they would use weapons of mass destruction for terror if they could get them.

Unfortunately, the international community’s PKO-plus treatment regime has not and probably cannot be applied in this region. In the first place, UN PKOs require major-power agreement, but, for example, the Syrian war has engaged the United States and Russia on opposite sides, at least concerning Assad (to this point).17 And even if an operation might get support from the Security Council in principle, there is typically great reluctance to send missions in the absence of a formal peace agreement and invitation by warring parties: the model is “peacekeeping” much more than “peacemaking.” This has been a barrier for UN and other third-party missions in all regions, but it may be more so in the MENA region given the number of significant regional powers engaged in intense competition there.

More important, even when Security Council political agreement is feasible, the fact that any foreign peacekeeping troops will surely act as a recruiting card for jihadis poses a major obstacle. Their rallying cry is to expel foreign influence. And finding capable peacekeeping forces from the region itself is made highly problematic by the Saudi-Iranian struggle, which ramifies into a region-wide Sunni-Shia conflict.

For civil wars that either ended since 1990 or are still ongoing, Table 1 shows the proportion that got UN PKOs (at some point) for each region. The MENA region has the largest number of wars with no PKO and the smallest number with a PKO. The sole PKO case is the abortive UN Supervision Mission in Syria that operated for just four months in 2012, an exception that proves the rule. Both before and since the rise of a violent, transnational Sunni jihadist movement that has greatly raised the costs for third-party peacekeeping, MENA has not been fertile ground for internationally sanctioned third-party support to end civil wars.18

Table 1
Number of Civil Wars with and without PKOs, by Region, 1990–2014

Regions PKO No PKO
MENA 1 (5%) 20 (95%)
Asia 2 (12%) 15 (88%)
Sub-Saharan Africa 16 (53%) 14 (47%)
Eastern Europe/Former USSR 6 (60%) 4 (40%)
Latin America/Caribbean 3 (60%) 2 (40%)

Note: Includes civil wars that ended or were ongoing after 1989. Source: Author’s coding, available at

Before 1945, state-building was frequently a slow and often highly violent process. One can argue that, by contrast, the post-1945 UN system has done remarkably well as an experiment in the wholesale proliferation of the modern state form. The period has seen unprecedented, global advances in life expectancy and living standards, as well as widespread diffusion of electoral democracy and probably a significant improvement in human rights, on average. Many countries, including many new states, have been little affected by large-scale violence.19

But we are now seeing major pressures and strains for which the PKO-plus regime appears to be inadequate. This is mainly due to the rise of civil war, state collapse, transnational jihadism, and major and regional power proxy conflicts in the MENA region. The deeper roots stem from the failure of the Arab (and some other) republics to find, after independence, a formula for successful governance: that is, nonabusive, non-kleptocratic government that fosters and allows adequate economic growth. Transnational jihadi movements are a religious nationalist reaction seeking better governance and a sense of dignity. Unfortunately, they are also vicious and immoral in the extreme, and destined to fail as a governance model if they ever really get to try to implement their current vision.

The experience of the United States in Iraq and Afghanistan (and, for that matter, Vietnam) suggests that the problem of building a state that can finance and govern itself can only be solved by locals, in what may be a violent process. Third-party support for one faction or another, or for formal power-sharing between former enemies, may put on hold or even undermine effective state-building. While these pessimistic conclusions surely do not apply everywhere – see the general point about the successes of the UN system above – their relevance to a number of states in the MENA region is especially confounding for “the international community,” and most of all for the region’s people.

The international response should focus on delivering humanitarian relief where it is possible to deliver without making matters worse, and trying to help protect against spillover effects in contiguous states that are basically functional. Containing and degrading the Islamic State (and the like) is fine, but if the United States or other Western militaries do too much, this may effectively help sustain the movement as a terrorist threat by preventing it from failing or evolving on its own. It is hard to kill an ideology by bombing it. In the longer run, the problem is state-building, something that can only be durably accomplished by the residents.


1 These and subsequent civil war statistics are based on an updated version of the civil war list used in James D. Fearon and David D. Laitin, “Ethnicity, Insurgency, and Civil War,” American Political Science Review 97 (1) (February 2003): 75–90. The core criterion for magnitude is that the conflict kills at least one thousand people with an average of at least one hundred deaths per year. For details on other criteria, such as for marking starts and ends of conflicts, see ibid.; and James D. Fearon, “Why Do Some Civil Wars Last So Much Longer Than Others?” Journal of Peace Research 41 (3) (2004): 275–301. Figure 1 omits anticolonial wars, which have little effect on the number of wars but do make wars as a share of independent states (if we assign anti-colonial wars to the metropole) higher in the late 1940s and 1950s. Note that other civil war lists yield quite similar overall pictures; for example, Therése Petterson and Peter Wallensteen, “Armed Conflicts, 1946–2014,” Journal of Peace Research 52 (4) (2015): 536–550.

2 I have included Pakistan and Afghanistan in the MENA region here. Sudan and South Sudan are grouped in sub-Saharan Africa.

3 Since the number of states has greatly increased since 1945, the rate of civil war outbreak per state has significantly trended down.

4 These estimates have to be rough because there are a good number of cases that are difficult to designate as clearly “ethnic” or not.

5 Chronic weakness of many developing-country militaries can have political origins: Insecure leaders fear that better-resourced, more-competent militaries would pose a greater risk of coups. In effect, they trade off coup risk for insurgency risk. See Philip Roessler, Ethnic Politics and State Power in Africa: The Logic of the Coup-Civil War Trap (New York: Cambridge University Press, 2017); and William Reno, “Fictional States & Atomized Public Spheres: A Non-Western Approach to Fragility,” Dædalus 146 (4) (2017). Fearon, “Why Do Some Civil Wars Last So Much Longer Than Others?” gives a more developed version of the argument about shocks and civil war onset.

6 See Barbara F. Walter, “The Critical Barrier to Civil War Settlement,” International Organization 51 (3) (Summer 1997): 335–364; James D. Fearon, “Commitment Problems and the Spread of Ethnic Conflict,” in The International Spread of Ethnic Conflict, ed. David A. Lake and Donald Rothchild (Princeton, N.J.: Princeton University Press, 1998), 107–126; and Fearon, “Why Do Some Civil Wars Last So Much Longer Than Others?”

7 James D. Fearon and David D. Laitin, “Civil War Termination” (unpublished paper, 2007). Note that it is not unusual for a government that has essentially won militarily to grant minor or relatively cosmetic concessions in negotiations to formalize the end of conflict (for example, Guatemala in 1992). For an example of an autonomy war ended by decisive military victory, see Sumit Ganguly, “Ending the Sri Lankan Civil War,” Dædalus 147 (1) (Winter 2018).

8 Richard Gowan and Stephen John Stedman, “The International Regime for Treating Civil War, 1988–2017” Dædalus 147 (1) (Winter 2018). See also James D. Fearon and David D. Laitin, “Neotrusteeship and the Problem of Weak States,” International Security 28 (4) (Spring 2004): 5–43; and Stephen D. Krasner, ed., International Regimes (Ithaca, N.Y.: Cornell University Press, 1983) on the concept of international regimes.

9 Virginia Page Fortna and Lise Morjé Howard, “Pitfalls and Prospects in the Peacekeeping Literature,” Annual Review of Political Science 11 (2008): 283–301; and Bernd Beber, Michael J. Gilligan, Jenny Guardado, and Sabrina Karim, “Peacekeeping, Compliance with International Norms, and Transactional Sex in Monrovia, Liberia,” International Organization 71 (1) (January 2017): 1–30.

10 For examples, see Michael Doyle and Nicholas Sambanis, Making War and Building Peace (Princeton, N.J.: Princeton University Press, 2006); Fortna and Howard, “Pitfalls and Prospects in the Peacekeeping Literature”; Virginia Page Fortna, Does Peacekeeping Work? Shaping Belligerents’ Choices after Civil War (Princeton, N.J.: Princeton University Press, 2008); Michael Gilligan and Ernest Sergenti, “Do un Interventions Cause Peace? Using Matching to Improve Causal Inference,” Quarterly Journal of Political Science 3 (2) (2008); Lisa Hultman, Jacob Kathman, and Megan Shannon, “United Nations Peacekeeping and Civilian Protection in Civil War,” American Journal of Political Science 57 (4) (2013): 875–891; and Andrea Ruggeri, Han Dorussen, and Theodora-Ismene Gizelis, “Winning the Peace Locally: UN Peacekeeping and Local Conflict,” International Organization 71 (1) (January 2017): 163–185. Missions often viewed as having been effective include UNTAG (Namibia, 1989–1990), UNTAC (Cambodia, 1992–1993), UNTAES (Eastern Slavonia, 1996–1998), UNPREDEP (Macedonia, 1995–1999), MINUGUA (Guatemala, 1997), UNAMSIL (Sierra Leone, 1999–2005), UNTAET/UNMISET (East Timor, 1999–2005), and UNMIL (Liberia, 2003).

11 Work developing this argument includes Joshua S. Goldstein, Winning the War on War: The Decline of Armed Conflict Worldwide (New York: Penguin, 2011); and Human Security Report Project, Human Security Report 2009/2010: The Causes of Peace and the Shrinking Costs of War (New York: Oxford University Press, 2011).

12 For evidence on the latter point, see Aila M. Matanock, Electing Peace: From Civil Conflict to Political Participation (New York: Cambridge University Press, 2017); and Aila M. Matanock, “Bullets for Ballots: Electoral Participation Provisions in Peace Agreements and Conflict Recurrence,” International Security 41 (4) (2017).

13 See, for example, Mujib Mashalaug, “Afghan Chief Executive Abdullah Denounces President Ghani as Unfit for Office,” The New York Times, August 11, 2016.

14 Rod Nordland and Jawad Sukhanyar, “U.S.-Backed Effort to Fight Afghan Corruption is a Near-Total Failure, Audit Finds,” The New York Times, September 27, 2016. Anthropologist Thomas Barfield has added that since the nineteenth century, “successful Afghan rulers [have] specialised in dealing with foreigners by getting them to fund their states and equip their militaries.” See “Dr. Thomas Barfield Recommends the Best Books on Afghanistan,” September 16, 2010,

15 To summarize the calculations, let a be total Official Development Assistance (ODA) aid; g government spending raised by country taxes and other nonaid sources; ga government spending from aid in the form of direct budgetary support; and m military spending. From the World Bank’s data series for total Development Assistance Committee (DAC) aid and government consumption in constant dollars (DT.ODA.ODAT.CD and NE.CON.GOVT.CD), averaged for 2004–2014, I estimate that for the median low-income country, a/(m + g + ga) = 0.86. So g/a = 1/0.86 – m/a – ga/a. Thus, the ratio of nonmilitary government spending to aid from donors, g/a, is at best close to 1, since ga/a is on average about 0.15. Further, in low-income countries, the military share of GDP is typically 0.01 to 0.02, while aid as a share of GDP is about 0.11 (median). Thus, g/a = 1/0.86 – 0.15 – 0.015/0.11 = 0.88, implying that the share of aid in total nonmilitary spending by government and aid sources would be approximately a/(a + g) = 0.53.

16 For analyses of the problems of “alien rule” and third-party state-building that draw similar conclusions, see Michael Hechter, Alien Rule (New York: Cambridge University Press, 2013); and David A. Lake, The Statebuilder’s Dilemma (Ithaca, N.Y.: Cornell University Press, 2016). On the centuries-long path of state-building in Britain, see also Francis Fukuyama, “The Last English Civil War,” Dædalus 147 (1) (Winter 2018).

17 Barry R. Posen, in “Civil Wars & the Structure of World Power,” Dædalus 146 (4) (Fall 2017), argues that coordinated PKO interventions will become less common due to what he sees as increasing “multipolarity.”

18 Nor has Asia, which in our data appears to be partly accounted for by the much larger share of autonomy-seeking conflicts in this region (autonomy-seeking wars are in general less likely to get PKOs in our data). See also Michael Gilligan and Stephen John Stedman, “Where Do the Peacekeepers Go?” International Studies Review 5 (4) (2003) on the infrequency of PKOs in Asia.

19 Bruce D. Jones and Stephen John Stedman stress these points in “Civil Wars & the Post–Cold War International Order,” Dædalus 146 (4) (Fall 2017).