The Limits of Peace: Third Parties in Civil Wars in Southeast Asia, 1993–2004

Since 1989, civil wars have constituted the most prevalent and dangerous form of armed violence in the world. Civil wars occur everywhere, but they affect different parts of the world disproportionately. For example, the number of all warring dyads has declined or remained low in Europe, Latin America, and the Middle East.

In the context of civil war a dyad means a government versus a rebel group. There can, of course, be several dyads as well. Thus, the total number of dyads for a period or a region is likely to be higher than the number of conflicts. In Southeast Asia, ten conflicts involved only one dyad but two conflicts involved more than one dyad. In the rest of Asia, six conflicts involved only one dyad, and four involved more than one. In Europe, twelve conflicts involved only one dyad and only one conflict consisted of more than one dyad. In the Middle East, two conflicts involved only one dyad and two involved conflicts with more than one dyad. Finally, in Africa, sixteen conflicts consisted of one dyad, and fifteen conflicts had more than one dyad.

Africa has seen a dramatic decrease in the number of warring dyads since 2001. The trend in Asia, however, in the last fifteen years, is that between fifteen and twenty dyads have been actively fighting in a given year (Uppsala Conflict Data Program [UCDP] 2006a), meaning that conflicts are protracted, either because of a comparative lack of initiatives for conflict management or prevention, or because such initiatives, if undertaken, have not had much success.

The temporal dynamics of civil wars also differ between regions. Between 1946 and 2004, civil wars in Southeast Asia (we define the region as comprising Indonesia, Laos, Myanmar, Thailand, Papua New Guinea, Malaysia, Vietnam/South Vietnam, the Philippines, and Cambodia) last 8.29 years on average. The average civil war duration for the rest of the world is only 3.94 years.¹ For these reasons alone, we believe that this part of the world should receive particular scrutiny.²

As Figure One shows, almost one‐fourth of the world's low‐intensity armed conflict dyads (thirty‐two out of a total of 128 dyads) in the period of 1993–2004 were found in Asia as a whole. Fourteen of these dyads were in the Southeast Asia region.

While some civil wars may be harder to manage than others (revenge, retaliation, and issues of identity and valued resources make such conflicts harder to resolve), it is interesting to note that in the region we study there have been few formal third‐party attempts in the context of regional civil wars. Figure Two reveals the paucity of third‐party efforts in Southeast Asia and the rest of Asia.³ This figure includes all third‐party efforts, including bilateral talks and other types of measures such as indirect talks, “good offices” (in which a third party provides warring factions with a neutral place to meet), arbitration, and peacekeeping operations by third parties. Of the 311 attempts made in Asia, 234 of them were undertaken in conflicts in Southeast Asia.

Southeast Asia receives just about the same amount of third‐party attention as the Americas, even though Southeast Asia has almost twice as many conflicting dyads. Figure Two also shows that more than five times as many third‐party attempts are directed to the Middle East than to the conflicting dyads in Southeast Asia despite the fact that the Middle East has fewer dyads than Southeast Asia.

Of the fifteen conflicting dyads in Southeast Asia, twelve received some form of third‐party attention. Three dyads in Myanmar did not receive any attention,⁴ and overall, the many internal conflicts in Myanmar did not receive much third‐party engagement.

In short, Asia clearly deserves special attention. In this article we focus specifically on Southeast Asia, where most Asian civil wars take place, and where they tend to be long. They also tend to recur after fighting has stopped. By one count, between 1946 and 2004 there were twelve recurrences of war in Myanmar after fighting had been stopped for at least one year. The war between Indonesia and the Aceh rebels recurred twice.⁵ These wars are intractable. The terminations are unstable and recurrence is highly likely. On top of all this — and perhaps adding to the problem in the first place — third parties have paid relatively little attention to Southeast Asia.

In this article, we catalog third‐party involvement and outcomes in the region and then present results from simple tests that indicate the form of involvement that has been most linked to agreements between warring parties. First, we discuss the wars in the region and note some of the basic elements of their intractability. Next, we define relevant third‐party management terms and concepts. Then we look closely at management efforts in the region using an original data set. We conclude with a discussion of policy‐relevant implications from our findings.

We are able to do this sort of analysis with the help of a new data set on conflict prevention in low‐level disputes collected by staff at the Conflict Data Program at Uppsala University (see UCDP 2006b ).⁶ We have further merged these data with data collected at Canterbury University on characteristics of conflicts in Southeast Asia (see UOC 2006).

Between 1993 and 2004 there were fifteen active government–rebel low‐intensity civil war dyads (i.e., internal conflicts where one group was in conflict with the government over legitimacy or territory). These are listed in Table One. Some were active for the whole period but most were inactive for part of the period (inactivity is measured as fewer than twenty‐five battle‐related deaths per calendar year) or had recently had a full‐scale war, defined here as a higher‐intensity conflict that involved more than one thousand deaths. The Moro Islamic Liberation Front (MILF)–government dyad in the Philippines falls from the sample in 2000 when the event turned into a war (with more than one thousand deaths). We have found some important commonalities in these conflicts that we believe are relevant to the role and the success rates of third‐party actors.

The dyads in Southeast Asia are fairly asymmetrical, by which we mean that one party in conflict, invariably the rebel group, is small and usually well outgunned and outmanned. A majority of the rebel groups in the region during the study period fought wars in peripheral territories (by these we mean border areas). These wars can be characterized as “sons‐of‐the‐soil” events (see Fearon 2004; Weiner 1978). Sons‐of‐the‐soil conflicts are typically rural‐based civil wars that are low in intensity but have a tendency to last a long time. Indeed, there have been many such wars in Asian countries, including Sri Lanka, India, Myanmar, Indonesia, China, the Philippines, and Papua New Guinea (see Fearon 2004). Sons‐of‐the‐soil wars involve a peripheral ethnic group that is fighting for autonomy or secession. The group is inextricably tied to the territory it occupies; tensions mount when there is a valuable resource in the territory and the government adopts a policy of transmigration into the territory (e.g., Indonesia and West Papua).

According to James Fearon (2004; see also Fearon and Laitin 2005), sons‐of‐the‐soil wars are difficult to resolve because of credible commitment issues. Each side expects the other to renege on any possible agreement. For example, if the rebels become relatively stronger, the government might negotiate and offer greater autonomy, but later when the state gains the upper hand in the region it can easily renege on such deals because no outside authority has power to force it to keep its commitment (see Nilsson 2006). The government is unlikely to force an all‐out confrontation because the rebels do not directly threaten the central government. The government would like to eliminate the rebels, but this is militarily difficult because the rebels' knowledge of the local terrain gives them an advantage. Therefore, a relatively small band of rebels can sustain a long war. Such wars are intermittent with occasional cease‐fires and peace agreements. If the groups are able to bring international attention to their plight because of human rights violations, ethnic cleansing, and/or genocide, this may strengthen their hands and possibly stimulate third‐party engagement.

Fearon (2004) thus successfully accounts for persistent civil wars in which the rebel force can be quite small. Many of these wars might well be thought of as internal rivalries (see DeRouen and Bercovitch, forthcoming). Small guerilla units can hide from the government, and, therefore, they benefit from rugged terrain. They are weaker than the government, but the latter cannot find them. As the rebel insurgency grows so does the probability of being turned in by deserters or the rural peasantry (e.g., Che Guevara's rebel army in Bolivia.)⁷ According to Fearon (2005), it may not be worth the effort to capture or kill a few more rebels if this may require increased military presence in the region.⁸ Thus, Fearon has explained conditions for a long‐term civil war that becomes stalemated.

Conflicts in the Southeast Asian region usually take place in areas with transportation difficulties and seasonal weathers (i.e., monsoons) that may correspond to the notion of “rugged” terrain. They are also low in intensity, which may work against attracting third‐party attention. International reaction may be triggered only if major action against civilians is reported by outside media or by civil society organizations. If third parties do become involved they are likely to be faced with uphill efforts in ending these protracted wars.

In contradistinction, wars over control of national governments (e.g., El Salvador) are expected to be more severe in some phases — and include explosive initial efforts (see Fearon 2004) by the insurgents — but shorter and more tractable. These wars are not as directly affected by structural factors such as secession, resources, or sons‐of‐the‐soil dynamics. In the context of the Cold War, many of these conflicts attracted armed intervention (e.g., Laos, Vietnam, Malaysia) that shaped their outcomes.

Among the conflicts in our sample, the Cambodian rebels (United National Front for an Independent, Neutral, Peaceful, and Cooperative Cambodia [FUNCINPEC]) fought for control of the national government. It was periodically intense. With the aid of the United Nations and diminishing great‐power patronage from China, the war came to an end. The wars in Myanmar, on the other hand, have been mainly sons‐of‐the‐soil variety. Various regional ethnic groups have been involved in very long and intractable wars with the military junta that runs the country. These wars have sometimes terminated with cease‐fires or simply low activity, but they typically reemerge after a brief respite.

The Aceh–government dyad in Indonesia is another example of a sons‐of‐the‐soil war. The Gerakan Aceh Merdeka (GAM) rebels in Aceh fought for years for greater autonomy and/or secession. The Aceh region is rich in natural gas reserves, which has also fed the conflict as the rebels have fought for what they perceived as “their” natural resource. Controlling the resources has been important for the government, which has been willing to invest in the struggle to keep control, at the same time preventing other ethnic groups from making similar demands.

The conflict in Papua, New Guinea was also a sons‐of‐the‐soil event tied to a large copper mine on the island of Bougainville. The inhabitants of Bougainville — although ethnolinguistically diverse themselves — are ethnically different from those on the mainland. Thus, a low‐intensity conflict ensued with the Bougainville Revolutionary Army able to stop mining operations, the government incapable of taking back control, and the rebels pushing for secession.

Civil wars in Philippines have both religious and ethnic‐secessionist undertones. The religious struggles involve several separatist groups based in the southern island of Mindanao fighting for autonomy or secession. A similar situation exists in the southern Patani region of Thailand where we have an assortment of rebel Muslim separatists classified under the heading “Patani insurgents.”

In the next section we address the issue of how a third party, of any sort, can become involved in such conflicts, and what it can achieve if it is involved.

Given all the difficulties and complex nature of such conflicts, how are they to be terminated or managed? Clearly, negotiations between the parties would not be workable. Many states would avoid any overt form of intervention in a conflict that can legitimately be seen as essentially internal. The only viable and effective way to deal with such conflicts is to initiate some form of peaceful third‐party activity (see Crocker, Hampson, and Aall 2005) Third‐party intervention is particularly useful in the context of intrastate conflicts (see Bercovitch and DeRouen 2005). Mediation can help the parties overcome many of the difficulties and information barriers they face. Mediation can help each side save face while at the same time adopting an exit strategy. Because it is noncoercive and nonbinding, mediation is a particularly attractive policy option for parties in internal conflicts.

It will be useful to define certain terms and concepts as used in this article. A third party is defined as a party that is involved in either helping the warring parties to reframe issues that define their conflict, to reduce overt conflict behavior, or to regulate other matters relating to the interaction between the adversaries (see Bercovitch 1997; Wallensteen 2007). The involvement of a third party may sometimes benefit one of the warring parties more than the other, or, in the extreme case, the third party may even become militarily involved against one of them.⁹

Given the multiplicity of conflicts in the world today, it is hardly surprising that numerous third parties are willing and eager to get involved in conflicts, both internal and interstate. We divide third parties into one of the following categories:

• a state, defined here as any political entity found in the List of Independent States (Gleditsch and Ward 1999);

• an intergovernmental organization, a grouping of states with formal statutes and a formal name (i.e., the United Nations, Association of Southeast Asian Nations, etc.); and

• any other type of third party, which includes nongovernmental organizations (NGO), independent individuals, or religious organizations (see Bercovitch and Schneider 2000).

Each category of third parties has its own characteristics and modes of interventions, and these interventions can take different forms. For example, we have classified talks into four types: indirect, direct, bilateral, and unclear talks.

By indirect talks we mean talks where the warring parties do not talk face‐to‐face but indirectly through the third party acting as an intermediary or mediator. The mediator relays information from one party to the other. In order for an event from our data set to be coded as mediation, it must be substantially validated that the third party actually transmits information from one warring party to the other; shuttle diplomacy is an example.

By bilateral talks, we include talks between the third party and one of the warring parties. Bilateral talks differ from indirect talks in that we do not have considerable reason to believe that the third party is passing along information between the warring parties when going from one party to the other. In many of the bilateral talks, the third party is only talking to one of the warring parties, most often the government. The goals of the third party include passing along information and perhaps simply exploring the positions of the parties. These bilateral talks can include official visits in which the conflict is discussed.

When the combatants meet face‐to‐face with a third party present, we have coded that as direct talks.

Finally, when the character of the talks is uncertain, we have coded the talks as unclear, that is, there is not sufficient information to say whether the talks were indirect, direct, or bilateral.

In addition to the different types of talks, the data set includes information about third‐party events that can be observed and measured, such as fact‐finding missions, good offices, arbitration, permanent observers, and peacekeeping operations. These are different forms of third‐party activity, each with its own logic, structure, and set of behaviors.

In a fact‐finding mission, a delegation seeks to establish the facts of the conflict, for example, whether human rights abuses, instances of violence, or violations of cease‐fires have taken place. In good offices events, a third‐party country offers the warring parties the opportunity to meet on safe and neutral ground. It does not actively engage in direct talks with the parties but only facilitates the talks by providing locations and facilities and other logistical and material support.

In arbitration, a third party issues a binding decision on a matter, for example, an international court ruling, after both parties have agreed that the third party will decide. Permanent observers are deployed by third parties with representation and with the stated purpose of preventing and/or managing the conflict. Finally, we have also included peacekeeping operations in the dataset.

In this section of the article, we look at the record of third‐party intervention in the conflicts listed in Table One. To reiterate, we only look at third‐party involvement in low‐intensity warring dyads that have not yet experienced war, which is defined as a conflict leading to at least one thousand battle‐related deaths in a year. Thus the focus here is on preventive actions aimed at managing conflict and reducing the likelihood of escalation.

Figure Three displays the most active third parties in the region. We find that The Humanitarian Dialogue Centre (HDC) in Geneva, an NGO, has been the most active third party in the region.¹⁰ Several regional actors are also among the most active ones. In addition, major actors such as the United States, the United Nations, and small countries with a strong tradition of peacemaking such as Switzerland and Sweden are also taking active roles in Southeast Asia.

The Geneva‐based HDC, an NGO devoted to conflict resolution, proposed to mediate in the Aceh, Indonesia conflict in 2000, and became the most frequent mediator in talks there between the parties. The HDC mediated direct talks and also engaged in bilateral talks with the government of Indonesia and the exiled GAM leadership in Sweden. All of HDC's efforts were directed toward Aceh.

The United States's involvement in the region, confined to bilateral talks, was mainly directed toward the conflict in Aceh (eight events of engagement) and the conflict in the Philippine region of Mindanao (fifteen events). In the Philippines, the United States engaged in talks with both the government and (mostly) the rebel Abu Sayyaf Group (ASG) concerning the taking of hostages, which is a common occurrence in the conflict over Mindanao. The United States also increased its third‐party presence in the conflict following the start of the “war on terror” in late 2001, engaging the government in talks on the conflict and on perceived links between Muslim groups in the Philippines and the Taliban movement in Afghanistan.

Australia and the Solomon Islands were the most important preventive diplomacy actors in the Papua New Guinea conflict over Bougainville in the time period of the study. The Solomon Islands acted as a mediator, engaged in bilateral talks with both of the warring parties, and gave good offices at several occasions. Australia engaged the government in bilateral talks, often supporting the government's position, and facilitated talks and provided good offices for conferences and high‐level meetings between the parties. In 1997, after the conflict in Papua New Guinea turned into one of low military activity (i.e., the violence diminished to less than twenty‐five battle‐related deaths, which is the UCDP threshold for inclusion), New Zealand and Australia became active in facilitating an agreement, but this activity falls outside our data set, which only records preventive activity when conflicts are above the threshold for a minor armed conflict.

The majority of the efforts undertaken by the United Nations and Thailand were directed at the many conflicts in Myanmar. Both third parties were engaged in bilateral talks with the government side. These talks did not so much address the conflict with the opposition All Burma Students Democratic Front as they did the general situation of democracy and human rights in Myanmar.

The majority of the Malaysian measures (thirteen out of seventeen) were directed toward the conflict in the Patani region of Thailand, which borders Malaysia. Malaysia held several rounds of bilateral talks with the Thai government on issues of border security, border patrols, and economic development for the Muslim areas in southern Thailand. Malaysia also criticized heavy‐handed government actions against the Muslim population.

Japan's efforts focused on the conflicts in Cambodia and in Aceh. In Aceh, Japan was part of the “Quartet,” a coalition that also comprised the United States, the World Bank, and the European Union, which cosponsored the peace process and put pressure on the Indonesian government to return to negotiations when these faltered. Countries such as Switzerland and Sweden also contributed to help avert war in the region. A notable number of good offices were given by Switzerland (all the efforts made by the Swiss were good offices) and Sweden. (Geneva is the home base of the HDC and Sweden harbored the exiled leaders of the rebel group, GAM.) The presence of the GAM leaders in Sweden also prompted the Swedish and Indonesian governments to hold a series of talks, both on the conflict and on the Indonesian demands that the leadership be extradited to Indonesia to stand trial for committing “terrorist” attacks.

Figure Four displays the different types of efforts taken in Southeast Asia for the period of 1993–2004. It shows that bilateral talks are undoubtedly the most common approach to handling low‐intensity, persistent civil wars in the region. The most common strategy seems to be to maintain dialogue with one warring party at a time. The results also indicate that it is more common for third parties to engage in bilateral talks with the government side than with the rebels. The category talks in Figure Four includes indirect, direct, and unclear types of talks. Eighteen of these events were direct — that is, the parties met face‐to‐face — in contrast to indirect talks where the parties communicated only through a mediator. Most of the direct talks concerned the conflict in Aceh, Indonesia (thirteen of them). All of these talks were mediated by the HDC. Some direct talks were also held in response to the conflicts in Bougainville, Papua New Guinea (five), Cambodia (three), and Mindanao, Philippines (one).

Most of the good‐offices efforts concerned the conflicts in Bougainville, Papua New Guinea, and the conflict in Aceh, Indonesia. In Bougainville, Australia facilitated most of the good offices, and in Aceh, Indonesia they were facilitated mostly by Sweden or Switzerland.

The fact‐finding efforts took place in Indonesia, Myanmar, Papua New Guinea, the Philippines, and Thailand. The third parties engaged in these events were, for the most part, NGOs such as Amnesty International, the HDC, and other regional and local organizations.

Only one peacekeeping mission was deployed in an armed conflict in the region: the Truce Monitoring Group (TMG), comprising peacekeepers from New Zealand, Australia, Fiji, and Vanuatu, was sent to Bougainville, Papua New Guinea. No arbitrations or permanent observations took place in the region.

Table Two presents the type of third‐party engagement in each of the conflicts in the region.

The conflict in Aceh received the most attention by third parties, as shown in Table Two. The parties to the conflict engaged in a serious peace process from 2000 to mid‐2003, with the HDC as the main active third party.¹¹ The relatively numerous efforts (seventy‐three) to help resolve the conflict in Aceh were undertaken during a five‐year period (1999–2004).

In contrast, Myanmar received the least attention, with only twenty‐six combined third‐party efforts undertaken during a ten‐year period (1994–2003). As stated before, the territorial conflicts in Myanmar received very little or no attention at all from the international community, as shown in Table Two. The third parties that did engage themselves in Myanmar were the United Nations (ten attempts) and Thailand (nine attempts). A few attempts were also made by Australia, the United States, and Amnesty International. In Myanmar, many of the bilateral talks held with Thailand as a third party dealt with how to handle the refugee situation in the border region of Myanmar and Thailand. The U.N. was also concerned about displaced people and the general humanitarian situation in the country.

In Bougainville, Papua New Guinea, the parties met without a mediator in 1993, after three years without any talks at all. In the beginning of 1994, bilateral talks between the government and third‐party regional states were held. Direct negotiations brokered by the Solomon Islands did not start until later that year. When the rebels declined to attend a peace conference and at the same time withdrew from the peace process, the situation escalated. In 1995, talks resumed, but the momentum had been lost. Formal negotiations finally restarted in July 1997 in New Zealand. In late 1997 the parties agreed on the imposition of a Truce Monitoring Group and drew up plans for a high‐level meeting of the leaders of the warring parties. The peace process continued through 1998, and the conflict was settled in 2001 within the framework of the Bougainville Peace Agreement.

In Mindanao, Philippines, third parties did little initially to prevent the conflict from escalating. But in 1996 a peace agreement was signed between the government and the largest warring faction in the conflict, the Moro National Liberation Front (MNLF), but neither of the other two rebel groups, the ASG nor the Moro Islamic Liberation Front (MILF), was a party to this agreement. Instead the government entered into direct talks with the MILF regarding a cease‐fire. The talks were interrupted and restarted several times in the next few years. In 1998, the MILF and the government signed a cessation of hostilities and agreed on further talks. Because of the situation's instability, the government and the MILF signed another cease‐fire agreement in January 1999 and once again agreed to start formal talks. In 2000, the ASG became involved in talks with the government following a host of hostage‐takings. The hostage situations drew international attention, and Libya, Malaysia, and the United States engaged the government in a series of bilateral talks on these issues throughout the year. This pattern of talks and negotiations was repeated in 2001 and 2002. In 2003 and 2004, the U.S. entered into a number of bilateral talks with the Filipino government on action against the ASG, but no further negotiations with the ASG were held.

In Thailand, the first year of conflict concerning the Patani region saw almost no talks related to the conflict. When the conflict intensified in 2004, however, so did the preventive activity, and the Malaysian and Thai governments held bilateral talks on how to thwart the insurgency. Throughout 2004 additional summits with regional and international organizations were held, and the Patani insurgency was debated. A number of fact‐finding missions were active during 2004, one launched by European countries under the auspices of the European Commission, one that comprised Muslim nations, and one conducted by a human rights commission in Thailand.

In Cambodia, regional third parties such as the Association of South East Asian Nations (ASEAN), Japan, and Thailand were among the most involved outside actors. Because of its history as a former colonial master in Cambodia, France also engaged actively in the conflicts. Talks involving these third parties started in July 1997, when FUNCINPEC, a royalist political party, renewed its armed struggle against the government. In 1998, negotiations produced some progress with both FUNCINPEC and the government of Cambodia announcing unilateral cease‐fires in February, the latter after talks with EU and French representatives. In late March, the exiled FUNCINPEC leaders returned to Cambodia, following talks between the United Nations, European diplomats, and the warring parties. Later that year, the talks led to an agreement where the parties again committed themselves to the 1991 peace agreement.

One obvious indicator of third‐party effectiveness is the achievement of an agreement; another is the duration of such an agreement. We treat outcome as a dichotomous measure coded here for months in which a mutual agreement was signed. Mutual agreements include any agreement (formal or informal) between the warring parties that regulates either the incompatibility of the conflict (i.e., it resolves the key disagreements among the parties) or that regulates any form of conflict behavior (such as withdrawal of troops). While the occurrence of an agreement at approximately the same time as third‐party involvement does not prove causality, we can nevertheless show some degree of association. The data collected here can be used to identify the forms of involvement associated with agreements.

A good place to begin our analyses is with bilateral talks and mediation. Talks are coded as bilateral if a third party speaks to only one of the warring parties and there is no evidence of information passing between warring parties. We also code this variable for a period of two months following the talk based on the expectation that talks might have a delayed effect.

Mediation follows the same coding rules as direct talks, but only when the third party is an intermediary and shares or facilitates the exchange of information between warring parties. We found twenty‐two months in which mediation took place. The HDC was the most frequent mediator with twelve events. Other mediators were France (two), Solomon Islands (three), ASEAN (one), the U.N. (one), the Red Cross (one), a Catholic priest (one), and two other NGOs (one each).

A cross‐national time‐series data set was created, with months as the time variable and dyads as the panel. In this way we have also been able to capture months in which the dependent variable (agreement) did not occur. Using simple cross‐tab measures of association, we found no evidence of a statistically significant relationship between bilateral talks and agreements (χ² = 0.23, P = 0.63, n = 672). We found, however, a strong positive relationship between mediation and agreements (χ² = 22.15, P = 0.000, n = 672).

We also analyzed the specific actions of the United Nations and the United States. U.S.‐sponsored bilateral talks were not significantly associated with agreements (χ² = 1.97, P = 0.16, n = 672). Most of these U.S. bilateral talks were with governments. The U.N. fared no better with its bilateral talks (χ² = 0.005, P = 0.95, n = 672).

We argue, therefore, that third‐party efforts to bring warring parties to agreement should be viewed as cumulative processes: impact can most likely only be expected over a period of time. Thus the fruitless attempts by the HDC to deal with the situation in Aceh may still have served as a basis for the negotiations conducted in 2005 by the Helsinki‐based Crisis Management Initiative. The enormous impact of the Indian Ocean tsunami cannot be overlooked, however. Similarly, the many attempts in the Bougainville conflict, spanning more than a decade from the first cease‐fire agreement in 1990, may have served to acquaint the parties with the opponent's interests and perceptions, ultimately resulting in an agreement with lasting potential.¹²

Another striking feature of the data analyses is that many of the mediation attempts involved third parties from outside the region. To some extent these were actors in the adjacent region (Australia, New Zealand, and the Solomon Islands), but largely they were from even more distant regions. Many of the talks were actually based in smaller countries in Europe such as Switzerland, Sweden, and later, Finland. This complements the Norwegian involvement in the conflict between the Sri Lankan government and the Tamil Tigers, which included a Nordic cease‐fire monitoring mission. The 2005 agreement on Aceh included EU observation.

The fact that there were few regionally based third‐party actions in the area reflects the sensitivity of the territorial conflicts. The involvement of neighboring countries would have given rise to fear of other forms of intrusion. Thus, small, distant, even NGO activities are seen as preferable. Such efforts, however, lack resources to reward the parties or even monitor the transitions from war to peace. Thus, the inability to develop regional third‐party efforts may be an additional explanation for why so many territorial conflicts in the region become particularly protracted.

Southeast Asia experiences more internal conflicts and fewer indigenous interventions than any other region of the world. This alone makes conflict in the region particularly important. What we find is that where a third party does intervene in a regional internal conflict, it is mostly an outside third party, and more often than not, an NGO, rather than the government of a state. Clearly, the characteristics of the conflicts in the region militate against success. Asymmetric, government‐versus‐rebels conflicts that tend to become protracted and complex are not ideal conflicts for intervention. The relative disparity in power between the disputants and the nature of the issues make many local states and organizations quite reluctant to toy with the idea of mediation. This is why, we believe, outside NGOs or international organizations are the most likely third‐party interveners.

Our findings indicate that actors in the region are all too well aware of the special features of internal conflicts in their region and are thus understandably not too happy about embracing a mediatory role. There is thus a pronounced division of labor internationally in the area of international mediation, where contiguous neighbors refuse to mediate, but distant organizations and NGOs take such mediation on. And they do so even though the conflicts they intervene in show no signs of being in a “hurting stalemate” (Zartman 2000). (By “hurting stalemate” we mean a situation where parties to a conflict experience increased costs that compel them to search for different, nonviolent, ways of dealing with their conflict.)

This article focuses on Southeast Asian internal warring dyads that have not yet escalated to full‐scale war. While Southeast Asia experiences the most frequent incidence of civil war, it receives the lowest level of attention from the international community in comparison to other regions. This article uses an original data set on third‐party involvement in Southeast Asian civil wars from 1993 to 2004. We note that there are many warring dyads in the region, and that their conflicts are costly, relatively intractable, and defy many attempts at termination. Like other civil wars, once these begin, they are difficult to terminate and even more difficult to resolve. The “sons‐of‐the‐soil” idea provides a good theoretical explanation for the temporal dynamics of most of these conflicts. Asymmetry and conflicts over contraband produce high numbers of fatalities and make the conflict costly and increasingly likely to recur. It is precisely such conflicts that we would expect international and regional organizations to pay the most attention to, but this is not the case.

Southeast Asian civil wars attract a disproportionately low number of third‐party management efforts. The type of issue in conflict (e.g., contraband, resources) and the nature of the parties in conflict (e.g., asymmetrical power resources) may well explain why so few regional states or organizations are keen to intervene. Such conflicts do not lend themselves to outside intervention, and that may be one major reason why we see them ending temporarily, only to become violent again.

The region has also experienced comparatively few multilateral peacekeeping missions, and that may well be another reason why so many civil wars reoccur. (See Doyle and Sambanis 2000 on the relationship between peacekeeping and longer peace.) As noted previously, in conflicts involving rebel groups it is hard to get a credible commitment to peace from a government or a rebel group. Many of the conflicts we examined are of this kind, and many outside parties are all too well aware of the difficulties of managing such conflicts.

We find that neither the United Nations nor the major powers are in a position to manage or terminate civil wars in Southeast Asia. Perhaps the conditions are just not those in which the parties feel they have suffered enough (what we may call a “hurting stalemate”), or perhaps they are just not conducive to third‐party intervention. When intervention does occur in this context, it is mostly by outside organizations or NGOs. These are actors that have the least to lose and most to gain from success. Their stakes in any particular, evolving, or enduring outcome is low. Their mode of intervention is not invasive but rather mostly facilitative, with an emphasis on encouraging negotiations and dialogue between the parties. This is a low‐cost, low‐stakes third‐party intervention, and this seems to be the form that typifies interventions in the region.

Third‐party interventions are neither as numerous nor as effective in this region as they should be, and this is precisely why they are needed the most. This study demonstrates that — in the absence of a multitrack diplomacy based on local, regional, and international resources, and a varied set of tools ranging from peacekeeping to postconflict reconstruction — civil wars in Southeast Asia will continue to pose a serious threat to regional and international security.