Non-inclusive ceasefires do not bring peace: findings from Myanmar

ABSTRACT Based on conflict data, interviews and media monitoring, this study of Myanmar’s non-inclusive ceasefires develops a four-step argument about the effect of ceasefires in complex conflict systems. First, non-state armed groups rarely co-ordinate their actions strategically. This makes it easy for governments to obtain ceasefires with some groups while fighting others. Second, when ceasefires ensure armed groups’ survival, they mostly hold. Third, non-inclusive ceasefires do not reduce a country’s overall level of violence, since fighting tends to escalate with excluded groups. On this basis we conclude that non-inclusive ceasefires do not present a viable alternative to an inclusive peace process.


Introduction
The term 'complex conflict system' denotes a situation in which two or more armed groups co-exist with a government's armed forces. In such systems, armed groups sometimes fight, sometimes not, and some may cooperate with the government against other groups. In 2019, the Uppsala Conflict Data Program (UCDP) listed the following countries as having more than one active internal state-based or non-state armed conflict: Afghanistan, Burkina Faso, Egypt, Ethiopia, India, Kenya, Mali, Mexico, Syria, Ukraine, Yemen and Myanmar. 1 Myanmar is the only country to have fulfilled the definition of a complex conflict system all the time since 1948, with new and old groups fighting or co-existing with the government. In most such systems (Somalia and Yemen are exceptions), the national army is powerful enough to dominate, although it does not have a monopoly of violence. In Myanmar, the army, navy and air forces have around 350,000 troops at their disposal, and in addition 85,000 police and multiple militias. 2 By early 2020, the country's 21 non-state armed groups were estimated to have a combined strength of just about 65,000 fighters. During that year, however, many groups utilized the security gained from ceasefire arrangements to recruit new troops. By the time of the military coup in February 2021, the estimated number of non-government troops had therefore risen to 85,000 (Appendix A). In 2015, eight groups signed a so-called Nationwide Ceasefire Agreement (NCA) with the government, and two more joined in 2018. Five others have bilateral ceasefire agreements. However, as of October 2020, four important groups stood without any such agreement: the Arakan Army (AA), Kachin Independence Army (KIA); Ta'ang National Liberation Army (TNLA); and Myanmar National Democratic Alliance Army (MNDAA). 3 For two years, from December 2018 to November 2020, the Tatmadaw focused on fighting the AA, while mostly avoiding clashes with the others. In November 2020, right after the national elections, the Tatmadaw and AA agreed on an informal, temporary ceasefire. After its 1 February 2021, coup, the Tatmadaw tried to uphold this comprehensive yet fragile system of ceasefires, so it could concentrate on repressing city-based protests and emerging armed resistance movements.
A ceasefire is conventionally defined as a unilateral, bilateral or multilateral verbal or written pledge to temporarily or permanently abstain from armed fighting. It may be separate or part of a broader agreement and is often seen as part of a peace process. A non-inclusive ceasefire is one that does not include all armed groups in a country. In a thoughtprovoking article, Marika Sosnowski proposes a broader definition of ceasefire as 'the codification of a certain military and political state of affairs during wartime'. 4 In our view, this definition is too broad, as it removes the main defining element: the pledge to refrain from using armed force. However, it is still fruitful because it opens avenues for creating ceasefire typologies not just based on duration but on context and content, such as power disparities and territorial control. Ceasefires sometimes codify relations between armed groups and governments that have not fought one another for a long time. Paul Staniland has therefore suggested that the concept 'armed conflict' be replaced with 'armed politics' or 'armed orders', covering active conflict, ceasefire and coexistence, all at the same time. 5 This article defines ceasefire as a unilateral, bilateral or multilateral pledge to refrain from using force, and to uphold a certain military and political state of affairs.
No comprehensive peace agreement has ever been reached in Myanmar, so we cannot find any positive evidence to support the finding in quantitative, comparative research that negotiation and implementation of comprehensive agreements is the surest way to peace in complex conflict systems. 6 Myanmar has, however, a rich experience with non-inclusive unilateral, bilateral and multilateral ceasefires, which allows us to measure the effect of ceasefires on a national conflict system.
Stimulating literatures have emerged on how to make peace in complex conflict systems, 7 and on ceasefires in internal conflicts. 8 Much research has been done on Myanmar's non-inclusive ceasefires from 1989 to 1995. 9 Research is also catching up on Myanmar's round of ceasefires in 2011-15. 10 The November 220 elections, which Daw Aung Suu Kyi's National League for Democracy (NLD) won in a landslide, led the military (Tadmadaw) to prepare for a coup. A new situation emerged that increased the Tatmadaw's need to uphold existing and establish new ceasefires while seeking to consolidate its new regime. We use this study of Myanmar's noninclusive ceasefires to address questions raised in the comparative literature and have three main findings to report: (1) Each group fends for itself, allowing the government to divideand-rule: When several armed groups struggle for the independence or autonomy of their homelands, survival is their overriding concern. They aim to maintain support from their constituents, recruit fighters, fund themselves, acquire arms, enjoy freedom of movement, and protect their bases. Each armed group pursues these goals for itself and rarely coordinates strategically with others, except if it depends on them for protection, funding, training, or weapons. This has made it easy for the government to obtain ceasefire agreements with some groups while fighting others.
(2) Most ceasefires hold: Once a ceasefire has been agreed, it is likely to hold if local leaders benefit from it and retain support from their constituents. The leaders of Myanmar's ceasefire groups have mostly been allowed to keep their arms, control some territory and profit from legal and illicit business opportunities. When armed groups benefit in this way, they are unlikely to break a ceasefire. (3) Non-inclusive ceasefires have not reduced the violence: Instead they have led to increased fighting between the government and nonceasefire groups, between ceasefire and non-ceasefire groups, and one-sided violence against civilians. Non-inclusive ceasefires have therefore not decreased the overall level of violence in the system.
On this basis, we argue that non-inclusive ceasefires do not represent a viable substitute for a strategy aimed at an inclusive peace agreement. These findings are based on research on the period 1989-2020. We may add, however, that developments in 2021 tend to confirm them. Since the February 1 military coup, there has been an upsurge of violence. Yet most of the established ceasefires have remained in force, in spite of calls from the new multi-ethnic National Unity Government (NUG) for a general uprising and armed resistance movement. Although some ethnic groups (Chin, Kachin, Kayin) have been well represented in NUG, most groups have continued to fend for themselves. Some have resumed fighting with the Tatmadaw, notably in Kachin, Chin, Kayah and Kayin states. In Shan state, the UWSA and NDAA have upheld their bilateral ceasefires, while the Shan State Army-South (RCSS) has fought against the Shan State Army-North (SSPP) and the TNLA instead of against the Tatmadaw. In Rakhine State, the AA has so far stood by the informal ceasefire it obtained with the Tatmadaw in November 2020. The new actors since the coup are hundreds of newly established localized People's Defence Forces (PDFs) among the ethnic Bamar majority who have taken up arms against the junta, and receive protection and training from some of the established ethnic armed groups.
We begin by presenting our methods. Then comes an overview of Myanmar's complex conflict system. Next, we describe Myanmar's two rounds of non-inclusive ceasefires (1989-95 and 2011-15). Thereafter, we go through each of our findings and conclude. Finally, we reflect on the relevance of our findings for other countries.

Method
In order to contribute to the literature on conflict endings, we use UCDP data and definitions, 11 with one modification: We combine the two categories 'state-based' and 'non-state' conflict in one definition of internal armed conflict as the use of armed force between a government and an armed group, or between non-state armed groups, resulting in at least 25 battlerelated deaths in a calendar year. 12 Myanmar's conflict system consists of multiple conflict dyads, which have sometimes been active, with a minimum of 25 battle deaths in one calendar year, and sometimes inactive, with few or zero fatalities (Appendix A).
To supplement the UCDP data, we use the Township-based Conflict Monitoring System (TCMS) of the Myanmar Institute for Peace and Security (MIPS), which records conflict events (clashes and incidents) as its main indicator of acceleration or deceleration. 13 Our project has sought to streamline the MIPS and UCDP definitions, so data from Myanmar can be used in international comparisons. While MIPS has not initially provided estimates for battle deaths or fatalities, which are the UCDP's main indicators of conflict intensity, MIPS has a richer differentiation between various kinds of clashes and incidents. 14 As of now, the TCMS does not include historical data from before 2016, while UCDP estimates go back to 1989.
Our study builds on both quantitative and qualitative data from 1989 to January 2021. It is informed by available Burmese and English-language literature, analysis of UCDP and TCMS data, observations from fieldwork, social media monitoring and semi-structured interviews undertaken over a period of three years with more than one hundred conflict actors and participants in Myanmar's peace process, 52 of whom were armed group leaders or Tatmadaw officers (Appendices B). Anonymized interview transcripts have been encrypted and stored offline. A master file with summaries is available on request. 15

Myanmar's complex conflict system
Myanmar has the world's most durable complex conflict system. In its ethnic borderlands, at least 21 armed groups are either engaged in fighting the government or enjoy a ceasefire. In contrast to countries like Syria and Iraq, all these armed groups have names, acronyms and known leaders, and their fighters often operate in uniform. This is part of an effort to act like states, with flags, parades, national days, courts, health services, schools, taxation systems, etc. Myanmar has experienced very few terrorist attacks. It is normally easy to establish who took part in a clash or stood behind an incident. Most groups have both a military and a political wing. However, this dynamic changed after the 1 February 2021 coup as more than 150 small localized militant groups emerged to resist the military junta.
Ethnic minority leaders have called repeatedly for a federal political settlement to end armed conflict, and many of them are inspired by the 1947 Panglong agreement. In February 1947, in the little town of Panglong in southern Shan State, General Aung San (the founder of Myanmar's armed forces and the father of Daw Aung San Suu Kyi), made an agreement with leaders of the Shan, Kachin and Chin ethnic groups to form a Union of Burma. One year later, when Burma achieved independence, Aung San had been assassinated, and the Panglong agreement lost force. A vision cherished by ethnic autonomy and democracy activists is to let Panglong inspire the creation of a democratic federation, recognizing the ethnic groups' right to self-government.
Since Panglong, Burma/Myanmar (the name change occurred in 1988) has had three constitutions, adopted in 1947, 1974 and 2008, 16 but not a single inter-ethnic agreement was negotiated between 1947 and 2015, when the Nationwide Ceasefire Agreement (NCA) was signed by leaders of eight armed groups (see Appendix A) and three representatives of the Union government (President Thein Sein, the Speaker of Parliament Shwe Mann, and the Commander-in-Chief of the armed forces, Senior General Min Aung Hlaing). In 2016, after Aung San Suu Kyi had assumed power as State Counsellor, she committed her government to the NCA and would only accept armed groups as full participants in her Union Peace Conferences (UPCs) if they first signed the NCA. Her peace process was called '21 st Century Panglong', with three UPCs being held during 2016-18 and a fourth in August 2020. 17 These served as platforms for presenting demands. 18 Multiple agreements were reached on uncontroversial points, and there were talks about federal principles. Yet, little progress was made on substantial issues.
Each of Myanmar's fourteen regions and states has its own government and elected assembly (hluttaw), but they have few resources and little power. Their chief ministers are appointed by the president. Ethnic parties demand that they be locally elected. Yet, if Myanmar were to devolve substantial powers to its regions and states, they risk being engulfed in ethno-nationalist agitation, with more instead of less violence (cf. Ethiopia). As in other countries with a complex conflict system, policies pursued since colonial times have ethnicized Myanmar's politics. Every national ID card indicates religion and ethnicity. 135 groups are recognized by law as 'national races' (taingyintha). 19 Myanmar's seven regions are dominated by the Bamar majority, while its seven states are named after ethnic groups: Mon; Kayin (Karen); Kayah (Karenni); Shan; Kachin; Chin; and Rakhine (Arakan). In addition, there are five recognized 'autonomous zones' (Danu, Pa'O, Palaung, Kokang, Naga), one 'autonomous region' (Wa) and four 'Special Regions' along the border to China. The 2-3 million Rohingya Muslims are not recognized as a 'national race' partly because this would give them a right to an autonomous zone. The ethnic divisions define the country's conflict system.
Since World War II, Myanmar has been ridden by conflict. Some armed groups have disappeared, while others have emerged. Most of the time, the conflicts have been inactive (with little or no fighting). In many cases the government has refrained from challenging the armed groups' control of certain areas, while punishing them if they moved outside of those areas. Many groups have kept their arms and territories by agreeing to transform themselves into formally recognized militias or border guard forces under the command of the government army, while others have remained independent. The Tatmadaw has never had a monopoly of violence in the highlands but has steadily increased its presence there with new garrisons connected by roads and helipads.
There have been ups and downs in the level of conflict. In 1948, the Tatmadaw was threatened by communists in the west and ethnic Kayin in the east, who advanced against the capital Yangon and came close to taking it. In the 1950s, the Chinese Civil War spilled over to Myanmar when nationalist Chinese troops took refuge in Shan and Kachin. In 1961, when the Chinese nationalists withdrew to Thailand, the drug lord Khun Sa's Mong Tai Army (MTA) built a realm of its own in the east. During the long dictatorship of General Ne Win, who seized power in 1962, the Tatmadaw faced rebel armies of many kinds. The communists concentrated on the Chinese borderlands, where they recruited fighters among the local minorities. New ethnic armies were formed by the Kachin in the north and Rakhine in the west, and several smaller groups have since founded armies of their own. When the communist rebellion ended in a 1989 mutiny, its ethnic fighters formed six new ethnic armed organizations, with strong links to China and organizational structures built on the communist model.
What are the armed groups' goals? Since 1989, when the communist rebellion collapsed after more than forty years of struggle, all but one of the country's armed groups have been ethnically defined. The exception is the All Burma Student Democratic Federation (ABSDF), which was formed by students fleeing from an army massacre in Yangon on 8 August 1988. From the beginning, the ABSDF depended on assistance from the Karen National Union (KNU) and, for some time, the Kachin Independence Organization/ Army (KIO/KIA). The ABSDF was always weak and never gained any hegemony over its ethnic hosts. From 1992 onward, the KNU led a coalition (National Council for the Union of Burma) aiming at regime change by means of combining a mass uprising with armed struggle. Its goal was a democratic federation. Generally, however, the ethnic armed groups have not aimed to change the central government. They fight for their homeland and want to endow it with state-like institutions, revenue streams and armed forces. They try their best to behave like states. 20 The groups most successful at achieving these goals are two communist successor groups: the United Wa State Army (UWSA) and the National Democratic Alliance Army in Mongla (NDAA). For the last thirty years, they have run their territories with little interference from the Union government. When the military government offered them ceasefires in 1989, its intention was to carry out local development projects. However, after some roads had been paved, and Buddhist temples built on hilltops, the state stepped aside and allowed four so-called Special Regions to administer themselves. The UWSA and NDAA, although de facto independent, do not seek formal independence. The goal of the Wa is a separate ethnic state within Myanmar, connecting two territories under its control (today they are parts of Shan State). 21 This meets resistance not only from the Myanmar government but also from a successor group to Khun Sa's MTA, the Restoration Council of Shan State/Shan State Army-South (RCSS), which insists on Shan unity, and the NDAA in Mongla, whose land is between the two Wa territories. Mongla, a multi-ethnic township with many Chinese inhabitants, has tried to ethnicize itself by claiming to be the homeland of its largest group, Akha, as this may give it the right to a status as an autonomous zone. 22 From 2011, a goal for many armed groups was to be recognized as partners in peace talks. During 2016-21, however, it was difficult to engage the government in genuine talks, as it had two rival heads: the State Counsellor, Daw Aung San Suu Kyi, 23 managed the economy, foreign policy, public health, education and other civilian affairs, while the Commander-in-Chief, Senior General Min Aung Hlaing, decided on matters of national security in the widest meaning of the term. 24 Formally, both operated under the authority of the president, who wielded little de facto power. Under the 2008 constitution, military-civilian coordination is meant to be ensured by the National Defense and Security Council (NDSC). While Aung San Suu Kyi was State Councellor, however, the president never convened the NDSC, where the military held a majority. 25 There was thus little coordination between the top-level civilian and military decision-makers.

Government-driven ceasefires
Myanmar's ceasefires came about in two rounds: in 1989-95 under Senior General Than Shwe's military junta, and in 2011-15 under the constitutional government of General Thein Sein, who was picked by Than Shwe to become President in 2011. In 1989, when the Burmese communist leaders fled to China after their Wa fighters had mutinied, Myanmar's military regime was in trouble after its 1988 crackdown on a nationwide mass uprising, centered in Yangon. General Ne Win had withdrawn from power after realizing that his autarkic economic policy had failed. A new military State Law and Order Restoration Committee (SLORC) opened the country to foreign trade and investments and organized elections to a National Constitutive Assembly. Yet, SLORC refused to convene it when the National League for Democracy (NLD), led by the 45-yearold Aung San Suu Kyi, won a majority of seats. Amidst repression of the democratic opposition and under pressure from the international community, SLORC turned to ceasefire to stabilize Myanmar's borderlands.
During 1989-95 SLORC reached informal ceasefires with six sizable groups and 35 smaller ones, most of which agreed to either disband or transform into government-controlled militias. The architect of the ceasefires was General Khin Nyunt (1939-) and his Military Intelligence (MI), which he led from 1983 to 2003, whereafter he briefly served as prime minister. 26 In 1989, he engaged informally with various local leaders. Peng Jiasheng, who had controlled Kokang for two decades as a member of the Communist Party of Burma, played a key role. He became the first of the former communist leaders to agree to a ceasefire and established what would become the Myanmar National Democratic Alliance Army (MNDAA). He also helped Khin Nyunt obtain agreements with the Wa, Mongla and the Shan State Army-North, as well as a small Kachin communist group, which later became a governmentcontrolled militia. By 1995, Khin Nyunt had created a system of ceasefires including all the main groups in Shan State, except the important MTA of Khun Sa. In 1994-95, the Tatmadaw also obtained ceasefires with three noncommunist groups: the Karenni National Progress Party (KNPP), the New Mon State Party (NMSP) and the Kachin Independence Organization (KIO).
Khin Nyunt's ceasefire deals had six characteristics: no international third parties were allowed into the negotiations; no discussion of any political settlement was permitted; the groups could keep their arms; group territories were demarcated; group leaders were afterwards tacitly allowed to engage in legal as well as illicit businesses; and, with one exception, all deals were solely oral. The only one that was written was with the KIO. Its ceasefire was signed in 1994 in the presence of a select group of witnesses, but the terms were kept secret; even members of the KIO Central Committee were not allowed to see them. 27 Shortly after signing, the KIO kicked the ABSDF out of Kachin State. 28 The ceasefires provided ethnic leaders with economic rewards in terms of business opportunities connected to the opening of trade with China. One scholar would brand this as 'ceasefire capitalism'. 29 Its four cornerstones would be drugs, timber, jade and casinos.
On the government side, the ceasefires were handled uniquely by the Military Intelligence. No regular military units were involved. This led to difficulties, and in 2004, Than Shwe purged Khin Nyunt and dissolved the Military Intelligence. Responsibility for maintaining the ceasefires was transferred to the chief of Military Security Affairs (MAS), whose staff were infantry officers with little understanding of ethnic sensitivities. Personal relations soured, and some ceasefire groups were not pleased when the junta put pressure on them to transform into government-affiliated militia or border guard forces. As of 2011, when General Thein Sein was elected president by a new National Assembly, most of Khin Nyunt's ceasefire system remained intact, although he himself was in jail. He was released in 2013.
Soon after assuming power, Thein Sein's government initiated a fresh round of talks, this time with the aim of reaching signed ceasefire agreements. He assigned two teams to contact the armed groups. One was led by railway minister Aung Min, who later became a minister of the president's office. He became the point man for the peace process and enlisted young intellectuals to help him, mostly from Egress, a non-governmental capacitybuilding think-tank, reinforced by educated people returning from exile. They eagerly embraced this chance to play a role under Aung Min's wings and established a Myanmar Peace Centre in Yangon. In 2011-13, their efforts led to bilateral ceasefires with 14 groups, many in continuation of Khin Nyunt's old deals.
Aung Min's peace process included the groups that had refused to become militias, and the two most powerful southern groups: the KNU and RCSS. The government also held talks with the KIO, whose ceasefire broke down on 9 June 2011, and obtained an agreement with it to 'reduce hostility' on 30 May 2013. Formally, this was not a fully-fledged 'ceasefire' and it merely provided a temporary break in the fighting. (Yet we have included it in Appendix A.) Hostilities continued on and off until July 2018, when both sides ceased to attack each other in Kachin State, although they did not reach any agreement. After the military coup of 1 February 2021, fighting in Kachin State resumed.
Thein Sein's government refused to acknowledge three smaller groups: Peng Jiasheng's MNDAA, whose 1989 ceasefire had broken down in 2009; and two new groups, which had never had a ceasefire: The Ta'ang National Liberation Army, Palaung (TNLA) and the Arakan Army (AA). 'We should not resurrect dead tigers', one senior Tatmadaw officer remarked. 30 Although the government allowed these groups to be represented in a Nationwide Ceasefire Negotiation Team (NCCT), it would not let them sign the NCA unless they first disarmed.
There are some essential differences between Myanmar's two rounds of ceasefires. Khin Nyunt's 1989-95 'elite pacts' divided territories and resources for a long period of time, so that Than Shwe's military junta could consolidate its power and manage a transition to constitutional government. The armed groups received territory and economic advantages but no political talks. 31 By contrast, the 2011-15 ceasefires were driven by a government hoping to terminate Myanmar's armed conflicts altogether. Thus, once the new bilateral ceasefires had been signed, the government engaged in a broader dialogue with the aim of negotiating a nationwide ceasefire as a first step to a comprehensive peace. 32

Finding 1: each group fends for itself, to the government's shortsighted advantage
In our pre-2021 interviews with armed group leaders, we were struck by how little they knew about each other. Our interview form included a set of questions that revealed the level of knowledge about other groups than one's own (Appendices B, questions C 1-4, D 2, F 1-2). It was surprisingly low. For example, the ABSDF leaders were better informed than others about the KNU and KIA because they had operated under their wings but knew little about the UWSA or NDAA. On reflection we found the low level of knowledge understandable. If one looks at Myanmar's armed groups from the outside or from a Yangon or Naypyidaw perspective, one may assume that they constitute a movement with a shared vision and a national urge to cooperate and form alliances. However, if you lead a local group, you must concern yourself primarily with financing, recruitment, command and training, organization, maintaining local support and guarding against threats to your existence. Groups far away in other parts of the country have little impact on your basic interests, unless you have business deals with them or rely on them for weapons and training. For armed group leaders it is therefore rational to concentrate on their local concerns rather than aligning themselves with other groups' behavior. Hence, they rarely base their strategic decisions on what other groups have done. 33  Seden Akcinaroglu adopts the birds-eye view and sees alliance as an optimal strategy for weak rebels. 35 From time to time, some of Myanmar's most politically minded ethnic leaders have thought likewise and have tried to coordinate the military and diplomatic strategies of the various groups. Yet, their alliances have always been ineffective. As the secretary of one alliance exclaimed in 2013: 'It's always the same: whenever the government talks peace, we begin to separate.' 36 Myanmar's armed groups have mostly remained passive when fellow groups have come under attack, thus allowing the Tatmadaw to concentrate offensives against other groups. Virtually all groups have been willing to talk with the government and have responded independently to offers. This has allowed the government to decide when to negotiate ceasefires with whom and avoid inclusive solutions.
A list of failed alliances from before 1991 may be found in Martin Smith's account from 1999. 37 In the 1990s, the most important ones were the National Democratic Front (NDF), formed in 1976 by twelve non-communist groups, and the Democratic Alliance of Burma (DAB), established in November 1988 to coordinate the struggle of the NDF and the ABSDF. In 1994, amidst the DAB's attempt to coordinate military campaigns in support of the KNU's struggle for survival, the KIO broke ranks and made its ceasefire deal with the junta. This killed the potential of the DAB.
In 2011, when the Kachin ceasefire was about to collapse, the KIO took the initiative to form a United Nationalities Federal Council (UNFC) with 14 members. They planned to build a Federal Union Army and coordinate their military pressure on the government. However, the UNFC never launched a coordinated offensive, and was unable to establish a combined leadership with authority over its members. At first, the government ignored the UNFC while negotiating bilateral agreements with its members. This weakened the UNFC, and in 2013, when the government agreed to collective talks for the purpose of pursuing a political dialogue, little came out of it. All important talks remained bilateral. The main force behind the UNFC was the KIO, which in April 2014 came under renewed attacks by the Tatmadaw, 38 but in 2015, the KIO's decision not to sign the NCA did not prevent other UNFC members from signing.
A study by David Brenner, built on fieldwork in areas controlled by the KNU and KIO, does not cite any evidence that the two groups learned from each other. Brenner confirms that when either group sent someone to Chiangmai in Thailand to take responsibility for UNFC coordination, that person lost access to genuine decision-making. 39 Brenner addresses the paradox that the Christian-led KIO welcomed a resumption of armed struggle in 2011 after a 17-year-long ceasefire, while shortly afterwards, the Christian-led KNU signed its first ceasefire, after more than 60 years of armed struggle. He sees no causal connection between the two opposite outcomes but finds that they were dictated by internal dynamics within each group. The KIO leaders who had benefitted from the 'capitalist ceasefire' 40 had lost support from the Kachin grassroots, and a new generation demanded action. The KNU leaders who opted for ceasefire, in spite of some grassroots opposition, saw accommodation with the government as a chance to reinsert themselves as leaders inside Myanmar at a time when many Kayin refugees returned from Thailand to their home country and sources of external support dried up. 41 The ceasefire became a means for them to withstand the challenge from the KNU's activist Brigade 5, whose commanders had remained inside Myanmar.
Under the military junta, the armed groups had often accused the government of pursuing a divide-and-rule strategy. So, when Thein Sein came to power, they demanded collective negotiations aimed at a nationwide ceasefire. In October 2013, the government's chief negotiator Aung Min allowed 16 armed groups to form the Nationwide Ceasefire Coordination Team (NCCT), which spent 17 months negotiating a draft NCA. The UWSA, NDAA and SSPP, however, did not take part in the negotiations. Neither did the RCSS, although it sent an observer to the talks. A draft was agreed on 31 March 2015. However, before signing it, the armed groups held a summit at the KNU headquarters in Lawkhila, which refused to accept the draft. A new negotiation team was formed, and in August 2015, it agreed on a revised text with the government. The question was then if all groups would be invited to sign. The KIO insisted on an all-inclusive approach, but the government refused to let the MNDAA, AA and TNLA sign. The KIO then decided not to sign either, while the KNU and RCSS signed along with six smaller groups (see Appendix A).
The KIO claimed that the reason it did not sign was the government's refusal to take an inclusive approach. Two other reasons may have been just as important. First, the KIO anticipated a victory for Aung San Suu Kyi's NLD in the 8 November 2015, elections, and most likely expected this to help it get a better deal. 42 Second, according to respondents in the Kachin capital Myitkyina, the main reason why the KIO did not sign was its internal situation. There was widespread dissatisfaction with the 1994-2011 ceasefire, which had brought corruption and plunder of natural resources, and there was a power struggle in the KIO between an older and a younger generation. Under these circumstances, and due to a decentralized, anti-authoritarian tradition within the Baptist community, the KIO needed to consult its grassroots before signing. The government, however, wanted a signing ceremony before the elections, and did not give the KIO time to carry out its consultation. 43 The story of how the KIA's bilateral ceasefire broke down in 2011 after 17 years and failed to be renewed through the NCA shows what may happen when a ceasefire provides economic benefits mainly to a small elite without making room for reforms to benefit broader segments of the local society. The ceasefire then becomes a cause for local resentment. From 2012, something similar played out among the Kayin, many of whom resented the KNU top brass' collaboration with the government in a peace process that did not seem to bring much result. 44 What about the three groups that were not allowed to sign? In February 2015, the MNDAA had initiated a failed offensive to retake the Kokang capital Laukkai, with help from the AA and TNLA. They sought to demonstrate that they could not be ignored. Instead, the offensive convinced the government that the three groups were unreliable. This was decisive for the non-inclusive character of the NCA.
In November 2016, a new but temporary 'Northern Alliance' of the KIA, MNDAA, TNLA and AA launched a joint offensive to occupy Mongko, a strategic border town. The Tatmadaw responded with much firepower and high mobility and conducted punitive attacks on KIA and TNLA bases. Consequently, the number of armed clashes escalated in 2017 and the first half of 2018. Then suddenly, in July 2018, the KIA and Tatmadaw stopped fighting in Kachin State. The KIA now left the burden of struggle to the other members of the Northern Alliance, which thus became militarily defunct, although it continued to have talks with the government. The AA, MNDAA and TNLA proclaimed a new Brotherhood Alliance, which in August 2019 carried out a joint raid against government positions in northern Shan State. Apart from this raid, however, the MNDAA and TNLA did little to relieve the Tatmadaw's pressure on the AA (Figure 1), which survived a life-and-death struggle in the north-central parts of Rakhine State from December 2018 to November 2020, when an unofficial ceasefire was agreed. The short history of the Northern and Brotherhood Alliances confirms how each group fends for itself. At the time when the KIA still hosted and trained the AA leadership (May 2016-October 2017), the KIA was fighting government forces. However, when the AA had built enough strength to confront the Tatmadaw in Rakhine State, the KIA respected the Tatmadaw's request that it stay within its designated zones in Kachin State. During 2018-20, this allowed the Tatmadaw to focus on its fight against the AA in Rakhine. In 2021, the situation was reversed. Now the AA benefitted from a truce while the KIA was once again engaged in heavy fighting.
In 2017, seven northern groups (UWSA, SSPP, NDAA, KIA, AA, TNLA, MNDAA) formed a diplomatic alliance, the Federal Political Negotiation and Consultative Committee (FPNCC). With backing from China, it sought talks with the Myanmar government. The government refused to recognize the FPNCC but was willing to talk to its members individually.
Myanmar's government has always been able to negotiate peace with some while fighting others. Governments facing several armed challengers can combine strategies of divide-and-conquer with divide-and-concede. 45 Myanmar's military has done just that. While fighting some groups, the government has let others keep their arms and territory, tax their constituents, build state-like structures and profit from legal and illegal businesses. The army itself has profited even more. In 2021, when an almost nation-wide protest movement arose against the military coup, a new attempt was made to form a broad ethno-political alliance with a National Unity Government and a new federal army. For the first time since the early 1990s, the KIO and KNU were fighting the Tatmadaw at the same time. It remains to be seen if, this time, the ethnic armed groups are able to break their divisive pattern and co-ordinate strategically. Until now, even without a monopoly of violence, the Tatmadaw has been able to dominate a system that has never seen more than partial peace.
Not only have Myanmar's armed groups failed to act in unison, they also have not based their decision-making on learning from other groups' experiences. Among the 52 armed group leaders and Tatmadaw officers we interviewed from mid-2017 to late 2019, no one said that their ceasefire decisions were influenced by other groups. When asked about strategic decision making, one veteran leader admitted: 'We hardly considered the other groups' relationships with the government. What matters most is the interest of our own group.' 46 Another ethnic leader stated: 'We decide according to our own historical background'. 47 One former KIO leader said their decision to pursue ceasefire with the junta in 1994 was driven by 'our desire to bring peace and stability . . . our decision was not influenced by other groups' ceasefire or continued fighting.' 48 Exceptions occur when a group depends on another group or when two groups have previously belonged to the same group. The Chin National Front (CNF) has compensated for its military weakness by playing bridge-builder, and this has enhanced its awareness of what other groups think and do. 49 Members of the ABSDF have good knowledge of the KNU. The KNU, DKBA and KNPP know and watch each other, since the DKBA is a splinter group from the KNU, and the KNPP is nearby. However, our respondents from these groups have little knowledge about the northern groups. 50 The TNLA and AA know the KIO, which helped them come into being. Since 2018, however, they depend more on the MNDAA and UWSA.
As Myanmar's strongest non-state armed group, the UWSA has influenced its neighbors NDAA and MNDAA. 51 It has also learnt from observing the Tatmadaw's treatment of other groups. Notably, when the Tatmadaw conquered Kokang in 2009 (see below), the UWSA feared similar attacks on itself, and began to provide arms to the KIA, TNLA and AA. This, however, was a selfprotecting strategy, not an example of inter-group solidarity or learning from other groups.
On 28 September 2016, the UWSA used force to impose its will on the NDAA by seizing five of its military outposts after the NDAA leaders had been too eager to participate in the government's peace process. 52 How the UWSA wields its influence over other groups is difficult to gauge, but it is clear that the smaller groups try to protect their freedom. Each armed group strives to behave like a sovereign state in the international system. The anarchic character of Myanmar's system makes it hard for the armed groups to coordinate. While this helps the government to divide-and-rule, it makes it impossible to defeat the Tatmadaw or negotiate an inclusive peace. If the government should want peace, it would have to negotiate in multiple directions on terms reflecting local power constellations.

Finding 2: most ceasefires hold
Perhaps surprisingly, since ceasefires in many countries tend to be temporary, in Myanmar, once a ceasefire has been agreed between the government and one or several armed groups, it has mostly been respected by both sides. As can be seen in Figure 2, between 1989 and 2008, 39 of 40 ceasefires held. The exception was an agreement between the government and the KNPP, which broke down in 1996 because of a quarrel over valuable logs (Appendix A). The next breakdown happened in Kokang 2009, when the Tatmadaw fell out with MNDAA leader Peng Jiasheng after twenty years of effective ceasefire. Figure 2 shows that only three out of 40 bilateral ceasefires from the Khin Nyunt period broke down, while all 18 agreed under President Thein Sein held until the 2021 military coup. 53 However, no less than 33 ceasefire groups bowed to the junta's pressure before 2010 and allowed themselves to either be disarmed or transformed into government-controlled militias or border guard forces. Under Thein Sein's presidency, the five ceasefire groups that remained from the previous period signed new formal ceasefires, along with 13 new groups. As already mentioned, four significant groups remained outside the ceasefire system: AA, KIA, MNDAA, and TNLA. Why were the ceasefires so resilient? A key reason may be that they were negotiated and agreed on locally by the parties themselves. They were not the results of external pressure, and there was no external mediation (except by local religious leaders and businessmen). We are not aware of any global statistical comparison of self-driven and externally driven ceasefires, but Karakus and Svensson find that in Syria, informal arrangements negotiated by parties internal to the conflict have more often been respected than those mediated by externals. 54 None of Myanmar's ceasefires have been mediated by externals.
A reason mentioned by several of our interview respondents is personal trust. Khin Nyunt and his staff made friends with ethnic leaders, and they were bound by a sense of personal obligation. However, if this were the key explanation, the agreements should have broken down in 2004, when Khin Nyunt and his Military Intelligence were purged. Yet, five more years passed before the Kokang ceasefire broke down and two more years before it happened in Kachin state. A more likely reason why ethnic leaders respected the ceasefire agreements is the stakes they received in exchange for keeping the peace: control of territory; opportunity to collect taxes; access to lucrative businesses; and permission to keep armed troops. 55 The government forces also had good reasons to uphold the ceasefires. Military companies and senior military officers benefitted economically from exploiting natural resources in the ceasefire areas and the army could concentrate on fighting the non-ceasefire groups. Meanwhile, the military junta got the stability it needed to carry out its plan for transition to a 'discipline-flourishing democracy'. Moreover, because of international sanctions, it did not have sufficient resources to carry out infrastructural or other modernization efforts in the ceasefire areas. Hence, it was convenient to leave those areas in control of local armies. Ultimately, however, the long-term intention of the military government was to persuade all non-state armed groups, once the transition to constitutional rule had been achieved, to either dissolve or be transformed into government-controlled militias.
The flaw in Khin Nyunt's system was that it brought corruption, exploitation of resources, abuse of power, and sometimes local struggles between government-controlled militias and non-state armed groups. There was no transparency and no political reform. This undermined the reputation of the ceasefires among a new generation of ethnic youth, seeing themselves as victims of a shameful system often referred to as 'neither-war-nor-peace'. 56 After the second round of ceasefires in 2011-12, some groups again obtained tangible advantages. The RCSS could dispatch troops to northern Shan, and some of the smaller groups found it easier to hire more troops when they no longer risked combat. Wearing a uniform allowed such troops to take actions that would otherwise give them problems with the police. Yet, the Thein Sein government's 2011-15 ceasefires did not provide economic advantages on the scale of Khin Nyunt's. As one of our respondents explains: 'In my opinion, there was no offer from the (Thein Sein) government. In the past, incentives for doing business were offered in return for ceasefire as the government did not want to solve the conflicts politically. Now signatory groups get a right to meet, but they are not given privileges to do business'. 57 However, at least until the 2021 military coup, a combination of safety from attack by government forces, permission to keep arms and uniforms, and access to Union Peace Conferences seem to have been enough for the ceasefire groups to uphold the agreements they signed. A certain number of clashes occurred because of lack of clear boundaries between territory controlled by the government and areas controlled by non-state forces, but these did not escalate into general fighting.
Apart from allowing IDPs to return to their villages, however, the ceasefire system did not bring much benefit to the population at large. As was the case under the junta, not much was done in Aung San Suu Kyi's time as State Counsellor to consult the population or encourage former fighters to engage in nonviolent politics. A groundswell of resentment against the Tatmadaw remained in ethnic areas, and non-ceasefire groups gained strength. This explains the apparent paradox of our third finding that although ceasefires have generally held, they have not brought more peace to the system as a whole.

Finding 3: non-inclusive ceasefires have not reduced the violence
If most ceasefires hold, a government should be able to gradually pacify a country by adding more ceasefires. Evidence from Myanmar does not support this expectation. Its non-inclusive ceasefires have increased rather than decreased the overall level of violence because they have been followed by more fighting with excluded groups, as well as one-sided violence against civilians and fighting between non-state groups.
The first round of ceasefires in 1989-95 pacified the Chinese border, while boosting conflict along the border to Thailand. As a centralized institution with a national outlook, the Tatmadaw could move troops from one front to another and deprive the MTA and KNU of their support base with its infamous four-cuts strategy, which cut off funding, food, intelligence and recruits by forcing the civilian population to move into government-controlled areas -or flee into the jungle or to Thailand. The UWSA also fought the MTA.
Through 1996-2008, the northern ceasefires survived while the KNU rebuilt itself. After Khun Sa had surrendered in 1995, remaining elements of the MTA set up the Restoration Council for Shan State (RCSS). The Tatmadaw used the occasion to establish garrisons in many new places. The strengthening of the Tatmadaw's territorial control helped its transition to semidemocratic governance, with a constitutional referendum in 2008, national elections in 2010 and the establishment of General Thein Sein's government in 2011 but did not lead to ethnic peace. Instead, there were new bursts of fighting and one-sided violence, this time mainly in the north and west.
Several causal mechanisms were at work. In the first round of ceasefires, it had been divide-and-rule. The government made economic and territorial concessions to some while subduing others. In the second round, since ceasefire had now become a ticket to participation in government-led peace talks, small, excluded groups were eager to prove themselves in battle in order to obtain a ceasefire. They thus launched raids called 'offensives'. The well-established excluded groups (UWSA, KIA) had territories to protect. They sought to deflect pressure from themselves by training and providing weapons to smaller groups. On its side, one of the ceasefire groups (RCSS) took advantage of having signed the NCA to move into new territories, which brought it into conflict with two non-ceasefire groups (TNLA and SSPP). And finally, in response to the total exclusion of the Muslim Rohingya from Myanmar's political system, a new armed group was formed: the Arakan Rohingya Salvation Army (ARSA). When it launched poorly armed raids in 2016 and 2017, the Tatmadaw responded with massive ethnic cleansing of Rohingya villages, forcing more than 700,000 to flee to Bangladesh. ARSA remains excluded from Myanmar's conflict system, not just by the government but by the other armed groups as well. It must have come as a surprise for the Tatmadaw that its ethnic cleansing did not generate more support from the Buddhist Rakhine population. In north and central Rakhine State, the Rakhine Buddhists instead rallied to the cause of the AA, which managed to build popular support for an amazingly effective insurgency. 58 With hindsight, we can see that two choices made by the military government during 2008-10 had fateful consequences. The first was to take the Kokang capital Laukkai by force in 2009, due to a suspicion that a local weapons factory provided weapons to armed groups in northeast India. In reaction to the Tatmadaw's attack, the MNDAA split in two factions. One decided to work with the Tatmadaw, while the other, led by Peng Jiasheng, fled to China. Since then, he has lobbied Chinese authorities for a tough approach to Myanmar. The MNDAA built a new army in China and tried in 2015 to retake Laukkai. In 2016, it formed the short-lived Northern Alliance with the KIA, AA and TNLA. During 2018-20, the AA and TNLA were the Tatmadaw's main adversaries (Figures 1 and 3).
The other fateful choice was to ignore the KIO's autonomy proposals in the consultations preceding the adoption of the 2008 constitution and to prevent a KIO-supported political party from fielding candidates in the 2010 elections. Although this did not immediately provoke a resumption of hostilities in Kachin State, it alienated the Kachin population. The KIA moreover feared to suffer the same fate as the MNDAA and face a demand to become a militia. In 2011, a small incident led to the breakdown of the Kachin ceasefire. By then, the KIA was well under way with training the TNLA and AA.
These two choices led not only to much fighting in Kokang and Kachin but also to the rise of the two armies that fought the government most ardently during 2018-20: the AA and TNLA.
Myanmar's NCA from October 2015 has been the main attempt so far to stake out a comprehensive peace process, based on principles derived from the armed groups' interest in ethnic self-determination, the NLD's democratic aspirations, and the Tatmadaw's quest for national unity. 59 The NCA is a partial, non-inclusive peace process agreement, since it includes less than half of the country's armed groups, only two of its seven strongest ones -the KNU and RCSS -and only some 20% of the non-state armed groups' combined troop strength (Appendix A). 60 The main groups that remain outside the NCA are the UWSA (which has a bilateral ceasefire agreement with the government), the AA (which agreed to a temporary ceasefire in November 2020) and the non-ceasefire groups MNDAA, KIA and TNLA.
After the NCA was signed, there was an increase in violence between the Tatmadaw and non-signatory groups. In 2016-17, the Tatmadaw conducted massive one-sided violence against Rohingya civilians, and engaged in fighting all four members of the ephemeral Northern Alliance. The battle-death estimates for these conflicts are uncertain, 61 but if we supplement them with MIPS' count of clashes, we get an impression of the escalation of violent conflict in Rakhine and northern Shan (Figures 1 and 3). There have also been multiple clashes between non-state groups. They have not reached the level of 25 battle deaths in a year but must be included in our analysis since they are related to the signing/non-signing of the NCA: signatories have fought against non-signatories. In 2021, all ten NCA signatories condemned the coup and some of them resumed their armed struggle.

Conclusion and comparisons
We have shown (1) that armed groups in Myanmar's complex conflict system have been unable to coordinate their strategies and rarely study or emulate one another's strategic decisions unless these groups established a dependent relationship. This weakens a hypothesis put forward by Quinn et al., with inspiration from the work of Mark J. C. Crescenzi, that peace may spread through a conflict system through reputational learning from one armed group to another. 62 When one group makes peace, they claim, others follow suit. We find little support for this in Myanmar.
We have further demonstrated (2) that Myanmar's ceasefires have generally held. We ascribe this, as far as the 1989-95 ceasefires were concerned, to the fact that armed groups could keep their arms and control certain territories for a long period of time and were allowed to engage in lucrative businesses. As for the fact that the 2015 NCA was also mostly respected by its signatories until 2021, we ascribe this to the function that ceasefire had as a ticket to the formal peace process. Yet, (3) Myanmar's ceasefires have not reduced the overall level of violence. Rather, violence increased after each round of ceasefires, as their non-inclusive nature stimulated fighting between the Tatmadaw and non-signatory groups, armed conflict between non-state armed groups, and one-sided violence against civilians. We have shown how this has played out statistically ( Figure 4) and have pointed out the escalatory mechanisms. One is divide-and-rule. Another is the urge felt by weak excluded groups to prove themselves in battle. And a third is the propensity for strong groups to divert government pressure from themselves by training and weaponizing groups that operate elsewhere.
On the policy level, these findings indicate that a peace strategy based on non-inclusive ceasefires holds little promise. While we cannot refute Desirée Nilsson's contention that a non-inclusive peace 'is possible', we find it unlikely in Myanmar. 63 We thus disagree with Lwin Cho Latt et al., who argue that the principle of 'all-inclusiveness' became an obstacle to Myanmar's peace process. We agree though, that peace building reforms do not need to wait until an inclusive agreement has been reached. 64 Importantly, our findings strengthen Quinn et al's point that 'negotiating and implementing a comprehensive peace agreement is an effective systemic-level conflict reduction strategy'. 65 We cannot provide positive support for this finding since no such agreement has been reached or implemented in Myanmar, but can give negative support by showing that non-inclusive agreements do not seem to constitute a viable alternative.
Is it possible to reach an inclusive peace agreement in Myanmar? Yes, but it requires a sustained effort by a government driven by a quest for peace. Since the armed groups have overlapping territorial claims, have several conflicting goals and have not so far been able to coordinate their military and political strategies, the onus of peacemaking must be on a representative union government. The International Crisis Group was right in 2020 to recommend that the government at that time should engage in political dialogue and negotiations with all the country's ethnic groups, 66 with the aim to establish participatory Legend: Figure 4 shows how two rounds of ceasefires (1989-95 and 2011-15) were initially accompanied by a spike and then by much vacillation in the number of battle-related deaths and victims of one-sided violence. When figures from 2021 become available they will show a new spike in one-sided violence after the 1 February 2021 military coup. Source: Uppsala Conflict Data Program (UCDP) datasets v. 20.1 (best estimates), and preliminary 'candidate figures' for 2020. The (low) 2020 figures will not necessarily conform fully to the final release of the data for 2020.
institutions where each group could work for its goals. With its 1 February 2021, coup, the Tatmadaw has put the non-inclusive ceasefire system to a new test. In order to concentrate on repressing popular resistance in towns and densely populated areas, it needs ceasefire with the armed groups in the periphery, but some of them have joined the resistance movement, providing it with training and protection, and have resumed active armed struggle. Are our findings relevant for other countries? In some respects, perhaps. Cambodia, Indonesia and Northeast India have all had longlasting internal conflicts. In Cambodia, a cycle of warfare, massacres and genocide ended after a comprehensive agreement was reached in Paris in 1991, with provisions for UN monitoring and a temporary UN administration overseeing elections and the formation of a coalition government. After some time, the strongest of Cambodia's armed groups, the Khmer Rouge, withdrew its support for the agreement and resumed its armed struggle. By then, however, it had lost international support and been marginalized domestically. By 1998, it succumbed to internal strife.
Indonesia had been plagued by several long-lasting rebellions as well as communal strife when the Suharto regime fell in 1998. The ensuing democratization was accompanied by increasing levels of violence. Seven years later, a new Indonesian government signed a comprehensive peace treaty with the strongest rebel group, the Free Aceh Movement. Significantly, the government also ordered the withdrawal of government-sponsored militias from other conflict areas and carried out decentralizing reforms. 67 This opened institutional venues for unarmed political struggles. Only one area, West Papua, is still plagued by armed conflict.
Northeast India has a situation that not only resembles Myanmar's, but is shared with it, since the same ethnic group (Naga) live on both sides of the border. 68 It could be interesting to compare India and Myanmar's approaches to counter-insurgency, 69 and not just focus on conflict onsets, duration and endings, but also on armed co-existence. The goal of a peace-seeking government is not necessarily to defeat its armed groups or persuade them to disarm, but to absorb them and incorporate them into its institutions. This blurs some of the distinctions used in quantitative conflict research and may also reduce the relevance of Myanmar's experience for countries such as Afghanistan, Syria or Yemen, where ceasefires are often short-lived tactical arrangements that only briefly interrupt intense warfare affecting the survival of the national government.
We hope that context-sensitive studies may be undertaken in several countries to see if our findings apply: Are there countries where nonstate armed groups form effective alliances, or do they, like in Myanmar, fend for themselves? Are ceasefires more likely to hold when armed groups are given lasting stakes in them? Can non-inclusive ceasefires contribute to peace in other countries with complex conflict systems, or do they stimulate fighting with excluded groups there as well? At any rate, the safest road to peace in a complex conflict system is surely to negotiate an inclusive peace agreement that establishes permanent institutional structures for non-violent political struggle. Notes 1. 'Active' means a minimum of 25 battle-related deaths in one calendar year. See Pettersson andÖberg, "Organized Violence, 1989-2019." 2. Selth, "All Going According to Plan?" 12.
3. An unofficial temporary ceasefire was agreed between the AA and Tatmadaw immediately after the 8 November 2020 elections, and upheld after the 1 February 2021 military coup. 4. Sosnowski, "Towards a Typology of Ceasefires." 5. Staniland, "Armed Politics and the Study of Intrastate Conflicts"; and Waterman, "Counterinsurgents' use of Force and 'Armed Orders." 6. Quinn, Joshi and Melander, "One Dyadic Peace Leads to Another?". 7. Quinn, Joshi and Melander, "One Dyadic Peace Leads to Another?"; Akcinaroglu, "Rebel interdependencies and civil war outcomes"; Nilsson, "Partial peace"; and Cunningham, "Divide and conquer or divide and concede. When two armed groups join forces in fighting a common adversary, UCDP codes it as a bilateral conflict between the leading group in the coalition and its adversary. 12. Unilateral ceasefires are not included in this study. 13 , in Naypyidaw, May 24-29, 2017, and in Rakhine April 20-24, 2018. Some groups, such as the ABSDF and CNF, helped the opposition by providing training but stopped short of breaking their ceasefire agreements. Actually, even the KIA's spokesperson refused to admit that its leaders ordered the KIA troops to fight alongside local resistance groups in the Sagaing Region. 35. Akcinaroglu, "Rebel Interdependencies and Civil War Outcomes." 36. Keenan, "Allied in War, Divided in Peace." 69. For the Indian approach, see Waterman, "Counterinsurgents' use of force," which is conceptually inspired by Staniland, "Armed Politics and the Study of Intrastate Conflicts."

Ethics
Research for this article was undertaken for a project led by Professor Erik Melander, University of Uppsala, and funded by the Swedish Research Council (Vetenskapsrådet). Approval of the research methods, including interviews, was given by the Uppsala Regional Ethics Committee by letter of 28 December 2018 (Dnr. 2018/035). All interviewees have given their informed consent to being cited in anonymized form. The semi-structured interviews were conducted with the use of a questionnaire (Appendices B and C). A list of interviewees has been kept separate from the interview recordings and transcripts, and an anonymized master file has been created for replication purposes. All interview transcripts and notes are kept in an offline encrypted disk outside Myanmar.

Disclosure statement
Before founding MIPS, Dr. Min Zaw Oo was involved in facilitating the negotiations leading to the 2015 Nationwide Ceasefire Agreement (NCA), and in monitoring the ceasefire. Mr. Ne Lynn Aung worked with Min Zaw Oo at the Joint Ceasefire Monitoring Committee (JMC) before joining MIPS. The authors do not consider this to present any conflict of interest.

Funding
Research at PRIO and MIPS for this article was funded by the Swedish Research Council (Vetenskapsrådet) through the Department of Peace and Conflict Research at Uppsala University.