The shadow of ‘the boys:’ rebel governance without territorial control in Assam’s ULFA insurgency

ABSTRACT This article leverages data from an oft-overlooked case of rebel governance – India’s United Liberation Front of Assam (ULFA) – to demonstrate the importance of de-centring territorial control as a prerequisite for rebel governance. ULFA neither controlled territory nor developed formalised bureaucratic institutions, yet its ‘parallel government’ held considerable sway over Assamese public life during 1985–1990, underpinned by its social embeddedness, influence upon media discourse and crucially its subversion of state structures, until its ability to limit state repression collapsed. The rise and fall of ULFA’s rebel governance illustrates the hybrid socio-political terrain upon which rebel governance is often laid.


Introduction
The rebel governance literature generally holds that 'territory enables control, and institutions and infrastructure enable governance'. 1 Recent works have challenged this assumption, highlighting that rebel governance may in fact be possible when rebels lack territorial control or formal political institutions. As Worrall notes: Some rebels may govern by taking and holding territory to create proto-states, others may not have sufficient resources to formally hold territory in such an overt manner and may instead look at temporary or temporal forms of governance, while others still may exert control from a distance, using psychological techniques or surveillance. 2 Moreover, governance may simply take place 'on the spot', or may co-opt and rely on hybrid institutional arrangements vis-à-vis the state. 3 However, the field has largely remained skewed towards powerful armed groups that have or are attempting to develop embryonic state structures, to the exclusion of rebel groups exercising minimalist forms of governance not reliant on territorial control. 4 This article introduces a little-known yet fascinating case of rebel governance -that of the United Liberation Front of Assam (ULFA) insurgency in Assam, Northeast India -to underline the importance of de-centring both territorial control and formalised institutions as prerequisites for rebel governance. Widely seen as a disorganised, limited outfit in comparison to the neighbouring Naga insurgency, which has received comparatively more attention in the rebel governance scholarship, 5 ULFA never established any formal territorial control. While some local units established rudimentary 'people's courts' in rural areas, ULFA rebel governance for the most part lacked formalised institutions. Yet during ULFA's heyday in the late 1980s, 'the boys', as they were affectionately known by sympathisers, constructed a 'parallel government' that 'really called the shots' across large swathes of everyday Assamese political life, at least until 1990. 6 ULFA was known for its aggressive moral policing and 'people's courts', its role in building schools, roads and flood defences and its 'Robin Hood'-like taxation of Assam's lucrative tea gardens. 7 How was ULFA able to operate its parallel governance structure? ULFA rebel governance was built around three key dimensions that negated both the need for physical control of territory and the development of formalised institutions. Firstly, the group was socially embedded in key strategic networks, emerging from common membership of the 1979-1985 Assam Movement against migration and the 'step motherly' (or exploitative and neglectful) treatment of Assam by the central government. 8 Through this social embeddedness, surrounded by an array of natural allies, the group was able to move freely across large parts of Assam, secure ad hoc coordination when opportunities to do so arose and conduct a combination of on-thespot, project-oriented and mobile governance activities that did not require a sustained presence based on territorial control. Emerging from this social embeddedness, the second dimension facilitating ULFA rebel governance was its strategic penetration of state structures. The emergence of a tacit understanding between ULFA and their ideological allies in the ruling Asom Gana Parishad (AGP) during 1985-1990 allowed ULFA to nullify police and broader state-level efforts against it. At the local level, this allowed ULFA to operate unchallenged and to coerce and co-opt elements of the state to direct service provision where needed. Again emerging from its social embeddedness, the third dimension concerned its ability to influence the media discourse to amplify the political effects of ULFA activities. This had two simultaneous effects. First, it created what Ashley Jackson terms 'atmospheric coercion', 9 a creeping, social-psychological form of influence in which the use or threat of violence created a 'fear psychosis' among the group's opponents and targets. 10 At the same time, however, it allowed news of the group and its activities to reach its broader constituencies, in doing so enhancing its existing social embeddedness. ULFA's governance activities, in the form of its moral policing, service provision and taxation -at least initially -often involved the coercion of opponents. This simultaneously enhanced ULFA's 'Robin Hood' image among its broader constituencies, which in turn contributed to the group's rapid recruitment across Assam throughout the late 1980s.
Ultimately however, ULFA's rebel governance collapsed under the weight of the group's expansion. Recruitment and training forced excesses in group taxation practices, drawing the ire of the central government and leading to the introduction of central rule in Assam and the accompanying deployment of the Indian Army. This forced ULFA underground, compelling it to retreat from much of its 'parallel government' activities. Thus, while ULFA's rebel governance flourished in the permissive conditions of the 1980s, its rebel governance was vulnerable when these three interlocking factors -in particular its subversion of state structures -were no longer able to create the right conditions.
Drawing on an extensive range of evidence and triangulating interviews, archival data and an array of secondary, 'grey', media and government documentation collected before, during and after fieldwork in Delhi and Assam in 2016, the article charts the rise and fall of ULFA's rebel governance and the centrality of its enabling social and political conditions in Assam. It leverages this data to demonstrate the importance of decoupling the study of rebel governance from the characteristics of 'stateness' while also deepening our understanding of the hybrid social and political terrain upon which rebel governance is often laid.
Sifting through secondary, grey and media materials allowed for the identification of patterns in how ULFA's 'parallel government' was conceptualised throughout the 1985-1990 period, but also highlighted gaps and important questions. Interviews with retired security forces members, politicians, administrators, journalists, civil society actors and academics then focused on these areas. Triangulating within and between these data sources thereby generated insights into the scope, depth and rationales of ULFA's governance activities during the period. 11 The article proceeds as follows. Section one interrogates the widely held assumptions regarding territorial control and political institutions in the rebel governance scholarship, highlighting how these have skewed case selection, with important consequences for the field. It then highlights recent challenges to these assumptions and the importance of exploring cases that do not conform to this pattern. Section two then introduces the ULFA insurgency in Assam and illustrates how the 'parallel government' of the 1980s was neither built on territorial control nor formalised administrative or bureaucratic institutions, exploring the reasons behind this. Section three outlines the three central conditions of social embeddedness, state penetration and media penetration upon which ULFA rebel governance rested. Section four illustrates how these dimensions facilitated ULFA's three core governance activities -service provision, moral policing and taxation -each of which in turn fed back into and reproduced ULFA's 'Robin Hood' aura. Section five then charts the collapse of ULFA's rebel governance. The pressures of an evergrowing cadre base however pushed ULFA into actions that undermined its ability to subvert state structures, as the central government deposed its allies in the state government and launched counterinsurgency operations. While ULFA's 'aura' and popular support endured beyond 1990, it was forced into a rapid retreat from much of its governance activities. The article concludes by reflecting on the implications of the rise and fall of ULFA's 'parallel government' for the relationship between rebel governance, territorial control and institutions more broadly.

De-centring 'stateness' in rebel governance
The field of rebel governance emerged as part of a shift away from framing civil wars as the collapse of order towards understanding conflict zones as 'ordered pieces of territory'. 12 The notion that rebels could construct and maintain order came as a counterpoint to the Hobbesian notion -once prevalent in the civil wars literature -that states were the sole providers of order and rebels the agents of disorder. 13 Rebel governance scholars have since unearthed a diverse 'archipelago of local rebel political orders' across conflict zones. 14 This scholarship has highlighted how group type, objectives and local contexts have shaped variations in the scope and depth of the provision of services, 15 the dispensation of justice 16 and the formation of local institutions. 17 Emerging as a counterpoint to the idea of civil war as the 'collapse' of order, much of this work understandably draws from the state formation literature to illustrate how rebels too can create order. Rebel governance projects are thus associated with embryonic states, 18 counterstates, or states within states. 19 This lingering state-centrism has meant that a number of assumptions have carried over from the state formation literature and become engrained in how the field understands the conditions and characteristics of rebel governance. This article builds on recent calls to relax two tightlyconnected assumptions; that to govern, rebels need to control territory, because territorial control enables the development of the institutions necessary to rule. 20 The first assumption contained within this statement is the centrality of territorial control as a prerequisite to governance. Most research on rebel governance essentially starts from the assumption that 'effective control over territory makes governance possible'. 21 The logic underpinning this assumption is that territorial control allows rebels to establish structures and institutions through which to interact with and solicit participation from civilians. This assumption has been built into the very definition of rebel governance, which Kasfir defines as 'the organisation of civilians within rebelheld territory'. 22 Most rebel governance research has thus essentially explored how rebels govern once they control territory. 23 This has had the effect of skewing the balance of cases studied towards those in which rebels exercise territorial control, while under-theorising those in which territorial control is contested, overlapping or non-existent.
The second, closely-related assumption emerging from this, as Loyle et al highlight, is that territorial control paves the way for 'institutions and infrastructure [to] enable governance'. 24 While recent scholarship has highlighted important variations in the depth, degree of institutionalisation and quality of rebel political institutions, 25 the assumption has generally held that rebels need to establish formal institutions if they are to govern. Given the frequency with which rebels develop (or attempt to develop) formal institutions, this assumption is an understandable one. As Karen Albert has highlighted, almost 64% of rebels in the post-war era created at least one governing institution. 26 With territorial control largely seen as a precondition of rebel governance, there is a tendency to focus on particularly 'state-like' cases with developed political, judicial and bureaucratic institutions, to the exclusion of cases where governance is informal, delegated or hybrid, comprising an assemblage of actors. 27 Recent works have made important calls to relax these assumptions. 28 Territorial control, rather than being seen as a critical determinant of whether governance can take place, should instead be understood as a spectrum along which the degree of rebel governance varies. 29 Rebels may control territory more firmly at certain times of day, month or year (monsoon season, for example). Degrees of control may vary from near-exclusive rebel control, to overlapping forms of states and rebel control, to highly minimalist forms of rebel presence. 30 These varying degrees of overlap, as works on wartime political order have highlighted, 31 can create an array of possible institutional assemblages of authority and governance that go beyond the notion of rebels establishing their own structures and institutions.
A number of works have started to highlight this array of possibilities. Ashley Jackson highlights how Taliban rebel governance predated territorial control through its 'creeping influence' into government-controlled areas, adopting a blend of 'atmospheric coercion, punctuated by occasional violence'. 32 Accounts of Hezbollah's rebel governance for example highlight how the group is both simultaneously enmeshed within state structures while operating its own independent governance structures. 33 In the Nagainhabited areas of Manipur, Northeast India, Thakur and Venugopal highlight how the main Naga rebel group, the National Socialist Council of Nagalim-Isak-Muivah (NSCN-IM), and local state structures operate parallel governance institutions with varying degrees of independence from and interdependence with one another depending on the specific sector of governance. 34 These cases highlight important degrees of hybridity, but nonetheless continue to focus on powerful rebel groups with welldeveloped governance structures and institutions. If we are to unravel the full spectrum of hybridity in territorial control and rebel institutions in rebel governance, we need to explore cases towards the minimalist end of this spectrum and analyse the conditions that underpinned and facilitated rebel governance practices.

The ULFA Insurgency in Assam, Northeast India
This article adds to this broadening understanding of hybridity in rebel governance by introducing the ULFA insurgency from Assam, the largest and most-populous state in India's understudied Northeast region. Unlike rebel governance in the neighbouring Naga conflict, which has received growing attention in recent years, 35 ULFA's 'parallel government' has surprisingly received no attention in the rebel governance literature. Indeed, unlike the main Naga rebel groups, which have developed governing institutions such as the GPRN and the Federal Government of Nagaland (FGN), ULFA has long been considered to be far more disorganised and lacking in guerrilla warfare skills. 36 This is perhaps due to the group's more recent history; since its heyday in the late 1980s, pressure from counterinsurgency operations and deepening internal fissures pushed the group away from being a mass-based insurgency towards a predominantly cross-border group alienated from its support base from the mid-1990s onwards. Yet while ULFA established neither any territorial control nor extensive institutions, the group governed large swathes of Assam during the peak of its insurgency, which occurred from the late 1980s until large-scale military COIN operations began in the early 1990s.
The ULFA insurgency itself originated in post-Independence mass movements to protect Assamese culture, language and economic resources against perceived threats from illegal migration and New Delhi's exploitative, colonial treatment of the 'unheeded hinterland'. 37 Fears that the Congress government had deliberately manipulated Assam's electoral lists to include illegal migrants in a bid to use them as a 'vote bank' for the Congress sparked the 1979-1985 Assam Movement, a large-scale social movement spanning a broad coalition of civil society and student organisations. State responses to the agitation, including forcing a controversial election in 1983, combined with lingering perceptions that Assam's natural resources were being exploited by Delhi, deepened secessionist sentiment within the movement. 38 The movement, spearheaded by the All Assam Students Union (AASU) and All Assam Gana Sangram Parishad (AAGSP), was mostly nonviolent and characterised by mass rallies, blockades and strikes, but violent clashes between protestors and police, communal violence including the notorious Nellie Massacre of 1983 and bomb blasts against opponents of the agitation led to a growing normalisation of violence that facilitated ULFA's rise. 39 Following negotiations between AASU, AAGSP and New Delhi, the Assam Accord was signed in 1985 and contained provisions to protect Assamese identity, language and culture. Leading agitation members then formed the Asom Gana Parishad (AGP), winning the 1985 state assembly elections with 67 of the state's 126 seats and becoming the incumbent state government. 40 However, lacklustre implementation of the accord's provisions deepened frustration with the central government, creating the conditions for ULFA's meteoric rise.
ULFA was formed in 1979 on the radical fringes of the Assam Movement with the aim of creating an independent, sovereign Assam. During the years of the agitation itself, the group generally maintained a low profile. ULFA tapped into the starkly visible economic inequalities of tea garden life in Assam, while recruiting from mainstream sections of the Assam movement. 41 The group established a series of camps (see Figure 1) and sought training from neighbouring Naga insurgents. From 1984-1985, the group gradually became more active, killing opponents of the Assam agitation, politicians  1986-1990. 44 associated with the 1983 elections, and conducting a series of daring bank raids. During 1985-1990, ULFA considerably intensified its activities, widening its targets to include state officials, economically influential individuals and members of 'outsider' business communities such as the Marwaris. 42 By the time military operations were launched against the group in November 1990, ULFA had built a well-armed organisation of approximately 1,000-3,000 cadres and was responsible for at least 113 deaths. 43 Enabling ULFA's 'parallel government' ULFA initiated an array of governance activities during the 1985-1990 period, conducting forms of service provision in outlying communities, setting up informal, rudimentary court institutions to punish immoral activities and taxing industries long considered exploitative, such as the tea gardens of Upper Assam. By 1988, 'ULFA's influence was visible in every sphere of public life in the Brahmaputra Valley', 45 described by former publicity secretary Sunil Nath as 'the government in those days'. 46 This situation changed rapidly in November 1990, when the central government dismissed the AGP, declared central rule and launched Operation Bajrang against the group. 47 While ULFA retained influence across Assam until well into the 2000s, its 'invincible halo' dissipated 48 ; much of the group's leadership and operational cadre base were forced to relocate to camps in Bhutan, Bangladesh and Myanmar. Becoming increasingly physically disconnected from its core constituents, ULFA's governance activities were 'quickly pushed into the background' in favour of a more straightforward campaign of military survival and cross-border strikes. 49 For this reason, this article focuses on the 1985-1990 period.
Although ULFA was able to extend its influence across large swathes of Assam, at no point did the group control territory, in the sense of a capacity to keep opponents out of a specific area. 50 Even in the upper-most districts of north-eastern Assam and its forest hinterlands, where ULFA's strongholds lay and where the state has long retained only a minimalist presence, the group never established an open presence in truly defensible 'liberated zones' that state forces were required to later retake. Instead, ULFA's camps, typically located deep within forested areas, functioned as sanctuaries that quickly dissipated after counterinsurgency operations were launched in 1990. While this indicates that territory in fact played some role in ULFA's approach, its governance activities were by no means confined to the forested areas around their camps and their immediate localities, but extended into and penetrated areas where the state had a far greater presence.
With the exception of limited, largely informal judicial institutions in outlying areas, the group did not establish widespread administrative or political institutions designed to formalise rebel-civilian relationships. For ULFA, a 1,000-3,000strong organisation in a state of 22.4 million, rebel governance instead relied on casting a shadow over Assamese social and political life. The group combined selective violence with widespread publicity from sympathetic local media to project an aura of power and authority, 51 exercising social rather than territorial control. For the group's foes, this created a 'fear psychosis' whereas for sympathisers and supporters it amplified the perception of the group as a 'Robin Hood' actor. This a-territorial, informal form of governance depended on its shared networks with former members of the Assam agitation. With agitation leaders holding the reins of state power in the state and with ULFA wellpositioned to ride the broader wave of pro-agitation sentiment, this allowed the group to 'infiltrate almost every level of governance and society', for a time dominating the Assam Movement and in doing so generating a far-reaching impact across Assamese society. 52 This section analyses these three closelyconnected conditions, before illustrating how they enabled ULFA's rebel governance in practice.

Social embeddedness
ULFA's governance was facilitated to a great extent by its ability to leverage pre-existing social ties with an array of allies across Assamese society. Although the organisation emerged on the radical fringes of the Assam Movement, the group's pro-secessionist ideology based on securing independence from colonial exploitation from the 'step mother at the centre' resonated ideologically with mainstream political discourse in Assam during the agitation. 53 Crucially, many of ULFA's early leaders shared the same activist networks with both other fringe and more mainstream organisations. Several founding ULFA members had for instance been members of the Asom Jatiyatabadi Yuva Chhatra Parishad (AJYCP), a radical student organisation critical of India's 'neocolonial' rule. 54 For local academic Nani Gopal Mahanta, given the overlap between the two during ULFA's formative years, 'practically there were no differences between the ULFA and the AJYCP'. 55 As members of ULFA participated in the agitation, they developed mutual associations with members of the All Assam Students Union (AASU), the main student body spearheading the agitation, and All Assam Gana Sangram Parishad (AAGSP), a political body constituted by the AASU. Each of these organisations were made up of broad coalitions containing elements of the Assamese urban middle classes as well as disaffected rural masses from the 'hinterlands'. 56 ULFA's presence within these pro-agitation networks not only meant that it shared a broad ideological space with large swathes of the Assam Movement, but crucially meant that it was able to mobilise and co-opt specific organisations and components of civil society to support and promote its governance activities while nullifying action against it.

Penetration of state structures
Arguably the critical factor enabling ULFA's rebel governance in the absence of territorial control and the development of governing institutions was its ability to effectively muzzle state repression and selectively leverage elements within the state as an institution and controller of (some) territory to augment its own governance efforts while projecting its own authority. As one journalist posted in Assam at the time recalled, while '[ULFA] did not have their own machinery, [.] they influenced and, in some areas, controlled the government machinery'. 57 This process, which combined elements of both co-option and subversion, took place at different levels within the state, ranging from the highest levels of state politics in Assam to localized penetration of the district and village-level administration and police. 58 The election of the AGP government in 1985 effectively extended ULFA's social networks into the corridors of power in Guwahati. As the political party of the agitation emerging directly from the AASU and AAGSP there was a social and ideological 'commonality' between the AGP and ULFA. 59 The group's General Secretary, Anup Chetia, and chairman Arabinda Rajkhowa were, for instance, members of the AAGSP during the early 1980s. 60 Indeed, elements of the AGP administration openly expressed sympathies for ULFA members. The then state government Home Minister, Brighu Phukan, stated for example that 'they [ULFA] are from among us'. 61 This ideological affinity did not create a clean, unproblematic state-armed group relationship. Rather, ULFA's penetration mapped onto and exploited the party's internal factional politics, particularly the rivalries between the Chief Minister Prafulla Kumar Mahanta and the Political Department led by Home Minister Brighu Phukan, to shape deal-making that, for ULFA, would at least paralyse the state government response. 62 As the AGP's tenure became plagued by dissatisfaction over its inability to implement the Assam Accord, 63 elements of the party increasingly saw opportunities to 'ride the tiger' of ULFA's growing popularity to extract concessions in implementing the Assam Accord from the central government. 64 ULFA's leadership, moreover, was aware that it could not grow and operate freely at the grassroots level without securing non-intervention from central and state-level security forces. 65 Therefore, through a combination of ideological overlaps, political incentives to tolerate ULFA activities, offers to take a cut out of extortion money and direct pressure from ULFA, 66 a tacit form of collusion and cooperation deepened between elements within the AGP administration and ULFA throughout 1985-1990. This cooperation ranged on a spectrum from direct forms of assistance to subverting efforts to counter it. During the Statefed rice scandal, for example, funds were misappropriated from a contract worth Rs. 20 million ($1,144,000 in 1990 U.S. dollars) to supply rations. When the then-Education Commissioner investigating the charges complained about the investigation's slow progress, the Chief Minister told him 'the money is for the boys'. 67 ULFA exploited this tacit arrangement to limit attempts to arrest its cadres or counter its activities. India's state governments are constitutionally responsible for law and order, and the Director General of the state's police is directly answerable to the Chief Minister. 68 This gives the political administration a great deal of influence over policing. The AGP administration refused to declare ULFA-affected districts 'disturbed' thereby sanctioning a security forces response, despite repeated requests from New Delhi. 69 Furthermore, ministerial interventions into policing created a culture of inaction against ULFA. As one police officer noted: 'whenever we got the ULFA boys arrested, they usually got released very quickly with the interference of the AGP ministers at that time'. 70 This was 'just a phone call away;' ULFA members were simply able to say 'it is our boy' and in doing so secure their release. 71 Police officers that were 'too aggressive' against ULFA were 'moved out', creating a culture that 'you are not expected to act'. 72 Later reports moreover disclosed that up to 500 individuals with 'close ULFA links' had been recruited into the police force during the AGP years, supplying intelligence to the group and even occasionally residing in major ULFA camps, 73 underlining the extent of ULFA's ability to subvert state attempts to counter it. By 1990, then, Governor D.D. Thakur lamented that: The loss of faith in the efficacy and the credibility of the government apparatus is so great that the distinction between ULFA, AASU and AGP, which existed at some stage, stands totally obliterated. 74 This combination of tacit arrangements with ideological allies and subversion and domination at the local level thereby effectively negated the need for territorial control. ULFA's influence upon elements of the AGP meant that it was able to subvert efforts to counter it at the highest levels of the state government. With this permissive environment established, it was then able to dominate local state assets through a combination of co-option, coercion and subversion, meaning it was able to utilise elements of the state administration to conduct its governance activities.

Casting the shadow: vernacular media and ULFA
Crucial to the development of ULFA's aura of social control was its use of highly selective, demonstrative -and therefore widely publicized -acts of violence to both appeal to allies within the Assam Movement as well as to intimidate opponents. Indeed, Nani Gopal Mahanta notes that during the early 1980s, ULFA's activities were particularly brutal or shocking, 'with the obvious objective of drawing attention' to the organisation. 75 During this period, ULFA entered the public eye by killing leading public figures opposed to the Assam Agitation, such as those responsible for the 1983 state assembly elections. A series of daring bank raids, such as a highprofile raid on a United Commercial Bank branch in Guwahati in 1985, amplified this 'demonstration effect' and brought ULFA's activities into focus. 76 To some extent, ULFA was able to benefit from elements of the Assamese vernacular press sympathetic to at least some of the group's actions. 77 However, as works on the relationship between the media and terrorism indicate, publicity does not have to be sympathetic to nonetheless support a strategy of simultaneously spreading fear and mobilizing support amongst key target populations. 78 ULFA's demonstrative acts of violence, its public moral policing (often in the form of public humiliation of offenders), and symbolic acts of forcing state officials to participate in service provision all attracted significant media attention. 79 This made ULFA's ideology, objectives and targets common knowledge. This enhanced ULFA's social control beyond the immediate audience by undermining confidence in and support for the government, intimidating a broader audience of opponents while appealing to allies and constituents in sympathetic communities across Assam.
ULFA recognized the centrality of shaping media narratives, establishing the position of publicity secretary and actively seeking to establish links with local journalists to convey the group's narratives. For instance, when ULFA encountered local resistance from an activist body highly critical of ULFA, the United Reservationist Minority Council of Assam (URMCA) -in June 1990, the group escorted a group of journalists to remote villages in Lakhimpur and arranged a press conference to ensure that the group's perspectives were conveyed in the press. 80 ULFA militants have also used coercion, issuing diktats to the press and intimidating journalists to ensure that particular messages are carried. 81 By combining these demonstrative acts with a strategy to engage local media, ULFA was able to spread a 'fear psychosis' amongst its opponents while generating a 'Robin Hood'-like image amongst those potentially sympathetic to its activities.

ULFA rebel governance
The following section highlights how the above conditions enabled ULFA's parallel government to function by analysing three core, closely interlinked aspects of ULFA rebel governance: service provision, moral policing and taxation. Analysing these three elements, which are common across both the practice and study of rebel governance, thereby serves to illustrate the functioning of rebel governance in the absence of territorial control. 290 A. WATERMAN

Service provision
A central part of ULFA's strategy to mobilise popular support and boost its 'Robin Hood' image in rural Assam was to engage in an array of development activities, ranging from building road and bridge infrastructure, building and maintaining flood defences, farming and irrigation projects and the construction of schools. 82 These forays into service provision were not the result of an established set of service-providing governance institutions designed by the group itself, nor did it create any formal governance infrastructure to supplant that of the state. Instead, ULFA usually coordinated welfare provision and construction work on individual projects by coercing, co-opting and directing an assemblage of non-state and local Indian state actors. This ability to leverage both local allies and the resources of the district-level state administration allowed the group to conduct 'on the spot' service provision that required neither territorial control nor significant investments in infrastructure. 83 For instance, ULFA frequently worked alongside the Asom Jatiya Unnayan Parishad (JUP), 84 a local development organisation formed in 1989, enabling it to engage in service provision 'in almost every part of the Brahmaputra Valley in Assam'. 85 The JUP is often labelled a front organisation of ULFA. While often denied, this charge does reflect the extent of the overlapping ties and shared membership between the two organisations and the strong overlaps between the ULFA and JUP areas of activity, which spanned large swathes of the Brahmaputra Valley of northern Assam (see Figure 2). The JUP was known for carrying out unimplemented government welfare projects where key development needs had already been identified by communities, 86 often in response to the regular damage wrought by the Brahmaputra Valley's flood seasons. 87 ULFA's influence on service provision essentially came through its association with the JUP, playing a key role in encouraging and supporting aid projects by both persuading, coercing and organising local labour, often involving entire villages. 88 In Biphuria, Lakhimpur district of northeastern Assam, for example, ULFA and the JUP mobilized at least 5,000 villagers from diverse communities to work on a flood defence project. In another example from 1990, ULFA's Nityananda Anchalik Committee organised around 100 locals to build an embankment near Golaghat Bridge. 89 ULFA was also involved with longer-term projects, including supervising the construction of local colleges and overseeing the development of cooperative farms that its JUP allies could then run in the longer-term. 90 Taking up unimplemented government projects based on existing, locallyidentified needs enabled ULFA to be seen to respond to local demands, while at the same time exposing the weaknesses of the state.
Since police interventions were discouraged to a great extent by the state administration, local state officials were forced to cooperate with ULFA projects on the ground, particularly in Lakhimpur and Dhemaji districts. 91 As one government official posted in an ULFA-affected district stated: For us, the choice was clear: toe the line prescribed by the ULFA leaders or get into trouble since it was clear that they had a tacit understanding with the political leadership. Whatever they demanded, had to be fulfilled. The requirements initially ranged from free use of telephone and official cars to improvement of roads or other facilities in a particular village. 92 In some cases, ULFA played a much more assertive role, forcing state police, local officials and other elites (such as tea garden managers) to participate in the labour of building river embankments, cleaning community houses and maintaining sanitation facilities and in doing so showcasing their local dominance. 93 These activities further amplified the 'Robin Hood' image which strengthened ULFA's legitimacy in the rural hinterlands. For example, when police took part in an ULFA-JUP-organised flood defence reconstruction project in Biphuria, Lakhimpur district of north-eastern Assam, local observers noted that 'it was a sight to watch how the local police went [.] to work in the site [.] all of us were delighted and could not stop appreciating the [ULFA] boys'. 94 In other examples from 1990, the wives of police officers and in one case a deputy commissioner of the district administration assisted in these activities. 95 While these service provision activities were only occasionally publicised, they consolidated local support for the organisation. Furthermore, some stories did appear in the vernacular press and thus continued to garner the image of a 'Robin Hood' actor beyond the area of activity, while undermining the credibility of the Indian state, until 1990.

Moral policing
Through its ability to tap into local social networks and subvert -and in some cases dominate -local state institutions, ULFA was able to regulate social life almost unchallenged in its areas of influence. Exploiting public discontent with the notoriously slow-moving nature of the Indian judicial system, the group issued diktats against and publicly punished social ills, including sexual harassment, prostitution, gambling, drugs and alcohol production and consumption, and bribery. 97 The scope of ULFA's judicial activities varied. In its rural strongholds, the group developed 'people's courts;' ULFA's only foray into institutional development, these limited, usually informal judicial institutions were designed to hear complaints about immoral behaviours and rapidly dispense justice through public acts of violence. According to Governor D. D. Thakur's report to the President in November 1990, these courts had supplanted the judicial institutions of the Indian state in Lakhimpur, Tinsukia and Dibrugarh. 98 Indeed, in Lakhipathar, near one of ULFA's largest camps, 'ULFA penetrated into almost all spheres of life', influencing social practices such as dress codes (banning clothes representing 'Indianness') and regulating the conduct of personal relationships. 99 Beyond its sanctuaries, the functioning of these courts relied on ULFA's subversion and co-option of the state; ULFA members in Dibrugarh for instance openly used the guest house of the State Electricity Board to run their court sessions. 100 Most of ULFA's judicial interventions consisted of ad hoc acts of humiliation or violence against offenders. For example, in Guwahati in December 1989, three youths suspected of selling illegal alcohol were made to kneel in front of a cinema in a display of public humiliation, while the police did not intervene. In other instances of highly demonstrative violent acts, dead bodies were left in public places, marked with ULFA logos and ULFA statements were released detailing the reported crimes of the accused. 101

Taxation
Recent works have noted that rebel taxation should be seen as 'a technology of governance that rebels deploy to resolve a variety of political, economic, and organisational challenges'. 102 ULFA's taxation practices, which predominantly targeted exploitative or 'outsider' businesses and industries such as the tea industry, sought to resolve two key challenges. First, the group's taxation sought to address organisational challenges; ULFA needed to finance its internal organisation-building, training and arms procurement, which was rapidly expanding and becoming increasingly expensive by the end of the 1980s. 103 Second, ULFA's approach to taxation is shaped ideologically by the group's emergence from the Assam Agitation, specifically through its ideological opposition to extractive, 'colonial' enterprises and exploitation and the disproportionate influence of so-called 'outsiders' in the local economy. The logic of ULFA taxation was essentially that since tea industrialists and outsider businessmen were at the helm of vast economic inequalities, and since international companies such as Williamson Magor extracted huge profits from Assam, these industries should be give back to Assam. 104 In this sense, ULFA's taxation practices are fundamentally linked to its attempts to regulate moral conduct.
ULFA taxation was therefore highly selective in its targets. ULFA militants would for instance collect intelligence from lower-level tea garden managers from within the community to determine local grievances and cases of managerial abuse, and use this information to leverage tax demands. 105 This selective practice differs for example from the nearby Naga armed groups whose taxes span the entirety of Naga society in a bid to support their claims as sovereign, separatist entities. 106 The practice of taxation itself was ad hoc and far from institutionalised during the late 1980s. The scope of the demands themselves fluctuated between requests for increasingly large sums of money to more redistributively-focused demands blurring the lines between taxation and service provision. This included, for example, ULFA demands that tea gardens upgrade local medical facilities or institute scholarships for disadvantaged youths. 107 To support its service provision efforts in conjunction with the JUP, ULFA also directly seized payments-in-kind such as manure, seeds and tools to aid its agricultural welfare projects, 108 while collecting financial donations from other local businesses.
Rather than formally controlling the areas where tea gardens resided, ULFA's taxation relied on its ability to generate atmospheric coercion, or 'fear psychosis' as it was popularly known in Assam at the time. Demands to tea gardens would for instance typically begin by taking away a tractor, paying a visit, or inviting tea garden managers to a meeting following a threatening phone call. 109 With the police unable to respond substantively, 'a simple telephoned threat, the display of weapons by tough-looking militants, "friendly advice" or even an occasional roughing up was enough to help local traders part with millions of rupees'. 110 Indeed, ULFA collected between Rs. 4-5 billion ($285,000,000) in extortion funds during the 1985-1990 period, underlining the extent to which this atmospheric coercion was successful. 111 At the same time, this strategy of coercion and violence, targeted as it was at symbols of wealth and exploitation in Assamese society, resonated with the group's 'Robin Hood' image of taking from the rich to give to the poor.
These underlying conditions and the governance activities enabled by them produced a self-reinforcing dynamic that contributed to the group's meteoric organisational growth during the late 1980s. ULFA's pre-existing social embeddedness within the broader ethnonationalist movement provided it with a broad support base upon which it could both rely on for preexisting social ties that would enable governance. Its ability to reach these constituents (and intimidate potential opponents) was enhanced by its ability to generate widespread vernacular media coverage. Its social embeddedness then enabled it to penetrate the state structures of the AGP government, preventing significant state responses against it.
ULFA's governance activities fed back into and reinforced these conditions. The group's service provision activities frequently co-opted allied organisations while forcing 'others' or state actors to participate in these projects, enhancing the group's 'Robin Hood' appeal to its constituents. Similarly, widely-publicised instances of moral policing and taxation of exploitative industries both intimidated potential 'others' while again enhancing the group's appeal to its support base. 112 This self-reinforcing dynamic contributed to ULFA's meteoric growth in its organisation, popularity and presence across Assam during the late 1980s. By 1990, the Governor of Assam D. D. Thakur estimated that the group had recruited up to 3,000 cadres. 113

Collapse
However, as the organisation rapidly grew, new tensions began to emerge within this interaction between the underpinning conditions and the governance activities that ultimately contributed to its rapid downfall. From 1988 in particular, the group's costs increased significantly. Its recruitment base was growing rapidly, it had established its base of operations in its forest sanctuary of Lakhipathar, and had secured an agreement to send a second batch of ULFA cadres to train with the Kachin Independence Army, an armed group operating in northern Myanmar. At the same time, ULFA's service provision in conjunction with the JUP was placing a growing financial burden on the organisation, while the ad hoc nature of these activities meant that financial collections to sustain them often lacked accountability. 114 The regularity and size of ULFA's demands thus increased, as did its propensity to use violence to enforce these demands. 115 The killing of Surendra Paul, Managing Director of Assam Frontier Tea Ltd. in April 1990 led to a dramatic flurry of negotiations between ULFA and representatives of the major international tea corporations in Assam. However, following complaints made by the tea corporations to Indian diplomats abroad, the central government increasingly became aware of the ULFA problem. Frustrated by the AGP's inaction, in November 1990 Delhi arranged to dramatically airlift tea executives out of Assam without informing the state government. 116 By the end of the month, the central government deposed the state government and declared 'President's Rule', before deploying the Army and launching counterinsurgency operations. This undercut the group's ability to limit state repression and in doing so facilitate its rebel governance. As a result, ULFA's ability to direct and coordinate welfare activities with and through the JUP and other local allies quickly dissipated in the face of Indian Army operations. 117 The three conditions of social embeddedness, state penetration and a conducive media environment created the conditions for the rapid expansion of ULFA's rebel governance activities. The swiftness with which ULFA's rebel governance efforts collapsed following the end of the AGP administration highlights the importance of the group's subversion of state structures as a critical enabler of its governance programme.

Conclusion
Building on recent calls to de-centre territorial control and political institutions as key conditions for rebel governance, this article has utilised the oftenoverlooked case of ULFA's five-year 'parallel government' in Assam in order to demonstrate the existence of -and the importance of studying -nonterritorial forms of rebel governance.
The ULFA case, by challenging the centrality of territorial control and the existence of formal institutions emerging from it, underlines the importance of going beyond notions of 'stateness' as 'the model and basis with which to compare rebel organisational structures' 118 and the importance of focusing on the social and political ordering structures enabling non-territorial rebel governance. Existing research has already highlighted the existence of differing forms of political order in rebel-held territory, how they relate to institutions and how they evolve over time. 119 As this article illustrates, this agenda can be extended into the realm of non-territorial rebel governance by better integrating the state and its interactions with both rebels and civilians. 120 Comparative research could build on this by overlaying rebel governance data with data on state-group armed orders to explore whether certain types of state-armed group orders are more conducive than others to the development of non-territorial rebel governance.
The article further demonstrates that although ULFA rebel governance was primarily non-territorial in nature, place and space still played an important role in how the group operated. Even in the absence of territorial control, insurgents require some degree of sanctuary from which to operate, whether in the form of clandestine camps in difficult-to-access or foreign bases, underground networks, sacred spaces, 121 or strongholds of support in cities. In the ULFA case, sanctuary was ensured by both nullifying the state response against it, by operating in rural communities in conjunction with local allies and by placing camps in remote jungles. This underlines the challenges of viewing sanctuary as a physical, territorially-based construct alone and supports existing calls to integrate the layered, textured nature of armed group sanctuaries into rebel governance research. 122 Further research should examine an insurgency's relationship to place and sanctuary, the reasons behind this and the opportunities and constraints this relationship places upon efforts to govern civilians. Here there are opportunities to better integrate notions of territoriality into our understanding of rebel governance, 123 particularly in thinking about how rebels relate to space and place, whether insurgent capacities match these conceptions, and whether and how these intersect with civilians' own understandings. 124 Lastly, ULFA's rebel governance raises questions of the relationship between temporality, the functioning of rebel organisations and pressures towards institutionalisation. Most of ULFA's governance interventions required only temporary interventions into local communities without the need for institutions, laying the foundations for local allies. However, as ULFA grew as an organisation, institutional funding pressures forced tax demands of ever-increasing regularity and size, ironically creating the conditions for the downfall of the parallel government. Using non-territorial rebel governance as a starting point, further research should draw greater attention to the nascent, intermediary stages of rebel governance, analysing the functional pressures that drive shifts towards institutionalisation and territorial control.