Government Responses to Gang Power: From Truce to War on Gangs in El Salvador

This article reflects on the challenges that governments face in dealing with illicit armed groups that have built up local power positions. It analyses an accommodating government response (a local truce) to gangs in the town of Nueva Concepción, El Salvador, and its aftermath that was characterised by a renewed crackdown on them. These responses did not manage to break gang power. While the local truce had led to some modifications in gang use of violence and the degree and forms of extortion they used, the ‘war on gangs’ that followed the local gang truce had a more disruptive effect, as it was not able to stop or break their control. The paper shows that different responses can take place simultaneously, affecting each other in intended and unintended ways. Recognizing the profound contestation about the ‘right policy mix’ in situations where state power is violently challenged, it also discusses some of the challenges and possibilities to combine accommodating, repressive and preventive approaches.


Introduction
Over the past 25 years, the street gang phenomenon in El Salvador has transformed into one of the country's main security challenges. After the civil war ended in 1992, gangs grew quickly and transformed from a sub-cultural phenomenon of small and loosely organized groups into organizations that are part of larger (globalized) gang networks that have their origin in the U.S. (Arana, 2005;Savenije, 2004Savenije, , 2009. Over time, gangs increasingly extorted local residents and businesses in large parts of El Salvador. Gangs are generally strongest in marginalized neighbourhoods of the larger cities, but the phenomenon has also spread to the countryside (Segovia et al., 2016). There are an estimated 60,000 gang members in El Salvador, of which some 10,000 are in prison. Gang violence is one of the most important causes of the extremely high homicide rates in the country. In 2015, El Salvador was Latin America's most violent country with 103 homicides per 100,000 inhabitants (Gagne, 2016).
It has proved extremely difficult for Salvadoran governments to contain the gang phenomenon. For decades, repression has been the dominant policy response towards gangs, but in March 2012, the Salvadoran government facilitated an agreement between the principal street gangs -called a truce (tregua) by the stakeholders -which led to a marked reduction of the homicide rates. However, the truce initiative also led to heated discussions about its effectiveness, legality and legitimacy. Critics claimed that gangs actually became stronger as a result of the truce, while supporters argued that gangs had already become so strong that involving them in the solution to the gang problem could prevent more bloodshed; they also believed that the gang leadership was genuinely interested in ending gang violence. However, as a result of societal and political opposition, the truce unravelled, and made way for a new wave of repression towards gangs.
While research on non-state armed groups in Latin America increasingly focuses on the local level (Moncada, 2013;Arias, 2017), there is only limited attention for the ways government responses play out at the local level and affect non-state armed groups in intended and unintended ways. This is also the case for research into the Salvadoran truce and its aftermath. This paper aims to contribute to a better understanding of the local level outcomes and dynamics of different gang policies. Using data gathered during fieldwork in a Salvadoran town with a strong gang presence, it analyses the local truce process and its repressive aftermath. It looks at the way in which different government policies played out at the local level, how these policies were experienced by local residents, and assesses how various policies affected local gang power. It also discusses some of the main challenges to transform or contain gang power, using different types of policies, and it draws conclusions about the wider significance of these findings for government policies dealing with illicit groups that challenge state power.

Governments and gangs -responses and liaisons
Government responses towards illicit non-state armed groups come in different forms. Schuberth (2014) makes a distinction between more accommodating, preventive and coercive (also repressive or conflictive) approaches. While accommodating approaches are based on the idea that negotiation, cooperation, or contact with gangs can contribute to the solution or mitigation of the problem, the assumption of coercive policies is that gangs are criminal actors that have to be countered by applying criminal justice and/or coercion. Despite the tensions between the assumptions undergirding accommodating and coercive approaches, it is not unusual to combine elements of these approaches, and peace processes in the past have shown that the two are not necessarily incompatible (Schuberth, 2014;Wennmann, 2014).
It is widely recognized that government policies towards illicit groups, such as gangs, can have profound and unintended consequences on these organizations, as well as the local environment in which they operate (Moncada, 2013 pp. 228-229;Pansters, 2015, p. 150). Repressive policies can lead to the transformation of illicit groups rather than their disappearance, to the hardening of relations between governments and these groups, and to higher levels of violence. This was clearly the case in El Salvador where a major critique on the zero-tolerance policies against gangs is that they contributed to the transformation of gangs, with prison leaderships playing an increasingly important role, and extortion (to sustain the gang) becoming more important (Cruz, 2011;Gomez-Hecht, 2013;Wolf, 2017). The main accommodating initiative in El Salvador (the truce process that started in 2012) led to an intense national discussion about the ethics of such an approach, the hidden agendas behind it, and the unintended effects on gang power. In addition, over the past decades a large number of preventive programmes has been implemented; it is now widely recognized that in order to counter the gang problem in Central America, social and preventive policies are important (Jütersonke et al., 2009). In El Salvador these programmes have led to prevention of young men and women from joining gangs, but they have not been able to significantly diminish the power of gangs. 1 It is clear that confronting gang power has been extremely difficult and most anti-gang policies have not been able to significantly contain the gang phenomenon. Three challenges stand out. Firstly, at the heart of the problem of street gangs and government responses to gangs is a contest about control, such as territorial control over neighbourhoods, and the capacity to collect tributes from local residents and enterprises (e.g. buses). Although gangs in Central America do not have political agendas, they (implicitly) challenge state power, as they contend with the laws and regulations of the state and promote alternative types of sanctioned behaviour (Migdal, 2001, p. 12). Secondly, during the past decades gangs have built up strong local power positions in numerous (mostly) marginalized neighbourhoods and their position is hard to dismantle in the short or medium term. This is not to say that non-state armed actors can-not be stopped, but rather that the problem is more protracted than often recognized. 2 Thirdly, as a result of the enormous challenge that street gangs pose to government power, there is a deficient capacity to formulate and implement a coherent set of policy responses, and to reach political consensus about the 'right' policies. The challenge that street gangs pose to the state has also led to policy efforts that challenge or are in conflict with the law.
The emergence of street gangs has to be understood within the larger political and economic context. Stephenson's (2013, p. 326) work in the Russian city of Kazan in the years following the collapse of the Soviet Union shows that gang power itself is 'highly responsive to the contradictions and problems of the social order, the power deficits that emerge in the larger political economy, and in certain circumstances can move to fill the power vacuum -at least at the local level'. In this regard, the deficiencies in, and problematic reform of, the Salvadoran security and justice institutions after the peace agreements of 1992 (Cruz, 2011) are of importance in understanding the opportunities for street gangs to grow so quickly. Although street gangs are the most visible manifestation filling the power void, profoundly affecting the everyday lives and security of the most marginalized sectors of Salvadoran society, they are not the only actors responsible for the country's high levels of insecurity and the dire state of the rule of law (ibid, [25][26]. The reluctance of powerful elites to reform state institutions, as well as the power of economic elites and the multiple linkages to organized crime (such as narco-trafficking) also contribute to protracted insecurity and instability (Aguilar, 2014;Silva, 2014;Wade, 2016).
In this context of problematic security and governance reforms, gangs have profoundly impacted local political orders. There is increasing attention for the capacity of criminal groups in Latin America to challenge state power with studies discussing alternative, or criminal, forms of governance (Arias & Davis Rodrigues, 2006), the existence and transformation of multiple sovereignties (Pansters, 2015), or the logic of non-state orders (Arias, 2013;Lawrence 2012). An important part of that literature looks at the power of drug cartels and the use of 'drug violence', especially in Colombia, Mexico and Brazil. From these writings it becomes clear that while organized crime is a sign and a result of state weakness, it rarely leads to state collapse (Arias, 2013, p. 264). Instead, a variety of localized orders come into existence, where state institutions, organized crime, and other actors cooperate or compete in diverse and complex ways (Arias, 2017). The relations between these actors differ from place to place, can be subject to (abrupt) change and do not necessarily lead to stable orders (Arias, 2013;Arias, 2017;Pansters, 2015).
Street gangs also have a capacity to establish liaisons with local government and other social actors. In the case of El Salvador, there are -despite a historical emphasis on police repression against gangs -multiple connections between local gangs members, local government officials, politicians and grassroot leaders (van der Borgh & Abello Colak, 2018, p. 6). These liaisons are often unstable and ad-hoc, but nevertheless provide some degree of stability. Examples of these connections became visible in the run-up to the presidential elections of 2014, as conversations took place between high level political leaders and gang leaders. For example, a video tape showed that representatives of the right-wing ARENA party met with gang leaders in the municipality of Ilopango to discuss, among other things, the possibility of a new truce (Labrador & Asensio, 2016). Interestingly, one of the participants of these meetings acknowledged that talks had taken place and stressed that politicians, mayors, and government staff need to dialogue with the local people 'who control the situation' (Labrador & Martínez, 2016). He also emphasized that a dialogue takes place almost on a daily basis. In a similar vein, the FMLN made their own alliance with gangs (Martínez & Valencia, 2016). Table 1 provides an overview of different types of government response towards gangs (repressive, accommodating, or none) and their formal and informal manifestations. The three types of government response in the upper row are based on Mitchell's (1981, p. 24) three types of social relationships (conflict, cooperation, isolation). The left column makes a distinction between formal and informal policies or actions. Formal policies are 'for the most part codified through publicly recognised rules', while informal policies are largely 'structured around implicit practices, social understandings, networks of interaction, and socially sanctioned norms of behaviour, which are neither officially established nor codified, but are commonly and widely accepted as legitimate' (Bagayoko et al., 2016, p. 7). 3 Indeed, Bagayoko et al. argue that the application of these definitions to real world situations is by no means straightforward and that informality can play a role in formal arrangements (ibid, p. 5). This implies that blurring between these types of policies can take place. This leads to six types of government responses to gangs. These responses do not necessarily exclude each other, but they relate in complex ways to each other, and can influence each other. 4 Since different government agencies are responsible for the various types of policy mentioned above, this also implies that at local level the 'policy mix' depends on the willingness and capacity to coordinate between different government agencies.

Methodology
This article is based on field work in the town of Nueva Concepción, El Salvador. The town shares a number of important characteristics with other marginalized and 'gang-controlled' municipalities, such as high levels of violence and insecurity, and the widespread practice of extortion. Gang activity and the local truce process in Nueva Concepción have particular characteristics. Only one of the larger gangs, MS13, is present, and the municipality has not been plagued by conflicts with other gangs (such as the Barrio 18 gang). However, clashes between groups of the same gang (called programas or factions in gang jargon) did and do take place. As will be discussed further on, the local truce process in this town started relatively late in comparison to other towns in El Salvador where the process had a larger local impact. The paper analyses how (changes in) government policies play out at the local level and how these affect local gang power. The focus is on the 'endusers' of security (Luckham & Kirk, 2013, p. 17), using the information residents provided about their experiences with gangs and anti-gang policies. Zooming in on two periods in which opposite approaches to deal with gangs were taken, I explore the differences in gang power as experienced by local residents.
Particular attention will be paid to one dimension of gang power: the capacity to extort. In the Salvadoran case, extortion has become the principal means of gangs to generate income for the survival of the gang (Gomez Hecht, 2013, p. 153;ICG, 2017). Extortion can be defined as the 'forced extraction of resources' for which services may be promised in return, but the latter is not necessarily the case (Varese, 2014, p. 350). It involves the use of coercion, including the threat or use of violence, to obtain money, property or services. While the threat and use of force is an important instrument of gang power, it should be stressed that gang power and control are more complex than this. A full analysis of gang power and control should, for instance, also include the capacity of gangs to liaise with other local actors (criminal and political) and the effects of decades of gang presence on the behaviour of residents. 5 Focusing on two periods, in which opposite approaches to deal with gangs were taken, the differences in gang power as experienced by local residents are explored. The research sought to understand how residents engage with gangs during different periods of government policies. Many of the topics discussed were considered sensitive; this was also the case with information about extortion. In this regard, several interviewees gave detailed information about their own experiences with gangs, including when and how they were extorted, while others reflected on the experiences of other residents. The experiences of extortion that were reported included one-off and monthly payments by businesses. Based on this information it is claimed that extortion continued during and after the local truce. However, the paper does not aim to give detailed information about the workings of local extortion practices. Therefore, this research provides an approximation of the changes in gang power and control as experienced by residents and policy makers during different periods of more accommodating and more repressive government responses.
Evidence from diverse sources have been triangulated to develop the analysis presented here. The sources include: (1) semi-structured interviews with local policy makers, politicians and residents who are not involved in gangs (including second or third interviews with the same interviewee); (2) analysis of some 30 newspaper articles, as well as a number of media clips about the local security situation; (3) policy reports (municipal) and other documents (e.g. court cases) with information about the situation in the municipality; and (4) research reports, investigative news articles, and academic works about the national security situation.
Interviews were conducted during six field visits between 2014 and 2017. A total of 25 persons were interviewed, some of them repeatedly, leading to a total of 39 interviews. The interviewees were selected through snowball sampling and included people of different age groups, sexes, and political backgrounds. The interviews were semi-structured, and generally combined questions on how people had experienced insecurity and violence during their lives, while discussing particular events, examples or processes related to the power of gangs in more detail. Respondents did not share the names of members of illicit groups or provide any information that might put themselves or others at risk, and were guaranteed anonymity. The majority (17) of the interviewees were living in town, and eight interviewees lived in other hamlets of the municipality. All interviewees had detailed knowledge about one or several of the other villages and hamlets where they did not live themselves. 6

Gangs in Nueva Concepción
Nueva Concepción is a relatively large municipality, covering 257 square kilometres. The entire municipality has approximately 35,000 inhabitants, of which 10,000 live in the municipal town. 7 Agriculture (rice, sugar cane, corn, beans, vegetables) and dairy cattle breeding have traditionally been the most important economic activities. 8 During the post-war years, small-holder agricultural production suffered under the liberalisation of El Salvador's economy (Sprenkels & van der Borgh, 2017), but the possibilities for agriculture in Nueva Concepción are still better than in many other parts of the department. However, the interest of young people to work in agriculture has steadily declined. Migration to El Salvador's capital San Salvador, and -much more importantly -to the United States, has become an important livelihood strategy. Today many thousands of Nueveños reside in the U.S. While the society and economy in Nueva Concepción underwent profound changes in the post-war years, the economic prospects for youth are bleak, and this partly explains the appeal that gangs have on a part of the youth, as they provide a strong sense of identity and some income.
Although the department of Chalatenango was one of the war-torn regions in the country, Nueva Concepción was far less affected by the war in comparison to the eastern part of the department that was controlled by the leftist guerrilla movement FMLN. During the war, the town was controlled by the police and the military. An extremely violent paramilitary group terrorized the countryside, killing alleged collaborators of the guerrilla and forcing civilians to join the patrols. There were some smaller camps of the guerrilla in the municipality, but there were only a few confrontations between the warring parties. The colonial town of Nueva Concepcion was considered relatively safe during the war years, and people from the conflict zones -especially in the eastmoved here and settled in new neighbourhoods. The peace agreements of 1992 set in motion a process of rapid demilitarization of politics (Lyons, 2005) which had profound repercussions for the municipality of Nueva Concepción. In the years after the agreements, the municipality was relatively calm and stable, suggesting that a transition towards democracy and peace had been set off. However, by the end of the century, it became clear that the building of a liberal constitutional order was being challenged and hampered by state and nonstate actors. Gangs were the most visible actor challenging the state, but organized crime and government corruption also contributed to a flawed local state capacity.
Gangs were not a new phenomenon to El Salvador and this was no different in Nueva Concepcion, where in the last years of the war new local gangs had popped up in the urban area, as well as in other villages and hamlets. Mara Transformer was one of the first groups; a local resident remembers that they did 'nothing else than to belong and to hang out. They were friends that had changed the way of greeting, of dressing.' 9 One of the important places for the Mara Transformer members to hang out was the so-called Calle sin Ley (Street without Law), where bars and brothels were located and occasionally fights might start at night. Someone else remembers that in 1995-1996 a young man came back from the U.S. to his village who introduced the gang culture. 'He had his clothes loose and all that. We went to the parties, and we had fights. Many boys who joined the gang are dead now, or in prison, or in the U.S.' 10 Local gangs in other parts of El Salvador had usually connected to other larger gang structures with origins in the U.S., which spread rapidly through the northern countries of Central America (Arana, 2005). However, where the competing gangs of MS13 and Barrio 18 had gained a presence in most cities and towns, this did not occur in Nueva Concepción, which has only one gang with two factions: the Hollywood Locos Salvadoreños (HLS) and the Fulton Salvadoreños Locos. 11 Around 2004, local cliques linked to these different factions, fought each other over the control of parts of the town, leading to violent clashes. At a later stage, hostilities diminished and an agreement was reached about the spheres of influence. Gang activity was initially strongest in Nueva Concepción, spreading later to other rural towns in the municipality.
The local factions of the MS13 are part of larger gang structures, which have become more hierarchically structured, and leaders of local cliques and factions maintain close relations with prison leaderships that are key in the gang structures (Dudley & Martínez, 2017). In 2012, the estimates of the number of gang members in the entire municipality of Nueva Concepción varied between 300 and 350, most of them male. These numbers include the lowest echelons of the gang membership -the homeboys that carry out orders, while it does not include the 'wannabies', often the youngest boys, who are not (full) gang members yet. 12 Disputes about local leadership take place frequently. 'I know them. Suddenly a gang member tells you he now is the chief, and has replaced someone else'. 13 While gang members used to be visible in the way they dressed, this is not always the case any longer and local residents do not necessarily know who is a gang member and who is not.
Gangs protect their 'turf' against other gangs, and they also seek to control the movements of certain people (e.g. police, visitors) through a system of posts (postes) and communication. 'They have a complicated system, like the police doesn't have.
[…] They know who is going where. They have their network of cell phones'. 14 It should be stressed, however, that gang presence is experienced in different ways throughout the municipality. Even in one and the same neighbourhood the experiences with gangs can be quite different, and not all people are confronted (in the same way) with violence of gangs and their extortion practices. The legitimacy of gangs is, however, extremely limited, which stands in stark contrast to the claims of, for example, Brazilian drug gangs that defend local residents from police brutality. In the words of a local resident: 'They have a degree of control, but actually they haven't. Because the people don't support them. They want to have control, instilling fear, but they don't have that because the people don't help them. The only thing that the people have is fear'. 15 After 2003, as a result of hard-handed policies and an increase in the number of imprisoned gang members, a transformation took place in gang structures. Among other things, gangs that were often led from within the prison system started to extort businesses on a large scale. Most interviewees argue that the municipality became extremely violent around 2006, which seems to be related to the introduction of extortion and the increasing involvement of gangs in narco trafficking. However, extortion started well before 2006. In 2003, sixteen local bus owners filed a complaint against five local gang members that had extorted them. 16 It is telling that extortion was initially resisted by resorting to the law, and that residents were willing to file a complaint. It is fair to say that in subsequent years the targeted use of violence against those not willing to pay extortion money instilled a deep fear. Interviewees all refer to the years between 2006-2008 as the period when violence increased, especially against respected prominent local businessmen who had apparently resisted paying. This wave of violence had a profound impact on the social fabric and feelings of insecurity. People felt insecure in their own town and stayed home after 6 or 7pm. The use of selective violence had a clear message to the rest of the community: 'Those who did not want to pay were killed. You see what happened to him'. 17 This has also led to an atmosphere where people do not talk about what they see and hear to the police or to each other. Thus, the selective and visible use of violence seems to have had a demonstration effect on the broader community with a view to make sure that residents hear, see, and remain silent (and pay).
Some of the interviewees reported that they had been extorted themselves, while most knew persons that were paying, including many ranchers, the majority of the businesses in town, and the drivers of buses and moto taxis. As will be discussed in the next section, it may be possible to negotiate the amount of extortion paid, but the threat and use of violence is always looming. While there were reports of local residents that had been forced to pay (usually larger) one-off payments, many businesses -such as ranchers, local shops, buses and moto taxis -paid extortion on a regular basis. And although all people paid extortion money because they were forced to, some did argue that in the absence of a viable police presence, paying gangs money made them 'feel safe'. This also points to the ambiguous relations between gang members and residents and how residents deal in pragmatic ways with gang power. This further reinforces distrust between residents, since it is feared and rumoured that some residents can rely on support of gangs.
Apart from extortion, the gangs in Nueva Concepción are also involved in the sale of drugs, which is entirely controlled by gangs: 'If they [the clique] would notice that someone else is selling drugs, they don't accept that. Drugs are sold for the clique. In Nueva Concepción there is no man or woman who sells marihuana who isn't connected to them. Extortion and marihuana are their income'. 18 There is no data on the income of these two sources, but arguably the income from extortion is bigger, because it takes place on such a broad scale. It also implies that putting a halt to extortion (which was -supposed to be -part of the truce) has had serious consequences for the income of the gangs, and might make the sale of drugs more attractive. It should, however, be emphasized that the range of illicit criminal actors in town is broader than the gangs. While gangs control the local retail of drugs, other persons are in involved in narco trafficking and organized crime. For example, the Cártel de Texis controls a route of illegal trafficking of drugs and cattle that crosses Nueva Concepción. As indicated by Arauz et al. (2011), at the local level there are networks that involve local gang members. 19 Moreover, on various occasions police officers have been arrested for involvement in drug trafficking or other illegal activities. 20 This involvement of some of the police officers has contributed to low levels of trust in the police in Nueva Concepción.

The Salvadoran gang truce
The local truce process and the backlash that followed should be understood in the context of changing national policy responses and discussions about the 'right response' to the gang problem. This section discusses the national level truce initiative, efforts to move it to the local level, and the discussions about the impact of the truce at the local level. In March 2012, while in the high security prison of Zacatecoluca, the leaders of the two principal gangs in the country -Mara Salvatrucha and Barrio 18 -agreed to a truce. The truce was mediated by Raúl Mijango, a former guerrilla commander and former Member of Parliament, and Monsignor Fabio Colindres, the bishop who serves the military, with the explicit support of the Minister of Justice and Public Security, Munguía Payés. Initially, a large part of the Salvadoran population distrusted the truce, but during the first year, it counted on considerable political support, since it led to a significant drop in the homicide rates (Whitfield, 2013;van der Borgh & Savenije, 2016). However, by May 2013, the truce started to unravel, and when a new government took office on 1 July 2014, the truce was largely considered to have failed.
During the first year of the truce, in early 2013, the facilitators presented a plan to consolidate the process by bringing it to the municipal level. The initiative, also referred to as the second phase of the truce, was called Municipalities Free of Violence (Municipios Libres de Violencia, MLV). It aimed to bring together relevant local stakeholders (including government officials, private enterprise, civil society and gangs), to reduce violence and criminal activities, to introduce community policing, and emphasized the need for local development and government investments in socio-economic projects (Whitfield, 2013). Indeed, the MLV was an effort to develop a territorial and more comprehensive approach in dealing with gangs; the facilitators and local stakeholders involved in the MLV argued that this process went beyond a truce between gangs, but should be seen as a pact within the municipality at large and as a local peace process. 21 At the local level, the initiative did not have a fixed format, but intended -among other things -to diminish violence and extortion and to integrate gangs into society. However, this also implied that there was no agreement stipulating the cooperation between government officials and gang structures at the local level, or any indication of how the rule of law would be restored (or better installed), how gangs would integrate into society, and how that process would be monitored and by whom.
A very limited number of municipalities were selected to start the MLV, and Nueva Concepción was one of the last to become involved (van der Borgh & Savenije, 2016, p. 29-30). As a result of this, the outcomes of the MLV in Nueva Concepción were less impressive in comparison to the municipality of Ilopango, which was the first one to join the initiative, and which is generally seen as the most successful attempt, or flagship, of the MLVs. 22 Ilopango is a municipality with a high gang presence, and the MLV process there showed both the possibilities to work with gangs as well as the risks and limitations.
While not all local cliques participated in the local process, the process led to a number of development projects for two gangs, 23 such as a road construction project in the municipality (working on the tracks that were under 'their control'), and a football pitch in 'their territories'. 24 The most tangible support from the central government was the allocation of 400 places in the PATY programme directed at the reintegration of unemployed youth (Sanz, 2013). Reportedly, these places were divided over members of the two gangs (ibid). Thus, gang structures were not dismantled, but instead gang members were given the chance to participate in government programmes, while the programmes were tailored to their territorial claims.
There has been ample discussion about the effects of the truce on the power of gangs. Writing about the case of Santa Tecla (a more successful MLV), Carballo (2014) argues that gang power increased during (and as a result) of the local truce process, while Zoethout (2015Zoethout ( , 2016 asserts that in the same municipality (in the neighbourhood of El Pino) the truce led to a higher degree of state presence (in the form of community policing and local government interventions). It is quite possible that both authors are right. In the case of Santa Tecla the local government had previously implemented programmes of violence prevention, including work in the neighbourhoods with a high incidence of gangs, which also involved some coordination with gang members. 25 The MLV project led to closer working relations with gangs. However, as a local staff member of the municipality stated, the connections between local gang members and imprisoned leaders was very strong. 26 It is therefore fair to say that even in the more successful cases of pacification, gang power was not 'broken', but rather that the existing coordinating mechanisms between government officials and local gang leaders were used and further developed in the framework of the MLV. This process led to modifications in gang power, with a stronger role for the prison leaderships. At the same time government presence increased in a number of neighbourhoods where they had previously been relatively weak, but this did not come at the cost of the presence and power of gangs.

The local truce in Nueva Concepción
The local truce process in Nueva Concepción started with local gang leaders approaching the mayor: 'They called me and said they wanted to talk to me, and that I should not be afraid'. 27 The mayor had a keen interest in violence reduction, which had been a key theme in his campaign for the 2009 elections. During the first term of his government (2009)(2010)(2011)(2012), he had already implemented preventive activities, such as recreational programmes. However, the initiative to start the process came from the gangs. It remains unclear why Nueva Concepción was selected. According to one of the facilitators, an imprisoned leader of one of the cliques was particularly interested in the process and may have pushed to include Nueva Concepción. It has also been argued that the choice for MLVs was related to narco interests of gangs and that the truce would be beneficial to them (Farah, 2016). However, the local cliques that were allegedly more deeply involved in narco trafficking seemed to be less interested in the process.
The mayor asked several religious leaders to act as facilitators, which was seen as a necessity to make the process work since 'gangs tend to respect church leaders'. 28 Many declined, but a pastor of a small local evangelical church accepted. The pastor together with a member of the Municipal Board became the key facilitators of the local process. They had conversations with local gang leaders and it was decided to sign a pact during a public event, which took place on 8 May 2013 in the presence of the local facilitators as well as the national facilitators of the truce and minister Mungía Payés. 29 The main promise of the gang leaders was to reduce homicides, extortion and other forms of crime, while the local government and facilitators would support gang members to reintegrate into society and look for funding for productive projects.
Although it was emphasized that the pact was signed between the gangs and the population, and the event was attended by hundreds of people, only a small number of local actors were directly involved: local government officials, the facilitators and the gang leaders. Local acceptance of the truce remained limited. In the words of one of the facilitators: 'Many people didn't like the process and instead were of the opinion that more repression was needed'. Another resident, who sympathized with the initiative, said that 'for the great majority, talking to gangs is a big sin'. 30 The reluctance and fear of working with gang members was also a reason for some of the members of a local committee working in the field of prevention to keep a distance from the local truce process. As a result, the work of that committee practically came to a standstill. And while the initiative was supported by the local mayor, a number of members from the municipal board (all from the left-wing party FMLN party) were very critical about support of the municipality for the local truce process and opposed investments in projects related to this effort. Lastly, the national police was divided about the initiative, and in Nueva Concepción the police was not involved in the initiative, but one of the facilitators briefed the local commander on the process on a regular basis.
Well aware of the scepticism towards the initiative, the facilitators emphasized that gang leaders were given an opportunity and not the other way around. One of them assured: 'We don't hand over the municipality to the gang'. They also emphasized that the local truce did not imply impunity for those gang members who still faced criminal charges. Neither did law enforcement agencies take into account possible implications of arrests for the local truce process. For example, just a couple of months into the local truce effort, two local gang leaders involved in the local truce and a person linked to the ranfla were arrested. But the process continued and new representatives were appointed by the gang. Nevertheless, the legality of the process was a major concern to the facilitators who emphasized that they wanted to participate 'without violating the law'. In this regard, there were some concerns about the Salvadoran law that criminalized gang members and which was ambivalent about contacts of non-gang members with gang members. 31 However, the facilitators thought that the support of the minister of Public Security and Justice, who even had attended the ceremony on 8 May, was reason enough to believe that this should be no reason for concern.

A mixed record
The results of the local truce in Nueva Concepción were mixed at best. A major consequence was the marked reduction of the homicides in the municipality. According to one of the facilitators this led to a more relaxed atmosphere which also fostered economic activity in the town. Indeed, physical violence vis-à-vis residents decreased, since, as one resident noted, 'gangs are more careful that there are no homicides here'. 32 Thus, there was a serious effort of national and local gang leaders to reduce physical violence vis-à-vis residents in the town of Nueva Concepción, and the gang leaders also took measures (vis-à-vis their own members) to make this work.
Although extortion did not stop during the truce process, it diminished, while the style of extortion changed. As discussed in the previous section, extortion had become widespread in the municipality and compliance was based on profound fear for violent repercussions. While in the period of the local truce some interviewees knew of cases where people no longer had to pay their monthly extortion, the most important change regarding extortion seemed to be that the use of brute threats had become less important. As one resident put it: 'Before, the gangs asked and that was it. But they started to negotiate.' 33 Thus, during the local truce process, extortion practices continued, while negotiations about the amount that one should pay (a possibility that had always existed) became more frequent. This also meant that the possibilities increased for local religious leaders and the facilitators of the local truce to request gangs to stop certain cases of extortion. This did not, however, always lead to a deal. Moreover, the changes mostly affected the town of Nueva Concepción, and to a far lesser extent the other villages and hamlets. One resident argued (in February 2015) that he had the impression that 'they are growing and are present in more places after the [local] agreements were signed'. 34 The local truce effort led to some changes in the use of violence and the form of extortion, but did not fundamentally alter the power of the gangs. Neither did it lead to more legitimacy, since gang control was still based on fear. Many respondents, including the ones sympathizing with the truce process, acknowledged that they were afraid of 'them'. Gang members were still hanging around, and in the villages and hamlets of Nueva Concepción they, for instance, appeared as uninvited guests at parties. It can, therefore, not come as a surprise that many evaluations of the initiative were negative. While most ar-gued that it had become calmer in the town of Nueva Concepción, they were rather disappointed about the outcomes of the local truce process. 'We hoped that the truce would have the same effects as the peace process in 1992, but it wasn't like that'. 35 Another person said: 'It was too little, too late; we need a more comprehensive process'. 36 And the doubts about the legitimacy of the process remained: 'There were changes, but you have to ask if it is correct to work with people who violate the law'. 37 Thus, while there were some changes in the behaviour of the gang and the relations with the local residents, these changes were rather limited. Three reasons can explain why this happened. Firstly, as already mentioned, the local truce started at a late stage of the MLV process, and did not count on a lot of support in the municipality. The effort in Nueva Concepción remained highly contested and was limited to talks between the facilitators and gangs about topics such as the use of violence and extortion, and about the implementation of a few development projects.
Secondly, the initiative depended on the willingness of gangs to change, while support within gangs for such a process was mixed. Moreover, it seemed difficult to explain what gang life entailed after the truce, and in the villages and hamlets gang members continued doing what they had always done, only modifying some of their practices. The pressure from the higher levels of the gang, which were communicated through a group of non-imprisoned leaders that had close connections with gangs (the ranfla that cooperated with the facilitators) was crucial to the process. 38 These persons intervened when there were issues with boys who disturbed the process. While this led to diminished violence vis-à-vis residents, this did not mean that local homeboys accepted the truce, and it did not lead to serious changes in the gang structure and practices of extortion.
Thirdly, the task at hand was too big for the local government and the local facilitators without support and commitment from the central government. It was clear that, in order for gang members to stop extortion practices (and the related threats and use of violence), alternative sources of income were crucial. The efforts to start projects for these youths largely failed, mainly due to a lack of funding. There had been promises of government funding for the MLVs, but these never materialized (van der Borgh & Savenije, 2016, pp. 28-30). The municipality made a small contribution to a local project for youth with no criminal charges, but this was a small amount of money (and highly contested within the board). There were only a few projects in which at best a few dozen gang members were involved. Many gang members were disappointed and complained that they had received nothing. But the few projects in Nueva Concepción also showed the challenges of socio-economic integration of gang members. For example, one interviewee argued that gang members did not want scholarships -'they want the fast money'. 39 The transition towards a different way of life seemed to require a far greater effort and investment, as well as a stronger presence of specialized agencies.

The backlash
The local truce process in Nueva Concepción started at a late stage when national support for the truce quickly eroded, and the extremely limited support from the national level deeply affected the initiative. Contact between local facilitators and gang members continued until August 2015, but it ended at the moment that the Constitutional Court ruled that gangs are terrorist organizations, and labelled not only gang members terrorists, but also 'collaborators'. 40 This raised serious questions about the legality of the task of the facilitators, and marked the end of the truce process in Nueva Concepción.
By 2015 repressive policies had gained the upper hand again. 41 The involvement of military personnel in public security tasks increased. There are strong indications that police violence against gang members has increased and led to summary executions (Valencia, 2016). Moreover, in March 2016 a number of new 'special measures' (medidas especiales) were announced, which cut off contact of gang leaders in seven major prisons with the gang members in the streets. 42 This was a conscious effort to break the influence of imprisoned gang leaders (Valencia Caravantes et al., 2016). Thus, in El Salvador a shift (back) took place from an unconventional and highly criticised 'accommodating' approach to a 'repressive' approach.
The repressive turn in government policies was clearly felt in Nueva Concepción. Interviewees were generally quite positive about the results and noted that the situation had become calmer, gangs had become weaker, crime had diminished, monitoring by gangs had diminished, and gang members had either been killed or gone to the mountains. Especially the deployment of military who patrolled with police had a marked effect at local level, in particular in the urban area. Many interviewees said that they felt safer and had greater trust in the local police and military. The fact that police and, especially, the military were showing their authority in a way they had not done before and which frightened gang members was appreciated. 'I have more confidence in the police now. Before, the police did nothing. Even public spaces were controlled by gangs'. 43 However, youth especially and residents living in neighbourhoods with a strong gang presence reported that repression of military and police also had turned against them. 44 Respect for police and army are clearly based on their ability to instil fear and to use force against youths. As an interviewee noted: 'Before, the kids just shouted at the police, but this changed one year ago [early 2016]. Today the police and, in particular, the army are respected'. 45 People also openly talked about the police who were now 'killing the gangs'. One argued: 'They don't do that in town, but in the countryside they kill them.' 46 Interviewees supposed that this had also happened when, on New Year's Eve 2016, five gang members were killed in a battle with the Special Reaction Forces of El Salvador (FES), when gang members were throwing a party. 47 However, this also led to new safety concerns. An interviewee who owned a ranch where young gang members used to hang out, asked the boys to 'not run away' when the police arrived, as he was afraid he would get caught in the gun fire. 48 While the repressive approaches had a more disruptive effect on gang power at local level, there is little reason to believe that this approach will end gang power, as past repressive approaches have indirectly contributed to the transformation and increasing power of gangs. Even now that current repressive approaches are clearly informed by a need to control 'territories' (by a waging 'war on gangs') and 'prisons' (by installing a special prison regime), this may well lead to a new transformation of gang structures and different relations of gangs with territory and residents. Such a transformation or adaptation seemed to be underway in Nueva Concepción, where gang presence became less visible in the town itself. One resident argued that 'there is still a strong and obscure force of youths that are organized. They are still present, at parties and vigils.' 49 Moreover, practices of extortion largely continued. Repression does not necessarily stop this dimension of gang power, since extortion does not require the visible presence of gangs, nor their de facto 'territorial' control. In addition, there were strong signs that, as a result of these policies, the problems had increased in other villages in the municipality. 50 Gang members had moved from the town of Nueva Concepcion to other places, where the presence of police and military was absent or weak. Also, several religious leaders reported that the anti-terrorism legislation deprived them of their possibilities to mediate between local residents and gang members.

Concluding remarks -government responses and gang power
This paper discussed how different government strategies to deal with gang power play out at the local level. The case of Nueva Concepción shows a number of lessons that are relevant for other cases where governments are confronted with non-state armed groups with territorial power. Firstly, it is unlikely that repressive or accommodating policies will be able to break the power of illicit groups in the short term. The local truce effort attempted to promote local peace processes and to transform the power of gangs, but this did not materialise, leading instead to modifications of gang control. While repression had always been part of the national government's responses to gangs, the government policies taken from 2015 onwards come close to a war against gang members, either imprisoning or killing them. That strategy affected the strength and visibility of gangs in Nueva Concepción, and had a more disruptive effect. However, increased military and police presence does not stop extortion practices of gangs either. This capacity of gangs to adapt points at a major challenge of governments dealing with illicit groups that have built up local power positions. While accommodating approaches are likely to lead to a type of hybrid arrangement, repressive policies can lead to rearrangements of gangs depending on their ability and flexibility to adapt to challenges to their power.
Secondly, accommodating and repressive approaches can be employed simultaneously, since the local truce process did not come at the price of repressive policies. Only one month after a local pact between gangs and the community had been signed, two of the local gang leaders were arrested. While this proved a challenge to the local accommodating strategy, these measures were not criticized by the facilitators. However, the anti-terrorism legislation (which is ambiguous about the possibilities to work with gang members), and the reported increase in extrajudicial killings were clearly at odds with the accommodating approach and did successfully block the local truce process. This is a reminder that mixed approaches are possible. In this regard, there is a need to limit and to 'control' military and police repression. Accommodating efforts need to be well thought through and tailored to local practices. Despite the resistance of many residents against accommodating efforts, the case of Nueva Concepción shows that local level practices of accommodation and mediation (e.g. by religious leaders) have existed for some time.
Thirdly, in situations of high criminal violence and protracted violence, where previous government policies have not led to the desired outcomes, the blurring of formal and informal practices to deal with gangs is very likely. The local truce process seemed to blur the line between the formal and informal; as it stimulated cooperation between local governments and gangs, and gangs had a role in the efforts to contain violence. When the truce -a highly contested initiative -unravelled, repression increased and summary executions were on the rise. While these executions are not part of formal government policy, the United Nations Special Rapporteur for Extrajudicial, Summary or Arbitrary Executions, Agnes Callamard, recently reported after a visit to El Salvador that she had 'found a pattern of behaviour between security personnel that could be considered as extrajudicial killings' (Reuters, 2018). This blurring of formal and informal practices -or informalisation of policy -is likely in situations of protracted violence, where previous policies have not led to the desired outcomes.
Fourthly, it is interesting to note that while it was the intention that social prevention projects would complement the local truce, these activities actually suffered under the local truce. The local committee promoting local preventive activities disintegrated, because some of its members became involved as facilitators in the local pact. Moreover, the socio-economic projects that gang members hoped for never materialized as a result of a lack of will by the national government, and a lack of resources and capacities of the local administration. Indeed, the provision of socio-economic projects for gang members proved to be a particularly sensitive point at the local government board, as a number of local representatives rejected these kinds of investments. This points to the challenge to legitimize reintegration programmes for ex-gang members. However, in order to deal effectively with the gang phenomenon, the socioeconomic needs of gangs members will have to be taken seriously and reintegration programmes are therefore of great importance.
The case study presented in this article shows that finding an optimal mix between repressive, accommodating and preventive approaches and which combines national and local efforts will be no easy task. This is not only the result of practical limitations, such as a lack of government resources, or the capacity of non-state armed groups to resist policies or to liaise with government actors and other local actors. It is also the result of the contestations about what the 'right policy' is, and whether informal responses are deemed legitimate or not. The lack of social and political consensus about what constitutes a legitimate and an effective response to illicit groups (such as street gangs) may well be one of the key challenges in most contexts of endemic violence. * * * Chris van der Borgh is a lecturer, researcher and consultant with the Centre for Conflict Studies at Utrecht University. He specialises in issues of peace and security, political order and international intervention. Recent publications focus on the politics of security provision in Kosovo and El Salvador.