Hiding violence to deal with the state

Introduction

In 2012, two major street gangs in El Salvador, Mara Salvatrucha 13 (MS-13) and Barrio 18, struck a truce credited with reducing homicide rates by more than 50% in one year. Although the initial conditions of the truce were unclear, evidence emerged that the government’s involvement was crucial for understanding its success (Martínez & Sanz, 2012). The outcome of the truce came as a surprise because previous efforts had not achieved similar results. This puzzle was similar to the one that bewildered observers in Medellin, Colombia in 2003. After years of implementing violence reduction strategies, the city witnessed a decline in homicides in the context of the demobilization of paramilitary groups, and after one group consolidated its control over most armed actors operating in the city.

For many years, Colombia and El Salvador recorded some of the highest homicide rates in the world (UNODC, 2013: 33). It is not surprising then that observers praised the success of the Salvadoran truce, and the Organization of American States (OAS) became a supporter of its second phase. But the truce also revived long-standing debates about the effectiveness of negotiating with criminals. This article addresses one question that lies at the core of such debates: under what conditions do negotiations between, or with, criminal organizations effectively reduce violence?

Extant criminological literature provides limited explanations for sudden violence reductions like those analyzed here. In Medellin, homicide rates dropped by more than 50% in one year. In El Salvador, homicides went from an average of 14 murders a day to six once the truce between gangs and government went into effect. Most scholarly work addresses violence reduction over large time spans (Blumstein & Wallman, 2006), providing room for structural explanations, such as changes in the economy and demographics; institutional interpretations, like government reforms or policing strategies; or behavioral changes at the community level, such as availability of guns and drugs (Eck & Maguire, 2006; Neumayer, 2003; Stamatel, 2009). The literature about sudden violence reduction resulting from pacts comes mostly from the scholarship on civil wars, as there is little expectation that governments will talk with criminals. In our cases, pacts involving criminal organizations remain the only known varying factor during the time the drop in homicides occurred.

We argue that truces and pacts can reduce homicides when (a) they involve the state as an administrator of violence reduction incentives and (b) criminal groups have achieved group cohesion and leadership that facilitate territorial control and strategic dependability. These conditions allow criminal organizations to regulate violence. When the state participates in negotiations involving criminal organizations, it often places a premium on homicide reduction. This creates incentives for gangs to reduce the visibility of violence and thus strategically extract concessions from the state, but this is only possible when they have effective leadership and organizational cohesion. By considering the visibility of violence, we can better understand criminal behavior during pacts, as the manipulation of visibility can be crucial for criminal leaders to enforce pacts within their organizations and receive judicial benefits while maintaining their activities. Paradoxically, pacts that successfully reduce homicides rely on violence for disciplining transgressions and violent behavior. To advance the argument, we compare the cases of Medellin in 2003 and El Salvador in 2012, when pacts with criminal actors succeeded.

The article shows how the limitations states face while dealing with criminals put a premium on negotiations that are aimed at reducing homicides, usually in exchange for judicial benefits. As a result, criminals have incentives not necessarily to abandon violence, but to make it less visible, so their violent behavior does not appear in homicide statistics. The result is that ‘effective’ truces and pacts can, in the end, strengthen the participating criminal actors.

Truces, pacts, and peace agreements

In some Latin American countries, truces between gangs and pacts between state and criminal actors have long been used as a violence reduction strategy (Knight, 2012). In countries where criminals control substantial territory, the state engages in negotiations as an alternative to costly wars and police crackdowns (Daly, 2016). Most negotiations are carried out surreptitiously as the political costs of disclosing dialogue tables are usually high for state officials. Recently, the growth of criminal organizations has made these practices more common. In the USA, with lesser participation of authorities, gang truces in Chicago and Los Angeles were not uncommon during the 1980s and sometimes they were credited with impressive violence reduction (Cotton, 1992). Yet there is little analysis of these processes, largely because negotiating parties rarely disclose the terms of their talks.

The lack of systematic research on these processes makes it difficult to assess their potential to reduce violence and to settle debates on their appropriateness. Pacts can be diverse: some may be initiated by gang members on their own, others are mediated and promoted by state or civil society groups, and others may result from implicit arrangements of accommodation (Muggah, Carpenter & McDougal, 2013). In gang-promoted truces, the behavioral expectation is the reduction of violence inflicted on each intervening party, while in pacts between the state and criminals the expectation is the general reduction of violence produced by those criminals. We discuss both cases within the same framework because when gangs exercise control that challenges the state, as occurs in many Latin American countries, states become key stakeholders in violence reduction. State actors can benefit from the public security gains of violence reduction in a pact or truce even if the behavioral expectation is not the same. ¹ We focus on pacts that are negotiated rather than on those that emerge as implicit accommodation because the line between those and corruption is blurred. We argue that the most successful homicide reduction pacts ² are those that include criminals with internal cohesion and leadership and that directly engage the state.

Criminological literature highlights that gang truces and criminal pacts have a limited long-term impact on reducing criminality because of their inability to address deep rooted causes of violence and, gangs (Klein, 1995: 149–150). Yet there is little basis to understand the radically differing outcomes that pacts can have. The literature on peace processes and demobilization, disarmament, and reintegration (DDR) efforts can, however, provide interesting clues to the analysis.

Criminals, gangs, and rebels

We focus on criminal organizations – structured groups of individuals that profit from engaging in illegal or semi-legal activities. There are multiple types and modes of operation among criminal gangs, and thus talking about them generically can be problematic. Some criminal groups are local, while others have transnational reach; some revolve around identities while others are purely opportunistic. We narrow our focus on groups that have evolved to what John Hagedorn (2008: 9–10) calls ‘institutionalized gangs’: ‘Institutionalization means that the gang’s show goes on despite changes in leadership (killed, incarcerated, or “matured out”); it has organization complex enough to sustain multiple roles of its members (including roles for women and children); it can adapt to changing environments (police repression or civil war); it fulfills some needs of its community (economics, security, services); and it organizes a distinct outlook of its members (symbols, rituals, traditions, sometimes called a subculture).’ We analyze these gangs alongside organized criminals for theory-building. In fact, our analysis implies that truces between neighborhood gangs may have limited impact in the overall violence of a city, while negotiations with larger gangs or powerful criminals are more consequential. ³

Extending the literature on peace negotiations to criminal actors as defined above is complicated because criminal gangs and rebel groups in civil wars have different goals. The goals of illegal actors are complex, but we can assert that a gang in Medellin is different from the FARC guerrilla. While criminal organizations are mostly motivated by profit and may confront the state only to coerce it to change its enforcement practices, rebels usually have a political or ideological goal such as conquering territory or accessing state power (Krause & Miliken, 2009). It is more difficult to negotiate with criminals without clear political goals. Additionally, rebels in civil wars are more likely to benefit from ending a conflict linked to their dissolution, while criminals may benefit from reducing violence, but not from dissolution, as they are motivated by profit. ⁴ These differences severely constrain states’ ability to openly negotiate with criminals, yet they may obscure three issues. First, many criminal gangs have close interactions with state actors that can range from open confrontation to collaboration (Arias, 2013; Moncada, 2013), and have crucial consequences for criminal behavior (Duran-Martinez, 2013; Snyder & Duran-Martinez, 2009). In consequence, even if they do not have political aims, criminals may show political dynamics comparable to those of rebels. Second, negotiations with insurgents, even when ideologically motivated, are never easy; they may be opposed by the population, and they are prone to violence return and spoilers, even after deals are struck (Walter, 2002). Third, state actors can frame the purposes of armed actors as political, or alternatively criminal, to fit particular violence reduction strategies.

The literature on peace agreements emphasizes the importance of credible commitments for the initiation of negotiations. Rebels may decide to negotiate when the costs of war are too high and the likelihood of success is uncertain and when they believe their counterparts will honor the agreements. These expectations are determined by the balance of power among armed groups, within the state, and by the number of warring factions.

Fragmentation complicates compliance among combatants and makes it more difficult to control spoilers (Greenhill & Major, 2006) or extremists (Kydd & Walter, 2002). Implementation of peace agreements may fail when combatants are unable to commit to and enforce deals (Walter, 2002). Findley (2012) argues that the same balance of power can have different implications in different stages of negotiation: fragmentation among the negotiating actors can facilitate cooperation at the initial stages of a peace process, but can hinder the implementation of accords. Fragmentation can also complicate negotiations by augmenting the possibility of spoilers during bargaining and by making monitoring combatants’ behavior more difficult. By the same token, not all criminal groups are able to effectively engage the state. It is when criminal organizations conspicuously challenge state authority and only when they have achieved some internal leadership and cohesion that they can compel the state to seek alternatives to reduce violence.

Regarding the balance of power within the state, one can infer by the same logic that a state homogenously committed to a peace process is likely to provide the credible commitment necessary for criminal actors to negotiate. In fact, the literature on peace negotiations suggests that state capacity is an important determinant of their success and sustainability (Gleditsch & Ruggeri, 2010; McBride, Milante & Skaperdas, 2011). Although the definition of state capacity is contentious and supersedes the purpose of this article, the key insight is that successful peace negotiations also require commitment by all relevant state actors, otherwise combatants have no incentives to negotiate or to maintain their peaceful behavior.

A pact can thus be effective in reducing homicides when two conditions are met: first, state actors must engage with a relatively homogenous position so they can provide incentives for criminals to moderate violence. State homogeneity can be reached either by a strong leadership or by the concurrence of interinstitutional policy agendas. Second, criminal gangs must have internal cohesion and legitimate leadership, which enable them to enforce behavior among members and facilitate territorial control. Both conditions are necessary.

A pact that does not engage the state, or that engages it partially, is likely to fail because police actions and the inattention to the structural conditions that lead to violence can easily motivate gang members to re-engage in violent behavior. For example, if a deal is implemented by a government agency but without the leadership of an elected official or the compliance of police forces, it is unlikely to generate incentives for criminals to reduce violence. Likewise, a pact subscribed by organizations with limited and contested territorial control is unlikely to change the dynamics of violence in a city because the contestability of gang hegemony may generate violence. To enforce a pact successfully, organizations have to assert supremacy vis-à-vis other criminal entrepreneurs in significant portions of territory and have to be able to implement the decisions taken by the chieftains. This is more feasible when there is a legitimate leadership (Decker, Bynum & Weisel, 1998). Such status does not come easily. In most cases, recognized leaderships and cohesiveness are forged in prior conflicts, which can be easily resumed by a single spoiler. Such fragility tends to elevate incentives to consolidate internal authority and turns the state into a stakeholder of one group’s hegemony. Hence, internal cohesion is crucial to enforce individual and collective behavior of criminal actors, but it also underpins the position of criminal organizations in relation to the state. We thus expect pacts to be more successful when they include a criminal organization that is hegemonic. If there are two contesting groups, pacts can be successful only if they include both organizations and they have comparable organizational capabilities.

Criminal pacts, just like peace agreements, are complex processes that change over time (Walter, 2002). The factors that facilitate an agreement may not be the same factors that facilitate implementation, and may even become a hindrance for it (Findley, 2012). Indeed, as we have anticipated, a paradox in criminal pacts is that they are most likely to reduce violence precisely when organizations are strong and this, in turn, constitutes their very problem. Criminal actors may be further strengthened during the process but their power is crucial to ensure compliance to the pact.

There is another potential explanation for the success of pacts in reducing homicides: third-party involvement in negotiations can mitigate commitment problems and provide successful monitoring mechanisms (Walter, 2002). We do not rule out the importance of this variable in explaining the success of agreements involving criminals. However our cases show that third-party mediation cannot account for sudden homicide reductions. In El Salvador and Medellin, the rapid homicide drops occurred without any significant third-party intervention. ⁵ Likewise, prior to these experiences in Medellin and El Salvador, there were pacts that involved third-party actors like churches and NGOs, but were unsuccessful. While third parties may contribute to success beyond initial homicide reductions, we anticipate that without the two conditions mentioned above, initial homicide reductions are less likely.

Visible violence and ensuring compliance

The peace agreement literature provides key insights for our argument but it has not integrated research showing how rebels use different forms of violence (Kalyvas, 2006). Yet, the analysis of violence visibility allows us to understand criminal behavior and incentives for violence reduction during pacts. Peace agreements aim to dismantle rebel groups opposing the state, often in exchange for concessions by the state. Violence in civil wars can serve multiple purposes – from gaining civilian allegiance (Kalyvas, 2006) to solving internal collective action problems (Staniland, 2014), but if the group engaged in negotiations can discipline its members, some motivations for violence are significantly reduced after the deal. Conversely, in pacts with criminals, expectations may be more limited for two reasons. First, states are in a difficult position to accept publicly that an organization has taken hold of territory or that it is a legitimate interlocutor. As a result, pacts are likely to be presented just as a violence reduction strategy. Second, for criminals, violence is not only a tool for discipline and political strategy, but an instrument to control illegal markets and solve market disputes. Hence, as long as lucrative illegal markets exist, some violence may be necessary.

In the criminal underworld, ‘aggressive regulative rules’ dominate the relationships between individuals and organizations (Jacobs & Wright, 2006: 123). Legitimacy and status within criminal organizations are attained by using violence. Law-enforcement institutions cannot regulate violence between criminal organizations through harsher penalties because those have little deterrent effect when criminals face threats by competing organizations and street identities (Decker, 1996). States are thus left with the possibility of negotiating violence if they provide judicial benefits for criminals. As a result, pacts privilege violence reduction but not necessarily the elimination of criminal activity. If criminals agree on a violence reduction pact, they have to reconcile their commitment to reduce violence with their need to maintain criminal activity, while keeping their organization relatively intact. Reducing the visibility of violence thus appears as a viable strategy because visible violence is easier for states to monitor.

The visibility of violence refers to whether criminals expose the attacks they carry out (Duran-Martinez, 2013). Using visible violence can be very useful to show power or inspire fear, but it can also attract unwanted law enforcement. Criminals may be less likely to display violence when they receive state protection or when they fear the state’s reaction to highly visible attacks. In those situations, criminals can minimize enforcement attention by burying or destroying bodies, but without abandoning violence altogether. A pact that provides benefits in exchange for reducing violence creates incentives for criminals to manipulate the visibility of violence. Manipulating visibility can also provide a tool for the criminal leadership to bargain compliance among the troops. Leaders may not ask the troops to forget grievances or to abandon illegal activities, but to do it in a way that is not recorded by authorities. As seen in the successful pact-driven homicide reductions in Medellin and El Salvador, many killings were hidden from public view. This does not necessarily mean that there was not some real homicide decrease. But as long as the organization and its illegal activities persist, it may require some violence to solve conflicts, and this violence is likely to go underground while the pact is in place. In sum, we expect pacts to be more successful when criminal gangs are capable of concealing some violence. They achieve this capability when they consolidate a significant degree of organization and leadership.

Pacts and truces in Medellin and El Salvador

We test our argument in Medellin and El Salvador. Both locations have experimented with varied approaches to reduce chronic levels of violence, including negotiating with criminals. Our unit of analysis is a pact episode. Because not all pacts have been successful, there is within-case variation in our dependent variable – success of pacts in reducing violence – that allows us to test the argument while holding constant key variables such as socio-economic conditions. In each location there has also been variation in our variables of interest, state involvement and criminal cohesion. This variation in the relevant variables permits inferences from our cases. Two differences between our cases may potentially threaten their comparability. First, Medellin is a city while El Salvador is a country. This aspect does not invalidate the comparison because given the history of Medellin and its political importance as Colombia’s second largest city, a multiplicity of national, regional, and local state actors have always been at play as occurs in the national context of El Salvador. Second, there is a long-standing armed conflict in Colombia that may affect negotiations with criminals. Armed conflicts provide opportunities for framing actors differently depending on whether the state wants to negotiate with them. For example, the Colombian government emphasized the political nature of paramilitaries to justify their demobilization despite their engagement in criminal activities. This does not invalidate the comparison because in both countries multiple actors beyond those involved in the pacts produce violence. Although El Salvador does not have a political conflict, there are other violent actors that could not be addressed through the pact.

The analysis is based on our previous research on violence in both locations. It directly stems from interviews with gang members, state officials, police officers, NGO workers, and journalists, conducted in Medellin in October 2010 and in San Salvador in May 2012. ⁶ We also utilized interviews and information collected by the local press to contextualize the development of negotiations.

Medellin: A long history with pacts and negotiations

Medellin has a complex history where actors mixing criminal and political motivations have coexisted since the 1970s. It also has a long record of policies aimed at dealing with violence, ranging from reactive to redistributive policies addressing the root causes of crime (Moncada, 2013). At different times, local and national governments and civil society leaders have negotiated with armed actors, or brokered pacts between them (Gil, 2013), with the aim of reducing violence.

Until 2003, most negotiations had limited impact on violence, either because they included few actors within a larger and fragmented criminal world, or because they had a limited engagement with the state. In 2003, a negotiation between state and criminals led, for the first time, to a city-wide decline in homicide rates. This outcome derived from the unified participation of local and national authorities and from the consolidation of the negotiating criminal organization’s leadership. Criminal control provided resources to make a credible promise of violence reduction and to regulate visibility when violence was required.

Paramilitary consolidation and successful negotiations

Towards the late 1990s, paramilitary groups started to consolidate by co-opting small gangs and defeating guerrillas, a process greatly aided by the government’s military operations, such as Operación Orión in 2002. Until then, no armed actor had completely dominated Medellin because it was difficult to co-opt multiple armed groups as successfully as paramilitaries did. One can argue that this consolidation resulted from the militarized structure of paramilitaries and thus that what matters for pacts is the type of organization rather than its level of cohesion. However, for a long time, paramilitary groups competed in Medellin before they effectively established control.

In 1998, the paramilitary structure Bloque Metro entered the city with the aim of defeating guerrillas. A former paramilitary soldier explained: ‘self-defense groups sent a leader to meet with the visible leader of a given banda or combo […] and recruited, armed, and trained many of the members of these gangs’ (Verdad Abierta, 2011). Gangs that refused to join were eliminated. According to a community organizer, in the late 1990s ‘the bandas were crushed with blood, there were fliers distributed in convenience stores showing that there were others coming to exert social control’. ⁷ This co-optation of gangs and weakening of guerrillas strengthened paramilitaries. In 2000, a fight between two paramilitary factions, the Bloque Metro and a faction led by Diego Fernando Murillo, aka Don Berna, generated a new wave of violence that was won by Murillo. He successfully established leadership over criminal and paramilitary factions. In his own words ‘at some point we had total control over Medellin […] we coopted those groups [bandas and combos] and they became part of our organization’ (Laverde Palma, 2013). By 2003, the city had fallen under the control of one criminal structure with the ability to discipline most criminal operators.

In 2003, structures commanded by Murillo joined the demobilization of paramilitary groups initiated by President Alvaro Uribe, which generated incentives for paramilitaries to reduce violence in exchange for judicial benefits. National and local governments coalesced around the need to have a successful demobilization and, for the first time, the state provided credible incentives for violence reduction. The negotiation with paramilitaries lay at the intersection between a DDR process and a pact with a criminal organization because paramilitaries were a key player in Colombia’s conflict, but they also engaged in criminal activities. In fact, many of the 3,270 people demobilized in Medellin belonged to local combos rather than paramilitary groups (Human Rights Watch, 2010: 10). Using drug trafficking networks and gangs involved in multiple illegal activities such as extortion was key for establishing a monopoly in the city, and after demobilization these illegal activities remained strong (Rozema, 2008: 450).

The blurry line between the political and criminal dimensions of paramilitaries was a factor both in favor of and against the political feasibility of the negotiation. To legitimize the negotiation the national government emphasized the political nature of paramilitaries. However, due to the criminal engagement of paramilitaries, the negotiation was as controversial as a pact with a straightforward criminal group would be. Success depended on collaboration between government levels: the national government controlled the negotiation, but the local government controlled the reinsertion of demobilized soldiers (Alonso & Valencia, 2008). Between 2002 and 2007, homicide rates decreased by 81% and reached a historic low of 34 homicides per 100,000 inhabitants in 2007 (see Figure 1). Murillo’s leadership enabled such a reduction because it provided the necessary cohesion to regulate violence, but the decrease would have not been achieved without the judicial incentives provided within the demobilization process. The fact that areas like the departments of Córdoba and Nariño saw increases in violence after demobilizations (Nussio & Howe, forthcoming: 5; Spagat, 2008: 5) supports the idea that the demobilization alone could not account for Medellin’s decline. In Cali, where paramilitaries were not as successful in controlling local criminal groups and demobilizations were smaller, or in Bogota where local and national governments were not aligned, demobilizations had a smaller impact on homicide. In 2003, after the first demobilizations, homicides declined by 46% in Medellin while Cali saw a 15% increase. Between 2002 and 2007, homicides declined by 21% in Cali but by 2007 the homicide rate was twice that of Medellin. The demobilization succeeded where the state engaged homogeneously and criminals were cohesive.

OPEN IN VIEWER

Figure 1.

Homicide rates in Medellin, 1997–2012 (per 100,000 inhabitants)

To sustain the homicide reduction, Murillo and his structure turned to other forms of violence, punishing the violation of the rules imposed by him. If violence was necessary, it was carefully hidden, for example by throwing bodies in rivers, or burying them in mass graves. In the words of a civic leader ‘bullets were over, but they kept killing with machetes, with knives. […] Apparently everything was calm, but it was a lie’ (Verdad Abierta, 2013).

While homicides declined, disappearances and other non-lethal violence such as displacement increased. Sexual violence increased from 843 reported cases in 2003 to 1142 in 2007 (Mora, 2008). Statistics about hidden violence are difficult to obtain although according to victims’ testimonies and paramilitaries’ confessions, it is estimated that in one location alone at least 300 corpses of disappeared people were buried between 2002 and 2005 (Pressly, 2014). The proportion of homicides with blunt objects and knives increased from 4% in 2002 to 14% in 2005, possibly reflecting an attempt to divert attention because such homicides are easily associated with non-organized violence (Comisión Nacional de Reparación y Reconciliación, 2011). According to a local police commander, ‘don Berna ordered not to kill, and nobody killed. I met a kid who knew his mother’s assassin but he did not do anything because he knew he could be killed [by don Berna’s people]’. ⁸

The pacification in Medellin has been attributed to socio-economic reforms and military operations like Orión. The military operations facilitated the consolidation of paramilitary power that was crucial for the effectiveness of the pact. Unlike previous military efforts, the success in defeating guerrillas also reflected greater state cohesion. Thus, military operations were not necessarily a separate factor in the pacification. Rather, they reflected state coordination and contributed to the cohesion of the paramilitaries, the two key elements of the pact’s success. Likewise, Medellin’s urban transformation contributed to declining violence but it cannot explain the sharp decrease that started with paramilitary demobilization and ended with the breakdown in Murillo’s control.

Murillo’s control was key for the effectiveness of the demobilization in reducing violence, but it was also a risk for the process because it allowed him not only to regulate homicides but to maintain criminal activity. As evidence of Murillo’s criminal enterprise mounted, it became impossible for the government to sustain his judicial benefits. He was extradited to the United States in 2008 and convicted on drug trafficking charges, and one of the two key factors sustaining the violence reduction collapsed. Homicide rates doubled as two of his heirs fought to fill the vacancy. In 2010, they attempted a truce but, without incentives, it only lasted a few weeks and did not reduce homicides. According to a local official ‘Sebastián and Valenciano (the two heirs) had the prestige but not the ability to rein in criminality Don Berna had’. ⁹

The police eventually captured Valenciano and Sebastián and since then homicides have declined. In July 2013, two criminal factions (Urabeños and Oficina de Envigado) attempted a new truce that, according to some observers, reduced murders by 30% (McDermott, 2013). Pacts of ‘no aggression’ are still brokered by civic groups (García, 2013) but the local government denies participation. Violence reduction may be temporary if negotiations with criminal actors engage only a fraction of them or a segment of the state. Peace may thus be vulnerable to changes in criminal structures.

The Salvadoran history of violence and gangs

El Salvador has also a complex history of violence. From 1980 to 1992, the country endured a civil war that left more than 70,000 people dead and thousands of Salvadorans displaced and maimed. The end of the war in 1992 did not stop the violence. Following the peace accords, the country experienced an exponential increase in violence to more than 100 homicides per 100,000 inhabitants (Cruz & González, 1997: 956). The surge was attributed to the aftermath of civil war as it yielded thousands of unemployed combatants, wide circulation of weapons, and social trauma (Cruz & González, 1997: 959–960).

Street gangs, locally called maras, became visible in the early 1990s when hundreds of migrant youths returned to Central America from Southern California. These groups constitute a network of cliques associated with the identity franchises of street gangs from Southern California, but whose behaviors were autonomous from the US syndicates (Cruz, 2010: 382). These transnational gangs are actively present in urban centers in Canada, United States, Mexico, Guatemala, El Salvador, and Honduras. There are two separate networks operating in El Salvador: MS-13 and Barrio 18 (Savenije, 2009: 95).

According to the Salvadoran government, by 2013 there were nearly 29,000 mara members in El Salvador, organized in nearly 2,000 different cliques (Santos, 2013). Police figures indicated that 90% of the extortions were carried out by either MS-13 or Barrio 18 (Diario La Página, 2013). Their extractive criminal activities cover the national territory with leaders managing operations from prisons. Maras extort money from convenience stores, transport unions, and small business owners. According to the National Board of Small Business, 70% of small companies are extorted by these gangs (Vasquez, 2013), while one of the main transport unions in the country reported that in 2013 its affiliates had disbursed approximately US$36 million to maras in ‘security taxes’ (Ramirez, 2013). Extortions constitute the main source of revenue for gang affiliates and their associates and families (Cruz, 2010: 382). In 2007, Demoscopia (2007: 57) estimated that a single gang member was able to collect around US$1,250 weekly.

In the early 2000s, the Salvadoran government embarked in crackdowns on street gangs called the mano dura programs. These policies contributed to a surge of crime. Murder rates went from 37 in 2001 to 56 per 100,000 in 2006, and youth gangs morphed into more organized crime-type groups (Cruz, 2011: 151–152). By 2011, the government estimated that gangs were responsible for up to 90% of national murders (Whitfield, 2013: 10). When the truce between the two major gangs and the state was being negotiated in late 2011, the per capita murder rate had reached nearly 70 per 100,000 (UNODC, 2013: 126).

Since 2000, attempts had been made to strike some deals between gangs and the government (Whitfield, 2013: 9). In several cases, ceasefires held temporarily at the community level, but truce-committed gang members always faced the prospect of bumping into an enemy not involved in the pact. ¹⁰ The enactment of the mano dura policies wrecked many truces during the 2000s and without state support, pacts collapsed. Government crackdowns pushed several gang members away from community prevention programs and stirred new conflicts (Savenije, 2009: 145–146). The massive imprisonment of youth gangs provided gangs with opportunities for reorganization as national syndicates. Simultaneously, crackdowns on maras increased the cohesiveness of cliques around gang franchises. Despite the diversity of interests of barrio cliques, MS-13 and Barrio 18 became national organizations (Cruz, 2010: 391).

The gang truce

In March 2012, the Salvadoran Minister for Public Security announced that the main gang organizations had called a truce. The statement came as a late admission after a news website revealed that the government had secretly negotiated the reduction of homicides with gang leaders. The pact between leaders of MS-13 and Barrio 18 led to a drastic reduction of homicides. The violence drop was in exchange for the government’s decision to transfer gang leaders from a maximum security prison to regular penitentiaries (Martínez et al., 2012). Allegedly, the bargain included an offer from government officials to disburse thousands of dollars to gang leaders. Although authorities initially denied such negotiations, the Security Minister finally conceded publicly that the truce resulted from a government strategy to cut murder rates (Martínez & Sanz, 2012).

Two facilitators appointed by the Minister of Security carried out negotiations. ¹¹ Although the President publicly refuted any role of his administration in the bargaining proceedings, the Minister of Security confirmed that approval for negotiations came directly from the Salvadoran presidency. The talks yielded initial results with the prison transfer of 30 gang leaders. The movement was followed by an extraordinary drop in the national homicide count. Daily homicide averages were nearly 14 during the weeks preceding the transfer; in the days after the relocation of gang bosses, the average homicide count dropped to almost six (Martínez & Sanz, 2012).

A series of events facilitated the negotiations leading to the truce. First, the confinement of national bosses of gangs in a single facility enabled exchanges between the main organizations years before government negotiations. Those intergang talks allowed gang organizations to decree curfews to paralyze the national transportation system and put pressure on the government. Second, the government’s effort to retake control over the prison system isolated gang leadership. Thus, gang leaders envisaged truce negotiations not only as a means to persuade the government to ease off prison conditions, but also as a way to restore their power. Third, the appointment of the former defense minister as security minister transformed the interaction between government and gangs. As a former military general, the minister was able to bank on political capital and dodge allegations of being soft on criminals. Finally, the increase of gang violence in the months following the appointment of the security minister precipitated a reconsideration of violence reduction strategies. After the failure of previous mano dura policies, authorities viewed dialogue with gangs as the only feasible alternative (Martínez & Sanz, 2012).

Following the truce, the landscape of violence dramatically changed. As shown in Figure 2, by December 2012, the homicide rate dropped to 41.2 per 100,000 inhabitants: a 43% decrease (Valencia, 2015). The negotiations between the government and gangs resulted in the most dramatic murder reduction since the civil war. However, the impact of the ceasefire on other crimes turned out to be less dramatic and more controversial. In particular, statistics on extortion did not show a significant decline; rather, transport unions and other trade organizations reported a significant rise in the ransom monies demanded by gangs (Ramirez, 2013).

OPEN IN VIEWER

Figure 2.

Homicide rates in El Salvador, 2000–14 (per 100,000 inhabitants)

The truce’s success in reducing homicides garnered attention from the international community. Gang leaders requested that the government allow the participation of the OAS as verifiers of a new phase of talks between the two gangs and the Salvadoran government (Organización de los Estados Americanos, 2012). Also, a list of demands was published to maintain and expand the truce. They included suspension of police activity in gang territories, repeal of anti-gang laws, elimination of social cleansing groups, eradication of torture by security forces, improvement of prison conditions, pardon of convicted gang members in retirement age, and establishment of social reinsertion plans for gang members (Sanz & Martínez, 2012).

The OAS proposed steps to advance the pacification effort. By late 2012, government negotiators proposed the gradual institutionalization of the truce in specific territories through the establishment of ‘Sanctuary Municipalities’ (Municipios Santuario). These municipalities would be violence-free zones where authorities and NGOs would promote violence prevention and employment programs for youth. Two weeks later, gang leaders released a statement embracing this proposal and announced that they would carry out the plan (Valencia, 2012). However, the implementation of the peace zones was riddled with problems. Community organizers faced inconsistent support from government institutions, while some gang members continued participating in criminal activities.

The paradox of homicide reduction while gangs continued participating in criminal activities reveals one of the main points of our argument. At the center of the negotiations was the perceptible homicide reduction, not disbanding gang organizations or cessation of criminal activities. The truce hinged on the visibility of gang violence. The regulation of homicides and the manipulation of violence visibility allowed gang leaders to bargain with the government and to maneuver the inner workings of the truce. In a scenario where the government was incapable of stopping youth gangs from murdering each other, gangs could only enforce the ceasefire by resorting to violence. In practice, this contradiction was resolved not by moderating the threat of violence, but by reducing its public perceptibility. This feature was made clear to one of the authors in an interview with two Barrio 18 members. Asked about how gangs enforced the truce, they responded: ‘If somebody violates the truce, we not only go after him, but also after his relatives […] and we do not kill them: we “disappear them”.’ ¹² More important than reducing violence was concealing homicides. The eventual discovery of mass graves across the country seems to provide support for that statement (Dudley, 2014). Moreover, one year after the truce, the police reported that the number of disappeared people had risen from 545 in 2012 to 1,070 in 2013, a 96% increase (Rauda Zablah, 2013).

The visibility of violence was crucial for the government. The reduction of killings granted authorities political capital as they were able to claim success in the fight against crime, even as other felonies remained high and syndicates refused demobilization. For some time, government spokespersons were able to promote the truce as an effective model to bring peace and security (Martínez & Sanz, 2012).

In May 2013, the truce suffered a major setback when the Salvadoran Supreme Court ruled that the appointment of the public security minister – and the government’s architect of the truce plan – 18 months earlier was unconstitutional because the Constitution banned military officers from holding any position in citizen security institutions. Although the general was later reappointed as national defense minister, his removal from the citizen security office broke up the government’s homogeneous policy and placed the truce in uncertainty. Soon after, the new minister of security took a critical posture against the truce asserting that negotiations were not a state strategy, as it had allowed gangs to consolidate their criminal activities (Castro Fagoaga, 2013). By early 2014, murders per month were back at nearly pre-truce levels as the pact had already collapsed, and early projections indicated that gang activities would push homicide rates to more than 90 per 100,000 in 2015 (The Guardian, 2015).

Conclusion

Criminal violence is a major challenge in several Latin American countries. Persistently high homicide rates in countries like Brazil, Colombia, El Salvador, Honduras, Mexico, and Venezuela have led observers to advocate for alternative responses to violence. In 2012, El Salvador presented a model of abrupt violence reduction as a result of the government’s gang negotiation strategy. This success spurred discussions among policymakers on the effectiveness of negotiations with criminal groups.

By comparing El Salvador and Medellin, we found that criminal pacts can produce swift violence reduction when (a) the state or its representatives are active participants of the deal by providing tangible incentives to criminal groups and (b) criminal organizations achieve organizational cohesion and leadership that facilitate effective territorial control. These conditions allow gangs to regulate violence and its visibility. In our cases, pacts collapsed and visible homicidal violence resumed once criminal organizations lost cohesion through the dismantling of their leadership (Medellin) or the government’s response lost its consistency through the dismissal of the architect of negotiations (El Salvador).

Our findings show conditions that facilitate the success of pacts in reducing homicide but do not imply that negotiations with gangs and criminals are the solution to increasing violence in the region. Rather, they highlight how criminals can manipulate the visibility of violence rather than reduce violence altogether, and how governments can benefit from such behavioral changes that effectively show up in statistics even if other forms of violence persist. They also highlight the importance of considering the self-sustaining nature of pacts and the paradox that pacts can strengthen criminal actors. Thus, it is crucial to judge the success of pacts beyond murders; otherwise pacts are prone to be short-lived and highly vulnerable to changes in criminal leadership or to public opposition when evidence of persistent criminal activity accumulates.

Our findings may apply to places where criminal groups generate significant violence, are connected to social and political structures, and challenge the state’s territorial power. Interrupting cumulative cycles of violence can be a sensible relief strategy to allow development of comprehensive security policies and reconstruct social, economic, and political networks. Truces and pacts can be viable strategies, but their dynamics and implications have to be analyzed systematically to avoid potential risks. We suggested one possibility to expand such analysis. More research is required to test our findings in other contexts, analyze the conditions under which pacts become politically feasible, explore the role of actors such as elites and civil society, and analyze additional dimensions of success such as sustainability, effective demobilization of gangs, and reduction in different forms of violence.

Finally, we believe that the lessons drawn from the analysis of the visibility of violence can be applied to civil wars, especially those where illegal markets play a central role, as they highlight the possibility that violence can be manipulated to enforce negotiations.