Explaining Patterns of Urban Violence in Medellin, Colombia

Latin America is one of the world's most violent regions, with 40 of the 50 most violent cities, but with only 8% of the world's population, and a staggering 33% of global homicides. At the forefront of these high levels of violence are gangs that are more flexible and persistent than previously thought. This paper provides a discussion on gangs in one Latin American city, Medellin, Colombia, where different non-state groups have contributed to changing patterns of homicide rates. The paper presents preliminary findings to show how, despite the city experiencing a 90% reduction in homicide rates in less than 25 years, violent non-state groups have become embedded as part and product of their environment, acting as coherent, logical and functional players, linked to the structural inequalities and institutional fragility of the larger society.


Introduction
Latin America is one of the most world's most violent regions, with 40 of the 50 most violent cities [1], but with only 8% of the world's population, and a staggering 33% of global homicides [2], a situation that has been described as a "murder epidemic" [3].At the forefront of these high levels of violence are gangs ( [4], p. 1) that are more flexible and persistent than previously thought ( [5], p. 409).
This paper provides a discussion on the specific case of Medellin, Colombia, examining the complexities of the different non-state violent groups (gangs linked to narco-trafficking, militia groups and paramilitaries) in this environment and the relationship they have with their host communities.Preliminary findings are presented to show how these groups have become embedded as part and product of their environment, acting as coherent, logical and functional players linked to the structural inequalities and institutional fragility of the larger society.The paper begins with a discussion of the emergence of violence in Latin American cities, showing how levels of urban poverty and inequalities are linked to the presence of gangs in these environments.Following this, a discussion is provided on gangs in the Northern Triangle (Honduras, Guatemala and El Salvador) and Rio de Janeiro, Brazil, highlighting the type of relationship the groups have with their host communities and the wider society.

Rapid Urbanisation and the Emergence of Violence
The world is currently undergoing the largest wave of urban growth in history.In the early 1800s, approximately 3% of the world's population lived in cities and two out of every ten people lived in an urban area ( [6], p. 1).In 2011, there were 480 cities with populations exceeding one million as compared to 80 cities in 1950 ( [6], p. vi).It is widely believed that rapid urbanisation and the marginalisation of specific groups is one of the contributing factors to a rise in organised forms of urban violence involving gangs ( [6], p. 2).The nature and amount of this violence vary considerably across countries and regions, according to different contexts, such as their history, social structure and state-society relations.
There is no agreed definition of urban violence as it is a "dynamic and evolving phenomenon" that often features "interconnected forms of insecurity" and encompasses physical and psychological harm against persons with its manifestations ranging from homicide to other forms of victimization ( [6], p. 20).Urban violence can involve crime against property, burglary, mugging, murder, assaults, rape and organised crime linked to drugs or political activities ( [7], p. 7).While the direct costs of urban violence are normally measured through economic indicators, such as the cost of the impact of the crimes on individuals and society, many of the indirect costs of urban violence are intangible, but still have dramatic effects on people's wellbeing in terms of their livelihood, eroding both financial and social capital ( [8], pp.[8][9]. While there is "nothing certain" about the onset of violence in an urban environment, rapid urbanisation and urban population density appear to have a strong relationship with high homicide rates ( [6], p. 15).However, there is evidence of rapid growth in cities with comparatively less impact on levels of urban violence such as Beijing, Tokyo, Dhaka and Mumbai ( [6], pp.14, 38; [9], p. 15).Nevertheless, it is widely believed that the lack of ability and preparation of national and subnational governments to absorb rapid urban population growth is a contributing factor to a rise in organised forms of urban violence ( [6], p. 2) such as in Colombia and Brazil ( [9], p. 46).Additionally, if state institutions are weak or lack the ability to exert control over a city, they can be considered "illegitimate" by citizens and different non-state informal institutions may emerge to fill the "institutional power vacuum" ( [5], p. 169; [6], p. 49), resulting in state institutions sharing control with these informal institutions ( [10], p. 4).
Furthermore, as cities have experienced rapid growth, levels of urban poverty and inequality have risen [11,12].These inequalities are now seen as "one of the most important variables" in explaining increasing levels of urban violence ( [6], p. 45), especially as they are associated with the growth of gangs in Latin American cities ( [13], p. 1).These gangs are one of the most obvious manifestations of high levels of urban violence.

Gangs in Latin American Cities
There is no unified definition of a gang.There is a consensus in conflict literature that in contrast to other violent non-state actors, gangs do not seek to overthrow the state ( [14], p. 369).Gangs typically use systematic and purposeful violence, such as homicides, to maintain their control over territory, people and economic activities [14,15].Other characteristics indicate they are not fighting "for" anything but themselves, with their behaviour patterns suggesting they are "clearly motivated principally by their own interests, rather than the active promotion of any form of collective good" ( [15], p. 231).The behaviour patterns of the gang members suggest they are more concerned with survival than specific political changes.
At the forefront of the high levels of violence in Latin America are gangs ( [4], p. 1).The motivations for individuals joining these groups are not "clear cut".Members are often young, poor, uneducated males who see their membership as a source of power, prestige and profit ( [6], p. 47).From the gangs, individuals who are marginalised from society can obtain a type of acknowledged social identity from their membership ( [16], p. 28), which they would otherwise not have from the wider society.In situations of entrenched institutional and social marginalisation, the gang is often a key institution and part of how residents "endow" their world with form and meaning [17].
Research on urban Latin American gangs portrays members as poor, disenfranchised young males, living in the urban margins, who are born into chronic inequality, elitism and exclusion [4].In Latin American cities "inequality and exclusion are woven into the fabric of the region" [4] and economic changes may only "exacerbate existing inequalities and create new ones" instead of improving conditions ( [5], p. 407).In these environments, youth access to employment can often become a matter of "dead-end, low-wage jobs" [17].This results in a threat to the identities of these young males who are consequently viewed as "being a loser, of being nothing" ( [18], p. 79).They may then turn to "alternative identities" (such as gangs) for a sense of meaning and belonging.Therefore, it is important not to view gang members in Latin American cities as "irrational criminal actors".Rather, they can be seen as individuals reacting against the wider exclusion with their practices of purposeful and selective violence being interpreted as linked to their social and economic exclusion ( [5], p. 408).As a result, gang members in Latin American cities normally have a high number of risk factors.They are often young, unemployed, marginalised males, with their larger social environments often consisting of structural inequalities and institutional fragility.
Currently the majority of empirical research on gangs in Latin America focuses on the Northern Triangle (Honduras, Guatemala and El Salvador) and Brazil ( [4], p. 2).In these environments, there appears to be a "complex and opaque relationship" between organised crime and gangs, however, because of the lack of reliable data, this is often "assumed rather than demonstrated" ( [4], p. 2).Furthermore, there is no consensus as to what exactly constitutes "organised crime", with the United Nations Convention against Transnational Organised Crime [19] providing a useful definition of an "organised criminal group."This is defined as a structured group of three or more persons, existing for a period of time and acting in concert with the aim of committing one of more serious crimes or offences . . . in order to obtain, directly or indirectly, a financial or other material benefit.The transnational dimension is then applied when operations extend beyond borders.This definition, whilst broad, attempts to capture the essence of the phenomenon, being the seriousness of the offence and their profit-driven nature and allows it be applied in wide context.
In the Northern Triangle, gangs are comprised of pandlillas and maras and organised crime networks are mainly controlled by transnational criminal organisations, such as the Zetas and the Sinaloa Federation [20].However, reliable information about gangs is "extremely scanty" with issues of under-reporting, deficient data collection, issues of political interference and many sensationalist media accounts ( [4], p. 4).The broad distinction between panadillas and maras is that pandillas are more localised, home-grown gangs, whereas maras have more transnational roots, with both mainly initially involved in small-scale, localised crime and delinquency, such as petty theft and muggings ( [4], p. 5).In 2012, the UNODC highlighted how these groups are often comprised of "largely uneducated, marginal youth" who "often lack basic skills such as functional literacy, the ability to drive a car and are not skilled in the use of weapons and are unpredictable".They often engage in "periodic demonstrations" of violence to maintain control of their territory, resulting in drawing "unwanted attention" from local authorities ( [20], p. 29).Members of these groups often have "a particular vocabulary, signs, tattoos and initiation rituals" which means more than often their identity is known throughout the community ( [20], p. 29).Furthermore, because of their "periodic use of violence" they "generally do not engage in any pretense of being community servants" and because of their unpredictable behaviour, they play a very "little role in transnational cocaine trafficking" ( [20], p. 7).More recently Rodgers and Baird [4] discussed how there is "clear evidence that both pandillas and maras are becoming more and more involved in drug trafficking and dealing over the past 15 years" particularly because of their reaction to "Hard Hand" (Mano Dura) policies in the early 2000s.However, due to limited reliable data this relationship is yet to be fully understood.
In contrast to the maras and pandillas in the Northern Triangle, gangs in Rio de Janeiro, Brazil since at least 1982 ( [21], p. 33) have had a strong connection to organised crime networks.Gangs in this city initially took advantage of "widespread distrust of the police" and the abandonment of the favelas by state welfare services and became "interwoven into the social fabric of their local communities" ( [22], pp.59-61).For example, they provided "law and order", "protection" and "social welfare services" for their host communities, such as financing child day-care facilities and giving handouts for cooking gas [21,23], resulting in them having a "Robin Hood mystique" around them ( [24], p. 63).More recently, these gangs have been known to organise "parties and community social events" (Interview in [25], p. 31) and provide certain welfare services, such as "organising with residents associations the construction of local infrastructure" ( [25], p. 31), acts which are perceived by the community as a "public good and act of benevolence" (Interview in [25], p. 31).Arias [26] argued how these types of services are provided by these actors, not because of a lack of state presence but rather, because the presence of the state does not meet the "most basic needs of the population", particularly in regards to guarantees of security, welfare and dispute resolution ( [27], p. 24).
The discussion on the Northern Triangle and Rio de Janeiro suggests a relationship between gangs linked to organised crime and their host communities, a phenomenon recently described by Rodgers and Baird as a type of "localised form of sovereign power" [4].Currently, there is no holistic approach to understand why gangs linked to organized crime appear to provide services in lieu of state institutions.Previously, Kruijt and Koonings [27] highlighted how the "historical state withdrawal of poor communities" (see [28]) in Latin American cities left "governance voids" which the gang ultimately filled.However, Arias ([26], p. 2) argues that these groups are not constructed in the absence of the state, but rather in coalition with corrupt elements of state institutions, which "give gangs a degree of localised power to fight off state agents trying to repress crime".Current research appears to show gangs linked to organised crime having a "strong and highly organised" relationship ( [4], p. 13) with their host community, enabling these actors to become "embedded as part and product of their environment" and not separate from it [18].
Attention now turns to the city of Medellin, Colombia, which has transformed from being the "most dangerous city in the world" to having over a 90% reduction in homicide rates in less than 25 years.A significant feature of Medellin has been the different types of non-state violent groups (gangs linked to narco-trafficking, militia groups and paramilitaries) that have contributed to changing patterns of homicide rates in the city and the relationship these groups have with their host communities.

Methodology
The measurement of violence across contexts is problematic because of not only differences in definitions of forms of violence but also because of inconsistencies in reporting.The most widely used indicator for measuring levels of urban violence is homicide rates, defined as the intentional killing of a person ( [9], p. 13), however these statistics are notoriously unreliable due to under-reporting, difficulties in interpretation and lack of reliability of data ( [8], p. 7).These quantitative methodologies also fail to capture how people experience violence, therefore there is an increased use of complementary qualitative methodologies, which give a voice to people's perception of violence ( [8], p. 7), looking at levels of insecurity, fear and victimization rates ( [9], p. 13).This paper therefore draws from both qualitative and quantitative sources to discuss patterns of violence in Medellin.
Fieldwork for this study was carried out in 2014 and 2015 in Medellin.During this time, over 18 semi-structured interviews were conducted with government officials, former government officials, academics, business leaders and members of community and civil society organisations who had valuable knowledge about the emergence and decline of violence in Medellin.The semi-structured interviews allowed a number of topic-related questions about the patterns of violence in Medellin and pre-determined supplementary questions to be asked.
Participants for this research were found using two non-probability techniques, purposive sampling and snowball sampling.In purposive sampling, samples were drawn strategically from actors who were actively involved or have valuable knowledge about the reduction of violence in Medellin.The snowball sample technique was used to complement the purposive sampling, by interviewing actors recommended from the purposive sample.This method is the "most effective" in accessing hidden and/or hard to reach populations [29] and can be "especially useful when the aim of the study is explorative, qualitative or descriptive" [30].Sampling bias was avoided as the original sample was taken from an array of actors [31].
In addition to interviews, uncounted hours were spent in "participant observation", including public security seminars, public debates, national crime conferences, security discussions with business leaders, academic debates, a university course and a focus group with local researchers.Data on homicides rates in Medellin are taken from the government organization, System of Information for Security and Cohabitation (Sistema de Información para la Seguridad y la Convivencia) (SISC).This organisation analyses data on violence in the city collected from the Colombian National Institute of Medicine, Legal and Forensic Sciences (Legal y Ciencias Forenses Instituto Nacional de Medicina) (INML), National Police of Colombia, Colombian Criminal Investigation (Seccional de Investigación Criminal) (SIJIN) and the Colombian Technical Investigation Team (Cuerpo Técnico de Investigación) (CTI).

From the Most Dangerous City in the World to the Most Innovative City in the World
Colombia (see Figure 1) has regularly been in the top three most violent countries of Latin America and in 2015, according to the Global Peace Index, was the 11th most violent country in the world [32].Colombia has over 5.7 million internally displaced people, therefore making it second only to Sudan in terms of population displacement [33].In 2013, more than 75% of its population lived in urban environments [34].

From the Most Dangerous City in the World to the Most Innovative City in the World
Colombia (see Figure 1) has regularly been in the top three most violent countries of Latin America and in 2015, according to the Global Peace Index, was the 11th most violent country in the world [32].Colombia has over 5.7 million internally displaced people, therefore making it second only to Sudan in terms of population displacement [33].In 2013, more than 75% of its population lived in urban environments [34].The city of Medellin experienced rapid growth from 1951 when the population tripled in size in little over 20 years ( [35], p. 148).The new migrants, escaping the violence in rural areas or looking for a better economic life in the city, often arrived lacking resources to obtain housing [36] forcing them to "illegally" occupy the hillsides by purchasing dubious land titles from pirate developers ( [37], p. 7).Local political parties were not equipped to represent the interests of the growing masses of the urban poor and to provide adequate resources for their development ( [35], p. 148).Rather, the police and army units were "sent out routinely to demolish these hillside settlements" ( [38], p. 77).Furthermore, the new residents were seen as possessing "an aura that each inhabitant was, by de facto, a criminal" and a threat to urban society ( [39], p. 8).The new migrants in the slums and squatter settlements were the most vulnerable and the least secure members of Medellin's population.
Gangs had existed in Medellin since at least the 1960s.They were comprised of small numbers of individuals and had low visibility ([39], p. 115).However, by the end of the 1980s, different types of violent non-state groups had emerged in the city, all contributing significantly to the increasing homicide rates.Firstly, the number of street gangs (galladas) grew.They were comprised of young males who assembled as a reaction to their economic and social exclusion from society ([39], p. 115).Secondly, because of the overcrowding and lack of control in the lower socio-economic neighbourhoods, groups of hooded vigilantes (encapuchados) undertook "social cleansing" of drug addicts, prostitutes, criminals, homeless people and other low-status people ( [40], p. 39).Thirdly, there were small gangs of young people (chichipatos) addicted to a form of crack cocaine (bazuco) who terrorised neighbourhoods, robbing stores and homes to satisfy their addiction ( [40], p. 45).Fourthly, there were militia groups originally affiliated with the M-19 nationalist guerrilla organisation, who had initially provided a type of community organisation.Finally, Colombia had become the main cocaine exporter to the US and the epicentre of the eponymous, Medellin cartel ([41], p. 17) which used gangs (bandas) and trained assassins (sicarios) to protect and advance cartel business.
By the end of the 1980s, in certain neighbourhoods, the militia groups were perceived as the The city of Medellin experienced rapid growth from 1951 when the population tripled in size in little over 20 years ( [35], p. 148).The new migrants, escaping the violence in rural areas or looking for a better economic life in the city, often arrived lacking resources to obtain housing [36] forcing them to "illegally" occupy the hillsides by purchasing dubious land titles from pirate developers ( [37], p. 7).Local political parties were not equipped to represent the interests of the growing masses of the urban poor and to provide adequate resources for their development ( [35], p. 148).Rather, the police and army units were "sent out routinely to demolish these hillside settlements" ( [38], p. 77).Furthermore, the new residents were seen as possessing "an aura that each inhabitant was, by de facto, a criminal" and a threat to urban society ( [39], p. 8).The new migrants in the slums and squatter settlements were the most vulnerable and the least secure members of Medellin's population.
Gangs had existed in Medellin since at least the 1960s.They were comprised of small numbers of individuals and had low visibility ( [39], p. 115).However, by the end of the 1980s, different types of violent non-state groups had emerged in the city, all contributing significantly to the increasing homicide rates.Firstly, the number of street gangs (galladas) grew.They were comprised of young males who assembled as a reaction to their economic and social exclusion from society ([39], p. 115).Secondly, because of the overcrowding and lack of control in the lower socio-economic neighbourhoods, groups of hooded vigilantes (encapuchados) undertook "social cleansing" of drug addicts, prostitutes, criminals, homeless people and other low-status people ( [40], p. 39).Thirdly, there were small gangs of young people (chichipatos) addicted to a form of crack cocaine (bazuco) who terrorised neighbourhoods, robbing stores and homes to satisfy their addiction ( [40], p. 45).Fourthly, there were militia groups originally affiliated with the M-19 nationalist guerrilla organisation, who had initially provided a type of community organisation.Finally, Colombia had become the main cocaine exporter to the US and the epicentre of the eponymous, Medellin cartel ( [41], p. 17) which used gangs (bandas) and trained assassins (sicarios) to protect and advance cartel business.
By the end of the 1980s, in certain neighbourhoods, the militia groups were perceived as the legitimate authorities ( [42], p. 67), because they (unlike the state) provided a semblance of a governing body and had close ties to their communities ( [40], p. 96).Concurrently, the criminal bandas grew in size with estimates of over 150 different groups [43].The criminal bandas recruited young males from the poorer neighbourhoods who were attracted to these groups not only for the economic benefits but also because their membership provided them with a social circle and a sense of prestige in their neighbourhoods ( [44], pp.85, 140).
In 1991, because of violent conflicts between these different violent non-state groups and between them and state actors, Medellin was portrayed as the "most dangerous city in the world" with over 6500 homicides that year.Medellin's homicide rate (381 per 100,000) was four times the national homicide rate of Colombia (90 per 100,000) and almost 40 times the United Nations definition of epidemic violence of 10 per 100,000 ([45], p. 5).Homicide rates for children from 1986 to 1991 had leapt from 171 to 1021 persons ( [46], p. 30) with up to 17 young poor males being assassinated each day.At the time, this level of urban violence was only exceeded in the world by Beirut, Lebanon, which was in a civil war.
The death of the leader of the Medellin cartel, Pablo Escobar, in 1993, fuelled hopes of declining levels of violence in the city.However, the remaining criminal actors learnt to coordinate themselves in a less hierarchal way into "networks of independent nodes" making their networks less visible to "guarantee the effectiveness of their operation" ( [40], p. 168).The criminal bandas learnt from the militia the importance of providing security functions and enforcing local norms through acts such as "social micro-assistance programs" like "taking drunks home and resolving family conflicts" ( [39], pp.125, 130).The militia continued to be perceived as the legitimate authorities in their communities, but some had begun to abuse their power as regulators of social action and were attracted to the "easy money" of narco-trafficking, leading to the increasing blurring of the militia/criminal dichotomy in the city ( [41], p. 22).
By the late 1990s, a right-wing paramilitary group, the United Self-Defence Forces of Colombia (Autodefensas Unidad de Colombia) (AUC) took an interest in the city.This group was funded by narco-traffickers ( [47], p. 8) and their interest in the city was because of its strategic geographical location and the large presence of violent actors who offered "a wealth of military resources" ([41], p. 26; [48], p. 437).The AUC entered Medellin under the rural paramilitary unit, Metro Block (Bloque Metro) (BM) initially targeting communities where residents "complained" about abuses from the militia groups ( [40], p. 183).Don Berna, the leader of the criminal group Oficina de Envigado, joined the AUC, although it is unclear as to his motivations.However, his involvement did create a powerful network of paramilitary and criminal elements ( [41], p. 26).Don Berna created a new block called the Block Cacique Nutibara (Bloque Cacique Nutibara) (BCN) and informed the bandas who were already working for him that they now worked for the BCN and they used their power to eliminate most of the remaining militia groups from the city, with the exception of the neighbourhood, Comuna 13.
At the end of the 1990s, the situation had slightly improved from the crisis of 1991 with a reduction from 381 homicides per 100,000 to 167 per 100,000 in 1999 [49].However, the majority of homicide victims continued to be young males ( [38], p. 72).Approximately 8000 young males were members of the different violent non-state groups in the city (militia, paramilitaries or bandas) ( [40], p. 125).By this time, it had become increasingly difficult to categorise the different violent non-state groups in the city.Bandas were generally associated with narco-trafficking, militia groups tended to be associated with left-wing politics and paramilitary groups were generally affiliated with the AUC.However, these categories "did not do justice to the fluid nature of illicit activity" in the city ( [40], p. 340).Furthermore, Baird [13] found from 40 life-history interviews with youths who sometimes called themselves gang members and sometimes paramilitaries, that not one interviewee claimed they joined gangs for political purposes.A government official with significant knowledge on the different violent non-state groups, who was interviewed in 2014, commented on the changing patterns of the allegiances of these young males.This official indicated that they were not motivated by a particular political cause, but instead by the socio-economic benefit offered by being a member of a group: "A lot of young people who were part of the militia changed to work in narco-trafficking.There were even some young people from the ELN [the left-wing guerrillas] who became part of the BCN [the right-wing paramilitaries] . . .For example, there was a leader who worked with Don Berna [the leader of the largest criminal organisation and the BCN], who was originally from the ELN.However, he ended up working as a paramilitary leader . . .Just look at how complicated the problems of violence are in Medellin!" [50] In 2002, the newly elected President, Alvaro Uribe, commenced his administration under the premise that the country needed a new security policy, which he called "democratic security" ([51], p. 6), adopting an approach of force against the left-wing guerrillas and negotiation with the right-wing paramilitaries.The approach of force was especially noted in October 2002 when the national government sent a "strong message" to the remaining militia groups in Comuna 13 when launching Operation Orion, resulting in the deaths of civilians, state forces and militia members ( [52], p. 458).The operation was a short-term success as although the militia groups left the city, the power vacuum was soon filled by the BCN, who had "quietly provided intelligence and tactical assistance" to state forces [53].This resulted in 70% of the city coming under the control of the AUC by December 2002 ( [52], p. 461).
In 2003, negotiations started between AUC combatants and the national government, which was the first time a national government had negotiated with right-wing paramilitary groups, and a pilot demobilisation program began in Medellin.More than 850 combatants pledged to re-enter society as "peaceful civilians" ( [40], p. 190) and to receive training on legal and social issues, medical and psychological assistance and cash allowances from the local government ( [48], p. 443).This program gained national and international recognition as "an innovative model of intervention to deal with situations of complex violence" ( [54], p. 205) particularly as homicide rates had fallen to a dramatic low of 57.3 per 100,000 in 2004 [49].Furthermore, to avoid undermining this "innovative" peace process, the national government began to use the term "criminal bands" (bandas criminals) (BACRIM) to distinguish between paramilitary groups and organised crime groups in the country ( [55], p. 5).
At the time of the demobilisation program, a new independent local mayor (Sergio Fajardo) was elected and his new government begun to substantially increase public investment and extend public services into the lower socio-economic neighbourhoods under Integral Urban Projects (Proyectos Urbanos Integrales) (PUIs) which brought together transport, education, health, and security infrastructure ( [51], p. 6).The local government further developed the Participatory Planning and Budgeting Programme aiming to bring government and planning closer to the citizenry and civil society, as well as legitimating local government and promoting transparency ( [45], p. 33).These urban development policies became known as Social Urbanism (Urbanismo Social) and included initiatives that invested in infrastructure for the poorer areas of the city and iconic architectural projects to address the "historical social debt" owed to these marginalised neighbourhoods ( [45], p. 5).
Among the results of the PUIs were the creation of 125,000 square metres of public spaces.Neighbourhood parks increased from three to 17 in nine neighbourhoods, 3439 new jobs were created and 290 programs were provided in education, health, social welfare, recreation and sports ( [56], p. 77).The participatory programme enabled community kitchens and football parks to be built in the communities ( [45], p. 33).There were funds to finance 1200 scholarships for young people to pursue university studies ([57], p. 9).Finally, local communities living under violence were given "opportunities to develop a preference for legal modes of participation" ( [58], p. 152).
As shown in Figure 2, homicide rates decreased in 2005 to 35.3 per 100,000 and to 34 per 100,000 in 2007 [49], almost 90% below the rates of 1991.Studies in 2011 even indicated that the interventions in the northeast neighbourhoods directly correlated with the reduction in homicides in this area [59].Social urbanism was seen as a success ([9], p. 79), the city became known internationally as the "Medellin Miracle" [6,45,60,61] and selected as the "Most innovative city in the world" in 2013 by Citi Bank and Wall Street Journal for "the infrastructural developments perceived to have contributed to the rapid decline in violence" [62].However, in 2008, under the new mayor (Alonso Salazar) homicide rates increased to 45.6 per 100,000 and 94.4 per 100,000 in 2009, slightly decreasing to 86.3 per 100,000 in 2010 [49].One of the reasons for this increase was the extradition of Don Berna to the US.This was because for the previous three years, Don Berna had a monopoly over all criminal actors in the city ( [38], pp.85, 61) referred to locally as "Donbernability" [63].Homicide rates had remained stable, as there was no fighting over the control of territory in the city.
Following the extradition of Don Berna there was a power vacuum in the criminal world of Medellin.Violent conflicts followed, not only for a new leader of the Oficina de Envigado, but also because a new group Los Urabeños, loosely linked to the AUC, used this as an opportunity to expand their growing drug trafficking network and take over the Oficina de Envigado ([55], p. 5).Furthermore, following the demobilisation program, there was evidence to suggest some of demobilised combatants simply merged into bandas with only "a vague anti-subversive ideology" and their motivations clearly dominated by their economic interests ( [40], p. 341).With these groups in conflict, by 2011, local sources estimated there were between 100 and 300 separate groups in the city linked to the Oficina de Envigado or Los Urabeños, with the majority of the members younger than 35 years old [64].As noted by a leading ex-government official interviewed in 2014, this increase in violence showed that despite the increased investment in the poorer neighbourhoods, these groups still had a significant presence in the city.
"In this time, shooting started again in Medellin, because these criminals were left without a defined structure, without a leader.There was no one to pay their salaries."[65].
There is little formal understanding of the modus operandi of the Oficina de Envigado and Los Urabeños, because they purposefully organise themselves in independent nodes ( [55], p. 5).However, Insight Crime [66] reports how the Oficina de Envigado and Los Urabeños are linked to the larger organised transnational criminal structures of BACRIM.There are then three levels to these transnational criminal structures, all dedicated to the different tasks required in narco-trafficking.For example, in the first layer are the "Board of Directors", made up of military heads and senior drug traffickers.In the second layer are regional lieutenants, controlling geographical areas and providing drug trafficking services.The third layer then consists of "charge offices" (oficinas de cobro), which provide the workforce and most of the sub-contractors.It is in this layer that the criminal gangs of Medellin operate.These bandas 1 are hired to carry out specific tasks, such as extortion, car theft and local drug dealing.They can use the name of their parent organisation and can call upon the larger network for support if they encounter problems with other bandas or the police [55].Government However, in 2008, under the new mayor (Alonso Salazar) homicide rates increased to 45.6 per 100,000 and 94.4 per 100,000 in 2009, slightly decreasing to 86.3 per 100,000 in 2010 [49].One of the reasons for this increase was the extradition of Don Berna to the US.This was because for the previous three years, Don Berna had a monopoly over all criminal actors in the city ( [38], pp.85, 61) referred to locally as "Donbernability" [63].Homicide rates had remained stable, as there was no fighting over the control of territory in the city.
Following the extradition of Don Berna there was a power vacuum in the criminal world of Medellin.Violent conflicts followed, not only for a new leader of the Oficina de Envigado, but also because a new group Los Urabeños, loosely linked to the AUC, used this as an opportunity to expand their growing drug trafficking network and take over the Oficina de Envigado ([55], p. 5).Furthermore, following the demobilisation program, there was evidence to suggest some of demobilised combatants simply merged into bandas with only "a vague anti-subversive ideology" and their motivations clearly dominated by their economic interests ( [40], p. 341).With these groups in conflict, by 2011, local sources estimated there were between 100 and 300 separate groups in the city linked to the Oficina de Envigado or Los Urabeños, with the majority of the members younger than 35 years old [64].As noted by a leading ex-government official interviewed in 2014, this increase in violence showed that despite the increased investment in the poorer neighbourhoods, these groups still had a significant presence in the city.
"In this time, shooting started again in Medellin, because these criminals were left without a defined structure, without a leader.There was no one to pay their salaries."[65] There is little formal understanding of the modus operandi of the Oficina de Envigado and Los Urabeños, because they purposefully organise themselves in independent nodes ( [55], p. 5).However, Insight Crime [66] reports how the Oficina de Envigado and Los Urabeños are linked to the larger organised transnational criminal structures of BACRIM.There are then three levels to these transnational criminal structures, all dedicated to the different tasks required in narco-trafficking.For example, in the first layer are the "Board of Directors", made up of military heads and senior drug traffickers.In the second layer are regional lieutenants, controlling geographical areas and providing drug trafficking services.The third layer then consists of "charge offices" (oficinas de cobro), which provide the workforce and most of the sub-contractors.It is in this layer that the criminal gangs of Medellin operate.These bandas 1 are hired to carry out specific tasks, such as extortion, car theft and local drug dealing.They can use the name of their parent organisation and can call upon the larger network for support if they encounter problems with other bandas or the police [55].Government officials and members of civil society organisations groups interviewed in 2014 and 2015 commented on the connection between the bandas in Medellin and the Oficina de Envigado and Los Urabeños, highlighting both the complexity and highly organised nature of these groups: "Medellin has a network of very complex structures.About 200 different structures in the city in which between 8000-10,000 people are involved.These structures have been maintained in the city."[67] "In Medellin we could have about 350 bandas who are related to either the Urabeños or the Oficina . . .There could be about 13,500 members and that is just in Medellin."[68] "We [the government organisation] say that the organised delinquency in Medellin is not any type of delinquency, because the delinquency in Medellin is one of the most structured, where some of the most economic resources are moved, because they continue to be narco-traffickers."[69] "Selling mobile phone calls in public, street sellers in the centre of the city are all controlled by the bandas . . .Here is one example, a person steals some petrol from a pipe, but this person needs security, they call the security 'carritos' who warn you when the police are approaching.This person also needs a driver.All the people involved in this group are paid and they all need access to weapons, just in case someone else wants to take over control of this group.There are a lot of these types of groups in Medellin."[70] "The money that is made from selling a stolen cell phone is part of the larger criminal structures and this money is divided between members of the combo . . ." [71] Furthermore, one government official highlighted how the structures of these organised criminal groups have changed significantly from the modus operandi of the Medellin cartel in the early 1990s.Rather than vertically integrated and hierarchal organisations with a clearly defined command structure these groups now operate in less visible networks ( [55], p. 3).This not only shows how organised crime in Colombia is resilient but also is quick to adapt to changing conditions.Additionally, as highlighted by this official, members of these groups are not normally identified outside their host community as being a member of a banda, which is linked to the need to operate in less visible networks.
"The structures stay in the city, you can kill the boss of whatever combo and they just reorganise and more and more rise up . . .It was easier in the 1990s to identify the boss of a particular group because it was Pablo Escobar and his sicarios.But now these actors have transformed and it is more difficult to identify them . . .The only people who can identify members of a combo are the ones who live in the community, but for people who do not live in that community it is very difficult to identify them."[71] Currently the "business model" of the Oficina de Envigado and Los Urabeños is co-operation, with negotiation preferred to violence [55].The best example of this is the pacto de fusil, an agreement between the Oficina de Envigado and Los Urabeños signed in July 2013, which divided territory in the city, lifted invisible borders, enforced the "death penalty" for rule breakers ( [72]; [73], p. 146) and orders of 1 Throughout some literature on gangs in Medellin, these groups are referred to as combos or bandas, however some literature on gangs in Medellin does separate these groups as criminal bandas and delinquent combos.For reasons of clarity, this paper is only using the term banda, however, the particular term that a participant used has not been changed.The difficulty in classifying these groups shows an example of the complexities and varied interpretations of criminal groups in Medellin.
assassinating more discreetly to "make victims disappear" ( [74], p. 99).The effect of this agreement was noted immediately with a reduction in homicides from 90 (in May 2013) to 60 (in August 2013) [49].Most respondents interviewed in 2014 and 2015 mentioned this pacto de fusil with a member from a civil society organisation providing detailed information, highlighting the capacity of these groups to control homicide rates in the city: "There is a pact between the different illegal groups, groups you can call bandas or paramilitaries . . .This pact means that they can't take their territory of another group and whilst this respect is kept, there will not be confrontations . . .This pact means that no one is permitted to assassinate people and if they do, they must do it outside of their neighbourhood or outside of Medellin or dismember the body and throw them into the Medellin river.Importantly there is no authorisation to have dead people in the neighbourhoods in Medellin . . .If you need to kill anyone you need to ask permission from the armed actors . . ." [75] This pacto de fusil therefore suggests that any reduction in homicide rates in the city are "partial and fragile" ( [76], p. 12), "subject to outside factors" and that homicides continue to be "organised, selective, have specific patterns and are directed towards specific populations" [77].For example, in 2015 under the new mayor (Anibal Gaviria) the city celebrated the "lowest homicide rates in 35 years" [49].However, this celebration took place in the shadow of events such as four dismembered bodies in September 2015 [78] and a triple homicide in November 2015 with victims found with evidence of torture before they were assassinated [79].These events have been linked to the continuing presence of the criminal bandas in the city because of their use of purposeful and selective violence.During interviews conducted with government officials and civil society organisations, the fragility of the homicides in the city was constantly highlighted, with the following comments from former and current government officials typical of all participants interviewed in 2014 and 2015: "There are places that are not on the police radar and then they 'explode'.For example one neighbourhood was relaxed and then two weeks ago there were 8 homicides."[80] "In the last couple of weeks there has been an increase of homicides in an area called Castilla.This is because there is fighting between three bandas over the control of drug outlets . . .This shows that every now and then when there are violent episodes there is not an underlying calm, there is not tranquility and there is not security, there is a subtle control . . .The armed groups have control over the people in their communities, which is a very violent control.Because they are the owners of the lives of people and they decide who lives and dies on a daily basis . . .The homicides are just the tips of the icebergs."[67] Therefore, instead of using homicidal violence as means of controlling their host communities, these groups have learnt new ways of control, adopting non-lethal violence such as extortion, threats and social control to maintain their authority ( [81], pp.8-9).For example, forced urban displacement has increased "considerably" since 2008 with 5395 victims in 2014 according to Medellin Como Vamos ([82], p. 65).This type of violence consists of threatening residents if an area or building is not abandoned within a specific period.It is a method recognised as a more effective "mechanism of territorial control as it does not leave the same physical evidence as homicides" ( [56], p. 7).A respondent with knowledge on forced urban displacement interviewed in 2014 related this increase to the growing presence of the criminal bandas: "In the 2006 and 2007 Medellin Como Vamos report [which evaluates changes in quality of life from different socio-economic zones] combos were only mentioned as a small threat to insecurity.However from 2008 to 2011 combos appeared as a threat and there was a relationship between this appearance and the increase in forced urban displacement . . .The combos had not been using force and then they started to." [71] Not only are the actors learning new ways of maintaining their control, but also there is evidence they are benefitting socially and economically from organisational arrangements set up by the local government, such as participatory budgeting, where they use intimidation to force community leaders to contract their services ( [83], p. 9).Any resistance from these leaders often leads to threats of violence or assassinations.For example, a community leader known for making complaints against the actors" use of intimidation in the participatory budgeting process was shot 25 times in May 2015, the third community leader to be brutally assassinated in less than a month in 2015 [84].Such events demonstrate how these groups use selective and purposeful violence to maintain their control.
The motivations for individuals to join these criminal bandas are not well defined and there is little understanding as to how many individuals actually join these groups.However, research indicates banda membership is linked to larger structural inequalities.For example, Medellin is often regarded as one of the most unequal cities in Latin America, noted in 2014 with a Gini coefficient of 0.526 (increasing from 0.506 in 2013) ( [83], p. 7).Recent qualitative research by Baird [13] found young males from the poorer neighbourhoods were "seduced" by the banda lifestyle, because any "alternative and dignified livelihood opportunities are hindered" for these individuals.From interviews with gang members, Baird [85] found that for these individuals, there is a "perpetual lack of legitimate opportunities and social mobility" with half of the members interviewed explicitly referring to "limited opportunities borne out of poverty as a causal factor leading them to gang membership" [85].A former government official and a member of a civil society organisation highlighted this lack of legitimate opportunities for individuals in the poorer neighbourhoods: "Young people [from the poorer communities] finish school and realise they don't have access to employment opportunities.For this reason young people don't want to study anymore, there is no motivation to stay in school.They would have liked to have done something after studying, but they realise that in this society, they can't . . ." [70] "If you are 14 years old and you see 50% or more of 20 year olds in your neighbourhood unemployed and you realise that the possibility of getting into University is about 1 or 2 out of every 100, which really in your perception is zero . . .Then this lack of opportunities for many young people is what pushes them into a violence.The armed group pays you economically and you get some social recognition which you don't get from society."[86] Furthermore, the poorer neighbourhoods of Medellin continue to have high unemployment (around 20%) and underemployment rates ( [83], p. 17), which present significant obstacles of social mobility for individuals from these neighbourhoods.Research by Abello-Colak and Guameros-Meza [83] found how the criminal bandas then take advantage of this lack of access to employment by offering employment at higher salaries in the form of drug distribution, extortion, selective assassinations or trades that are part of the broader money laundering operations, such as bakers, private security and credit-loan collectors.Not only does this form of employment offer economic benefits for the members but also their role in the banda provides a sense of meaning and purpose for these marginalized individuals, which they are unable to find from the larger society because of their lack of social mobility.A government official and a member of a civil society organization interviewed in 2014 highlighted the difference in salaries offered by the criminal bandas in comparison to the "minimum salary" employment these individuals are normally offered in the formal economy.
"It is easier to be in a combo where your salary is more than 1 million pesos a week ($300 USD) rather than killing yourself working in a manual labour where they pay you a minimum salary [$195 a week].Because from a plaza de vicio [drug dealing corner], the minimum that they will make is 4 million pesos ($1200 USD)-which is five times the minimum salary!" [71] "In a banda there are some people who earn eighty thousand pesos ($240 USD) a week.Others will earn 1 million pesos ($300 USD) or 2 million pesos ($600 USD) or 5 million pesos ($1500 USD) a week.The boss of a large combo will earn up to 30 million pesos ($9000 USD).This is without counting their monthly bonuses . . .!" [68] Lastly, similar to research on gangs in Rio de Janeiro, in Medellin residents have continuous interactions with these actors in arrangements for "service provision" ([87], p. 19).They are "perceived as a legitimate actor" ( [81], p. 23), especially if residents perceive benefits from their presence ( [76], p. 11) and as "key to their daily survival" ( [83], p. 11).While these actors do not appear to provide welfare services, such as the gangs in Rio de Janeiro do, they do however provide certain security and conflict resolution services in their host communities.Residents rely on these actors to resolve issues ranging from "small quarrels to serious offences" in homes, neighbourhoods, workplaces and schools, as the formal mechanisms are perceived as "time consuming, more expensive, less effective or inaccessible" ( [83], p. 12).For example, recent research from Medellin suggests these actors take "advantage of the low collective efficacy" in communities to solve "problems of social order" [88].Furthermore, residents may prefer using these actors to the police "because of their efficiency and effectiveness in regulating and resolving security problems", despite the "host communities awareness of the negative effects of this relationship" [88].During interviews conducted with not only with academics, government officials, civil society organisations and community leaders but also during participant observation, these themes were highlighted, with the following comments typical of participants interviewed in 2014 and 2015: "The criminal actors have a social base in their communities . . .The community is either scared of these actors or they have some affection towards them . . .They either live with them or are obligated towards their control . . .They provide a large service . . .They help to regulate everything . . .One group helps me study and another group provides me with security so that I am able to go and study!" [68] "Unfortunately, because this city has a long history of illegal actors and criminal organisations, this means that, for some people, in their territory security is provided by an armed actor.These residents don't care if this security is provided by public forces, paramilitaries, militia groups or narco-trafficking groups.It could be any of these groups and the most important reason is because they provide security."[67] Currently, there is no precise approach to understanding why gangs linked to organised crime appear to "give back" to their host communities in the form of social welfare or security services instead of pursing their own short-time goals of maximising their wealth.For example, in their study on Medellin, Abello-Colak and Guarneros-Meza [86] highlighted how the gangs relationship with their host community could be linked to their "contestation and continuous interactions community members have with these groups in arrangements of service provision" rather than the absence of the state.On the other hand, Wolff [25] on studying the gangs in Rio de Janeiro used three rationalist assumptions to attempt to explain their relationship with their host community, employing elements of literature on civil war and counterinsurgency.Firstly, many political philosophers (such as Hobbes, Locke, and Weber, amongst others) agree that populations have a "preference for both social order and material welfare over disorder and destitution and communities will attempt to seek out some form of political authority to provide these benefits when they are lacking".Secondly, "criminal enterprises will seek to establish political and social control of territory in the absence of an effective authority that can mitigate the costs of illicit market competition" as noted previously with the Sicilian Mafia [89].Lastly, drawing from Stathis Kalyvas's [90] theory of irregular warfare, this relationship is a type of logic of territorial control, as criminal groups will invest in improving their relationship with their host communities in order to reduce the cost of control, which by coercion alone may be "exorbitant".These developing approaches allow somewhat of an understanding as to how gangs linked to organised crime groups may interact not only with their host community but also with the larger community and the state.However, further research is needed on this phenomenon as further understanding of this relationship will advance theoretical understandings of criminal behaviour in many ways.

Conclusions
The Medellin experience shows a reduction in the most serious type of violence, homicides, from appallingly high levels to moderate ones.However, this significant reduction does not mean the perpetrators of violence have disappeared, rather they have remained and restructured.These groups are still present in poorer neighbourhoods of Medellin where they provide certain services and are often perceived as legitimate actors.The non-state actors in Medellin are very capable of acting violently and sometimes do, however not to the extent of previous territorial conflicts in earlier years, when homicides were the preferred mode of resolving disputes.More recently, these groups have adopted practices such as the threat of violence and forced urban displacement.While homicides rates have declined enormously in the city, the structures of these violent non-state groups are still in place and capable of taking action.
While certain socioeconomic factors have improved for poor communities, these structural inequalities and fragile or disinterested state actors account for the acceptance of the embeddedness of non-formal actors as legitimate actors in Medellin's poor communities.These actors provide a type of service provision to their host community, in the form of employment and security services, comparable to the actions of gangs linked to organised crime in Rio de Janeiro.Further research is needed to fully comprehend this relationship to not only understand the changing patterns of urban violence in Medellin, but also to advance theoretical understandings of criminal behaviour on a global scale.