Abstract
Violence is commonly viewed as an inherent attribute of the drug trade. Yet, there is dramatic variation in drug violence within countries afflicted by drug trafficking. This article advances a novel framework that explains how the interaction between two critical variables, the cohesion of the state security apparatus, and the competition in the illegal market determines traffickers’ incentives to employ violence. The analysis introduces a generally overlooked dimension of violence, its visibility. Visibility refers to whether traffickers publicly expose their use of violence or claim responsibility for their attacks. Drawing on fieldwork in five cities in Colombia and Mexico (Cali, Medellin, Ciudad Juárez, Culiacán, and Tijuana), 175 interviews, and a new data set on drug violence, I argue that violence becomes visible and frequent when trafficking organizations compete and the state security apparatus is fragmented. By contrast, violence becomes less visible and less frequent when the criminal market is monopolized and the state security apparatus is cohesive.
In 2009 Mexican authorities captured Santiago Meza, a criminal known as “El Pozolero” (the Stew Maker). Upon his capture, Meza confessed that for over ten years he and his collaborators disintegrated in acid more than 300 corpses while working for drug traffickers in Tijuana. Although Meza’s capture and his methods were quickly blended into the brutality that characterized Mexico’s violence since 2006, his testimony revealed a puzzle: for years, his job was to eliminate the evidence of violence, and beginning in 2008, however, he was specifically instructed to, instead of hiding bodies, place them in the middle of a well-transited street, along with a note. Why would criminals decide to make violence visible, if doing so can attract enforcement attention that can be detrimental for their business, as occurred to Meza, who was captured shortly after exposing his work?
Violence is commonly viewed as an inherent attribute of the drug trade. Yet, as the case of El Pozolero suggests, there is dramatic variation in drug violence across countries similarly afflicted by drug trafficking. Mexico experienced an upsurge in violence despite being home to trafficking organizations for over six decades, and in Colombia violence has declined despite persistent trafficking. There is also puzzling variation within countries with similar levels of drug trafficking: even though Ciudad Juárez and Tijuana experienced spikes of violence since 2006, the spike of about 700 percent in homicide rates in Ciudad Juárez in the course of one year was far higher and longer lived than in Tijuana, where homicides increased by about 200 percent. Certainly, one does not expect identical violence trends across cities, yet differences in this magnitude raise the question, what explains such variation across major sites of drug trafficking?
This article addresses these questions by comparing five cities experiencing contrasting patterns of drug violence over the past three decades (1984-2010): Cali and Medellín in Colombia and Ciudad Juárez, Culiacán, and Tijuana in Mexico. I argue that the interaction between two critical variables—the cohesion of the state security apparatus and the amount of competition in illegal drug markets—determines the incentives and opportunities of drug traffickers to employ violence.
To characterize varied forms of violence in drug markets, and to understand the impact of, and reactions to them, this article disaggregates violence along two dimensions: frequency and visibility. The case studies will show that the frequency of violence increases as the illegal market shifts from monopolistic to competitive, and the visibility of violence increases as the state security apparatus shifts from cohesive to fragmented, because cohesive states are more likely to credibly commit to protect or, alternatively, to prosecute criminals. Criminal actors, in turn, may refrain from using visible violence that can trigger state reactions when they receive credible state protection and fear losing it, or when they believe that the state can dismantle them.
To develop this argument the first section of this article conceptualizes drug violence and highlights its complex variation. The second assesses existing explanations for drug violence. The third lays out a political economy framework that explains how state and market determine drug violence. The fourth demonstrates the argument with case studies from the five cities, and the last section concludes.
The Multiple Dimensions of Drug Violence
Drug violence, defined as lethal violence that emerges in the functioning of drug markets, can involve civilians, state officials, and criminals alike. It constitutes an increasing threat to governance and citizen security in places as diverse as Afghanistan, Bolivia, Brazil, Guinea Bissau, Honduras, and Myanmar, among others. Colombia and Mexico represent paradigmatic cases of this phenomenon as they have been at the front line of the war on drugs and suffer the deleterious consequences of violence: in Mexico, as explained by Shirk and Wallman (2015), drug violence caused at least 60,000 deaths between 2007 and 2012, 1 and in Colombia drug trafficking continues complicating a five-decade-old conflict.
These prominent examples suggest that we might expect violence in drug markets to be frequent, because in illegal markets that lack state regulation, violence is deployed strategically to gain market share, solve contractual problems (Andreas and Wallman 2009; Friman 2009), and address disciplinary and succession issues (Reuter 2009). Or we might expect violence to be a rare strategy because it can attract enforcement attention, and criminals might therefore prefer to limit its use (Reuter 1989). To understand variation in violence, we thus need to evaluate its frequency, that is, the rate at which violent events occur. However, as the case of El Pozolero suggests, the full “repertoire” of violence includes another dimension, its visibility.
Visibility refers to whether criminals publicly display the evidence and/or claim responsibility for their attacks. 2 Visibility is a function of the methods employed by killers, the number of victims per attack, the type of victims, and whether criminals claim responsibility. An intelligence official I interviewed acknowledged the importance of visibility while describing the violence deployed by current trafficking organizations in Colombia: “The BACRIM [criminal gangs] dismember bodies with machetes, of course, they are not as sophisticated as before when they used chain saws […] Now we see another type of violence, a new profile of the killer. There is more interest in hiding, but the new structures are absolutely violent.” 3 Other officers echoed this idea, observing that as a response to enforcement actions, traffickers now prefer to be “traffickers in silence.” 4
The five cities analyzed have experienced contrasting changes in the frequency and visibility of drug violence. For example, since 2006, violence in Ciudad Juárez became more visible as criminals claimed responsibility for beheadings and killed public officials in a pattern that contrasted with the less visible disappearance of people prevailing in the early 1990s. Likewise in Colombia, between 1984 and 1993, during a period known as narcoterrorism, violence in Medellín was more frequent than in Cali. Over time frequency in Medellín decreased, whereas it increased in Cali even though the large drug trafficking organizations (DTOs) of the 1980s had been dismantled in both cities. Table 1 illustrates the patterns that emerge from combining the two dimensions of violence 5 and the puzzling variation that is uncovered not only across cities but also within the same city over time with a two-dimensional assessment of violence.
Types and Cases of Drug Violence.
Existing Research on Drug Violence
The relationship between illegal markets and violence has been crucial both in the study of criminal organizations and in the study of civil wars and conflict. Illegal markets like drug trafficking are expected to fuel violence, as a result of either greed in civil wars or the increasing power and extension of organized crime. Existing research about the drug trade may overlook the variation described earlier because it often expects violence to be a natural by-product of drug trafficking. Analyses that emphasize the increasing threat of transnational organized crime (Naim 2006), that link the existence of lootable goods (like drugs) to the duration of civil wars (Fearon 2004; Ross 2004), and that include illegal profits as a distinctive component of “new wars” (Kaldor 2001, 102) tend to create a direct connection between illegal markets like drug trafficking and violence, which may obscure important variation.
There has been an increase in studies unpacking the connection between illegality and violence (Andreas and Wallman 2009; Snyder and Duran-Martinez 2009) and partially in response to violence in Mexico, there are prominent analyses explaining why drug violence becomes extreme (Astorga and Shirk 2010; Lessing 2012; Osorio 2013; Rios 2012; Rios and Shirk 2011). The present study contributes to this research by analyzing multiple violence patterns, avoiding the selection bias that can emerge when inferring causes looking only at extremely high violence, that is, truncating variation in the dependent variable.
Studies of the illicit drug trade offer potential explanations for the frequency of violence that can be labeled as (1) economic, (2) focused on international policy, and (3) focused on structural economic and social processes.
Economic approaches contend that certain aspects of the drug trade chain (e.g., distribution), certain goods (e.g., cocaine), and higher volumes of trade generate more profit and, as a result, more violence (Castillo, Mejia, and Restrepo 2013). The rationale is that profitable markets are likely to produce more violence, as they create incentives for criminals to access the market and protect their turf violently (Reuter 2009). In other explanations, the rationale is the opposite that shrinking profits spur conflict because to maintain profits criminals need to eliminate rivals (Goldstein 1997). Purely economic approaches account for crucial market dynamics yet, they are limited in explaining why violence can be associated with both high and low market prices and with both shortages and ample supplies of drugs.
Explanations focused on international policy argue that violence in drug markets is a function of the global drug prohibition regime and the US led “War on Drugs,” which has emphasized militarized responses to drug trafficking (Youngers and Rosin 2005). These explanations argue that as enforcement associated with prohibition increases, the monetary stakes, competition, and consequently the violence associated with drug markets increase too (Gootenberg 2012), as when relatively lenient US narcotics policy toward Mexico shifted in the mid-2000s, triggering militarized approaches toward traffickers (Serrano 2007). These arguments show how prohibition constitutes the backdrop against which criminalization and violence in drug markets flourish. Yet, a focus on international policy does not account for variation in violence within countries facing similar international pressure.
Finally, explanations focused on socioeconomic change link processes like globalization, rapid urbanization, and economic transformation to criminal violence. These explanations have been essential to understanding the upsurge of urban violence in Latin America in the 1990s (Kruijt and Koonings 1999; Moser and Rodgers 2005; Davis 2010) but overlook relatively nonviolent illegal markets as well as variation in violence across cities undergoing similar socioeconomic transformations.
Overall, these explanations provide crucial insights, yet they do not suffice to explain variation in drug violence across time and space. Most important, no existing analysis of drug violence employs systematically the visibility dimension.
A Political Economy Framework for Drug Violence
A political economy framework, which understands drug violence as resulting from interactions between states and illegal markets, goes beyond purely economic explanations that link profits in illegal markets to violence or that assume violence to result from market illegality alone. It also builds on, and contributes to, literature on civil wars and literature on drug trafficking and illicit markets.
Research on civil wars analyzing how changes in local balances of power between states and insurgencies shape strategies of violence (Arjona 2010; Kalyvas 2006) provides crucial insights for the argument. This work is seldom concerned with criminals, because they do not aim at controlling the state, yet a focus on local struggles for power can also explain the violent behavior of criminal organizations, which may challenge state sovereignty as much as insurgencies. By analyzing diverse interactions between states and criminals, which can range from confrontation to collaboration (Arias 2013; Lee and Thoumi 1999; Moncada 2013), my analysis can contribute to understanding violence when the state tolerates, promotes, or bargains with nonstate violent actors (Staniland 2012). The state’s active role in shaping violence dynamics has sometimes been overlooked in the civil war literature.
The literature on illicit markets illuminates how enforcement actions aimed at dismantling illegal organizations can shape illegal markets and violence (Andreas 2011; Friman 2004). Enforcement actions may dislodge organizations by killing or capturing leaders, thus creating possibly violent internal power struggles (Friman 2004; Rios 2012). This article expands the focus of this literature on individual enforcement actions by looking at the security apparatuses that underlie these actions and by unpacking violence into two dimensions.
The Logic of the Argument
In illegal markets, criminals may use violence to solve disputes given the absence of legal mediation. Violence can also signal toughness: the more violent an organization, the less likely that competitors will try to overpower it or that members of the organization will cheat or defraud. The more visible violence is, the more likely that the toughness and power of the organization will be communicated to the public, which is to the criminals’ advantage. Yet, violence also has drawbacks, such as scaring away nonviolent partners and, especially, attracting police attention (Gambetta 2009, 34).
If criminals believe the state can effectively attack them, they may prefer to hide violence to minimize the risk of being targeted and prosecuted by police action. 6 Hiding violence can thus become a tactical adaptation to avoid state attention. Criminals may also prefer less visibility if they receive state-sponsored protection that is dependent on their “peaceful” behavior (Snyder and Duran-Martinez 2009). True, state-sponsored protection can also guarantee impunity when criminals commit violent acts, but even a weak corrupt state may be forced to react to visible violence. Therefore, if criminals believe the state can protect them effectively, they lose incentives to expose violence, which could undermine that protection by obliging enforcers to respond to the potential public outrage generated by visible violence.
By contrast, if criminals do not have credible protection, or do not fear state action, they lose incentives to hide and gain incentives to signal their power and pressure the state and their rivals through visible violence. In the absence of predictable protection, refraining from using visible violence does not guarantee that criminals will be protected against enforcement. They may decide to use visible violence or simply abandon the extra steps necessary to hide it. 7
The state security apparatus determines the government’s ability to credibly enforce the law or, alternatively, to protect criminal actors. It refers to the elected authorities and enforcement agencies (police, army, and intelligence agencies) that determine security and antinarcotics policies in any location, and it can be cohesive or fragmented. Enforcement efficacy depends on the ability to coordinate enforcement actions and thus should increase as power within the security apparatus is more cohesive. Likewise, the provision of reliable protection to criminals depends on the ability of state actors to guarantee that no other state authority will enforce the law against criminals, and this is more likely within a cohesive security apparatus. Conversely, as the security apparatus fragments, coordination and successful enforcement become more difficult, and protection becomes less reliable. Fragmentation can make enforcement less effective 8 because it complicates coordination even when enforcement actions proliferate. Criminals may have more opportunities to bribe public officials, but they also face more possibilities of prosecution, and corrupt deals become costly, as they require the acquiescence of more people.
The visibility of violence thus depends on the state security apparatus. A cohesive security apparatus is likely to reduce the visibility of violence because it makes state protection more reliable or enforcement more efficient. By contrast, a fragmented security apparatus is likely to increase the visibility of violence because it makes protection less predictable or enforcement less effective.
Furthermore, I argue that the frequency of violence depends on the number of organizations competing over drug trafficking in a given city. When the illegal market is monopolistic, violence is likely to be less frequent because traffickers do not face turf disputes and may resort to violence only to discipline their members or to punish transactions gone wrong. By contrast, when the illegal market is competitive, with disputes for territory among two or more organizations, violence is likely to be more frequent as criminals try to drive out competitors not by lowering prices but by using force against them (Friman 2009; Skaperdas 2001, 187; Skaperdas and Konrad 2012). 9 Disputes for territorial control are unlikely to be solved with a single homicide, and violence may nurture retaliation that is unlikely to stop before competitors are eliminated or significantly diminished.
Paradoxically, a cohesive state more able to enforce the law can also successfully protect criminals. The results are the same (less visibility), but the mechanisms are different. A cohesive protector state deters criminals from using visible violence that could force the state to act. A cohesive nonprotector state provides a credible threat, that is, criminals know the state can harm them and they try to avoid state attention by reducing visibility. Visibility thus depends on the relation between state and criminals, but it does not mean that it is used only against the state: once the incentive to hide violence disappears, criminals may decide, depending on their interests, to use visible violence to eliminate rivals, to retaliate against or pressure the state, or to scare away civilians, rivals, and enforcers.
It is crucial to note that criminals have to consider both frequency and visibility when making decisions about violent strategies. In other words, if the market is competitive and disputes emerge, criminals are likely to deploy violence, but they simultaneously decide, depending on the conditions of the state security apparatus, whether that violence should be publicly exposed or should rather be kept hidden.
Jointly, the security apparatus and competition in the drug market determine patterns of violence. A fragmented security apparatus creates conditions for visible violence, but such violence may not emerge if there is no market contention. Conversely, market contention can increase frequency, but if the state is cohesive, such violence may be hidden. 10 It is possible that market competition may also affect visibility and state cohesion may affect frequency. Holding market forces constant, frequency may increase if criminals take advantage of state fragmentation to redress nonmarket grievances. Likewise, holding state cohesion constant, competition can spur visibility because criminals benefit from a violent reputation when they need to eliminate rivals. However, if competition creates incentives for visibility but criminals fear enforcement or losing protection, they may prefer to hide. And if criminals take advantage of a fragmented state to generate more violence but they enjoy a market monopoly, violence is likely to be sporadic because the power of the organization deters revenge and retaliation. Thus, although each independent variable can affect both visibility and frequency, its effect is predominant on one dimension of violence.
Cohesion in the State Security Apparatus
Cohesion depends on three components: intergovernmental relationships (vertical relations among government levels), interagency relationships (horizontal relations across enforcement agencies), and time horizons of elected and enforcement officials. Fragmentation increases when there are conflicts between government levels (e.g., the president and the mayor), conflicts across enforcement agencies (e.g., the military and the police), and when time horizons of officials are short due to processes such as rotation and purges in police forces or greater electoral competition.
The three components are interrelated but do not necessarily covary. A purge within a police agency affects time horizons by increasing uncertainty about who is a cop and about how long someone can remain within the purged agency, but it may not increase conflicts between enforcement agencies or government levels. A change in one component of the security apparatus is thus likely to increase the visibility of violence even if other components remain stable, and the more changes along the three components, the larger the effect on visibility.
Cohesion of the security apparatus thus depends on institutional arrangements that distribute government responsibilities such as regime type (centralized vs. federal), on bureaucratic dynamics that determine rotation and responsibilities of enforcement agencies, and on political and electoral dynamics and processes of democratization. For example, democratization can produce fragmentation by multiplying the political forces that influence policy making or that may provide protection. As documented in this special issue, the declining hegemony of the institutional revolutionary party (PRI) in Mexico undermined the state’s ability to protect criminals (Snyder and Duran-Martinez 2009) and political competition also affects the likelihood of enforcement (Dell 2014; Osorio 2013). Political competition thus impacts cohesion in the security apparatus, but it is not the only relevant factor. Electoral competition can co-occur with cohesive security if political actors cooperate beyond party lines, as described below in the case of Medellín. In other cases, fragmentation may derive from deploying new enforcement agencies that add an actor to an existing security configuration, as occurred in Mexican municipalities where military deployments exacerbated state fragmentation. Of course, more parsimonious analyses using just one component, such as electoral competition, could also help explain changes in violence. Yet a broad notion of the security apparatus illuminates important variability, for example, that not all municipalities that saw a military intervention in Mexico increased violence in the same proportion, or as Shirk and Wallman (2015) point out, that the timing of democratization does not necessarily coincide with upsurges in violence.
State and Market Interaction
Actions of the state security apparatus can shape the criminal market. For example, captures and enforcement pressures can transform, fragment, or create leadership vacancies within criminal organizations (Dorn, Murji, and South 1992, 35; Friman 2004). Heightened law enforcement can force cooperation among criminals who would otherwise work individually and create pressures for territorial expansion, but it can also force organizations to reduce their size to evade attention.
Conversely, the criminal market can force the security apparatus to reform in its effort to confront criminality and violence. In Colombia in the 1980s, when drug violence threatened the state, security agencies, especially the police, were unprepared for antinarcotics operations and plagued by lack of coordination and corruption (Camacho-Guizado 1994; Casas Dupuy 2005). The need to confront criminal organizations motivated crucial transformations in Colombian security agencies.
These complex interactions provide the basis for understanding the dynamics of the argument: where drug trafficking emerges within a cohesive security apparatus, violence is likely to be relatively hidden, and thus law enforcement institutions are less likely to react and transform themselves in response to violence. This continuity contributes to predictable relations between state and criminals and thus to relatively stable configurations of violence. By contrast, if drug trafficking emerges within a fragmented state, where protection is unpredictable, visible violence may emerge, the state is likely to react, and relations between criminals and the state are likely to change constantly, generating volatile patterns of violence. In sum, when state institutions change we are more likely to see transformations in violence than when the state is cohesive and stable.
Case Studies
This section substantiates the argument by examining periods of violence from Cali, Culiacán, Ciudad Juárez, Medellín, and Tijuana. These cities have been home to major DTOs throughout the entire study period (1984–2010) and highlight the variation in violence that occurs over time and both across and within territories. Given the limited development of theories on drug violence and of data amenable for large-N testing, cases were chosen to represent variation in the dependent variables, frequency, and visibility (Lieberman 2005). However, cases also show variation along independent variables (cohesion in the security apparatus and competition in the illegal market) and other key variables such as size of the illegal market and international influence. This method elucidates causal connections to better specify theories that could then be quantitatively tested.
The unit of analysis is a period of violence within a city. I use subnational comparative analyses (Snyder 2001) across two countries and systematic paired comparisons to identify the simplest combination of variables that produce violence in each case and to provide greater generalizability to the findings (Lankina 2012). Pairing cities in the same country that experienced contrasting patterns of violence provides a most-similar comparison. Pairing similar patterns of violence in different countries provides a most-different comparison. Finally, pairing periods of violence within each city provides an even stronger, longitudinal most-similar research design. I present in detail cases that illustrate the controlled comparisons and show how variation in illegal market competition combined with differing degrees of cohesion in the security apparatus creates four possible combinations of frequency–visibility: low frequency–low visibility (LF-LV), high frequency–low visibility (HF-LV), low frequency–high visibility (LF-HV), and high frequency–high visibility (HF-HV).
The data were derived from over fifteen months of fieldwork in the five cities and 196 semistructured interviews, including forty-two police, enforcement, and intelligence officials, thirty-four politicians and public officials, twenty-three journalists, forty-three human rights and nongovernmental organization workers, ten representatives of international organizations, and thirty-eight academics. Because of security concerns, I conducted few interviews (seven) with active criminals. This limits an analysis of criminal incentives, but I use judicial records and other first-hand testimonies to fill this gap. I also created a data set that includes 6,497 violent events coded from the main local newspaper in each city for selected years. 11 Among the features coded that are relevant to visibility are the methods, the type of victims, and the number of victims for each violent event. Table 2 summarizes the information in the data set to illustrate variation in visibility. Cases of low visibility entail less deliberate “publicizing” of violence, fewer collective (three or more victims) attacks and public targets, and no claims of responsibility. Cases of high visibility entail more publicly exposed methods, collective attacks, public targets, and claims of responsibility. To assess frequency, I used homicide rates (see Figure 1), coding as high-frequency cases with rates over 20 per 100,000. 12
Frequency of violence (homicides per 100,000). Source: In Spanish Instituto Nacional de Estadistica y Geografia (INEGI; Mexico), Medicina Legal (Colombia).
Visibility in Colombian and Mexican Cities.
Note: Boldface values indicate high visibility. For each violent event, each of four variables was given a value of 1 (low visibility), 2 (medium), or 3(high). These values were summed and then divided by the number of variables for which there was information. This yielded an index of visibility ranging from 0 to 3 (0 was given if no information was available for a variable). The second-to-last column indicates how many attacks in each city had scores of 1.5 to 3. High-visibility city-years were defined as having more than 100 high-visibility events. Details provided in the appendix.
aCar bombs and explosions, mutilation or incineration in public spaces, corpses with notes. bNotes and messages do not necessarily entail a direct claim of responsibility, that is, an organization signing. Only events with a direct claim are included.
The operationalization of independent variables combines data from official statistics, bibliographic material, government documents, newspapers reports, judicial records, and interviews. To assess cohesion in the security apparatus, I combined indicators of electoral competition such as electoral volatility and win margins for local executive elections, indicators of intergovernment relations (party coincidence between president, governor, and mayor), and indicators of interagency relationships by coding whether military operations or rotations occurred in a given year. To assess market competition, I combined sources to determine how many criminal or armed organizations had territorial presence in each city-year. Triangulation of sources was essential to compensate for the recall problems that emerge in interviews referring to distant events. Table 3 summarizes the values of key variables for each city-period. 13
Indicators for Each City-period.
LF-LV Violence
This pattern occurs when the state is cohesive (e.g., low electoral volatility and high win margins) and the market is monopolistic (one armed group tends to dominate). Market monopoly reduces conflicts for territorial control, and state cohesion motivates criminals to use less visible violence because they fear losing state protection or because they fear state prosecution, or both. Violence may emerge to discipline members, to eliminate potential competitors, or to regulate “public threats” such as low-level criminality, drug users, or social “undesirables,” but it is usually hidden.
A LF-LV configuration existed in Ciudad Juárez between 1984 and 1994, when the Juárez DTO (aka Carrillo Fuentes Organization) controlled the city. It also existed in Tijuana between 1984 and 1987 when the Arellano Felix organization (AFO) established a criminal monopoly and weaved networks of protection within security agencies. The mechanism limiting visibility in both cases was state protection. The power of the Attorney General’s Office (PGR) and its police branch, the Federal Security Directorate (DFS), along with PRI hegemony, expressed in high win margins and party coordination among levels of government, limited confrontations among enforcement agencies, and motivated criminals to hide violence. In the words of a former DFS agent who worked in Juárez, “The Attorney General’s Office coordinated everything, sometimes even controlled the army. There was order even in the relation with the state’s attorney. There were executions, but the bodies disappeared, they were not thrown out in the streets.” 14
One may explain these two cases as resulting from a broader Mexican context of PRI domination, but local variation in illegal market competition and in the security apparatus explains differences in the timing of violence: Tijuana experienced an earlier spike in frequency than Ciudad Juárez, because the AFO faced earlier challenges to its monopoly than the Juárez DTO. Similarly, as explained below, in 1996, Tijuana experienced radical changes in law enforcement, which generated visible attacks that Juárez did not experience until later.
A key illustration for LF-LV is Medellín between 2003 and 2007. Despite many differences between Ciudad Juárez and Tijuana, on one hand, and Medellín, on the other, the same factors played out in creating similar configurations of violence. Once the most violent city in the world at the height of Pablo Escobar’s criminal power, when the homicide rate reached a historic high of 381 deaths per 100,000 inhabitants in 1991, Medellín underwent a sharp decrease in homicide in the mid-2000s. Between 2003 and 2007, its rate decreased by about 81 percent and reached a historic low of 34 homicides per 100,000 inhabitants in 2007. 15 As frequency declined, so did visibility. Murders were carefully hidden, for example, by throwing bodies in rivers or burying them in mass graves. A monopoly in the illegal market and a security apparatus that had become more cohesive generated the conditions for LF-LV.
During this period illegal markets fell under almost monopolistic control of Diego Murillo, aka Don Berna who also disciplined low-level criminality and gangs. Berna started as a member of the Medellín DTO, then became its enemy, and then successfully mutated into a paramilitary leader who consolidated a criminal monopoly by co-opting some gangs, eliminating rivals, and winning over an initially allied but then rival paramilitary faction, the Bloque Metro (Comisión Nacional de Reparación y Reconciliación [CNRR] 2011; Alonso, Giraldo, and Sierra 2007). Berna’s success resulted from his ability to win subsequent wars against criminal and political competitors. As one expert puts it, Berna “was the winner of all the wars in the city.” 16 He controlled every major form of organized crime, from extortion to local drug distribution. As Berna himself puts it, “We co-opted groups so they became part of our organization. In [Medellín’s] neighborhoods the social fabric was lost; there was a lot of vice. We needed an authority to regulate the situation” ( Queríamos que Castaño llegara a la Presidencia 2013). Such control implied disciplining, regulating, and brokering conflicts among criminal gangs: “With the subordination of bandits, nobody could move without Berna’s permission.” 17
While Berna’s power was growing, by the early 2000s, the security apparatus in Medellín had changed notably, becoming more cohesive. In 2002, enforcement operations in shantytowns that deployed military and police forces reflected high coordination among enforcement agencies. 18 In 2002, there were eleven military operations (CNRR 2011, 76), and two of them—Operations Mariscal and Orión—were notable because police and army operated jointly along with the judicial police Departamento Administrativo de Seguridad (DAS) and the Attorney General’s Office. Unlike urban military operations in the late 1980s, these operations were highly coordinated. The army general who commanded joint enforcement operations emphasized that teamwork had been key in all them ( Desarticulada en Medellín Red Logística de las FARC 2003).
Another essential factor for state cohesion was collaboration between national and city governments. In 2002, the national government of President Alvaro Uribe (2002–2010) started a demobilization process for paramilitary groups. The process benefited not only paramilitaries but also many drug traffickers disguised as paramilitaries, and Berna, who became a paramilitary leader, entered the peace process. In November 2003, as commander of the Bloque Cacique Nutibara, he demobilized along with 868 men, many of whom were members of local gangs rather than paramilitary soldiers (Human Rights Watch 2010).
Mayor Sergio Fajardo was neither part of the peace negotiation nor a member of Uribe’s political coalition, but he supported the process that he defined as “an exit door from delinquency.” 19 This situation created a strategic concurrence of interests that eliminated conflicts between national and local governments: the local government did not question the negotiation, and the national government did not intervene in local security initiatives. The success of the process depended on collaboration between government levels: the national government had the political control, but the city controlled the reinsertion of demobilized soldiers (Alonso and Valencia 2008). In the words of a police officer, “The order was not to follow them [criminals], because there was a process with the president and the mayor’s office. There was political resistance to bothering them […] because there was a peace process ongoing.” 20
As became evident later, while engaged in the peace process Berna continued controlling illegal markets, but he knew that, in principle, the state was not going after him if he did not use excessive violence, because both local and national authorities needed a successful peace process. Violence reduction thus derived from two interrelated processes: the monopolization of the criminal market, and increasing state cohesion, 21 which created a shared interest between state and criminals to reduce violence.
Although the pacification in Medellín has been attributed to other factors, chiefly an urban transformation that improved social indicators during Mayor Fajardo’s administration, the argument here still holds, especially when considering that in 2009 homicide rates doubled (from forty-five in 2008 to ninety-four) without a major change in social conditions. Rather, the new spike in frequency occurred as accusations of Berna’s continued engagement in illegal markets became too difficult for the government to overlook, and he was extradited in 2008, leading to a fragmentation of the criminal market and to a violent confrontation between his successors.
HF-LV
This, a very common configuration, occurs when the drug market is competitive and thus generates violence aimed at gaining market control, but the state security apparatus is cohesive. The incentive for traffickers to make violence visible is reduced (or the desirability of hiding it is greater).
Cali (1989–2010) and Culiacán (1984–2008) displayed high homicide rates but never generated as much government concern as their country counterparts, Medellín and Ciudad Juárez. Criminal organizations employed frequent violence to gain market control, discipline members, deter betrayal, and even to enhance their reputation with the public by eliminating “social threats.” Yet, state-sponsored protection motivated them to hide violence. State protection guaranteed that criminals could kill with impunity, but they also knew that visible violence could force the state to attack them. Cali and Culiacán illustrate how although state protection can help eliminate criminal rivals, cohesive states can also maintain arrangements in which they do not favor one group over others.
Culiacán, the capital of the state of Sinaloa, has been crucial in the history of drug trafficking in Mexico (Astorga 2005). By 1986, Culiacán had a very high homicide rate of 66 per 100,000 inhabitants, which remained above national averages throughout the decade. This persistent high-frequency violence throughout the 1990s and 2000s resulted from confrontation between criminal factions. Most early drug traffickers were born in Sinaloa and their common origin facilitated certain continuity in criminal leadership but at the same time generated strong disputes among leaders who operated in an oligopolistic fashion (Astorga 2005, 166).
Culiacán’s frequent violence was also characterized by low visibility. As shown in Table 2, collective violence was uncommon: before 2009, there were few events with more than two victims, and a high percentage of violent events (62%) did not produce fatalities but only left people wounded. Most homicide victims were unlikely to attract public attention: peasants, nonprofessional workers, and delinquents. Finally, a high proportion of violence occurred in rural areas, specifically in Tierra Blanca, a poor neighborhood connecting the city with the mountains (La Sierra). A higher percentage of homicides occurred in the outskirts and rural areas of Culiacán (37 percent) than in the other cities compared in this study (the average for all other cities was 16%). A significant amount of violence was thus hidden from public attention. In fact, there were places in the city known as “botaderos” (dumpsters) for corpses, for example, the “Cerro del Tute” in the 1970s and 1980s and in the 2000s “La Costerita” road. 22 It is understandable that traffickers hid violence in these locations connected to land they owned, because few people mingled around them.
If confrontations between criminals explained frequency, stable state protection reduced their incentives to use visible violence. The concentration of political power allowed close relations between traffickers and authorities in Culiacán. The historic power of Sinaloan governors, 23 who dominated local mayors, 24 limited intergovernmental conflicts, and along with the stability of the PRI in elections, contributed to a cohesive state. Sinaloa was a PRI stronghold (Alvarado 1997), and even though in the early 1990s opposition parties fared well in elections, this was not enough to challenge PRI domination. Gubernatorial power and PRI dominance explain how, despite frequent (but low visibility) violence and increasing accusations of corruption of its political class, neither motivated sustained enforcement actions in Culiacán. This situation remained stable until 2008, when an all-out war between two criminal groups and state fragmentation, mainly through military deployments, generated an HF-HV configuration.
As in Culiacán, in Cali homicide rates since 1991 were always above national rates but violence was not very visible. In the 1980s, the Rodriguez Orejuela brothers, leaders of the Cali DTO, established tight control on illegal markets, but between 1989 and 1994 other criminals challenged them. Competition continued after the Cali DTO was dismantled in 1995 because disputes emerged among its heirs. The protagonists of turf disputes changed when a leader was imprisoned or killed, yet the confrontations that generated frequent violence continued, although with low visibility due to predictable state protection.
The Rodriguez Orejuela brothers were reputedly reluctant to use violence. But they did not renounce violence altogether, and when it was necessary, they hid it (Chepesiuk 2003, 63), dumping bodies in rivers, for example, or removing fingerprints from corpses so that they could not be identified. In the words of an army colonel who pursued the Cali DTO, “In Cali there were silent deaths, without noise, without paraphernalia.”
25
Alongside methods that concealed evidence, in the 1980s, traffickers avoided high-profile victims, attacks with multiple victims, and attacks against public officials and the state, as one of the Rodriguez Orejuelas himself explained: Mr. Pablo Escobar […] sent us a letter asking for our collaboration, physical and economic, to start a war without barracks against the state and its institutions and representatives, including the justice sector. Our response, because of our beliefs and because of our respect, not only towards institutions but also towards citizens, and because of our families’ peace, was absolutely negative. (Rodriguez and Rodriguez 2005, 58)
The Proceso 8000, a large-scale judicial process initiated in 1995, documented links between the Cali DTO and state officials, illustrating traffickers’ ability to establish symbiotic relations with state authorities. Politicians indicted during the Proceso 8000 included governors, mayors, and senators who played crucial roles for the DTO, from creating facade businesses to laundering money, to supporting legislation favorable for traffickers.
The capture of the Cali DTO leaders in 1995 increased the prominence of groups from nearby rural areas, chiefly the Norte del Valle DTO. Turf battles became more intense. According to an intelligence agent in Cali, “Since 1996–1998 the dynamic has always been the same, a fight for territorial control.” 27 Yet, violence remained of low visibility because the security apparatus did not change dramatically, criminals continued colluding with the state, and thus with some exceptions, they preferred to limit violence that could force state attention. An army general observed that “the Cartel del Norte del Valle assimilated the experience of Cali and resorted more to bribes at the municipal level.” 28 Throughout the years the state remained cohesive and prone to collusion with criminals. In fact, a 2008 United Nations Development Program report identified the penetration of drug trafficking into the political life of the region as a major hindrance to human development.
A plausible alternative explanation for HF-LV pattern in Cali and Culiacán is the close links between civil society and traffickers who at some points enjoyed legitimacy among the public that could have motivated them to hide violence. Yet, those links cannot explain why, for example, in Cali the Norte del Valle traffickers seemed less popular than the Rodriguez Orejuela group despite very similar violent tactics, or why in Culiacán a decline in traffickers’ legitimacy among the population followed, rather than preceded, an increase in visibility.
Cali and Culiacán represent a most-different systems comparison: two cities in different countries and with different socioeconomic histories share a pattern of violence driven by the same factors—a cohesive state and a competitive market. The persistent high frequency in Cali even when the profitability of drug trafficking seemed to be declining in Colombia further illustrates the need to go beyond mere profitability to explain violence.
LF-HV
This pattern results from a monopolistic market and an increase in fragmentation within a state that used to be cohesive and once protected criminals. It emerges when new enforcement agencies are deployed in a territory or when rotation and purges reduce time horizons of enforcement agents. Traffickers then use visible violence mainly to retaliate against the state and to force it to refrain from enforcing the law. This pattern is uncommon and tends to be very short-lived because it is highly unstable and can easily evolve into HF-HV if the state or the market further fragments.
Tijuana in 1996 to 1997 illustrates such configuration, with high-profile killings that included at least eight high-ranking government officials, twelve police officers, and the editor of Zeta, a magazine that openly denounced corruption and trafficking (Moore 1997). Up to that point, the AFO had enjoyed state protection, even though political competition had increased with the election of Mexico’s first non-PRI governor in Tijuana’s state of Baja California. Although the AFO faced increased challenges from competitors, it had confronted them with low-visibility violence and with strategies that eliminated the evidence of attacks, such as those deployed by El Pozolero described earlier. In 1996, the security apparatus fragmented significantly as a result of purges in enforcement agencies, but market competition did not change substantially. As a result, visible violence emerged, but frequency, measured through homicide rates, remained stable (20 per 100,000 in 1996 compared to 21 in 1995 and 19 in 1997).
Most visible killings in 1996 were retaliatory responses to anticorruption policies that destabilized protection arrangements and followed an uncommon announcement by a member of the National Institute for the Combat of Drugs (INCD) denouncing extreme corruption in Mexican antinarcotics agencies. This accusation led to the appointment of a member of the opposition party Partido de Accion Nacional (PAN) to the Attorney General’s Office (PGR) who then purged about half of the members of the Tijuana State Attorney’s office. The newly appointed director of the Federal Police in Tijuana echoed denunciations of corruption in September 1996, and not surprisingly he was assassinated a few days after swearing in for his position. The purges increased state fragmentation and temporarily reduced traffickers’ incentives to hide violence. Enforcement operations proliferated but due to fragmentation, they were not threatening to criminals who had reason to believe that they could force the state to change its policies and re-establish protection arrangements.
The AFO employed visible violence in 1996 and 1997 to retaliate against state enforcement and to deter enforcement agencies from continuing operations. The strong market control the organization enjoyed guaranteed that at least in the short run, frequency would not escalate. However, the situation changed in the following years as market competition increased and pushed homicide rates upward.
HF-HV
The HF-HV pattern results from a competitive market and a fragmented state along all three components (conflicts among levels of government and between enforcement agencies, and short time horizons of public officials; indicated by low party coordination, military operations, and low win margins). This pattern emerged in Ciudad Juárez in 2008 to 2011, as criminal competition reached extreme levels and the state was deeply fragmented.
Between 2007 and 2008, Ciudad Juárez experienced an unprecedented increase of 713 percent in homicide rates. Visibility also increased as criminals adopted such methods as beheadings, carried out collective murders, and attacked police forces. A territorial dispute with the Sinaloa DTO destabilized the monopoly the Juárez DTO achieved during the 1980s and spun out of control as both groups outsourced violence, actively recruiting youth gangs to wage war. Local police calculated the existence of at least 460 gangs, and one of the largest, the Barrio Azteca alone, had about 3,000 members (Castillo 2009, 309). The Sinaloa DTO employed local gangs known as Los Mexicles and Los Artistas Asesinos and created an armed branch known as Gente Nueva, and the Juárez DTO employed Los Aztecas and La Linea. As a result of this growing number of armed actors, violence mushroomed quickly.
The fragmentation of the security apparatus underlaid the increase in visibility. After President Felipe Calderon declared a war against trafficking organizations, he sent troops to Ciudad Juárez in 2008. The army deployment, as well as the constant confrontation between the mayor and the governor of Chihuahua even though both belonged to the PRI, created a highly fragmented security situation as the governor did little to support the mayor’s security policies. 29 When city and state governments changed in 2010, confrontations between mayor and federal forces ensued. By 2010, the National Human Rights Commission declared that violence in Juárez resulted from lack of coordination among enforcement agencies ( Violencia en Juarez por falta de coordinación 2010) and the US consul in Ciudad Juárez also expressed concerns about the lack of trust between government levels and enforcement agencies ( En Juarez ejército se dedico a ver pugna entre carteles 2010).
At the same time, networks of protection broke down, thus reducing incentives to avoid visible violence, as traffickers were no longer sure that if they received protection from one government sector, they would not be prosecuted by another. Traffickers declared a war against police officers who refused to collaborate with them or whom they considered to be protectors of their rivals. In February 2009, the Juárez police chief quit after several officers were slain and traffickers issued threats stating that a police officer would be killed every forty-eight hours until the police chief quit. In 2010, a total of 149 law enforcement officials from municipal, state, and federal agencies were killed, and the bodies of some federal and municipal police appeared with notes such as “I’m a cop at the service of ____” (Author’s data set) in a pattern of accusations and rumors that linked the federal police with the Sinaloa DTO and the municipal police with the Juárez DTO and La Linea.
The same factors that generated the HF-HV situation in Ciudad Juárez also created a HF-HV in Tijuana as the AFO split into two warring factions and the state fragmented after military deployments in 2007. Homicide rates increased by 245 percent in 2007 and visible attacks proliferated. Yet in Tijuana the HF-HV episode was shorter lived because the state recovered its cohesion through the achievement of a militarized unified command and coordination arrangements among the mayor, governor, and president. 30 Additionally, one organization—the Sinaloa DTO—gained market control more rapidly than in Ciudad Juárez. The difference between these border cities, both crucial drug trafficking routes and with similar sociocultural influences, shows that the fragmenting impact of military interventions may vary depending on other components of the security apparatus. It also shows that a market’s strategic location and profitability are not enough to explain violence. One aspect that could be further explored is that in Tijuana criminals did not outsource violence to youth gangs as they did in Juárez. According to Julian Leyzaola, an army colonel who served as police chief in both cities, “in Tijuana crime was organized, specialized […] there were gangs but not linked to crime. [The traffickers] had their own killers and gangs were used only sporadically,” whereas in Juárez “the fight between cartels is a fight between gangs, a constant battle.” 31 Outsourcing thus could explain cases of extreme violence, when both frequency and visibility are very high.
Conclusion
Drug violence can be better understood by analyzing interactions between states and criminal actors rather than either actor in isolation. The public discussion about Mexican violence tends to be sharply divided between those who see it as a result of poorly planned government policy, and those who portray it solely as a result of drug market transformations. Both accounts downplay key interactions between the state and the organized crime groups. Analyzing drug violence as a joint function of state cohesion and market competition provides a strong explanation for the striking variation in drug violence seen both across space and over time.
A focus on visibility unveils variation that is not apparent by looking at frequency only. It shows that apparently peaceful conditions should not be equated with absence of violent criminality, as these may actually be situations where one criminal group has established a state-protected monopoly of illegality and prefers to use only low-visibility violence. Violence-reduction policies thus should not stop only when the most visible violence has decreased. This approach to violence also recognizes the potential contradictions between criminals’ interest in using violence to show power, or hiding it to avoid state attention. Attending to visibility also has implications for how we collect and analyze data on violence, and for obvious reasons, most information on less visible violence is not easily available and requires extra work to gather.
The explanation I advance provides a theory and a method that can clarify causal dynamics through detailed process tracing. It should be augmented by research focused on disaggregated components of the state security apparatus, fine-grained assessments of criminal competition, and large-N analyses to test the generalizability of the theory. Another area where more research is warranted is the targets of violence. This article’s analytical framework has implications for whom organized criminals are likely to target. In periods of fragmented law enforcement, violence is likely to be aimed at state officials if a single group controls criminal markets. In periods of high competition and fragmented law enforcement, criminals are likely to target state officials as well as civilians and rival criminals. In periods of high competition but state cohesion, violence is likely to affect civilians and criminals more than state officials and public figures. Finally, the concept of visibility is useful in drawing attention to the fact that a given victim category may—predictably—be targeted at different times with differing degrees of visibility: in the 1990s, Mexican traffickers preferred to “disappear” their rivals, but the targeting of rivals became much more visible in the mid-2000s.
In sum, a political economy framework challenges the idea that high levels of crime and illegality always reflect state absence and highlights the importance of analyzing relations between state security agencies and political actors to understand violence in illegal markets.
Footnotes
Author’s Note
Ana Arjona, Nicholas Barnes, Yanilda Gonzalez, Javier Osorio, and Hillel Soifer provided excellent comments on earlier versions of this article. Enrique Desmond Arias and Eduardo Moncada also provided key insights. Two anonymous reviewers and JCR editors also provided critical feedback. All errors remain mine.
Acknowledgments
I thank Richard Snyder, Peter Andreas, and Pauline Jones-Luong for their constant support. I also thank issue editors Joel Wallman and David Shirk for their valuable insights and support. I owe David Shirk credit for this article’s title.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This article is part of a dissertation funded by the US Institute of Peace, the International Dissertation Research Fellowship of the Social Science Research Council SSRC, the Drugs, Security and Democracy Fellowship of the SSRC and the Open Society Foundations, and the Watson Institute, the Center for Latin American and Caribbean Studies, the Global Program in Development, and the department of Political Science at Brown University.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
