Narco-Insurgency: Charting Gang-Violence in Mexico

This study seeks to understand more fully the most recent spike in drug-related violence in Mexico that began in 2006 and continues to this day. This research began with the hope of uncovering discernable patterns in the spatial distribution of the violence that has erupted over the last five years. Specifically, the goal of this study is to uncover those variables most relevant to the number of executions observed at the municipal level among the different cartels operating in Mexico.  Thus, this study will analyze micro-level literature on civil war faction violence in conjunction with economic theory on firm behavior, framing the cartels as little more than revenue-maximizing drug entrepreneurs.

Indeed, many insights into this conflict and reasonable predictions regarding its future direction derive from the body of civil war literature.  Although this particular struggle is not considered a civil war by the traditional definition, the existing literature on civil wars informed the author’s decision to incorporate specific variables into the study—namely the frequency and locations of the executions.  These variables were tested using a new data set that holds the geographical coordinates and population statistics for 1,147 municipalities in Mexico, covering a population of 99 million people out of the nation’s 114 million total. The data set spans a period of five years between 2006 and 2010.


In December 2006, Felipe Calderon assumed the office of the Mexican presidency. As his first major act, he declared war on Mexico’s drug cartels, immediately instigating rampant violence across the country. Since Calderon’s declaration, drug-related violence in Mexico has skyrocketed, claiming the lives of over 36,000 Mexican citizens, including such high profile government officials as the acting director of Mexico’s federal police force.[1] The vast majority of these victims are involved in the drug trade themselves, but the number of uninvolved civilians dead and missing continues to rise precipitously. Even within the context of Mexico’s long history as a smuggling route to the United States, the recent rise in drug-related violence is an unprecedented phenomenon.

Mexico’s role in international trafficking has traditionally been that of a ‘bootlegger,’ moving contraband from other Latin American countries to U.S. consumers using routes that have been developed since at least 1910.[2] Eventually, the nation’s northwestern states would become the most important distribution network of narcotics headed to the United States.[3] Gabriela Recio credits prohibition in the United States as the primary reason for the establishment of these illegal markets and Mexico’s long record of professional smuggling and enduring drug distribution networks. More importantly, Recio highlights the fact that the northwestern states of Mexico now have approximately 100 years of experience developing and improving these drug distribution networks.[4]

These long-lived narco-networks have begun to pose a serious security threat to the most basic functioning of Mexico’s government across a broad swath of the country and is not only confined to the northwest. Perhaps the application of  comparative analytics, using lessons from a different country with a very similar plight, will help to illustrate the nature of the drug-related security threat facing Mexico.  Johan Engvall, developing further upon the security framework of Buzan (1991) and Buzan, Waever and De Wilde (1998), presents the case of Tajikistan to explore the impact of the drug trade on national security and stability.[5] Tajikistan shares many similarities with Mexico in terms of the drug trade. Tajikistan is home to some of the most lucrative drug routes in all of Asia, due to its geographic position between Afghanistan (the world’s largest grower of opium) and Russia (one of the largest global consumers of heroin). In short, Tajikistan is the Mexico of Central Asia, situated perfectly between the growers and consumers of narcotics, acting purely as the distribution network for these two groups. Insight into the relation between criminal drug gangs and state-ensured security in Tajikistan therefore can offer a lens with which to view Mexico’s drug war.

Using Engvall’s framework, drug trafficking poses a security threat to governments through several mechanisms. Those mechanisms most relevant to this discussion refer to the security of the population and the legitimacy of institutions. In a democratic society, the state relies on the population for stability and revenue in most scenarios, so a physical threat to this base of support threatens these needs. In Mexico, President Felipe Calderon has stated publicly that cartels have levied taxes on locals in areas they control.[6]  Clearly, this conflicts with the general definition of a state where the government is the only body to oversee taxation. It also has been estimated that the drug violence in 2008 reduced the country’s GDP by 1% from increasing the cost of doing business due to the need for more security, as well as reduced investment because of a weakened rule of law.[7] Thus drug violence clearly can act through economic channels to undermine the capacity of the state to govern by reducing its revenues and the incomes of the population.  The most relevant danger to Mexico of Engvall’s institutions is the institutional decay that arises from strong organized drug gangs. The prevalence of cartel drug-money in Mexican politics undermines the democratic and political institutions of Mexico, and ultimately reduces the legitimacy of the Mexican state. When organized criminal gangs can effectively infiltrate governmental institutions and exert influence, the functions of government are necessarily distorted—diminished in their capacity to distribute public goods, especially security.[8]

Despite these tangible threats to the public’s security and the intensity of the drug violence in Mexico, academic researchers have, with a few notable exceptions, largely ignored the conflict.[9]. The civil war literature could be of significant value in helping to uncover patterns in the cartel violence but, as of yet, has not been considered in current discussions. The recent episodes of violence affecting Mexico are observed quite frequently in the world at large. The lessons from other episodes of conflict, specifically those regarding the geographical distribution and the intensity of the violence, are equally applicable to Mexico’s recent unrest. Some may argue that the units of interest between Mexico’s conflict and other civil wars are characteristically different.  But, a host of civil war literature argues that many rebel groups are composed wholly of profit-maximizing opportunists, not unlike Mexico’s cartels.[10]

Part of the reason conflict researchers have not thoroughly studied the violence in Mexico is because Mexico’s violent organized elements are criminals who lack obvious political motives. And, though they may target political figures and institutions with violence; their motives are purely profit Engvall points out that issues concerning drugs and crime are traditionally not considered security issues and are not studied in political science or security studies.[11] The majority of civil wars have stopped far short of the 36,000 battle deaths observed in Mexico. With the violence so out of hand, officials in the United States, including Secretary of State Hillary Clinton[12] and the Undersecretary of the U.S. Army, Joseph Westphal,[13] have even begun to refer to the conflict in Mexico as an ‘insurgency’. Indeed, many similarities may be found between Mexico’s recent spike in violence and observed civil wars and insurgencies (defined here as armed rebellions by an unrecognized sovereign entity that engages in organized and coordinated acts of violence against a recognized sovereign authority).

The goals and motivations of the participants mark the key distinction between drug wars and civil wars. Participants in drug wars, unlike civil wars, lack political goals and motivations, seeking only material profits from their illicit activities. Though achieving the maximum amount of drug profits may necessitate coercive acts against the state, in the form of bribes or violent intimidation, drug dealers do not advocate the wholesale destruction of the state or the creation of a new state.

One obvious similarity is the factional nature of civil wars and drug conflicts. In civil wars, government and various rebel factions violently contest each other for territorial gains that hold political and material benefits. The underlying logic of territorial contestation in civil wars–that factions compete violently for territory that has value to the participants—also applies to drug conflicts.[14] In this study, micro-level civil war literature on territorial contestation and faction violence is leveraged in order to understand the spatial distribution of violence as well as its intensity throughout Mexico.

Additionally, in the study of civil wars, political scientists and theorists have traditionally stressed normative grievances as the impetus for political upheaval and civil violence. With the work of developmental economists Paul Collier and Anke Hoeffler explanations regarding the onset of conflict have tended to focus on the opportunities afforded to rebel organizations to violently oppose states.[15]  Though political grievances such as frustrations involving ethnic divisions (that precede the onset of conflict), have proven generally necessary for conflict onset, the environmental contexts in which civil wars occur, and which delineate opportunities for rebels to organize, have dominated scholarly discussions.  Further research by economists has stressed a correlation between “greed” and “conflict onset” to argue that rebels are motivated mostly by private gain, suggesting that these actors are little more than common criminals or run-of-the-mill bandits.[16] This framework fits perfectly into discussions of drug conflicts in general and Mexico’s recent spike in drug-related gang-on-gang violence in particular. This similarity relaxes the distinctions between civil wars and drug wars regarding the motivations of actors and contribute further to understanding drug conflicts using the literature on civil wars.  The literature on civil wars has also extensively addressed questions of violence and their spatial diffusion. The following works present several theories to the geography of violence in conflict as well as their intensity that apply to the recent case of gang violence in Mexico.



When applying economic models of civil war, rebellion is considered a criminal act where the actors seek only economic gains. In the case of Mexico, the actors are exactly these sorts of criminals that are motivated purely by economic benefits. In Mexico, it is estimated that the drug trade produces between $6 billion and $15 billion annually in revenues.[17] These revenues finance and sustain the criminal enterprises throughout Mexico and provide the cartels with few if any reasons to cease their efforts to extract rents from the drug-trade. In this way, the presence of drug markets in America and an abundant supply of drugs from South America creates ample opportunity for cartels to form. Thinking of this interaction in terms of a typical business decision, the cartel, acting like a firm, enters the illicit drug distribution market because the profits are greater than zero. And so long as there is profit to be made, a cartel will exist to facilitate the exchange of materials across borders. Firm theory also holds that profits will only approach zero with the introduction of new competitors in the market. But, given the peculiarities of this particular business and the actors involved, competition between cartels assumes the form of gang-on-gang violence rather than price competition. Given this unique model of competition, it is likely then, that the presence of multiple gangs within a region should be correlated with higher levels of executions. Conversely, the presence of only one gang, without competition, should suggest fewer instances of gang-on-gang violence.

Much of the literature linking resources and civil conflict seeks to understand the implications of the links between resources, conflict geography and intensity. In his study of the relevance of resources to thirteen civil wars, Ross notes that resource presence increased the casualty rates in several of the cases, while showing mixed effects in the majority.[18] Ross hypothesizes several mechanisms by which resource presence may have increased the intensity of these conflicts.  The most relevant to Mexico holds that resource wealth tends to increase the casualty rate in civil wars as the value of the resource-rich land is much higher than other parts of the country that lack resources.[19] Applying Ross’s theory to Mexico shows a higher concentration of executions and violence in the areas where land is more valuable. In this particular case, value derives from the drugs that are smuggled and the routes that transport them. These drug routes tend to span the coastlines of Mexico’s Pacific and Gulf States as well as its international borders. As a result, municipalities that are located nearer Mexico’s coastlines and borders should experience higher levels of gang violence and executions.

Hypothesis One:        Municipalities closer to the US border should exhibit more executions than those farther from the US border.

Hypothesis Two:       Municipalities closer to the Guatemalan border should exhibit more executions than those farther from the Guatemalan border.

Hypothesis Three:      Municipalities closer to Mexico’s Pacific and Gulf coasts should exhibit more executions than those farther from Mexico’s coasts.

Furthermore, if the price of the drugs transported increases, the value of controlling these routes should increase as well, resulting in even more gang competition for these routes.

Hypothesis Four:       The price of the drugs being transported should be positively correlated with the levels of executions in a municipality.

In addition to Ross’s work on the location and intensity of violence in civil wars, Stathis Kalyvas has produced several hypotheses regarding the implementation of targeted violence in towns during the Greek civil war that may very well apply to Mexico. Kalyvas argues that selective violence will occur where one actor exercises a dominant but incomplete control over the area or town in question. In areas where control is dominant and complete, however, selective violence will be less common because there is no competition and no defection from the existing order.[20] This framework produces a rather counterintuitive finding that suggests levels of violence are reduced where sovereignty is shared between violent actors and also where one actor retains a dominant position of complete control.

From Kalyvas’s theories we can hypothesize a story regarding the spatial diffusion of gang violence in Mexico that complements our conclusions from the previous application of firm theory. First, we expect to see that where there is sole control over a territory, gang-on-gang violence should be limited and the levels of selective violence in the form of executions lower than for regions that support multiple gangs. But, where multiple gangs are present the impact is less certain because there is no obvious way to code for the relative strength of gang dominance.  Because sovereignty is unlikely to be shared between cartels, where there are multiple gangs and there is no sole dominance of one gang in particular, we should see an increase in the levels of violence in a community. Using the language of Kalyvas’s paper, cartels in Mexico then either exhibit dominant and complete control over a territory, suppressing violence; or a dominant but incomplete control, encouraging violence.

Hypothesis Five:        The presence of multiple drug gangs should increase the levels of executions observed in a city. Similarly, the presence of one gang should have a calming effect on the levels of executions observed in a city.

These two hypotheses by Ross and Kalyvas in a large sense inspired the following investigation in charting and discerning the patterns of violence throughout Mexico over the last five years. Though the wealth of literature regarding “greedy rebels” and the effects of resource presence on conflict also informed this investigation deeply, the following model sought to capture many of these relationships Ross and Kalyvas discuss regarding the location and intensity of violence.



The data to test these hypotheses combines several other unrelated datasets that held information relevant to this discussion. The dataset covers 1,147 different municipalities out of a total of 2,450 within the 32 States of Mexico. The dataset only includes those 1,147 municipalities provided by the Mexican government in their new database covering gang-related deaths in Mexico. Nearly 99 million persons live within these municipalities covered by the dataset out of a countrywide population of about 114 million persons in 2011.[21]

The data on gang-related murders was compiled by Mexican Federal Government agencies concerned with gang-related homicides.[22] An interagency working group was developed to collect and integrate the information into a countrywide database so that organizations and researchers might perform further analyses regarding the drug violence as this paper does.[23] A new indicator on “alleged deaths during criminal rivalry” was developed for the purposes of just such an investigation. In the government’s data, an execution is defined as a homicide where the perpetrator is a member of a criminal gang and when extreme violence was used against the victim, for example: decapitation, dismemberment, mutilations, or burns. The victim’s corpse must also satisfy two defining features in order for a violent death to be considered an execution: signs of torture or severe injury as defined earlier, and whether the corpse was found at a separate drop-site, or displayed some form of restraint. The use of a blanket is another indicator as this suggests concealment while transporting or dumping a body at an off-site location.  Finally, the new standard includes individuals with ties to a criminal group, and messages left claiming responsibility from rival criminal gangs. Using this working definition, the database has recorded some 30,821 executions between December of 2006 and 2010.

This dataset is the first tool exclusively covering the drug-related violence in Mexico and is the foundation for this investigation. In analyzing the data, there are many stages of validation and confirmation to eliminate errors and overlap as the collection and coding efforts span several government bodies.[24] The database is also updated as new information arises on particular episodes of violence, typically when mass graves are discovered. This constant updating ensures that the numbers of homicides presented in this investigation will not be exactly the same, but will unlikely be so wildly different as to undermine the significance of the statistical results. The database, however, also includes the violent deaths of individuals at a statewide level that could not be classified as occurring in a municipality or necessarily as a gang-related execution.  These deaths were coded as “No Specific” location and were dropped from the dataset used in this study. Again, these unspecified deaths numbered too few to significantly alter the results of this investigation. Also, missing from most of these related datasets is the numbers of missing persons that has risen as the violence progressed. For obvious reasons these “missing persons” are not included in the tally of drug-related homicides, but there appears to be a strong relationship between their numbers and the the frequent unearthing of mass graves throughout the country.

The rest of the investigation was based on the format of this government database that is organized by state and municipality. The data on the latitude and longitudinal coordinates of the municipalities was provided by using Google Earth. Google Earth allowed filtered searching by municipality, state and country, which limited the duplicate recording of locations for municipalities. The information on population was compiled by The National Institute of Statistic and Geography or INEGI. INEGI is an autonomous agency of the Mexican Government that records the country’s census every ten years. The population statistics for this investigation were from 2010, except for a few cases in the Mexican state of Oaxaca where population estimates in 2005 were recorded.

Once the geographic locations of the municipalities were identified, a program was used to calculate the distance between that municipality and the US, Mexican and Guatemalan border. The program works by identifying the closest distance between the municipality and the international border of interest. These distances could then theoretically produce some problems or inconsistencies as the shortest distance between the Yucatan and the US would be a line going through the Gulf of Mexico as opposed to the length of Mexico’s coastline, which is a more likely transportation route to the United States. The extent of this problem is uncertain, and presumably not so large that it undermines the findings, but it is present.

The data on gang presence, unfortunately, is less precise.  In the dataset, gang presence is coded as 1 for the presence of a single gang, 2 for the presence of multiple gangs, and 0 for disputed territory where dominance is unclear for a particular set of gangs. For example, a BBC report included a chart mapping the presence of the top seven cartels in Mexico in 2006.[25] A large problem recognized by this investigation regards the coding of gang presence. Gang presence was mostly coded at the state level even if certain municipalities within larger states might exhibit only single gang presence. In addition, a considerable amount of “eye-balling” was used to estimate gang presence at the state level. Given that state lines are not delineated in the BBC map, it was hard in several cases, namely the Gulf States and particularly in the state of Tamaulipas, to determine at which point the Zeta’s area of control began and the Gulf Cartel ended. Also, a gang was typically not present every part of a state, yet if the majority of the state hosts a gang, then the entire area is coded as its territory. Furthermore, one problem with using this map is the necessary assumption (for this study) that these lines of presence are fixed and do not shift. Given that so much violence has erupted, however, it is extremely unlikely that this map of gang presence in 2010 is the same as in 2006, but at least between 2006 and 2008 these lines were rather static.[26] Moreover, though this category of “disputed territory” was recorded in this report’s dataset, it was not used in the statistical analysis to compare the extent of gang violence because it was unclear what “disputed” territory constituted. Most likely the territory was “disputed” by multiple gangs, but without a definition, the data was recorded with the hopes that one day, the status “disputed” will be clarified. This is also not the only map available charting gang presence throughout Mexico. There are many other maps that chart this similar layout as found on the BBC’s website, even those that delineate state lines. These maps charting gang presence also include different sets of cartels and minimize the extent of this “disputed territory” indicator. Nevertheless, the other charts mostly appear to derive from the map offered in this report.

Similarly, the Trans-Border Institute has published the following chart in their 2010 report on the drug violence in Mexico.[27] This chart records the gang-on-gang violence in Mexico by group and fits well with the gang competition framework established earlier in this report. The chart is particularly telling and highlights how the presence of multiple gangs is responsible for some 63.9% of all the killings within Mexico between 2006 and 2010.[28]

The information on the US street price of a gram of marijuana and cocaine were compiled from different websites. The data on the price of a gram of marijuana came from High Times magazine’s online price index of marijuana. There are few if any sources that consistently index the US street price for Marijuana and likely none disaggregate the data by the quality of the Marijuana. Given the notoriety of High Times and their trust by growers nationally and internationally, it was selected as the study’s primary source.  The fall price averages for an ounce of Marijuana between 2006 and 2010 were used.[29] Because the price of marijuana seems to fluctuate with the seasons like most agricultural products, the prices compiled are not the best indicators. The average US street price of cocaine on the other hand, is readily available and regularly posted by the Drug Enforcement Agency of the United States. The 2010 National Drug Threat Assessment records the price movements of a gram of cocaine between March of 2006 and September of 2009. The average of these prices for 2006 through 2009 was used to register the average price for the years spanning this investigation’s dataset. For 2010 the assumption was made that the price of cocaine did not change which is highly unlikely. This index also includes the average purity of the cocaine, which should impact the price of the cocaine. Purity was not included in this investigation however. A table presenting the DEA information is below.[30]

With this information the dataset is now complete and we may begin to test these hypotheses. Using this model, four different regressions were run using executions as the dependent variable in all four of them. Tables 1-4 are simple OLS regressions. Table 1, tests how the presence of gangs influences the levels of executions in a municipality. Table 2, adds geographic variables, including latitude and longitude. Table 3, I adds the price of marijuana and cocaine as independent variables. Table 4, gauges how gang competition impacts the levels of violence as opposed to situations where only one gang is present – a test of Kaylvas’s hypothesis. To do this, dummy variables for states  were created where there was one gang present and states where at least two gangs were present. As mentioned earlier, the indicator for “disputed territory” was not included as the definition for this term was unclear. Population was used as a control variable throughout Tables 2-4.

The results are recorded below. The results for all of these variables are significant in the 99th percentile, but do not always point in the expected direction.




Variable |Executions1 Executions2 Executions3 Executions4


gangs | 1.291867   -2.8170182  -2.8136921

population |.00007818    .00008279   .00008272   .00008265

distborder |            -.04222502  -.04233203  -.04268322

distusa |              .0403162   .04038155   .04074405

distguatem~a |             .01155845   .00934688   .00910484

longitude               1.2320695   1.0371911   1.0443679

latitude |             4.3410092   4.4601582   4.5179709

marijuana |                        -3.4816897  -3.4830715

cocaine |                         .10082432   .10082976

gangs1 |                                    -4.4467117

gangs2 |                                    -5.5058706


Number of observations =    5720




homicides |      Coef.   Std. Err.      t    P>|t|


gangs |   1.291867   .5854125     2.21   0.027

population |   .0000782   3.48e-06    22.45   0.000


Number of observations =    5720




homicides |      Coef.   Std. Err.      t    P>|t|


gangs |  -2.817018   1.078084    -2.61   0.009

distborder |   -.042225   .0080881    -5.22   0.000

distusa |   .0403162   .0078879     5.11   0.000

distguatem~a |   .0115585   .0023459     4.93   0.000

population |   .0000828   3.67e-06    22.56   0.000

longitude |    1.23207   .2144801     5.74   0.000

latitude |   4.341009   .7940334     5.47   0.000


Number of observations =    5720




homicides |      Coef.   Std. Err.      t    P>|t|


gangs |  -2.813692   1.075044    -2.62   0.009

distborder |   -.042332   .0080659    -5.25   0.000

distusa |   .0403816   .0078659     5.13   0.000

population |   .0000827   3.66e-06    22.60   0.000

marijuana |   -3.48169   1.675852    -2.08   0.038

cocaine |   .1008243   .0216514     4.66   0.000

distguatem~a |   .0093469    .003057     3.06   0.002

longitude |   1.037191   .2752904     3.77   0.000

latitude |   4.460158   .7989491     5.58   0.000


Number of observations =    5720




homicides |      Coef.   Std. Err.      t    P>|t|


gangs1 |  -4.446712   1.823674    -2.44   0.015

gangs2 |  -5.505871   2.152837    -2.56   0.011

distborder |  -.0426832    .008072    -5.29   0.000

distusa |   .0407441   .0078726     5.18   0.000

population |   .0000826   3.66e-06    22.58   0.000

marijuana |  -3.483072   1.675819    -2.08   0.038

cocaine |   .1008298   .0216509     4.66   0.000

distguatem~a |   .0091048   .0030647     2.97   0.003

longitude |   1.044368    .275361     3.79   0.000


latitude |   4.517971   .8006335     5.64   0.000


Number of observations =    5720



 The following matrix illustrates whether the hypotheses were confirmed by the table as well as the level of statistical significance of these results.

Hypothesis and Table Matrix of Results Executions1 Executions2 Executions3 Executions4
H1  NA No*** No*** No***
H2 NA No*** No*** No***
H3 NA Yes*** Yes*** Yes***
H4 NANA NANA No***Yes*** No***Yes***
H5 Yes*** No*** No*** No/Yes***

* p<0.05, ** p<0.01, *** p<0.001

H1       This matrix should present the findings in a way that is most accessible to the reader. Looking to the first hypothesis: that cities closer to the US border should exhibit more violence; surprisingly, we find negative results. We see that the hypothesis is not confirmed and is consistent throughout all tables. The level of significance remains constant in the 99th percentile. Thus we observe that cities closer to the US border should experience relatively lower levels of executions. This finding is rather puzzling however given that some 50% of the drug related deaths in 2010 occurred in the states of Sinaloa, Tamaulipas and Chihuahua where Chihuahua and Tamaulipas both border the United States. In addition, the most violent city in Mexico for 2008 through 2010 is without question Ciudad Juarez in Chihuahua. It is possible that the relative levels of violence are increasing country wide throughout these 5 years and rising faster in those municipalities further away from the US border. Perhaps the second hypothesis may confirm this explanation as it measures the influence of the Guatemalan border at the opposite end of Mexico. In the future, robustness checks may be instituted into this model to measure at what distance from the US border are executions in a municipality positively correlated with the distance and provide some measure of significance. In a future study, it might be hypothesized that municipalities that fall within the mean distance of cities from the US border, throughout all of Mexico, confirming the hypothesis that executions increase as you approach the US border. In addition, the US border may be characteristically different from other borders with Mexico given the strength of US law enforcement agencies that monitor the border.

H2       The second hypothesis, that as one approaches the Guatemalan border the levels of violence increase, is not confirmed for any tables that apply. The coefficients point in the unexpected positive direction (that is as distance from the Guatemalan border increases, the level of executions increases) and are significant in the 99th percentile. This result also does not necessarily confirm the possibility that throughout the 5 year period violence has spread further south as a percent of overall drug related violence within Mexico, which I argued might be a reason we observe a positive relationship between the levels of executions and distance to the US border. This finding should relieve Mexican citizens nearest the Guatemalan border, whom we read from newspaper reports are currently experiencing higher levels of cartel and drug-related violence.[31] This conclusion, however, may not bode so well for Guatemalans on the other side of the Mexico border who are not recorded in this study.

H3       The third hypothesis: that as one approaches the Mexican border the levels of violence should increase is, on the other hand, completely confirmed by the results of the tables. All relevant tables confirm the expected negative relationship between the distance between a municipality and the observed levels of executions. The explanatory power of this measure of distance to the Mexican border, however, may be understated for this investigation. In addition to capturing the Pacific and Gulf coasts, this variable will also capture the same interactions of the US and Guatemalan borders captured in the previous two regressions which found the opposite relationship. Without the inclusion of these two borders in the variable it is likely the coefficients would have been even more negative. Perhaps in the future, one might try to only capture this interaction for states that do not share a border with the US or Guatemala, but that might restrict the sample and obviously bias the information too much. Either way, the results are highly significant and confirm the third hypothesis,

H4       Hypothesis four held that as the price of drugs increased so should the levels of violence. There were two types of drugs used in this investigation, marijuana and cocaine. In the matrix of results presented earlier, the first result refers to marijuana and the second to the results for cocaine. Here too we observe different effects. Throughout every relevant table the price of cocaine is positively correlated with the levels of violence in a municipality and is highly significant. The price of marijuana on the other hand does not seem to be negatively correlated with the levels of violence and is not significant at all for the levels of executions in a city. Considering that the price of marijuana has not fluctuated too much throughout the course of this conflict, this may not be so unsurprising. If it were possible to include the volumes of marijuana and cocaine transported between the borders perhaps we could see why the prices of marijuana have remained so stagnant while cocaine has increased so precipitously. Perhaps given the lower quality nature of marijuana trafficked from Mexico, there is some consumer substitution effect in favor of higher quality marijuana trafficked from Canada; it is unclear. More likely, the alternative methods of sourcing the price of marijuana conducted by this paper are to blame for the mixed and counterintuitive results considering that the volume of marijuana trafficked across the border, as derived from drug seizures at the US border, is much higher than the volume of cocaine trafficked.[32] Further, there is an endogeneity problem with the price of drugs and the number of deaths observed in a city. As the conflict progresses and the number of deaths increases, necessarily the price of cocaine reflecting this increased cost of business should push higher. And if, as this paper argues, as the value of cocaine increases so does the level of violence, we uncover a positive feedback mechanism whereby violence increases the price of cocaine and therefore increases the likelihood of violence. In this undesirable scenario, violence and government pressure increases the value of these routes by pushing the price of cocaine higher thus incentivizing more violence. Either way, from these particular results it appears the higher the price of cocaine the more violence will be observed, whereas the price of marijuana matters less or is negatively correlated with drug related violence.

H5       Hypothesis five holds that the presence of multiple gangs should increase the levels of executions in a municipality whereas municipalities with one gang present should have relatively fewer executions. As expected, before instituting more variables, we observe that as the number of gangs increases the levels of executions rises. This interaction is captured in the “gangs” variable in table one. These findings are also significant in the 99th percentile. The rest of the tables, however, do not confirm Hypothesis 5 and are also significant in the 99th percentile. After instituting more variables the “gangs”, and “gangs2” variables, exhibit strong and significant negative correlations with executions. In an almost complete reversal, the coefficient for “gangs” and “gangs2” dives lower, disproving that portion of Hypothesis 5. Given model one, the results are somewhat mixed with the relationship between multiple gangs and the levels of violence, but trusting more in the quantity of results displayed in models 2-4 that introduce more than one variable, it appears multiple gangs do not increase the levels of violence. The coefficient on the “gangs1” variable, on the other hand, points in the expected negative direction suggesting the presence of only one gang reduces the levels of executions in a municipality. These results are also incredibly significant as well, in the 99th percentile, confirming that portion of Hypothesis 5. The poor coding methods of gang presence of this investigation, however, heavily influenced these conclusions regarding Hypothesis 5 and it is this poor method that likely provides these counterintuitive results. Whereas gang presence is coded at the state-level executions are coded at the municipal level. The result is many cities and towns that experienced no executions, likely because of the low population size and little strategic importance of the town, were coded as harboring at least one gang, when these towns in reality did not. A future paper with a better method of coding gang presence at the municipal level would go a long way to empirically proving this positive association between multiple gang presence and the levels of executions. But until then, the results of this investigation seem to generally illustrate the dominance of more permanently fixed geographic factors, like city location, than the more fluid, less permanent factors like gang-presence.



These results confirm several of the five hypotheses offered earlier in the examination. Aggregated, this empirical evidence illustrates that municipalities closer to the Mexican border, but not the US or Guatemalan border, are much more likely to experience higher levels of drug related violence in the form of executions. This finding is further exacerbated so long as the price of cocaine rises, which it has precipitously and is expected to continue to do so. These price increases will make most cities vulnerable to higher levels of violence, though the effect will be disproportionate depending on whether someone lives near the Mexican border, but not necessarily the US or Guatemalan border. These findings mostly confirm the theoretical contributions of Ross’s work that hold violence will occur in more valuable regions. In Mexico, cities near smuggling routes on the coasts are more valuable to the cartels and so these cities experience more violence as expected under Ross’s framework. Though the results for these two international borders related variables do not confirm Ross’s hypotheses, this may simply illustrate that the US and Guatemalan international borders are not as valuable as hypothesized by this paper. The evidence regarding gang presence was not definitive either, but mostly suggested that the presence of multiple gangs did not have the expected positive impact on the levels of executions. Municipalities that host only one gang will experience suppressed levels of executions as hypothesized. Ultimately, this evidence might confirm the original Kalyvas hypothesis that when sovereignty is shared the levels of selective violence are lower. The investigation disregarded that notion, arguing that it did not apply to gang violence in Mexico between cartels that have few incentives or opportunities to share market control and access. Perhaps this restriction of the Hypothesis 5 was incorrect. As Kalyvas might argue, it may be that the amount of information made available to cartels when they are contesting a territory is sufficiently diminished that executions are less common. However, as mentioned earlier and in light of the table produced by the Trans-Border Institute that illustrates the quantities of the group-on-group violence, the different methods of coding for gang presence and executions is likely to blame for the negative correlations between multiple gang presence and the level of executions. Turning back to the literature on civil wars, though, hopefully this study might be able to derive some policy implications from these results that may reduce the levels of executions mapped in this study.

Given this realization that resource-financing opportunities substantially increase the likelihood of conflict onset, many new authors have sought to uncover the mechanism by which financing results in rebellion. Within the “greedy rebel” framework, the looting of local resources appears motivated purely by financial gain. This method of financing, however, impacts the nature of rebel-recruitment within the organization, impacting membership, structure and behavior.[33] Jeremy Weinstein points to a sort of rebel “resource curse” whereby the strength of rebel groups is undermined by resource abundance. Over time, the recruits join these resource abundant groups for the rents and not the cause, poisoning the ideological quality of the recruits. Weinstein’s interpretations of the interaction between resources and rebel group structures fits well with this discussion considering Weinstein focuses exclusively on material motivations—the only motivations that exist within Mexico’s drug-gangs. As such, much of the cartel labor force consists of low skill drug-mules, transporting drugs to and from the United States. These mules are involved with these cartels for the opportunity to make high sums of money in very short periods of time with minimal effort exerted, though admittedly the risks are quite high. The mules are likely to be poorer than the average Mexican and less skilled. They are also a fluid source of labor, resembling something close to migrant workers.

All these characteristics point to a low level of commitment from these recruits and may explain the recent setbacks of many of these cartels throughout Mexico in the face of government intervention by the highly motivated, trained and equipped military forces. The relative strengths of these two organizations, the cartels and the state, are skewed in favor of the state. In a military confrontation the state has the advantage in terms of expertise, resources, and hardware. With a continued sustained crackdown on the cartels the government should be able to dismantle their networks in the same way the Medellin Cartel in Colombia was eventually disbanded, but as Jorge Chabat, Mexican counter-narcotics policy expert, reminds us, the most sinister weapon the cartels have is the power to corrupt.[34] Even so, already there is evidence that La Familia, one of the most prominent drug gangs in Mexico has disbanded after continued pressure from the Mexican military. With the leadership of La Familia either killed or in jail, in 2011, signs were posted notifying the authorities and the communities of Michoacán that the gang has begun to dissolve itself.[35] This success is noteworthy and commendable, but also disappointing in one large sense; there are no alternative solutions to the application of force between the state and drug gangs.

The unique illicit nature of drugs introduces peculiar constraints on the termination processes of civil conflicts. Empirical analysis of the relationship between resource-financing and conflict duration tends to validate this notion of rebellion as a criminal enterprise or business venture, whereby rebels continue to oppose the government because of the high rents they collect.[36] In addition, the low opportunity costs of recruitment, given the relative poverty, the low skill and low commitment status of these recruits ensures an almost endless supply of traffickers and an extension of the conflict in Mexico.[37] These two tendencies, high rents and cheap labor, only serve to increase the duration of conflict. Studies by Fearon and Collier both confirm that conflicts tend to be longer where rebels have access to illicit financing.[38] Using thirteen case studies, Michael Ross also confirmed this positive relationship between resource wealth and the duration of conflict that included drug resources.[39] Svante Cornell estimates that where rebels relied on contraband for financing, the mean duration of conflict was 48.2 years.[40] Furthermore, the increases in the price of the commodities that finance rebel activity tend to prolong conflicts as well. This last point regarding commodity prices is very important in the case of Mexico as the price of cocaine has increased some 75.4% between January, 2007 and September of 2009.[41] These higher prices of cocaine can extend duration of the conflict in Mexico through several channels. Higher rents on cocaine increase the economic benefit of distributing those illicit substances and, as a result, cartels are much less likely to forego such profitable opportunities. In addition, higher prices of cocaine not only increase the motivation for rebel groups to continue opposing the government, but also increase their relative capacity to resist. Interdiction efforts, the preferred method of policing the supply of drugs, raise the price of drugs, which only increases the value of the routes and the level of violence. Without such high rents the cartels could not expect to resist the government, or each other for that matter, as much as they have already.



This research set out to explain the patterns of the gang-on-gang violence experienced in Mexico between 2006 and 2010 following President Felipe Calderon’s declaration of war on Mexico’s drug cartels. Utilizing the relevant civil war literature on greedy rebels, the influences of resource presence on conflicts, the spatial diffusion of violence and its intensity, this study has yielded several hypotheses to understand the patterns of Mexico’s gang violence. It appears that the levels of executions within Mexico are positively influenced by rising drug prices as well as proximity to the smuggling routes at the Mexican border.  However, executions do not seem to be positively correlated with proximity to the US or Guatemalan borders. Seemingly, these results may contribute to law enforcement efforts in countries or even municipalities facing similar gang driven drug-violence. In an immediate sense, these results do not point definitively to a more general southward migration in drug-related violence that would otherwise threaten Mexicans near Guatemala, though the majority of violence is concentrated on the Guatemalan side of the border already. Unfortunately though, the policy conclusions we derive from these results are damning for traditional US and Mexican counter-narcotics efforts, which focus mostly on drug interdiction efforts. These seizures of narcotics increase the price of drugs by decreasing their supply on the market as well as the costs associated with trafficking. Given the highly significant and positive impacts the price of drugs has on the levels of executions in Mexico’s cities, a new approach to fighting narcotics traffic seems necessary to avoid the collateral damage of heightened gang competition and the subsequent flurry of executions.


 BBC. 2008. “Crime ‘damages’ Mexico economy.”

BBC. 2010.“Calderon: Mexico drug gangs seeking to replace state.”

BBC. 2011. “La Familia drug gang: Mexico says cartel ‘in retreat.”

BBC. 2011. “Guatemala ‘drug lord’ Juan Ortiz Lopez captured by US.”

BBC. 2011. “Q&A: Mexico’s drug-related violence.”

BBC. 2010. “Clinton says Mexico drug crime like an insurgency.”

Buhaug, Halvard and Gates, Scott. 2002. The Geography of Civil War. Journal of Peace Research, Vol. 39, No. 4, Special Issue on Civil War in Developing Countries. Sage Publications, Ltd.

Central Intelligence Agency. 2011. World Factbook, Population of Mexico.

Chabat, Jorge. 2002. Mexico’s War on Drugs: No Margin for Maneuver. The American Academy of Political and Social Science.

Collier, Paul and Hoeffler, Anke. 2005. Resource Rents, Governance, and Conflict. The Journal of Conflict Resolution, Vol. 49, No. 4, Paradigm in Distress? Primary Commodities and Civil War. Sage Publications.

Collier, Paul and Hoeffler, Anke. 2004. Greed and Grievance in Civil War. Oxford University Press.

Cornell, Svante. 2005. The Interaction of Narcotics and Conflict. Journal of Peace Research, Vol. 42, No. 6. Sage Publications, Ltd.

Cunningham, David. 2006. Veto Players and Civil War Duration. American Journal of Political Science, Vol. 50, No. 4. Midwest Political Science Association.

Durnan, Michael and Peceny, Mark. 2006. The FARC’s Best Friend: US Antidrug Policies and the Deepening of Colombia’s Civil War in the 1990’s. Latin American Politics and Society, Vol. 48, No. 2. Blackwell Publishing on behalf of the University of Miami.

Engvall, Johann. 2006. The State under Siege: The Drug Trade and Organised Crime in Tajikistan. Europe-Asia Studies, Vol. 58, No. 6. Taylor and Francis, Ltd.

Fearon, James. 2004. Why Do Some Civil Wars Last so Much Longer than Others? Journal of Peace Research, Vol. 41, No. 3. Sage Publications, Ltd.

Fearon, James and Laitin, David. 2003. Ethnicity, Insurgency and Civil War. The American Political Science Review, Vol. 97, No. 1. American Political Science Association.

Fox News Latino. 2011. “Mexico Condemns Comment that U.S. Might Have to Send Troops to Fight Cartel “Insurgency”.”

Galeotti, Mark. 2003. Narcotic Knot. The World Today, Vol. 59, No. 12. Royal Institute of International Affairs.

Hegre, Havard. 2004. The Duration and Termination of Civil War. Journal of Peace Research, Vol. 41, No. 3. Sage Publications, Ltd.

High Times. 2008. “June 2008 THMQ.”

Humphreys, Macartan. 2005. Conflict, and Conflict Resolution: Uncovering the Mechanisms. The Journal of Conflict Resolution, Vol. 49, No. 4, Paradigm in Distress? Primary Commodities and Civil War. Sage Publications, Ltd.

Kalyvas, Stathis. 2006. The Logic of Violence in Civil War. Cambridge University Press.

Kalyvas, Stathis. 2001. New and Old Civil Wars: A Valid Distinction? World Politics, Vol. 54, No. 1. Cambridge University Press.

Kalyvas, Stathis. 2004. The Paradox of Terrorism in Civil War. The Journal of Ethics, Vol. 8, No. 1. Springer.

Lujala, Paivil and Gleditsche, Nils and Gilmore, Elisabeth. 2005. A Diamond Curse? Civil War and a Lootable Resource. The Journal of Conflict Resolution, Vol. 49, No. 4, Paradigm in Distress? Primary Commodities and Civil War. Sage Publications, Ltd.

Malkin, Victoria. 2011. Narcotrafficking, Migration and Modernity in Rural Mexico. Latin American Perspectives, Vol. 28, No. 4. Mexico in the 1990’s: Economic Crisis, Social Polarization, and Class Struggle, Part 2. Sage Publications, Ltd.

New York Times. 2008. “Mexico’s war against drugs is killing its police.”

Poire, Alejandro. 2011. Presentacion de la Base de Datos de Presuntos Homicidios Relacionados con la Delincuencia Organizado.

Patenostro, Silvana. 1995. Mexico as a Narco-Democracy. World Policy Journal, Vol. 12, No. 1. The MIT Press and the World Policy Institute.

Recio, Gabriela. 2002. Drugs and Alcohol: US Prohibition and the Origins of the Drug Trade in Mexico, 1910-1930. Journal of Latin American Studies, Vol. 34, No. 1. Cambridge University Press.

Regan, Patrick and Norton, Daniel. 2005. Greed, Grievance, and Mobilization in Civil Wars. The Journal of Conflict Resolution, Vol. 49, No. 3. Sage Publications, Ltd.

Rios, Viridiana and Shirk, David. 2011. Drug Violence in Mexico, Data and Analysis Through 2010. Trans-Border Institute, University of San Diego (Feb., 2011).

Ross, Michael. 2006. A Closer Look at Oil, Diamonds and Civil War. Annual Review of Political Science.

Ross, Michael. 2004. How Do Natural Resources Influence Civil War? Evidence from Thirteen Cases. International Organization, Vol. 58, No. 1. Cambridge University Press on behalf of the International Organization Foundation.

Ross, Michael. 2004. What Do We Know about Natural Resources and Civil War? Journal of Peace Research, Vol. 41, No. 3. Sage Publications, Ltd.

Snyder, Richard and Bhavnani, Ravi. 2005. Diamonds, Blood and Taxes: A Revenue-Centered Framework for Explaining Political Order. The Journal of Conflict Resolution, Vol. 49, No. 4, Paradigm in Distress? Primary Commodities and Civil War. Sage Publications, Ltd.

Steinberg, Michael. 2000. Guerillas, Drugs and Third World War-Making. Geographical Review, Vol. 90, No. 2. American Geographical Society.

Stratfor Global Intelligence. 2009. “Mexican Drug Cartels: Two Wars and a Look Southward.” Stratfor Global Intelligence.

Time Magazine. 2008. Mexico’s Narco-Insurgency.”

US Department of Justice. 2010. National Drug Threat Assessment 2010. National Drug Intelligence Center.

Weidmann, Nils. 2009. Geography as Motivation and Opportunity: Group Concentration and Ethnic Conflict. The Journal of Conflict Resolution, Vol. 53. Center for Comparative and International Studies.

Weinstein, Jeremy. 2005. Resources and the Information Problem in Rebel Recruitment. The Journal of Conflict Resolution, Vol. 49, No. 4, Paradigm in Distress? Primary Commodities and Civil War. Sage Publications, Ltd.


[1]Mexico’s war against drugs is killing its police.” New York Times, May 26, 2008.

[2] Gabriela Recio, “Drugs and Alcohol: US Prohibition and the Origins of the Drug Trade in Mexico, 1910-1930,” Journal of Latin American Studies, Vol. 34, No. 1 (Feb., 2002), pp. 22 (

[3] Ibid, pp. 36

[4] Ibid, pp. 42

[5] Johan Engvall, “The State Under Siege: The Drug Trade and Organised Crime in Tajikistan,” Europe-Asia Studies, Vol. 58, No. 6 (Sep., 2006), pp. 827,

[6] “Calderon: Mexico drug gangs seeking to replace state,” BBC, Aug. 5, 2010. These taxes are likely considered to be “protection” fees that keep gang members from vandalizing property.

[7] “Crime ‘damages’ Mexico economy,” BBC, Sept. 3, 2008.

[8] Johan Engvall, “The State Under Siege: The Drug Trade and Organised Crime in Tajikistan,” Europe-Asia Studies, Vol. 58, No. 6 (Sep., 2006), pp. 847,

[9] Viridiana Rios and David A. Shirk, “Drug Violence in Mexico, Data and Analysis Through 2010.” Trans-Border Institute, University of San Diego (Feb., 2011). This is one of the few academic reports on the drug-violence in Mexico that draws on the same data and databases as this report.

[10] Stathis N. Kalyvas, “New and Old Civil Wars: A Valid Distinction?” World Politics, Vol. 54, No. 1 (Oct., 2001), pp. 100,

[11] Johan Engvall, “The State Under Siege: The Drug Trade and Organised Crime in Tajikistan,” Europe-Asia Studies, Vol. 58, No. 6 (Sep., 2006), pp. 828,

[12]Clinton says Mexico drug crime like an insurgency” BBC, Sep. 9th, 2010.

[13]Mexico Condemns Comment that U.S. Might Have to Send Troops to Fight Cartel “Insurgency”.” Fox News Latino, Feb. 9th, 2011. Speaking at a University Audience, Undersecretary of the U.S. Army, Joseph Westphal, the second highest-ranking civilian official in the U.S. Army, referred to the most recent drug-related violence in Mexico as an “insurgency”. Undersecretary Westphal has since retracted his comments and apologized for misspeaking.

[14] Nils B. Weidmann, “Geography as Motivation and Opportunity: Group Concentration and Ethnic Conflict.” The Journal of Conflict Resolution, Vol. 53, (Aug., 2009), pp. 487-495.

[15] Paul Collier and Anke Hoeffler, “Greed and Grievance in Civil War.” Oxford Economic Papers 56, (2004), pp. 563.

[16] Ibid. pp. 103.

[17] Jorge Chabat, “Mexico’s War on Drugs: No Margin for Maneuver” Annals, AAPSS, 582, (July, 2002), pp. 137. That revenue figure represents between 1% and 3% of Mexico’s GDP.

[18] Michael L. Ross, “How Do Natural Resources Influence Civil War? Evidence from Thirteen Cases.” International Organization, Vol. 58, No. 1, (Winter, 2004), pp. 55,

[19] Ibid pp. 45.

[20] Stathis N. Kalyvas, “The Logic of Violence in Civil War.” Cambridge University Press (2006), pp. 174.

[21] CIA World Factbook, Population of Mexico. Estimates of the population of Mexico by July 2011.

[22] Dr. Alejandro Poire, “Presentacion de la Base de Datos de Presuntos Homicidios Relacionados con la Delincuencia Organizado.” (Jan., 2011). A Speech presenting the new database on drug gang-related murders.

[23] Ibid.

[24] Ibid.

[25]Q&A: Mexico’s drug-related violence.” BBC, February 1, 2011. This map was provided by Stratfor Security, a private global intelligence firm and published on the BBC’s website. The company is well respected as a security consultant with client lists that include governments.

[26] “Mexican Drug Cartels: Two Wars and a Look Southward.” Stratfor Global Intelligence. December 17, 2009.

[27] Viridiana Rios and David A. Shirk. “Drug Violence in Mexico, Data and Analysis Through 2010.” Trans-Border Institute, University of San Diego (Feb., 2011).   pp. 22.

[28] Ibid. pp. 22. I arrived at the number 63.9% by adding the figures from % of Total for all rows except “Other”. It is unclear why adding the “Other” row, however, still leaves the summation .2% short of 100%.

[29] “June 2008 THMQ” High Times, Market analysis, Current MIDS Index. This price index covers the average price of an average quality of marijuana by the ounce. From here I converted the ounces to grams to match my measurement unit for cocaine.

[30] National Drug Threat Assessment 2010, US Department of Justice, National Drug Intelligence Center, 2010.

[31]Guatemala ‘drug lord’ Juan Ortiz Lopez captured by US.” BBC News, March, 30 2011.

[32] The Department of Justice on their website details the quantity of various drugs seized at the US border annually. The volume of marijuana seized at the border is typically 50 times that of cocaine.

[33] Jeremy M. Weinstein, “Resources and the Information Problem in Rebel Recruitment”, The Journal of Conflict Resolution, Vol. 49, No. 4, Paradigm in Distress? Primary Commodities and Civil War (Aug., 2005), pp. 598,

[34]Mexico’s Narco-Insurgency”, Time Magazine, Jan. 25, 2008.

[35]La Familia drug gang: Mexico says cartel ‘in retreat’”, BBC, Jan. 26, 2011.

[36]The Duration and Termination of Civil War” Havard Hegre, Journal of Peace Research, Vol. 41, No. 3 (May, 2004), pp. 245.

[37] Ibid., pp. 246.

[38] Ibid., pp. 247. Cites Fearon (2004) and Collier (2004).

[39] Michael L. Ross, “How Do Natural Resources Influence Civil War? Evidence from Thirteen Cases.” International Organization, Vol. 58, No. 1, (Winter, 2004), pp. 52,

[40] Svante E. Cornell, “The Interaction of Narcotics and Conflict”, Journal of Peace Research, Vol. 42, No. 6 (Nov., 2005), pp. 753.

[41] U.S. Department of Justice, National Drug Threat Assessment, 2010, National Drug Intelligence Center, (May, 2010), pp. 28.

Leave a Comment

YRIS is a student publication, and Yale University is not responsible for its content.