Diego Valle's Blog

Interactive map of the drug war in Mexico

2012-01-25T12:18:00.001-06:00

Click on the image to visit the interactive drug war map. Or try the Spanish version

If you're interested at all in what's happening in Mexico you can't miss the interactive map of the drug war I just made. You can link directly to cities or whole regions within Mexico and post them to Twitter and Facebook by clicking on the "Share This Map" link at the bottom of the box. You can even compare 2007 México with 2010 México and switch between drug war-related homicides and total homicides (the ones from the INEGI). If you hover over the chart you'll get the monthly values and information on important events. To top it off you can export the monthly data to csv. You'll need a latest generation browser to use it.

To unclutter the map and following the lead of the paper Trafficking Networks and the Mexican Drug War by Melissa Dell, I decided to only show the optimal highways (according to my own data and Google Directions) to reach the US border ports from the municipalities with the highest drug plant eradication between 1994 and 2003 and the highest 2d density estimate of drug labs based on newspaper reports of seizures. The map is a work in progress and is still missing the cocaine routes, but hopefully I'll be able to add them shortly.

My assumptions in making the map:

Homicides that were registered with no date of occurrence were assumed to have occurred on the month they were registered.
The total homicide database has a cutoff date of the last day of the year for recording deaths, so for example, in 2009 there occurred 671 homicides that weren't registered until 2010 (most occurred in December). I adjusted the 2010 database assuming the homicides were under counted by the same percentage as they were in 2009. So instead of the 25,005 homicides in the database I'm showing the adjusted number of 25,679. I used a similar adjustement at the municipality level.
Even though the municipalities of Culiacán and Navolato are not officially a metro area I considered them one since they are only half and hour from each other and together have a million inhabitants.

The really cool thing about the map is that it makes it very easy to select regions of Mexico and link directly to them, which makes refuting mistaken claims by government officials, like the one Poiré made last year, a cinch.

In Jalisco, Colima, and Nayarit, violence was increasing systematically previous to the killing of Nacho Coronel at a rate of more than 1 death per week, or 6 extra deaths every five weeks. After his death violence remained at a high level, but it increased at a much slower rate, barely 1 death every five weeks.
[...]
Replicating the analysis with data from the state of Jalisco, the region where Nacho Coronel was killed, and even with data from Zapopan, we obtain the same results

--Alejandro Poiré, Nexos

But did violence really stop increasing near Guadalajara? This should serve as a warning not to extrapolate based on a few months of observations and underlines the importance of making the data available to researchers on a prompt basis to avoid embarrassing mistakes.

There are about a million things you can analyze with the maps:

The weird falls in homicides in Chiapas around the end 2007 and 2009.
Why did Sinaloa end up with more drug war-related homicides than total homicides from mid 2007 until the end of 2008? (the answer will be my next post)
The sudden increase in homicides in Nuevo León and Tamaulipas around the end of February when the Zetas and CDG (Gulf Cartel) went to war with each other.
Perhaps you heard that México is much safer than a certain other much smaller country or sub-country region and wish to only compare certain parts of México with a similar population/shape to the much smaller country/sub-country region.
The super secret link to the 2011 drug war-related homicides. (The size of the circles and the color scale used to fill them was annualized so it's on the same scale as the years for which full data is available, but the numbers shown on the map correspond to the Jan-Sep data).
Are the Zetas really the most brutal cartel as the Mexican Government and Stratfor assert? You can compare northeastern Mexico to Chihuahua and find out just how mistaken they are, both in terms of rates and total homicides.
How the Tubutama massacre is registered as having occurred in Tubutuma according to the homicide database, but in Saric according the drug war-related homicide database

A note about drug war-related homicides

I'd be very surprised if the Mexican government had the capacity to correctly count the drug war homicides. There were big differences between the homicide databases starting in 2009 in Tijuana and in 2010 in Juárez. I tend to think of the drug war-related homicides as an independent count of a subset of firearm/extremely violent homicides based on police records rather than death certificates (independent of whether organized crime was involved or not). Looking at the whole country there has been a steady increase in the difference between INEGI homicides and the drug-war related ones.

This is not to say that the data for INEGI is without errors, besides not having registered the mass grave in Taxco and the immigrant massacre in San Fernando, there has been a steady increase in deaths of unknown intent by external injury caused by firearm. In Mexico most accidents are by transportation, most suicides by suffocation and most homicides by firearm, so a quick and dirty way to see if a more in-depth analysis is needed is to look at firearm deaths:

This is not to say all the deaths were homicides, since it would be perfectly reasonable to expect that as the availability of firearms increases, the number of accidents involving firearms increases, but the evidence does suggest that there has been an important under counting of homicides and even more so of drug war-related homicides.

Deaths of unknown injury intent in all of Mexico that were by firearm:


Year	Cause	Deaths
2004	Firearm	399
2005	Firearm	431
2006	Firearm	373
2007	Firearm	520
2008	Firearm	494
2009	Firearm	737
2010	Firearm	1,063

You may also be wondering why there were more drug-related homicides than total homicides in the Frontera Chica, but not in the northern region Nuevo León. The most likely cause is that some homicides were registered as deaths of unknown intent in Tamaulipas, but not in Nuevo León.

Deaths by firearm that were not homicides in Tamaulipas:


Year	Cause	Deaths
2004	Firearm	46
2005	Firearm	47
2006	Firearm	33
2007	Firearm	39
2008	Firearm	54
2009	Firearm	45
2010	Firearm	251

A similar thing happened in Durango, it's most obvious in La Laguna. Deaths by firearm that were not homicides in La Laguna:


Year	Cause	Deaths
2004	Firearm	9
2005	Firearm	10
2006	Firearm	10
2007	Firearm	14
2008	Firearm	28
2009	Firearm	61
2010	Firearm	120

P.S. You can download the source at GitHub

Homicides in Mexico 2010

2011-12-28T01:00:00.000-06:00

Recently the Mexican government released to the public the final homicide data for 2010, and as you can see from the chart Mexico has suffered from a steep rise in homicides since 2008, and in 2010 reached the highest homicide rate in recent history.

The Mexican government had actually released preliminary homicide data a couple of months ago, but I though it looked incomplete so I didn't blog about it. For some reason (too much ponche?) the INEGI hasn't issued a press release with the final data --including more than 1,000 previously uncounted homicides. Hopefully they'll do it sometime soon.

Because of the cutoff date for registering deaths of December 31 there is always an under-counting of homicides during the last year for which data is available. Taking this into account, the number of homicides in Mexico for the last few years has been:

Year	Homicides
2004	9,347
2005	9,983
2006	10,426
2007	8,840
2008	14,175
2009	19,771
2010	25,800 (25,000 in the database)

México had an homicide rate of 23 in 2010. There has been a steady growth in the increase in violence each year since 2008: from 2007 to 2008 homicides grew by 5,300, the next year they grew by 5,600 and in 2010 they grew by 6000.

Problems with the data
Given previous problems with the recording of homicides I did some quick checks of the data and found the following issues:

The 55 bodies found in the mass grave in the municipality of Taxco were not recorded in the database (not even as accidents). In the drug war-related homicide database 55 deaths were recorded during the month of May.
There were an unusually high number of deaths of unknown injury intent in the state of Tamaulipas, and many parts of the state had more drug war-related homicides than total homicides.
In the muncipality of San Fernando 72 immigrants were killed in August, but only 51 deaths were recorded in the database for that month. In the drug war-related homicide database 89 deaths were recorded during August.

Obviously these problems deserve a more rigorous analysis.

The Charts

Since the median age in México went from 19 in 1990 to 26 in 2010 I figured I'd age adjust the data using the 2000 Mexican population estimates from the CONAPO/COLMEX

All the border states now suffer from historical homicide records:

For comparison purposes the most violent state in the US, Louisiana, had a homicide rate of 11 in 2010.

In southern and central Mexico violence rose in Morelos and to a much lesser extent in the State of México and the Federal District. Sadly, while Michoacán and Guerrero saw violence decrease in 2010, all the available information points to an increase in violence in 2011 as the capture of "La Barbie" and the split of La Familia Michocana with the Caballeros Templarios led to an unraveling of alliances between drug gangs.

The split between the Zetas and the Gulf Cartel led to an increase in violence in Tamaulipas and Nuevo León (and to a lesser extent San Luis Potosí)

Since the rise in violence in northeastern Mexico was so sudden it would point to at least some level of organization and vertical integration in the Zetas and the Gulf Cartel (only in México obviously).

In Nayarit, Colima, and Jalisco violence had started rising since 2008 when "El Mochomo" was captured, but it reached a whole new level in 2010 due to the kidnapping of one of the sons of "Nacho Coronel," the death of "El Chaguin," and the split of the Milenio Cartel in two. Nayarit actually had to start summer vacation early because so many shootouts were occurring.

Since there were no military operations in these states previous to the escalation of violence this would be evidence against the theory that the violence in México is mainly the byproduct of the military coming in and unraveling local crime markets by taking over police duties.

In northwestern Mexico violence kept on rising, and if you've been reading my blog you won't be surprised to know that the number of homicides rose in the municipality of Tijuana by about 10% from 2009 to 2010. Hopefully in 2011, with Leyzaola gone, violence will finally go down in Tijuana.

In Ciudad Juárez there were a total of 3,800 homicides (3,684 in the database), but there were 193 deaths of unknown injury intent, there are only 400-500 deaths by accidents each year in Juárez and 2010 wasn't anything out of the ordinary, so many of those deaths of unknown intent are likely to be homicides. Coming up with an exact number is complicated. Whichever the true number, the metro area of Ciudad Juárez with a population of 1.3 million had more deaths than the states of California and Texas combined.

What's weird about the 2010 homicide number in Juárez is that at first the Fiscalía del Estado put the number of homicides at around 3,000, but much later in 2011 corrected the number to 3,600. This is something to keep in mind with all the talk of Juarez becoming less violent in 2011. Numbers can be updated, and I wonder if the whenever the drug-war related homicide database is updated it will reflected the newer homicide data.

In both Chihuahua and Sinaloa homicides were the main cause of deaths beating Ischemic heart disease.

For calculating which "cities" in mexico are the most violent I used the metro areas as defined by the INEGI, adding a made up metro area of Culiacán-Navolato, plus those municipalities which are not part of a metro area but have a population of more than 200,000 people.

It's always tricky to do comparisons across countries since cities in the US are geographical concepts akin to localities in México but with local governance (municipalities would be the equivalent of counties), and as such, only a small percentage of the people present at a given locatility/city/town actually live there, so as a first approximation it is better to use the concept of an integrated economic region--the metropolitan statistical area. New Orleans-Metairie-Kenner had a homicide rate of 21 in 2010.

Ciudad Juárez is a clear outlier in female homicides, a female homicide rate of 58 is unbelievably sad. Since Juárez had a reputation for being a killing ground for women since before the drug war, I again did a comparison with the metro area of New Orleans (parishes of Jefferson, Orleans, Plaquemines, St. Bernard, St. Tammany, St. Charles, and St. John the Baptist) using data from the CDC, and while you can argue just how accurate the population estimates for both cities are and whether all homicides were recorded properly in Juárez, I don't think it' crazy to claim that before the drug war Juárez was safer for women than New Orleans.

The urban centers that saw the biggest increases or decreases in homicide rates with a log scale:

With so many bodies going unclaimed it is not surprising that the quality of the ages in the database measured as an excess of 0s or 5s has decreased. The most violent age groups:

Note the rapid rise in the homicide rate of 15-19 year olds.

Here's a comparison of urban and rural (those with less than 10,000 people) municipalities, age adjusted since I just don't want to measure whatever age difference exists between the different levels of urbanization:

Since the Mexican police has been publishing monthly homicide statistics for 2011 I decided to use the auto.arima function from the forecast package to predict homicides and drug war-related homicides in 2011. It may sound silly to use police homicides to predict vital statistics homicides, but the Mexican police is notorious for under-reporting homicides and manipulating the data they publish, even if it does follow the overall trend in violence. The results were between 13,000 and 19,100 drug war homicides and between 27,500 and 31,700 homicides in 2011

Hopefully in 2011 violence didn't rise quite as much as in the previous years.

P.S. You can download the code to clean up the homicide data from github or simply check out the cleaned up files from the data sets section of my website

Crime at the municipal level

2011-11-02T04:41:00.002-06:00

Click on the image to go to the interactive map

The Mexican government recently released crime data for 2011 at the municipality level. Sadly, it is no disaggregated by month, but beggars can't be choosers. To analyze the data I made an interactive map with d3 that includes the locations of the municipalities with the highest drug plant eradication and a 2d kernel density estimate of the location of meth labs based on newspaper reports.

The data is not without problems since Acapulco actually came out on top as the most violent urban center (metro areas or municipalities not part of a metro area with more than 100,000 people) in Mexico, and while violence has certainly increased in the port city during the last few months --it may even have been the most violent city in Mexico during August-- there is no way it is more violent than Ciudad Juarez taking into account the whole year.

Mexico has an epidemiological monitoring system, it is designed for transmissible diseases like influenza rather than violent deaths. There is always some delay in recording homicides since by law you have to perform an autopsy and the police are involved; however, Chihuahua is one of the few states that records homicides in a more or less timely manner, even if the data is still incomplete. Checking the data in the epidemiological system there were 1,176 murders in Juárez through July (the last month available) compared to 1,055 in the police database through August (one extra month).

To believe the SNSP data you'd have to posit that doctors are making up homicide certificates and recording them each month in the epidemiological system so that the monthly numbers match the final data. Homicides in Juárez are down from the same period in 2010, but the police case data from the SNSP is plain wrong.

A similar thing happened in 2010 when the SNSP reported 3,806 total murders in the state of Chihuahua compared to 4,427 drug-related (or organized crime) murders. Stuff like this probably happens because there are many victims per police case. It's pretty sad when doctors are better at recording crime than the police.

Durango, Durango is another municipality that shows a big discrepancy, with a preliminary 421 homicides (including the mass graves) in the epidemiological system and 143 in the police database.

Looking at other crimes it is not surprising that kidnappings were high in certain regions of Durango and Tamaulipas where mass graves were found. The Zacatecas, Morelia, and Villahermosa metro areas also have worryingly high kidnapping rates and are hot spots of drug cartel activity.

And if you want to look at historical homicide data you can't do better than this map by Stanford student Dorothy Kronick:

Map by Dorothy Kronick

P.S. What's happening in Atarjea, Guanajuato, by its location I'd guess meth labs.

One less general

2011-05-25T22:33:00.001-05:00

Former Mexican Army general Jorge Juárez Loera was shot dead last Saturday when getting out of his Mini Cooper after a traffic accident in Mexico City.

Juárez Loera had just retired on May 16th after turning the mandatory retirement age of 65. Before leaving his post he was an Oficial Mayor of the Secretary of Defense, the third most position in the Mexican military.

During 2007 he was the commander of the XI Military Region, headquartered in Torreón (part of the metropolitan area of La Laguna), as such he oversaw military operations in La Laguna:

Sources: Mortality Database SSA/INEGI, Segundo Informe de Labores - 2008 - SSP

In 2008 he commanded the controversial Joint Operation Chihuahua, which coincided with a rise in homicides all across the state of Chihuahua, particularly in Ciudad Juárez and the metropolitan area of Chihuahua, Chihuahua.

Juárez Loera rebuffed critics of the rising death toll by saying: "I would like to see reporters change their articles and instead of writing about 'one more death', they should say, 'one less criminal.' "

In July 2008 General Espitia replaced Juárez Loera as the head of J.O. Chihuahua. Four of the military bodyguards assigned to protect his wife were attacked while traveling in a Cadillac Escalade reported stolen in Durango. Two were killed.

Are there other areas where we see a steep rise in homicides after a Joint Operation? There was an increase in homicides after Joint Operation Culiacán-Navolato:

This presentation [pdf] by Alejandro Poiré mentions that the operation in Culiacán was expanded into Mazatlán in July. As in Culiacán and Navolato there was a subsequent rise in homicides:

Looking at the drug-related homicide rates we see a similar pattern. Noé Sandoval, the general in charge of Joint Operation Culiacán-Navolato, was promoted in December 2010 to be commander of the IV Military Region headquartered in Monterrey, Nuevo León.

Source: Reporte mensual sobre los niveles de delincuencia el Área Metropolitana de Monterrey, producido por la asociación civil Iluminemos Nuevo León con base en las estadísticas oficiales.

Since the Mexican government views violence among cartels as a sign of success, it should come as no surprise that Generals with records of failing to stop the rise in homicides are promoted.

Two more generals have suffered attempts on their lives: Tello Quiñones who was kidnapped and killed by Zetas from Morelia, Michoacán. Before retiring and serving as a security consultant to the police force of Cancún Tello Quiñones was the commander of the 21th military zone in Morelia, Michoacán during 2007, where one can only assume he battled the Zetas. It is interesting to note how unlike in Chihuahua and Sinaloa, military operations in Michoacán were correlated with decreases in homicides:

The other general who suffered an attempt on his life was Acosta Chaparro, he was shot in the abdomen during an attempted robbery in 2010. Acosta Chaparro was sentenced to prision in 2000 because he allegedly received money from Amado Carrillo Fuentes. However in 2007 he was exonerated of all charges and set free.

Tijuana is more violent than ever

2011-03-31T22:18:00.002-06:00

The security strategy implemented in Tijuana has been praised to the stars and is frequently portrayed as the way to beat the cartels. The L.A. Times wet so far as to claim that Tijuana's chief of police Julián Leyzaola Pérez was "the model for the kind of law enforcement muscle the Mexican government needs to battle organized crime."

Most of praise was due to claims of decreases in homicides in 2009 and 2010 compared to 2008. However, a close look at the recently released 2009 mortality database paints a completely different picture:

There was also a dark side to Leyzaola's tactics which included lots of heavy handed policing: instructing his bodyguards to shoot drug traffickers in the head, kicking the dead body of a cartel gunmen, and allegedly torturing police officers he suspected of being in the pay of drug cartels.

As you can see in the following chart Joint Operation Tijuana had no effect on the number of murders --even thought the army confiscated the weapons of the police and they had to arm themselves with slingshots. Then there was a rise in the number of homicides around New Year in 2008, an astoundingly steep rise that started around the second half of September 2008 and lasted until the start of December, and yet another steep rise at the end of 2009.

Not only did homicides increase, but also other types of crime (data for all of Baja California):

	2007	2008	2009	2010
Violent robberies	8,920	11,833	13,722	11,806
Robberies to banking institutions	13	52	98	41
Kidnappings	20	115	103	79

Leyzaola became chief of police on December 10, 2008, right when the first wave of violence was ending. He started implementing his reforms, which included greater cooperation with the army and purging the police department, in January of 2009.

The basic problem in Tijuana is that one of the high level operatives of the Tijuana Cartel switched teams and started operating with support from the Sinaloa Cartel. According to various newspaper stories "El Teo" broke off sometime during the first months of 2008 and later in the year started an all out war against the remaining members of the Tijuana Cartel to control all the trafficking routes in Tijuana.

Here's a look at the network structure of the Tijuana Cartel (work in progress):

Back when in the late 90s and the early 00s when Benjamin and Ramon Arellano Felix were the leaders of the Tijuana Cartel it had more of a hierarchical structure. However, as each Arellano Felix brother was captured or killed the cartel lost its strength. Eduardo Arellano Felix was the last of the free remaining brothers, and he was captured at the end of 2008. After the capture of Eduardo, Fernando Sánchez Arellano, son of Enedina Arellano Felix, became the leader of the Tijuana Cartel. With the cartel weakend it is not surprising to find it full of shifting alliances. "El Sillas" and "El Gordo Villareal" were the main gunmen tasked with battling it out with "El Teo".

The Sinaloa Cartel is more of a federation of cartels than a top down organization. It has various bosses, the most important one being "El Chapo Guzmán". The faction backing "El Teo" was the one headed by "El Mayo" Zambada, although according to Leyzaola "El Chapo" sent an emissary to try and bribe him --though he may just have said it to prove that not even the Mexican equivalent of Bin Laden could corrupt him.

I believe "El Teo" operated with some independance from the Sinaloans even as he depended on them for gunmen, safe houses to hide in, and drugs to cross into the US.

"El Teo" was captured in January 2010, and his successor "El Muletas" was captured in February, however his gang is far from finished. Two distinct factions of the remains of "El Teo" still operate in Tijuana with the backing of the Sinaloa Cartel: "El Guicho" and "El Aquiles".

In late 2010, one of the guys in charge of fighting "El Teo", "El Sillas", broke off from the faction of the Tijuana Cartel headed by Fernando Sánchez Arellano ("El Ingeniero") and started a war his former boss and the Sinaloa backed remains of Teo's gang, he kidnapped "El Mayo" Zambada's niece (she was rescued by the army) sparking yet another round of violence. He may also have been responsible for the 134 ton drug bust that belonged to the Sinaloans.

Since "El Mayo" Zambada felt it was safe enough for his niece to travel to Tijuana it is very likely he entered into some sort of agreement with Fernando Sánchez after the capture of "El Teo" to let his drug shipments through without violent confrontations.

If we look at the daily number of murders we can see some interesting patterns. In mid September a prision riot was started by the brother of "El Muletas" in which 19 people were killed (over different days). Violence skyrocketed after that: sixteen bodies were found with signs of torture scattered around Tijuana and then one of the leaders of the Tijuana Cartel, Eduardo Arellano Félix, was captured.

Violence rose again after the capture, and then a girl who supposedly was Teo's girlfriend was killed in a failed kidnapping attempted, as revenge El Teo killed people close to Fernando Sanchez and allegedly El Teo and El Ingeniero entered into a truce after that. This coincides with the fall in homicides in the early days of December. Violence then stabilized at an incredibly high level until December 2009 when it shot up again (homicides for December 2009 are under-counted by 25% because of the cutoff date for entering homicides into the database). According to Zeta Tijuana the increase in violence was due to the Federal Police stealing a meth shipment from Tijuana's airport and handing it to "El Sillas."

Then in early January 2010 "El Teo" was captured and by all accounts violence receded from the peak it had reached in December. The successor to "El Teo", "El Muletas", was captured in February. There's no data on homicides for 2010, but there is data from the drug-related homicide database:

Note: The fact that there were more drug-related homicides than total homicides during November (an other months) doesn't necessarily mean that some homicides were fraudulently not counted since I'm ordering the mortality database by date of occurrence, and it may well be possible that the drug related homicide database is ordered by the date the murder was registered (the drug-related homicide database is not documented very well).

Interestingly, if we look at the pattern of who certified the deaths and wether an autopsy was performed we find a structural change right when drug related and the total homicide databases start to diverge:

Just because the difference between all homicides and drug-related homicides increases after Leyzaola was named chief of Police doesn't mean he is responsible since the same pattern is apparent in Mexicali.

And the difference may have just occured because a lot more non-drug related crimes were committed while Leyzoala was chief of police. Though given the daily pattern of homicides I'm very skeptical of this. One way to test this hypothesis would be to wait until the 2010 data is released and see if there was a big drop in homicides after "El Teo" and "El Muletas" were captured, if there were one, we could assume most of the crimes in Tijuana were drug related and the government under-counted drug related homicides.

We can also compare the total number of homicides in 2008 and 2009 from the INEGI with other homicide sources:

From the INEGI (for the metropolitan area of Tijuana, the municipalities of Tijuana, Rosarito, and Tecate)
2008 - 908
2009 ~ 1250 (1176 in the database)

From the transparency portal of the Baja California State Government (for the metropolitan area of Tijuana)
2008 - 685
2009 - 615
2010 - 674

For all of Baja California from the National System of Public Security (SNSP, only available at the state level and provided by the Baja California State Police)
2008 - 853
2009 - 749
2010 - 883

From press reports (it's not clear if they are referring to the city of Tijuana or only the municipality)
2008 - 844
2009 - 655

From the INEGI (Tijuana Municipality only)
2008 - 782
2009 ~ 1105 (1066 in the database)

And the correlation between gun homicides and drug related homicides is different in Tijuana than in other parts of the country.

To make sure that homicides were not being misclasified as other types of violent deaths I also checked deaths by accident and suicides. Violent deaths other than homicide stayed constant throughout 2006 and 2009

These are some big discrepancies from the different sources of homicide data and given the praise garnished on Tijuana's security strategy it is of utmost importance to clarify them. Especially given that Leyzaola has now been named chief of police in Ciudad Juárez and his model has been exported to other cities.

My best guess is that Leyzaola just got lucky to be named police chief right when the truce between the cartels started and then for some reason homicide numbers based on police reports (including those in the drug related homicide database) were under counted furthering his mystique. But he had very little to do with the ebbs and flows of violence in Tijuana.

Drug War Hotspots

2011-03-16T21:33:00.011-06:00

It turns out it is really easy to customize maps with Fusion Tables and the Maps API so I recreated the above map of the number and rate of drug-related homicides by municipality or metropolitan area.

Drugs Enter Here:
Waypoints:
(Ctrl-Click for multiple selection)

Exit:

The legend is the same as in the first map: the green colors correspond to lower rates and the reds to higher ones. I capped the maximum rate at 50 to make it easier to distinguish the municipalities. Make sure you check out the terrain view.

Strengths and weaknesses of crime data in Mexico

2011-02-10T11:55:00.000-06:00

With so much data pertaining to the drug war released recently it's hard to keep track of it all. And as with all things in life there are different pros and cons associated with each of the datasets: The homicide data from the police (SNSP), the homicide data from the vital statistics (INEGI), and the different estimates of drug war related deaths from Reforma, Milenio, and the database of homicides presumed to have been committed by organized crime.

Homicide data from the SNSP

This data is based on police reports. The homicide rates calculated with it used to be higher until 2008 when they suddenly started being lower.

Cons

The numbers in the database correspond to the number of police reports ("averiguaciones previas") for the crime of homicide, not to he number of dead bodies. The reports may contain than one victim and furthermore they may be repeated.
In 2008 the SNSP gave incomplete data to the ICESI for the state of Chihuahua, then when the data for 2009 was released, they updated the 2008 data and again gave incomplete data for Chihuahua. It's understandable that there would be some delay given the incredible rise in homicides, but it's kind of fishy to forget to mention how incomplete it was. Sadly, the incomplete data was used for the homicide rates calculated by the UN.
In the state of Tamaulipas, during August 2010, Mexican marines found the dead bodies of 72 persons inside a ranch. The victims were immigrants from Central and South America, presumably killed by the Zetas. Yet during the month of August there were less than 70 homicides in all of Tamaulipas.
The State of Mexico had 3 months without homicides at the end of 1998
Starting January 2007 the number of homicides in the state of Mexico dropped by half from one month to the next.
In 1997, Yucatán, Aguascalientes, and Querétaro had incredibly high homicide rates, there’s probably an error in database.
The number of homicides in Tlaxcala before 2007 seem way to high.
According to the database there were no homicides in Tlaxcala during 2007. However the General Secretary of the state verbally reported to the ICESI that there were 42 homicides in 2007. Also in Tlaxcala, during 2006, there were an anomalously high number of kidnappings, probably the result of another error unless the smallest state in Mexico accounted for over 40% of the kidnappings in the entire country.
There's no data on homicides by firearm in the states of Baja California, Oaxaca and Tabasco, and it fluctuates widly in Guererro, and Jalisco.
The proportion of homicides by firearm was incredibly low in Chihuahua in 2008. That was the year of the joint operation in Chihuahua and it doesn't match the data from the INEGI. Basically the firearm data from the SNSP is useless unless you make lots of adjustments to it.
It looks like the different states send a bunch of excel files to the SNSP which then tallies them. With such an outdated way of collecting data it's no surprise the database is plagued with mistakes.

Pros

It is constantly updated and the data is only a couple of months out of date. Although the SNSP hasn't updated it's online download tool since September they have updated the pdfs with the crime data. (Here's a scrapper written in python to extract homicide data from the pdfs)

Homicide data from the INEGI

This data is based on death certificates compiled by the Mexican government.

Cons

The deaths of the Acteal Massacre in 1997 were registered as accidents instead of homicides, and certified by a forensic doctor ("medico legista") in Tuxtla Gutiérrez to boot. However, I did check some of the recent massacres and they were all in the database.
It takes a while for all the death certificates to be tallied and the data is usually more than a year out of date
The cutoff date of December 31 means the homicides for the last available year are under-counted by 4% (25% for the last month of the year)
There's a weird pattern of lesions of undetermined intent in Ciudad Juárez in 2007
In 2008 most newspapers reported the number of deaths in Ciudad Juárez as 1650, the number in database is 1610, the discrepancy is probably because newspapers tend to report not only the number homicides in Juárez but also in the adjacent municipalities. In 2009 there were 2,316 homicides in the database (about 2,375 taking into account the undercount), but press reports placed the number of homicides as close to 2,700.

Pros

This is the best record of homicides in Mexico. If you download the mortality database from SINAIS you'll get a daily record for all deaths in Mexico at the locality and municipality levels. You can also find out how many people died from firearms, poisoning, knife wounds, etc.

Execution tallies by the newspapers Milenio and Reforma

Cons

Given the low prosecution rates it is not surprising that the series between Reforma and Milenio differ. However the difference between the series should be random and starting in June 2009 Reforma shows a precipitous drop only to go back up again. Given that Chihuahua accounted for most of the drop it looks like Reforma missed some narco-executions.

I am thankful for Reforma's steadfast devotion to the task of tallying the homicides week by week, it matters so that people know what is happening and the government can't hide the magnitude of this tragedy. I've used their data before, but something went wrong in the state of Chihuahua in 2009.
Milenio went from counting more drug war related homicides than the government to fewer.

Pros

Constantly updated
Data from Reforma is available by week and at the state level

Crimes presumed to be linked with organized crime (drug war related murders)

This data is based on police reports, but filtered to only include those deaths presumed to be linked with organized crime (drug cartels) and exclude duplicate reports.

Cons

The government missed the incredible rise in murders in Ciudad Juárez right before the army arrived. There's also a discrepancy with the Baja California homicide data.

The government uses a definition of what constitutes an execution at odds with what the newspapers define to be an execution (all drug war related deaths)
Since the database contains the number of deaths and not police reports like the SNSP homicide data, we can compare it to the INEGI. A higher number of drug-related homicides than total homicides would indicate a serious problem with at least one of the databases:

There's a big discrepancy in the state of Sinaloa. The differences in other states are small and may be due to the deaths being ordered by date of registration instead of occurrence. I also only compared by state since it is unclear in the drug related homicide datase whether the municipality refers to the place where the murder took place or where it was registered.
If you add the values for all the aggressions, shootouts, and executions and compare them with the precomputed totals in the database, the values for April 2009 (Acaponeta) and October 2010 (Manzanillo) are off by one, which speaks volumes about the care with which it was compiled.
We can also compare the drug-related database to the SNSP homicide data. As I suspected the number of homicides in Chihuahua and Baja California was underreported:

Pros

The database is divided into executions, shootouts, and aggressions against the government. Although the definitions seem somewhat dubious: Shootouts can start as aggressions. And any homicide by firearm may be counted as an execution.
Contains recent data

Since I've suspect that homicides by firearm account for a big chunk of drug-related homicides I decided to compare the drug-related homicides in the most violent states (excluding Sinaloa because of the discrepancy) with the data from the INEGI:

The relationships between firearm homicides and drug-related homicides look linear...

except for Baja California, but that will be the topic of my next post...

The most violent metropolitan areas in Mexico

2011-02-09T13:38:00.000-06:00

I'm surprised at how much violence has increased in Mexicali and how at odds it is with the drug-related homicide data.

Year	Drug-Related Homicides	Total Homicides
2007	21	64
2008	39	106
2009	69	213

Maps of drug-related homicides

2011-02-04T16:42:00.003-06:00

Just by looking at the map it is obvious that the drug war is a dispute over the drug traficking routes to the United States; Mexico's drug-producing regions of the "Golden Triangle" and "Tierra Caliente," where lots of meth labs are located and opium poppies and mariguana is grown; the ports where cocaine and the precursors for producing methanpetimes arrive; and the routes to the airports in Cuernavaca and Mexico City (though maybe in Mexico City the local drug market is big enough to be important).

Hermosillo, Ciudad Obregón, Agua Prieta, Saltillo, Monclova, Ciudad Acuña and Piedras Negras seem to have relatively low levels of violence given their closeness to the U.S. Border. Is it because they are firmly under the control of a single drug cartel? The drug-related homicide rates for Tijuana and Mexicali are likely underestimated in this dataset given their high homicide rates.

Since the drug related homicide data was only released at the municipality level, I calculated the centroids of the municipalities taking the population weighted means of the latitude and longitude of the localities within each municipality based on the 2000 census.

These maps give a good idea of what has been happening in Mexico during the last few years:

2007	2008
2009	2010

There was an increase in homicides in 2006 in Michoacán and as a result of this at the end of the year President Calderón sent troops to pacify the region. The following year México had the lowest homicide rate ever, and then in 2008 the Sinaloa Cartel started a war with the Juárez Cartel, betrayed one of their high level operatives nicknamed "El Mochomo," and backed "El Teo" to try and take control of the drug traficking routes in Tijuana.

In 2009 Michoacán once more turned violent and the federal government again sent troops to pacify the region. A the end of the year the brother of "El Mochomo" was killed in Cuernavaca and this sparked a surge in violence in Sinaloa and Guerrero as the Sinaloa Cartel tried to destroy the remains of his organization once and for all. In 2010 the Gulf Cartel went to war with the Zetas which is why northeastern Mexico turned red.

When percentages mislead

2011-01-31T21:36:00.000-06:00

According to security spokesman Alejandro Poiré violence in Mexico is a regional phenomenon since 85 municipalities concentrate 70% of drug related homicides. While it is certainly correct to say that not all of México has suffered from the drug war, it is misleading to present the percentage of total homicides concentrated in a few municipalities as evidence. Municipalities have different populations and furthermore drug related homicides increased from 2,826 in 2007 to 15,273 in 2010 which makes it even harder to interpret the percentages.

A much better way to show the the concentration of violence is to calculate the percentage of the Mexican population living at or above a homicide rate. Since it is much more trustworthy than the drug related homicide database Poiré uses, I decided to use the recently released mortality database as my source.

Number of people living in a municipality with a homicide rate above 50:

2006 - 1,100,000
2007 - 850,000
2008 - 5,800,000
2009 - 9,100,000

Number of people living in a municipality with a homicide rate above 10:

2006 - 33,500,000
2007 - 26,500,000
2008 - 42,300,000
2009 - 54,900,000

While it is certainly true that violence is concentrated, the number of people suffering its effects has increased greatly.

Homicides in Mexico 2006-2009

2011-01-27T22:32:00.006-06:00

Just today the Mexican government released to the public the mortality database for 2009, and as you can see from the chart Mexico has suffered from a steep rise in homicides from 2008 onward and very likely reached the highest violence rate in recent history last year. Since the Mexican government also recently made available a database of homicides presumably linked with the drug war we can divide homicides into those related to the drug war an those that are not:

Most of the increase has been because of the drug war, though as Mexico becomes more violent there has been a small rise in homicides not related to the drug war.

By State:

For females:

The most violent municipalities:

Ciudad Juárez was again the most violent municipality with a total of 2316 (about 2375 taking into account that there wasn't enough time to record all homicides that occurred in 2009 into the database) which is much lower than the press reports of about 2700.

What stands out from the chart of the most violent municipalities for females is that Almoloya and Texcoco in the state of Mexico made the list (hint hint future gubernatorial candidates).

In this chart I show the maximum and minimum homicide rates from 1990 to 2006 (right before the drug war started) and compare it with the 2009 rate to put it in historical perspective:

Interestingly, the rate for Michoacán in spite of the drug war is nowhere near its maximum. The rate for 11 states is above or near their maximum from 1990 to 2006 and for 7 it is at or near their lowest rate (hint hint future presidential candidate).

The monthly totals:

The weekly totals:

If the data follows the same pattern as in 2008, homicides in 2009 were under-counted by about 4%, which is perfectly understandable since there isn't always time to enter into the database those homicides that occured during the last weeks of the year and the cutoff is December 31, 2009.

Taking into account the undercount for the last year the totals would be:

2006 - 10,429
2007 - 8,843
2008 - 14,175
2009 ~ 19,900 (19,121 recorded in the database)

which assuming a population of 111 million (there were 112.3 million Mexicans in the 2010 census) would give a homicide rate of:

which is of course quite different from the 16,117 homicides according to police reports, and their corresponding rate of 15.

The mean age for homicide victims in 2009 was 34.6 with a sd of 14.3 years (it hasn't changed significantly since 2006):

Men are proportionally much more likely to be homicide victims between the ages of 18 and 47:

The increase in homicides since the drug war went charlie-foxtrot was especially acute among 18-40 years olds:

Homicide victims by age groups:

Marital status of the homicide victims:

There was a big increse in the proportion of victims whose ocupation was unknown (maybe it's time to include "sicario" in the list of occupations):

The proportion of homicides that occurred out in the streets keeps on increasing:

The educational attainment of homicide victims:

When violence increases it's usally men who suffer the most, though according to newspaper stories there was a big increase in the proportion of women murdered in Ciudad Juárez in 2010, maybe the increase in the percentage of homicide victims who are men will stop next year?

The rate of increase in homicides commited with a firearm slowed down:

Right before midnigth is the most dangerous time in México:

(There was a mistake in the database, when the hour of death was unknow it was coded as having occurred at midnight, I discarded that value and used the average of the number of homicides between 11:00pm and 1:00am)

~~Sunday is the most dangerous day of the week~~There was no statistically significant difference between the days of the week:

Technical Note: The population for the first chart was interpolated from the CONAPO estimates for 1990-2000 and the 2010 census since the official estimates for 2010 were off by 4 million people. The state and municipality rates are the ones done by the CONAPO since creating new estimates would depend on calculating the correct fertility rates, the lower than expected immigration rates, the number of people who have left the most violent municipalities and why the 2005 population survey came out so wrong. The 2010 homicide rate was calculated from a linear regression based on execution rates.

P.S. The code is at my GitHub

Some problems with the Mexican mortality database

2010-12-22T21:50:00.001-06:00

I've written before about how I couldn't find the Acteal Massacre in the homicide database available at the INEGI. So I decided to check if the deaths that occurred as a consequence of the massacre were misclassified as other types of violent death,

and indeed they were. In the municipality of Chenalhó, where the massacre occurred, there were 44 accidentals deaths during December 1997. It is pretty clear that the deaths of the Acteal Massacre for some reason were classified as being accidental.

You can check it out for yourself by downloading the Mexican mortality statistics from SINAIS for 1998, and see that the deaths were classified as unintentional injuries with ICD-10 codes of:

W26 - Contact with knife (5 deaths)
X59 - Exposure to unspecified factor (7 deaths)
W34 - Accidental malfunction from other specified firearms (32 deaths)

Because of the discrepancy and because it is ground zero of the drug war, I also checked to see if there was any evidence that deaths by homicide in Ciudad Juárez were misclassified, and I found a very interesting pattern where lesions of unknown intent shot up as a percentage of homicides in 2007 (there 197 homicides vs 153 lesions of unknown intent) compared to other years.

This would tend to corroborate the testimony (Spanish) of "La Barbie" that the leader of the Sinaloa Cartel started sending his people to Ciudad Juárez in 2007, even if the official homicide statistics showed a slight decrease during that year. Interestingly, lesions of unknown intent started rising in 2004, the same year the brother of the leader of the Juárez Cartel was murdered.

The code is here

Recent developments in the drug war

2010-12-01T22:47:00.000-06:00

The Mexican government recently released data on all crime statistics from January 1997 to October 2010, as reported by the various police forces operating in Mexico. This information provides an unprecedented look at what has happened since the drug war began and allows for a frank assessment of the deterioration of security in Mexico.

For some states the data is only available up to August, but for most it is available up to October. The first thing to do is visualize the data to see if there are any mistakes. And there were, indeed, some mistakes:

The state of Chiapas reported 0 homicides during the month of January 2010, but in previous versions of the reports there were 96 homicides. Probably someone’s finger slipped while updating the database.
In the state of Tamaulipas, during August 2010, Mexican marines found the dead bodies of 72 persons inside a ranch. The victims were immigrants from Central America, presumably killed by the Zetas. Yet during the month of August there were less than 70 homicides in all of Tamaulipas. Probably the police haven’t had time to update the database to include this massacre.
Starting January 2007 the number of homicides in the state of Mexico dropped by half from one month to the next. Perhaps some bodies were swept under the bed?
In 1997, Yucatan and Querétaro had incredibly high homicide rates, there’s probably an error in database.
According to the database there were no homicides in Tlaxcala during 2007. However the General Secretary of the state verbally reported to the ICESI that there were 42 homicides in 2007. Also in Tlaxcala, during 2006, there were an anomalously high number of kidnappings, probably the result of another error unless the smallest state in Mexico accounted for over 40% of the kidnappings in the entire country.
The numbers in the database for some states correspond to the number of police reports for the crime of homicide, not tto he number of dead bodies. Because of the increase in multiple homicides, the numbers for Chihuahua and Baja California are very likely to be underestimated.

Because of issues involving the undereporting of crimes, in this blog post I will concentrate on reported homicides.

Special note about the population of Mexico: Preliminary data from the 2010 Census indicate that the population estimates made by the CONAPO (which I’m using) were off by about 4 million people, that’s a pretty big mistake. Depending on whether it was due to higher fertility rates or reduced levels of emigration from Mexico to the US, this likely means the homicide rates are a bit lower than shown here. Also lots of people have left Chihuahua, black swans and all that.

With all that said let’s look at the state level data:

All States Affected by the Drug War

Chihuahua

This is Mexico’s most violent state as a consequence of the war between the Sinaloa Cartel and the Juárez Cartel.

Sinaloa vs Beltrán Leyva (and later also Barbie)

The Beltrán Leyvas were brothers who used to be allied with “El Chapo” Guzmán, but in January 2008 one of the brothers, “El Mochomo”, was captured by the army and the rest of the family blamed their former boss of tipping off the government to his location. In response they allied themselves with the Zetas and started a war with the Sinaloans.

Later in December 2009, Arturo Beltrán Leyva, the leader of the group, was killed in a shootout in Cuernavaca. As a consequence of his demise one of their chief enforcers, “La Barbie”, split from the Beltran Leyva organization and started a war with the remaining brother. The Sinaloa Cartel also took advantage of this opportunity to try and take over their drug traficking routes once and for all.

One point I'd like to make is that showing the homicide rates for whole states hides the different geographical patterns. Violence started in different ways in the biggest cities of Sinaloa: Culiacán and Mazatlán. The start of the violence in Culiacán coincided with Joint Operation Culiacán-Navolato and in Mazatlán with the expansion of the operation to Mazatlán and Salvador Alvarado.

Likewise in Durango the patterns were different for the city of Durango and the Metropolitan Area of La Laguna (Torreón, Gómez Palacios and Lerdo).

Just recently 20 people were kidnapped and murdered in the tourist destination of Acapulco, Guerrero, because they were mistaken for members of La Familia.

In April the leader of the Beltrán Leyva’s in Nayarit kidnapped and killed the son of the Sinaloan boss in Jalisco, Nacho Coronel, afterwards violence rose steeply. When Nacho Coronel was killed by the army in July violence decreased somewhat in Nayarit but rose in Colima.

Zetas vs Gulf Cartel

The Zetas used to be the paramilitary group of the Gulf Cartel, but sometime in late February the Gulf Cartel entered into an agreement with the Sinaloa Cartel and La Familia Michoacana, they then declared war on the Zetas.

Michoacán

This state is where it all started, on December 11 2006, Felipe Calderón, president of Mexico sent troops to his home state to battle La Familia. After another rise in violence the government started Joint Operation Monoraca in 2009.

Baja California

The trend-line represents a robust linear regression because of the big outliers when “El Doctor”, the leader of Tijuana Cartel was captured, and the rise in violence right before “El Teo” was captured. The city of Tijuana has been praised as an example of successfully reducing violence. The police records only include data at the state level, but since the vast majority of murders in Baja California occur in Tijuana it is safe to say that violence has only increased with the drug war.

It is pretty sad when the strategy that’s been praised to the skies has only resulted in more deaths.

Recently, relatives of one of the leaders of the Sinaloa Cartel were kidnapped in Tijuana, the only way “El Mayo” Zambada would send his family to that city was if he had come to some sort of agreement with the Tijuana Cartel. My guess as to who was responsible for the kidnapping is that it was some splinter group from the Tijuana Cartel who couldn’t accept the new alliance.

As a bonus here’s an OpenHeatMap of the logarithms of homicide rates (You'll find a lot more maps of crime in Mexico over at Animal Político):

You can download the source code from my GitHub account

Femicides in Mexico 2006-2008

2010-11-11T13:25:00.000-06:00

Since the Mexican Human Rights Commission is working on a report of femicides in Mexico, but were only able to access data from the police forces of 18 states, I reran the program I used to analyze the homicides in Ciudad Juárez with data for all femicides.

The only problem is that since I'm using the death certificate database from the vital statistics system, and it takes time to gather all death certificates, I only have data up to the last day of December 2008 (the 2009 vital statistics are supposed to come out during late November/ early December)

The most surprising finding is that in spite of the fact that overall homicide rates are higher in Mexico than in the US, female homicide rates were lower in Mexico in 2007. Although of course, homicides rates for both men and women have increased since then due to the drug war.

Rates by state:

As a consequence of the drug war Chihuahua has the highest levels of femicides.

Monthly homicides:

The yearly totals:

The percentage committed with a firearm:

The distributions of the ages of homicide victims grouped by sex:

The percentage in each age group:

Given the low (though increasing) labor-force participation rates for women in Mexico it's not surprising that most are inactive:

The locations where the bodies were found:

Education:

Marital Status:

It would probably be an interesting topic to compare the marital status of femicide victims to that of the overall population controlling for age

The source code is available from my GitHub account

Mapping drug war related homicides in 2010

2010-11-09T05:39:00.003-06:00

There have been some very good visualizations of the Wikileaks data so I decided to create one of the drug war in Mexico

The above map was made using data collected by Walter McKay, mainly from El Universal and El Diario reports. The data is stored as a Google Map and is already structured to include the number of deaths. That saved me a lot of work!

Since the map contains the original newspaper story as metadata I performed a simple grep to determine if those involved in the aggressions were only civilians (both innocents and cartel gunmen) or whether some of those involved —though not necessarily killed— belonged to government forces.

However, there are some caveats to keep in mind:

I surely introduced some errors when categorizing the victims with a simple grep.
As Walter McKay explains in this blog post killings have become so routine that it is no longer possible to report them all. Newspapers will sometimes report that there were a total of XX homicides in Ciudad Juárez instead of separately reporting each incident.
The vast majority of incidents are only reported if deaths occurred.
To avoid detection, illicit dumping grounds have become a come way to dispose of bodies, this makes it even harder to get an accurate count and size for each event.
The press in Tamaulipas is under treat from the drug cartels to avoid reporting their activities. Compare the online edition of the Brownsville Herald with El Bravo de Matamoros the day after Tony Tormenta was killed in Matamoros. There's no mention of the 9 hour shootout between the army and the forces of the Gulf Cartel in the later. There have been far more incidents involving civilians than is obvious from the map.
Since most of Chihuahua's population is concentrated in its capital city of Chihuahua and in Ciudad Juárez, it doesn't look as violent as other states when in fact it is by far the most violent state. Also the central area of Mexico is the most populous.

But in spite of this I still believe that the map gives a good overall picture of what has happened this year. You can see how violent Sinaloa has become and how most of the deaths which involved the military have been concentrated in northeastern Mexico and Guerrero.

According the Mexican Department of Defense there have been a total of 565 civilians (both innocents and cartel gunmen) killed by military forces since the start of the drug war. The total for each state is the following:

Tamaulipas	122
Nuevo León	88
Guerrero	81
Durango	57

According to the Mexican military there have been a total of 191 soldiers killed, so ignoring the fact that the civilian deaths include innocents and a difference of one month in the periods reported, that would give a kill ratio of 3.5

Perhaps not surprisingly, going by press reports, most fatal incidents involving innocent civilians and the military have occurred in Tamaulipas and Nuevo León; however, let me emphasize that there are no official records of innocents killed.

According to the Mexican government most of the drug war deaths are the responsibility of the Sinaloa Cartel. And yet the cartels with the biggest presence in the states with the most civilians killed would be the Gulf Cartel and the Zetas, battling each other in Tamaulipas and Nuevo León; and the Beltrán Leyvas and the remnants of "La Barbie" organization battling in Guerrero.

And if you're wondering why there hasn't been much drug cartel activity in the central part of the border, here's a map of Mexican highways and an interactive one made with this spreadsheet and Openheatmap:

Here's the source code (please note that Walter McKay constantly updates the Google Map and the output you get will depend on when it is run)

The dead of Juarez

2010-11-04T01:32:00.011-06:00

Ciudad Juárez became the most violent city in Mexico as a consequence of the vicious war between the Juárez Cartel and the Sinaloa Cartel. So far more than 6,500 people have been murdered in this city since the war started. With the Juárez Cartel significantly weakened, 2010 is set to be the most violent year yet...

Here are some statistics of the dead of Juárez:

The violence took off 4 weeks before the army arrived. As Reforma reported, the Sinaloa Cartel had advance knowledge of the military operation, and the cartel temporarily stopped the war right before the arrival of the army. However, after the army took over the city, the number of murders inmmediatly started increasing in spite of (or due to) having thousands of soldiers patrolling the streets. Soon the number of murders reached new records.

The monthly totals:

and the yearly totals,

though if you look at the weekly chart you'll notice there wasn't any time to record all the homicides that occurred during the last few weeks of December 2008, so the data for that year is under counted by about 3%.

The median age for homicide victims in 2008 was 30 with a sd of 11 years:

Men are proportionally much more likely to be homicide victims between the ages of 18 and 35:

Homicide being mostly a male enterprise it should come as no surprise that the vast majority of homicides are committed against men.

Perhaps paradoxically the more violence there is, the lower the percentage of women who are victims of homicides. About 20% of homicides in the US are committed against women and in the UK it is about 30%, so the low percentage shouldn't be interpreted as a measure of misogyny.

The increase in homicides in 2008 was especially acute among 20-35 years olds:

Homicide victims by age groups:

Marital status of the homicide victims:

Not surprisingly, most of the victims were maquiladora workers:

There was a big increase in the proportion of homicides that occurred out in the streets:

The educational attainment of homicide victims:

An increase in the percentage of homicides committed with a firearm:

Don't go out in the evening and definitely stay home at night (do cartel hitmen have supper at 7pm?):

(There was a mistake in the database, when the hour of death was unknow it was coded as having occurred at midnight, I discarded that value and used the average of the number of deaths between 11:00pm and 1:00am)

P.S. The code is here

The 2009 homicide data for Chihuahua has been updated

2010-10-20T16:43:00.001-05:00

As I described in this post, the reported number of homicides[1] in Chihuahua during 2009 turned out to be incomplete. Guess what? The SNSP just released [xls file] an updated version of the homicide data, and Chihuahua went from having 2,523 homicides to 3,156. My estimate was 3,256, so I came pretty close. And the homicide rate for the whole country was 15, exactly what I predicted.

Score one for linear regression on World Statistics Day. Now we'll just have to wait until the INEGI releases it's numbers based on death certificates to see what the real number of homicide deaths was (3,600?).

Since the SNSP conveniently forgot to mention that the first numbers it released in 2008 and 2009 were incomplete, and nothing like this had ever happened before, the official homicide statistics are turning into a real mess of incompatible numbers. For example, this report from the ICESI uses the original 2008 data, and this study by Mexico Evalúa and this story in The Economist use the original data from 2009.

P.S. Here's the code

[1] More accurately the number of police reports, which may contain more than one victim

OpenHeatMap of the Drug War in Mexico

2010-10-20T16:03:00.000-05:00

OpenHeatMap is an open-source tool that lets you turn a spreadsheets into a maps. It is very user-friendly if you have lots of data because it allows you to zoom into the map, and if you're like me and can never remember the exact location of important cities, when you hover your mouse over a region it shows you its name and the exact value you assigned it.

But best of all Pete Warden just updated OpenHeatMap with maps of Mexico at the municipality level. I decided to take advantage of this and built a HeatMap/Choropleth of the murder rate in Mexico for the last three years (the drug war started December 2006).

The spreadsheet I used to build the maps looks like this:

mex_muni_code	value	time
01001	3.1	2006
01001	3.9	2007
01001	5.7	2008
01002	2.4	2006
01002	2.4	2007

The mex_muni_code column corresponds to the combined state and municipality codes, the value column is the homicide rate, and time is an optional column if you want the map to have a slider so you can compare homicide rates over time. Once the spreadsheet is ready all you have to do it is upload it and you're set to go.

In 2006 you can see how Michoacán turns red as La Familia wrests control of the state from the Zetas and also the end of the failed attempt by the Sinaloa Cartel to take over the drug trafficking routes of Nuevo Laredo. Notice how in 2007 the whole map pales considerably as it was the year with the lowest homicide rate on record, and how in 2008 the whole border region of Chihuahua turns bright red as the Sinaloa Cartel starts a war to defeat the Juárez Cartel.

You can also see a larger version of the map at OpenHeatMap.

The homicide rate was calculated with population data from the CONAPO and homicide data from the SINAIS. The data used to generate the heatmap is available as a Google Spreadsheet.

How the expiration of the assault weapon ban affected Mexico

2010-09-02T14:39:00.002-05:00

There has been a lot of attention paid to the role of US guns exacerbating the violence in Mexico. The assault-weapon ban expired on September 14, 2004, but with the recent spiraling of violence in Mexico the ban has attracted renewed attention. Just recently the Mexican President stood before the American Congress and blamed the assault weapon ban for the rising violence in Mexico, seemingly without proof. This post will try to clarify some of the issues surrounding the controversy.

I think it would be fair to say that the conventional wisdom by scholars who have studied the ban would be the one Christopher Koper, Daniel Woods and Jeffrey Roth stated: "We cannot clearly credit the ban with any of the nation’s recent drop in gun violence." In other words, the assault weapon ban had no significant effect—at least where the United States is concerned. And as far as I know there are no studies of the effect of the ban expiration on Mexico.

As is well known, gun control is politically charged issue, not only in the United States, but also in Mexico. For example, after two students died in a shootout between the army and cartel gunmen, the rector of what is arguably Mexico's best university stated:

For me [Rangel Sostmann] the solution is that Mexico and the United States must change their policies. And the policy right now is for the United States to send the money and weapons, and Mexico puts the dead and the drugs. And that ultimately is not going to work, it's not a question of driving the Army out, the problem is that you have to change the policies, in this case the United States policy.
—Rangel Sostmann, ITESM Campus Rector

Later an investigation by the National Human Rights Commission concluded that the army had used excessive force, impeded the investigation, planted weapons on the students, possibly tortured them for several minutes, and then shot them at point blank range. The day after the report was released, the ITESM bestowed upon the Secretary of Defense the "AdHonorem" award, given to persons whose works have contributed in an outstanding manner to the "the fight against injustice." As you can see gun control has become extremely politicized, to say the least.

With politics hopefully out of the way, let's look at the homicide data with a seasonal decomposition by loess:

As you can see from the charts there was an increase in the number of homicides and homicides with firearm that coincided with the expiration. The big drop at the start of 2007 was due to the military operations in Michoacan and Acapulco, and also to a temporary drop in the homicide rate of Mexico City (which had nothing to do with the drug war).

Since homicides and homicides with firearm can rise at the same time because the overall level of violence in society is increasing, say because of land reform or election disputes, we should also look at how the proportion of homicides with firearm has changed in Mexico:

It looks like the proportion of homicides by firearm started changing about a year after the expiration of the assault weapon ban. Mexico is a big country so it's worth looking at its different regions. There was a coding mistake in the homicide database (Chihuahua was classified as being in the North East), so I divided Mexico into its economic regions:

Notice how the proportion of homicides rose in North Eastern Mexico, though it would be fairer to say it was more of a temporary dip. Also, there was a big rise in South Central Mexico long after the expiration of the assault weapon ban, this is the most populous region of Mexico. In the South East there was a big increase in the proportion a month before the ban went into effect.

Now just the municipalities that border the United States:

The chart shows that the proportion of homicides commited with a firearm did change around the time of the expiration of the assault weapons ban, but well within the normal range. The proportion didn't rise outside the normal range until 2008. But remember how only the North East had a rise in the proportion of homicides commited with a firearm?

Now, let's look at the big border cities:

Aha, now we see that in Nuevo Laredo the proportion of homicides commited with a firearm started rising at the same time the ban expired. It's also a good idea to look at the number of homicides by firearm instead of the proportion:

Wow, a big rise in homicides by firearm in Nuevo Laredo that coincided with the expiration. Plus, before the ban the number of monthly homicides was low, which would explain why the variance in the proportion chart was so much greater before the ban.

But before we conclude that it was due to the assault weapons ban we have to take into account that "La Barbie", working for the Sinaloa Cartel, was trying to take over the drug trafficking routes of Nuevo Laredo after the capture of Osiel Cárdenas, the leader of the Gulf Cartel, in March 2003. And that as a consequence of the violence Vicente Fox started Operation Secure Mexico in June 2005 and sent troops to Nuevo Laredo:

It looks like there was indeed an effect independent of the capture of Osiel Cardenas and the arrival of the army.

Now let's look at the proportion of homicides by firearm in each of the states of Mexico and the Federal District. I used the strucchange package to help visualize the breakpoints (even though it's not quite correct to run a linear regression on the data that's what I used):

And now the interesting states:

The drug trafficking states:

The changes in Guerrero and Durango are clearly not significant, and are in fact slight dips. Since the state of Tamaulipas shows no signs of the rise in violence in Nuevo Laredo, it must have been a localized phenomenon, at least in that state.

Of all the states I have to say that only Nuevo León and Chihuahua look especially interesting:

However in Chihuahua there's another big confounding variable in that the leader of the Sinaloa Cartel ended his alliance with the Juarez Cartel by killing Rodolfo Carrillo Fuentes, brother of the leader of the Juarez Cartel a couple of days before the ban expired.

Supposedly after "El Chapo" Guzmán escaped from prison he began to rebuild his drug empire and sometime thereafter entered into a non-aggression pact with the Juárez Cartel, but on Sep 11, 2004, he ended the alliance by killing the brother of the leader of the Juarez Cartel, since neither the proportion or the number of homicides committed with a firearm changed in Ciudad Juárez, my guess is that the breaking of the pact had more to do with the rise in violence than the assault weapon ban.

Nuevo León also experienced a rise in firearm violence:

Nuevo León is right next to Nuevo Laredo:

We can also look at the suicide and suicide with firearm rates for all of Mexico:

No effect. There's a very interesting seasonal pattern, and the fact that the suicide rate has been rising while suicide with firearm rate has been decreasing would go against the conventional wisdom that the availability of firearms increases suicides. This, of course, doesn't concern the effect of expiration of the ban on assault weapons so I'll leave it at that.

Before we conclude that the assault weapon ban affected Mexico we need a plausible mechanism for how assault weapons would benefit the drug cartels and clear up some confusions and misunderstandings regarding guns:

90% of the guns in Mexico come from the US: Of those traced, about 90% came from the US. The keyword being traced. This isn't all guns that were confiscated in Mexico, not even a random sample, but a highly biased one. For example, according to GAO, 25% percent of traced guns were assault weapons, while according to the Mexican government 50% of seized guns were assault weapons. So it would be mistaken to draw inferences about where the total percentage of guns come from. However, given that thousands of guns have been traced to the US, it would be correct to say that the United States is one of the main suppliers of guns to Mexico.
Assault Weapons are "Military Grade": None of the assault weapons for sale to civilians in the United States are of military grade and with the exception of some collector items they are all semiautomatic. They deliver one round each time you pull the trigger. In a fully automatic weapon many bullets can be fired with one trigger press.
"Cop Killer" Guns: Some bullets can pass through body armor; however, they are not for sale to civilians. For some reason drug dealers took to calling the guns that can fire this kind of bullet "Cop Killers," even if they can only load them with regular bullets.

During the period in which the Assault Weapon Ban was in effect, the sale of any newly manufactured gun that met any of the following characteristics was prohibited:

Folding stock
Pistol grip
Bayonet mount
Flash suppressor
Guns specifically mentioned in the ban

The Assault Weapon Ban also prohibited magazines that held more than ten rounds. The law did not ban the possession or sale of pre-existing assault weapons or magazines. As a consequence pre-ban assault weapons and magazines rose in price until the ban ended.

In response to the ban, firearm manufacturers quickly retooled their guns so that they didn't meet the law's definition of assault weapon, and most of the banned features don't strike as particularly important. For example, I bet most guns smuggled to Mexico are disassembled and so a stock of any kind would be removed. And I've never heard of a cartel shootout that involved bayonet mounts.

However a recent analysis by the Texas Ranger's Border Security Operations Center highlights the importance of high capacity magazines.Unlike assault weapons, high-capacity magazines, which are used with many guns, have been selling briskly since the ban ended because prices have dropped considerably.
—The New York Times
It is very common to find live ammunition rounds strewn on the ground after drug cartel shootouts (examples: 1 2 3), reloading a gun when other people are shooting at you is not an easy task and cheap large capacity magazines are an advantage during long shootouts, like the ones that occured in Nuevo Laredo when "La Barbie" was trying to take it over from the Gulf Cartel. This would be one plausible mechanism through which drug cartels could have benefited from the expiration of the ban.

It's also a good idea to put the size of the effect (if indeed there was one) in the context of the drug war. Clearly they are separate processes:

Sadly for Nuevo Leon I have no recent data on the number of homicides by firearm, so instead I'm showing the overall number of homicides. The relatively small rise at the beginning of 2007 coincided with the start of the joint operation Tamaulipas-Nuevo León:

And for Nuevo Laredo I have no recent homicide data, which is a real pity since it wouldn't look ridiculously low, here's a chart comparing it to other border cities, you can see how it stands out in 2005 and 2006. During 2010, in Tamaulipas there were 338 drug related homicides during the first half of the year according to the Reforma newspaper, during all of 2005 there were a total of 351 homicides (all kinds). I can only surmise that the whole state of Tamaulipas and Nuevo Laredo are much more violent now than in 2005, even if the difference is not quite as stark as in other regions of Mexico.

I modeled the effect of the assault ban as a Zero-Inflated Poisson regression model predicting the number of homicides that occurred daily in Nuevo Laredo from the time Osiel Cárdenas was captured to the time the army arrived. The model with a breakpoint ocurring at the time the assault weapon ban expired proved superior to one without a breakpoint.

I think it is much easier to visualize the size of the effect with weekly data:

Certainly the assault weapon ban wasn't the only variable that contributed to the rise in violence: A weak and inept government, a large Mexican-American population willing to smuggle guns (Edgar Valdez Villareal "La Barbie" was born an raised in Texas), the corruption of the public security forces of Mexico, and the weakening of the Gulf Cartel with the capture of Osiel Cásrdenas also played a big part.

But the capture of Osiel Cardenas occured more than a year before the violence started, the army arrived more than six months after the violence started, there's a plausible advantage to having cheap high capacity magazines, and the rise in homicides coinicided with the expiration of the ban. I can only conclude that the expiration of the Assault Weapon Ban contributed to the violence in Nuevo Laredo with possible spillover effects in Nuevo León, and I've certainly updated my beliefs about the effect of the Assault Weapon Ban.

Clearly much work remains to be done, more complex models could be fitted to the data and I didn't answer all the questions of interest: I didn't perform a detailed analysis of the change in proportion of homicides by firearm in Nuevo Laredo or an analysis of what happened in Nuevo León, but given that "La Barbie", the guy in charge of taking over Nuevo Laredo, was caught on Monday, finding out the whether he took advantage of the expiration of the assault weapon ban should be a simple matter of asking him.

The code is here. (I used the excellent package Project Template to build the skeleton)

Myths about Ciudad Juarez

2010-08-18T22:11:00.047-05:00

Last year there were over 2,600 murders in Ciudad Juarez, and if the more than 1,800 murders so far this year are any indications, there will be even more murders in 2010. Ciudad Juarez is a scary place, but it wasn't always that way...

I learned from Noel Maurer's Blog that Ciudad Juarez used to have a low murder rate before 1993, but it rose steeply after the local leader of the Juarez Cartel, Aguilar Guajardo, died, and Amado Carillo replaced him. After Amado Carrillo had firmly established himself thing calmed down somewhat, but by then the violence had switched to a new equilibrium. But anyways, go read Noel Maurer's take on it

Somehow this caused Ciudad Juarez to become the poster city for lawlessness, famous for its hundreds of unsolved murders against women. This, despite the fact that were other urban areas that had higher homicides rates against women or that its overall murder rate was no different from the other big border cities. In short, the conventional wisdom was wrong and this blog post will hopefully convince you that:

The homicide rate in Ciudad Juarez was not different from other border cities.
The homicide rate for women, while certainly higher than average for Mexico, was lower than in other large urban areas and not significantly different from other border cities.
Ciudad Juarez is not poor or particularly unequal as measured by its gini coefficients relative to other Mexican municipalities.

The part about the femicides

The chart below shows that when the homicide rate started to increase in the early 90s, the proportion of homicides that were female also increased, which means that the female homicide rate increased even faster.

But how does Ciudad Juarez's femicide rate compare to other parts of Mexico?

Miguel Hidalgo is a relatively nice borough in Mexico City, in spite of its high homicide rate its inhabitants are more likely to be found protesting the killing of trees in front of the Lycée Franco-Mexicain than protesting femicides, and they count as their neighbor the richest man in the world. Toluca is the capital of the State of Mexico near Mexico City. The fact that there were areas more dangerous to women than Ciudad Juarez was well known to Mexican criminologists.

What about the other big border cities?

Ciudad Juarez wasn't significantly more violent to women than any other border city.

Why did Ciudad Juarez acquire such a bad reputation:

The big increase in murders when the Juarez Cartel switched owners. And since the homicide rate for women increased even faster than for men, people started freaking out, even if it was well below other cities in Mexico.
virtually all the victims were poor, young, slender women with cinnamon skin and long dark hair
—New York Times
Missing pretty girl syndrome. Young attractive women are given a disproportionate attention when they go missing. In thruth the age of the victims was no different in Toluca than in Ciudad Juarez, but various organizations played up the pretty young girl part. Most recently the MAC makeup line had to apologize for its cosmetics line quinceanera inspired by Ciudad Juarez.
The anti-globalization movement latched onto the killings to show the evils of free trade since they increased around the time NAFTA came into effect.

The part about the homicides

Now lets look at the homicide rate for both men and women:

Again, Ciudad Juarez is not different from other border cities. The big increase in Nuevo Laredo in 2005-2006 was due to the Gulf Cartel trying to take over the drug trafficking route or "plaza" of Nuevo Laredo.

In the next chart I chose to include the montainous and very rural municipality of Badiraguato since a lot of cartel leaders have come from this municipality in Sinaloa. There's some doubt as to whether the current leader of Juarez cartel, Vicente Carrillo, was born in Mexico City or Sinaloa but his uncle and former leader of the Guadalajara Cartel, Ernesto Fonseca Carrillo, was born in Badiraguato. Joaquin "El Chapo" Guzman, the leader of the Sinaloa Cartel, was also born in Badiraguato

Again, the fact that Ciudad Juarez wasn't the most violent city in Mexico was well known to Mexican criminologists.

The part about the social indicators

Ciudad Juarez is actually one of the riches municipalities in Mexico, and since GDP per capita tends to be highly correlated with all sorts of social well-being indicators, Juarez tends to rank higher than average for Mexico in all development indicators. Its population tends to be well educated (over 97% can read and write), and the state where it is located scores high on the Pisa exam.

Ciudad Juarez is unequal, but it is far from the most unequal municipality in Mexico. However there were two indicators where Ciudad Juarez stood out:

There were a lot of births to single mothers (single as in not partnered, as opposed to those unmarried mothers living in a domestic arrangement). Divorce also tends to be more common in the northern states, but it is still low by international standards.

All top 5 are cities in the state of Chihuahua.
Chihuahua along with Coahuila had the highest unemployment rate in 2009 (Ciudad Juarez is not part of the urban areas covered in the unemployment survey). But I'm sure part of the reason was the incredible rise in violence that started in 2008. Before the drug war, the whole border region had experienced the highest economic growth rates in all of Mexico.

Hopefully I convinced you that the current violence levels in Ciudad Juarez weren't due to a long process of social decomposition contrary to how the Mexican government is trying to portray it:

The deterioration of values and evidently the expression of violent criminality in Mexico, was not a phenomenon that appeared suddenly, or that occurred fortuitously in our country; it was the fruit of a very long process, that today is showing, precisely, this grave result for the country, but it wasn't something that happened suddenly.
—Felipe Calderon, President of Mexico

There's no question that the equilibrium of violence can shift over long periods but what happened in 2008 was a sudden shock, not the fruit of a long process. Reasons to believe otherwise have no basis in objective reality.

The code is here

Hacking the CISEN homicide data

2010-08-04T18:03:00.001-05:00

The director of the CISEN, the Mexican agency in charge of generating intelligence to safeguard Mexico's security, said in a presentation yesterday that the number of homicides linked to the drug war was a little over 28,000 since the start of the Calderon administration.

Guillermo Valdés didn't provide the raw data so I used Engauge to digitize the slide with the homicide data from the presentation:

And here's the original screen capture from the video:

In spite of the permanent [joint police-military] operations, homicides, kidnappings and extortions persist...

Notice how the y axis doesn't have a scale, but it didn't matter since the total sums to 28,228. I simply set the scale arbitrarily and multiplied the monthly numbers by a constant so that they summed to 28,228.

We can also compare the CISEN data to the number of executions as reported by the Mexican newspaper Milenio. From January 2007 to July 2010 Milenio reported about 3,900 less executions than the CISEN. Even so, the trendlines are pretty close:

There was a temporary drop in the number of executions with the start of the joint operations in Michoacan (December 11, 2006) and Acapulco (January 15, 2007). Executions stayed constant for the rest of the year and then rose again around the time Joint Operation Chihuahua started (March 27, 2009), and just kept on rising, with a brief respite during early 2009, which was probably due to the reinforcements sent to Ciudad Juarez. The big increase at the start of 2010 was probably due to the Gulf Cartel reneging its alliance with the Zetas.

If the CISEN can get this kind of data one has to wonder why the Mexican Government gave the ICESI incomplete homicide data for 2009.

The code is here and the data is available as a Google Spreadsheet

Update 5/August/2010: In a statement today the Mexican government put the exact number of drug-related homicides as 28,228 and clarified that what they term "executions" only include members of criminal bands, drug-related homicides as counted by the CISEN include both members of criminal bands and drug-related aggressions against public security forces.

Mexican newspapers, as opposed to the Mexican government, include all drug-related homicides in what they term executions. In this post I've elected to use the term "execution" as used by Mexican newspapers rather than the government since they were the first to publish their data.

I've updated the Google Spreadsheet data and some of the numbers in this post to reflect the exact 28,228 drug-related homicides according to the CISEN.

Mystery solved: The discrepancy in homicide data

2010-07-27T13:14:00.003-05:00

I've been complaining about how homicide statistics from police sources were too low in 2009, with the entire state of Chihuahua having less homicides than its biggest city. I was thinking of finding out if I could use the IFAI (Freedom of Information Access) to obtain the original CIEISP forms which the state police forces are supposed to fill out each month and send to the National System of Public Security (SNSP) for tallying, to see if there were any unusual patterns, but someone beat me to requesting the data.

You can see the request here, and download the excel file with the CIEISP forms here.

The main difference between the datasets is that the CIEISP forms usually report lower numbers than the data the SNSP gave to the ICESI. This is not surprising since the CIEISP forms have no recorded homicides in some states during the last months of the year (the request was made in December 2009), but the numbers for Chihuahua in both datasets are identical, with a total of 2523 homicides recorded. More importantly, in the months of November and December there were no homicides registered in the CIEISP forms, and the data for October looks incomplete. That's the reason Chihuahua had such a low homicide rate according to the police, the data they gave to the ICESI only includes 9¾ months of homicides.

Given that Mexican President Felipe Calderón and the former Attorney General Eduardo Medina-Mora are on record as stating that violence in Mexico is low, it is not surprising that a government agency would give misleading information to an NGO. I bet the UN, a couple of think-tanks and newspapers will use the lower number, the data will be considered "official", and then next year the SNSP will quietly update the data, just like what happened in 2008.

To estimate the homicides in the missing months I deleted the last three months from the CIEISP forms and predicted them from a linear regression. The predicted number of homicides in Chihuahua for the whole year was 3256. I still think that even with the missing months added in, the police data will be missing about 400 homicides (according to Milenio there were 3,687 narco-executions in Chihuahua), just like in 2008, but at least it no longer looks ridiculously low.

From the chart it looks as if the SNSP only gave "preliminary" data for Chihuahua, coincidentally the most violent state in Mexico. Here's a chart with the estimated data for Chihuahua, the rest of the states use the original numbers from the SNSP:

It looks really bad, Chihuahua saw its murder rate go from 76 to 96 (though likely more) and Durango's homicide rate more than doubled. Chihuahua in 2009 nearly had the murder rate of Ciudad Juarez in 2008!

We can also look at the execution rates reported in David Shirk's Drug Violence in Mexico Data and Analysis as tallied by 3 major Mexican newspapers: El Universal, Milenio and Reforma. For the most part the tallies of the newspapers are similar with Milenio usually reporting higher numbers than Reforma, and El Universal in the middle. The big exception is Chihuahua, where Milenio reported 1555 more executions than Reforma:

Number of narco-executions in Chihuahua:

Reforma	2,082
El Universal	3,250
Milenio	3,687

The numbers for Reforma look too low since in Ciudad Juarez there normally used to be about 200 murders a year, and in 2009 there were more than 2600, I'm pretty sure the vast majority of those extra murders were due to the war between the Sinaloa and Juarez Cartels—and that's only one city—so the number provided by Reforma looks too low. On the other hand there probably were about 3600 homicides in the whole state of Chihuahua and I doubt each and every one of them was related to the drug cartels, so the number provided by Milenio looks too high. I just split the difference and used the average.

Here's the percentage of homicides due to narco-executions:

Like I said the number of homicides in Chihuahua is probably an underestimate and the percentage of homicides which are linked to narco-executions is likely a little bit lower.

P.S. You can download the data and code from my Github account (the first time you run the program it will download a 3MB map of Mexico from GADM)

Homicide in North America

2010-07-14T11:52:00.012-05:00

I'm surprised by how similar the trends are (excluding the drug war in Mexico). There were big decreases in the homicide rate in all three countries starting in the early nineties, which then slowed down around 2000.

The homicide rates for Mexicans and Hispanics in the US are very similar, this is not surprising given that 2/3 of Hispanics are of Mexican descent (the median age in Mexico is 26 and among Hispanics 27). I wish there were data available for Hispanics before 1999 since there was a significant drop in the homicide rate in both Mexico and the US. If you look at the top chart the drop in the murder rate in the United States started slowing a couple of years before Mexico's and that's probably the reason for the bigger gap during the first years of the series.

Update: Here's the first image with a log scale

The municipalities with the highest and lowest homicide rates by sex

2010-07-09T10:26:00.017-05:00

The municipalities are ordered by their peak homicide rate and the dotted gray line is the average homicide rate over the 2005–2008 period. Since homicide is mostly a male phenomenon it's not surprising to find Ciudad Juarez is the most violent city for men.

As you can see, in Ciudad Juarez it's only with the drug war that murder rates against women skyrocketed. In spite of its stereotype as the world capital of femicides it was a relatively safe city. Place like Toluca near Mexico City, and Lazaro Cardenas in Michoacan used to have higher homicide rates.

I see lots of municipalities that are part of Guadalajara, Monterrey, or Tamaulipas, which sadly are not safe anymore; however, Merida is very safe and an incredible place to visit. If what you're looking for is beaches, you can't go wrong with places like Tulum or Los Cabos.

The most violent municipalities in Mexico (2008)

2010-06-22T00:11:00.003-05:00

The top six most violent municipalities are near the US border. Ciudad Juárez is in a class by itself with 113 homicides per 100,000 people.
José Azueta is the municipality where Zihuatanejo is located. Mazátlan, another popular tourist destination, also appears on the list.
Lázaro Cárdenas is the largest seaport in Mexico and ever since the ban on ephedrine, where most of the precursors to produce methamphetamines enter Mexico. Not to mention cocaine from South America.
Culiacán and Navolato are in Sinaloa
I'm surprised that Miguel Hidalgo appears on the list, since it includes some of the most affluent parts of Mexico City.

This chart was part of my analysis of the drug war in Mexico and the code is available from Github