Use the data at your own peril. Better yet, don't use it

The Continuing Problem of Mexico's Crime Statistics

By —— Bio and Archives--April 6, 2014

Guns-Crime-Terror-Security | Comments | Print Friendly | Subscribe | Email Us

The Continuing Problem of Mexico's Crime Statistics

With the latest release of Mexico’s supposedly official crime rate statistics, which include full data for 2013 as well as the first couple months of 2014, security experts need to—once again—be cautioned in the use of this information for drawing any conclusions about crime in Mexico.


I previously showed (backup version with figures) that Mexican crime data were, at best, dubious. For a number of crimes and regions since 1997 (when the dataset begins), we see sufficiently bizarre statistical trends that simply cannot be real. The latest database by the Mexican government fails to correct these obvious historical anomalies, and continues some clear problems.

For example, in 2008 the state of Morelos had exactly zero kidnappings. In 2009, this jumped to 33. As of 2013, the number was 150. The kidnapping rate increased almost six-fold over the last two years alone. Durango’s kidnapping rate increased steadily from zero in 2006 up to 4.73/100,000 population in 2010 (the highest in the country, more than 4.4 times the national average), and then decreased consistently to only 1.21/100,000 population in 2013 (below the national average of 1.44). Real data that accurately reflects the situation on the ground and the actual threats in these two states? Hard to say, but suspicious at best.

The extortion statistics surpass ridiculous. The state of Mexico apparently had zero extortions in 1997 and 1998. Then suddenly in 1999 the state had an astonishing 2,035 exortions, making up 60% of the entire nation’s total of 3,391 for that year. What happened in 2000? The state of Mexico magically returned to zero extortions, and purportedly didn’t have another extortion until 2012, when it suddenly reported 1,039 in 2012 (14% of the national total) and 1,668 in 2013 (almost 21% of the national total). Is this real? No chance. And because the number of extortions—for the very few years since 1997 in which they have been reported—in the state of Mexico comprise such large percentages of the national total, the national extortion rate is fundamentally flawed and useless for any time trend analyses. The state of Mexico isn’t the only one with an extortion data problem. Puebla didn’t have a single extortion between 1997 and 2010, then it had three in 2011. In 2012, it had 454, and another 258 in 2013. Is this real data going back to the start of the record in 1997? Impossible.

As I’ve noted in my prior article, the data for stealing a vehicle with violence takes the proverbial cake for absurdity. In Baja California, there were zero vehicle thefts with violence in 1997. In 1998, there were 29,143, with between 22,736 and 29,532 from 1999 through 2003 (making up over 40% of the national total). And then there were absolutely no vehicle thefts with violence in the Baja from 2004 to 2008. Between 2009 and 2013, the number rose to 361. Chance that this dataset is real? Zero. Oaxaca had no vehicle thefts with violence from 1997 to 2000 and from 2002 to 2010. In 2001, it had 197. Again, not likely real data.

So what happened to all those vehicle thefts with violence in Baja California outside the bizarre 1998 to 2003 timeframe? Well, a close look at the vehicle thefts without violence data tells us the likely answer. In 1997, Baja California had 10,345 vehicle thefts without violence. Then it had zero vehicle thefts without violence from 1998 to 2003, after which in 2004 it had 28,296 and stayed in the tens of thousands of vehicle thefts without violence per year range up to the present. So what happened? Looks like all the vehicle thefts without violence data in the Baja from 1998 through 2003 were dumped into the vehicle thefts with violence category, or vice versa. Who knows what went on with this dataset, but it has no credibility.

Given the unusual patterns that are evident within and between states and timeframes for these crimes, it appears the overall Mexican crime dataset has either been manipulated for political purposes, suffered from reporting incompetence and gross errors that cannot be reliably corrected for, or—more likely—both. It would be of great value to the security policy community to have reliable Mexican crime statistics upon which to make educated decisions and assessments. Unfortunately, these do not exist. Attempting to draw inferences from Mexico’s crime data will be as rife with errors as the underlying data.


Only YOU can save CFP from Social Media Suppression. Tweet, Post, Forward, Subscribe or Bookmark us

Sierra Rayne -- Bio and Archives | Comments

Sierra Rayne holds a Ph.D. in Chemistry and writes regularly on environment, energy, and national security topics. He can be found on Twitter at @srayne_ca

Commenting Policy

Please adhere to our commenting policy to avoid being banned. As a privately owned website, we reserve the right to remove any comment and ban any user at any time.

Comments that contain spam, advertising, vulgarity, threats of violence and death, racism, anti-Semitism, or personal or abusive attacks on other users may be removed and result in a ban.
-- Follow these instructions on registering: