Criminal fragmentation in Mexico
Criminal fragmentation in Mexico (DOI: https://doi.org/10.1017/psrm.2025.4) was written by Jane Esberg in 2026. It was published in Political Science Research and Methods (vol. 14).
The author codes the news stories shared on Borderland Beat, a Mexican narcoblog.
- Effectively a body of stories that has been curated by a subject matter expert
- Anonymity of a citizen journalist means there is less likely to be an underreporting effect
- Body covers 2009 through October 2020 (site shuttered due to lawsuit brought by an individual who had been reported on)
Body is coded to extract...
- organization name
- implicit assumption that any incident has a singular organization that is responsible; they act semi-independently
- 473 total organizations are coded
- organization type
9 organizations are coded as major cartels
cells collaborate with a major cartel
splinters broke away from a hierarchy
- differentiated from cells mostly by journalist notes
171 organizations are unaffiliated groups
- usually local, but really defined by the exclusion of the above characteristics
- by reflection, everything that isn't a major cartel can be considered a 'minor group'
- journalist-noted relations to other organizations
- municipality
- when an organization emerged
- first appearance in the body
- when an organization expands
- first appearance of an existing organization in a new municipality
23 organizations could not be tied to a municipality, so were dropped from analysis.
To understand fragmentation, the author matches this dataset to
a U.S. sanctions list; and
- a municipality-level dataset of mayors and "log hectares of poppy and marijuana seized, a proxy for drug presence (Ch, 2023)"
The intent is to model a 'kingpin strategy': what is the effect of going after cartel leadership. A municipality is 'treated' if a sanctioned cartel leader is removed, and remains treated for 3 years. The author uses difference in differences methods to estimate the effect of treatment on the number of organizations, split by major cartels vs. minor groups, and also splitting emergences vs. expansions.
- Control for electoral environment (if mayor is conservative), level of criminal activity, and two-way fixed effects for year and municipality.
- There is a significant and positive effect on the number of organizations in every category.
- Increases in numbers are also found before treatment, so there is not a clear causal relation.
Hypothesizing that drugs are a competitive market and so not the best indicator of organized criminal activity, the author matches to a municipality-level dataset of gas prices and logged pipeline length. Fuel theft (huachicoleo) is a prominent alternative to the drug trade. Regress on interaction of gas prices and pipeline length.
- Significant and positive effect on expansion.
Reading notes
Interesting approach to the topic. Haven't seen this method of data collection before. Effects are very small even if significant, I am not convinced that anything is proven conclusively.