Why Locking Up Cartel Members Does not Work

TL;DR

Addressing cartel violence, the paper asks whether punitive incarceration can meaningfully reduce cartel activity. It develops an age-cohort, agent-based model parameterized with Mexican data and a master equation for cartel dynamics $\dot{A}(t)=\rho A(t)-\eta A(t)-\theta A(t)-\omega A^2(t)$ to quantify lifetime cartel participation and potential. The results show incarceration averts only about $9\%$ of cartel potential for a cohort, with longer sentences offering marginal gains (to around $41\%$ of the original potential) and complete rehabilitation potentially reducing cartel potential to $36\%$; full rehabilitation could raise prevention to roughly $29\%$ beyond incarceration alone. The authors conclude that social reintegration strategies likely offer far larger reductions in cartel activity than harsher imprisonment, providing a practical policy direction and a modeling framework for evaluating reforms.

Abstract

One of the core strategies to reduce cartel violence is by directly targeting members with law enforcement. Whether targeting leaders, disrupting parts of the organisation, or incarcerating members, the purpose is to reduce the strength of cartels directly. Most security strategies result in increased incarceration rates. Yet its effectiveness in addressing organised crime remains unclear, particularly if it fails to prevent recidivism upon release from jail. Here, a model is constructed to quantify cartel participation across generations, where individuals are recruited, age over time, and exit cartels as victims of a homicide or due to incapacitation, or retirement. Incarcerating cartel members prevents less than 10% of cartel offences. Additionally, doubling penalties would reduce cartel members' potential by less than 5%, thereby challenging proposals for stricter rules. Yet, rehabilitation after prison, often neglected as an integral part of the security strategy, could be more effective in lowering cartel crimes.

Figures (9)

• Figure 1: Model for the flow of cartel members according to their age. a - Each week, a list of active members is updated by keeping track of members who are recruited (mostly young), who are killed, incapacitated, or retire. b - Composition of cartel members by age (vertical), between 2005 and 2045 (horizontal). Each stripe represents an age cohort (with the 1990 cohort highlighted). Younger cohorts are below older cohorts. We estimate that by 2025, 3% of cartel members are under 20 years old, 38% are between 20 and 30, 43% are between 30 and 40, 15% are between 40 and 50, and only 1% are over 50 years old. Of the 183,000 active cartel members in 2025, 4.8% are from the 1990 cohort. By 2035, active cartel members are expected to be approximately 240,000, including 1.5% from the 1990 cohort.
• Figure 1: Number of recruited members of the 1990 age cohort (vertical axis) when the maximum age for recruitment is defined as $u+\delta$ (horizontal axis). By repeating the same stochastic process, we observe departures in the number of active members across different age-at-recruitment distributions. For values of $\delta =0$, we obtain departures based on the original distribution, mostly between 15,000 and 16,000 members (right panel).
• Figure 2: Life course trajectory of the 1990 cohort. a - Share of the different states of the recruited individuals of the 1990 age cohort between 2005 (when that cohort experienced its first recruitment) and 2045 (when all individuals have exited cartels, either being killed, incapacitated, or retiring). b - Share of the recruits and the active members that belong to the 1990 cohort between 2005 and 2045.
• Figure 2: Observed distribution of sentences in prison based on the prison population survey ENPOL. Utilising that survey, we approximate the frequency with which each sentence is assigned. That distribution is then used to analyse time spent in prison.
• Figure 3: Modelled life after prison. a - Distribution of ending states after spending time in prison for all individuals from the 1990 age cohort who were incapacitated. Intervals are obtained by repeating the assignment of sentences to different cases. b - Impact of increasing incarceration penalties (horizontal axis) on the cartel potential of the 1990 age cohort (vertical). By multiplying each sentence by a penalty multiplier $\alpha\geq 1$, the cartel activities of the cohort vary. With the current penalties (with a value of the multiplier $\alpha = 1$), of the cartel potential of the cohort, 33% is averted due to homicides, 12% due to retirement, and only 9% is averted due to incarceration (so 46% of the years of cartel activity will be effective). By doubling the incarceration rate (with $\alpha = 2$), the prison prevents 17% of the cohort activities, reducing the share of activities averted by retirement and homicides, so the effective cartel offences drop from 46% to 41%. c - Impact of increasing social rehabilitation (horizontal axis) on the cartel potential of the 1990 age cohort (vertical). With $\beta = 0$, all the prison population rejoins the cartel upon release, and with $\beta = 1$, no one returns to the cartel, reducing its potential from 46% to 36%.