Criminal organizations are pervasive. In many cities globally, gangs serve critical functions by enforcing social order and property rights, often despite the presence of strong governments. Such conditions present a puzzle to researchers and policymakers who expect gang activity to diminish in the presence of greater state power. This paper aims to resolve this paradox, investigating how these so-called “duopolies” function, and what policies may be effective at curbing gang activity. 

The authors use Medellín, Colombia, as a case study in their analysis. Medellín is a large city that features an advanced economy and strong government at the local-, state-, and national-level. Nonetheless, virtually every low- and middle-income neighborhood in Medellín is also occupied by one of more than 350 small gangs, or combos, who typically engage in at least some governance activities, including prohibiting and punishing property crime, settling disputes between neighbors, enforcing rules, and, exchange, taxing locals. 

The authors begin their analysis by interviewing 118 criminal leaders and group members across 41 groups in Medellín. They supplement their interviews with data from a citywide survey covering nearly 7,000 residents and businesses, whom the authors query concerning state and combo rule in their neighborhood. They document the following concerning gang rule in Medellín:

  • While the state is the predominant provider of protection, combos are seldom far behind. In roughly a third of city blocks, residents report that their combo is the leading provider of security and dispute resolution. 
  • Combos typically collect security fees for their services. But while taxing residents involves coercion, many payees report they value the services and that they view the combos as legitimate. 
  • Several criminal leaders report that that they govern not for the direct profits, but rather because it protects their other business lines, especially drug sales. Providing neighborhood order reduces the need for routine police patrols and entry by special agents. When police do enter, residents who are loyal to the gangs are less likely to inform on gang members, and may even actively help them hide. 

Building on these results, the authors next turn to determining how combos respond to increased state presence. To do so, they exploit a 1987 policy change that altered policing jurisdictions in Medellín and generated quasi-random changes in the distance to local and state protection across different neighborhoods. The authors use their survey, which was administered in 2019, to measure the long-run impacts of this change on governance by both the combos and the state. They find the following: 

  • Not surprisingly, moving closer to police and municipal headquarters increases reports of state governance. On blocks assigned to be 400 meters closer to their police and municipal dispute resolution agencies (the median change), residents reported 12% higher state responsiveness to disputes and disorder. This suggests that proximity matters for projecting state power. 
  • On average, the combos responded to increased state governance by governing more themselves. Growing 400 meters closer to state headquarters increases reports of combo rule by 25%. This implies the combo is about twice as responsive as the state. This may be because the state takes several measures to distribute services as evenly as possible across the neighborhood, dampening the impact of distance. 
  • The authors use machine learning to predict neighborhoods’ proximity to future drug markets, using characteristics such as demographics and development levels. They find that on blocks with the lowest potential drug rents, combos do not respond to state presence by governing more. On blocks with the highest future potential profits, however, the combo response is dramatic. 
  • The authors find little evidence that state presence leads to greater economic development and in-migration, supporting their hypothesis that the primary mechanism connecting state and combo governance is the protection of rents. 

This research helps to explain why gang rule is a problem in “semi-strong states,” where governments are powerful enough to create illicit markets and challenge criminal gangs, yet too weak to eliminate them entirely.  The same often holds for neighborhoods, where stronger state presence crowds out gangs in higher-income areas, while conceding criminal activity in low- and middle-income neighborhoods.

Policy-wise, these results suggest governments will need to go beyond piecemeal state-building to counter gang rule, perhaps by trying to shape local norms of combo legitimacy, or undermining drug profits. The authors encourage policymakers to proceed with caution, however, as undermining drug profits could result in more ruthless, extractive, and violent organizations. 

More on this topic

Research Briefs·Mar 22, 2024

Fallow Lengths and the Structure of Property Rights

Etienne Le Rossignol, Sara Lowes, and Eduardo Montero
Places where land needs to be fallowed for longer periods are more likely to have communal property rights, both historically and presently. World Bank land titling interventions are less effective in places with longer fallow requirements, and longer fallow periods...
Topics: Development Economics
Research Briefs·Mar 6, 2024

Fiscal Rules, Austerity in Public Administration, and Political Accountability: Evidence from a Natural Experiment in Colombia

Maria Carreri and Luis R. Martínez
Limitations to fiscal spending in Colombia reduced overspending on public administration without affecting public goods, better aligning spending with voters’ preferences.
Topics: Development Economics, Monetary Policy
Research Briefs·Nov 29, 2023

Nothing Gold Can Stay: Artisanal Mine Certifications and Conflict Dynamics in the Congo

Hans Christensen and Samuel Chang
Conflict-free certifications for artisanal mines are associated with a 9.4% (16.3%) reduction in armed group–initiated conflicts (fatalities) within a 10-km radius of gold mines. After certifications, there is no aggregate reduction in conflict intensity in Eastern DRC territories, and conflicts...
Topics: Development Economics