La minería ilegal es prominente en todo el mundo, pero rara vez esta actividad se reporta ante las autoridades responsables de monitorearla. Colombia Mining Monitoring (CoMiMo) utiliza inteligencia artificial y tecnología satelital para localizar posibles puntos de ubicación de minas ilegales en Colombia. Usando esta tecnología, los investigadores revelaron la ubicación de minas a las autoridades locales y nacionales para medir su respuesta y determinar si esta información reduce o reubica la presencia de la minería ilegal.
New monitoring technologies can help curb illegal activities by reducing information asymmetries between enforcing and monitoring government agents. I created a novel dataset using machine learning predictions on satellite imagery that detects illegal mining. Then I disclosed the predictions to government agents to study the impact on illegal activity. I randomly assigned municipalities to one of four groups: (1) information to the observer (local government) of potential mine locations in his jurisdiction; (2) information to the enforcer (National government) of potential mine locations; (3) information to both observer and enforcer, and (4) a control group, where I informed no one. I use an independent expert validated dataset that measures gold mining to evaluate the effect of the intervention. I find that the effect of treatment is relatively similar regardless of who is informed: in treated municipalities, illegal mining is reduced by 11% in the disclosed locations and surrounding areas. However, when accounting for negative spillovers — increases in illegal mining in areas not targeted by the information — the net reduction is only 7%. These results illustrate the benefits of new technologies for building state capacity and reducing illegal activity.
Organised criminal activities, by their nature, are hard to measure. Administrative data are often missing, problematic, or misleading. Moreover, organised criminal activities are under-reported, and under-reporting rates may be greatest where gangs are strongest. Researchers hoping to quantify organised crime systematically face daunting challenges. Collecting information on organised crime is inherently a slow process of cautious trial and error. It will vary from city to city, and typically within a city as well. Dozens of qualitative and quantitative researchers have shown that this can be done with care, ethically, and with adequate protection for human subjects. What they all have in common is that they commit themselves to a place, and they all take their time. While there are risks, the benefits can be enormous. The information these investigators collect is often rare and invaluable. Officials and policymakers commonly have little insight into criminal organisations, with terrible consequences for policy, be it inaction, mediocrity, or adverse and unintended consequences. Here we draw on our experience in Colombia, Brazil, and Liberia of collecting systematic data on illicit activities and armed groups, in order to share our learning with other researchers or organisations that fund research in this area, who may find this useful for their own research. We address: first steps before asking questions, common challenges and solutions, and alternative sources. Our work thus far emphasises the relevance of deep qualitative work to identify local partners; the need for intense piloting of survey instruments and a close oversight of survey firms, ranging from how they hire enumerators to how they plan and implement field work; the power of using survey experiments to mitigate and measure measurement error; and the relevance of cross-validating findings with complementary data sources.
Organised crime poses one of the greatest threats to national security and development in the 21st century. Despite this, most policy, data collection, and scholarly research focuses on individuals and disorganised violence. Our work addresses several critical gaps in knowledge: 1. What are the incentives for gangs to engage in violence and socially costly behaviour? 2. Which are the trade-offs that practitioners face when deciding how to engage with organised violence? 3. What type of information do relevant decision-makers need to inform their policies? 4. Which are the most relevant tools for tracking down gang behaviour and use of violence?
Los barrios marginales y pobres de todo el mundo están ocupados por poderosos grupos criminales que cada vez reclutan más niños y adolescentes. ¿Qué factores llevan a las personas a unirse a estos grupos y cómo pueden los gobiernos evitar el reclutamiento? En Medellín, Colombia, una ciudad caracterizada por una alta presencia de grupos criminales denominados combos, unos investigadores están realizando un estudio para comprender el proceso de reclutamiento, identificar a los niños y los adolescentes en riesgo de ser reclutados y probar diferentes intervenciones para evitar el reclutamiento.
Las irregularidades electorales —actividades ilegales que buscan influir en las elecciones— a menudo amenazan las instituciones democráticas en los países de ingresos bajos y medios. En Colombia, los investigadores llevaron a cabo una evaluación aleatoria para medir el impacto de una intervención que alentaba a los ciudadanos a denunciar irregularidades a una ONG local y que variaba si los candidatos estaban informados sobre la campaña de denuncia. La campaña, realizada a través de Facebook, aumentó la cantidad de denuncias reportadas por los ciudadanos y redujo la cantidad de irregularidades observadas en las elecciones. Esta sencilla intervención fue más costo-efectiva que otros esfuerzos de monitoreo usados tradicionalmente.
Throughout the developing world, citizens distrust the police and hesitate to bring crimes to their attention—a suboptimal equilibrium that makes it difﬁcult for the police to effectively combat crime and violence. Community policing has been touted as one solution to this problem, but evidence on its efﬁcacy in developing country contexts is sparse. We present results from a large-scale ﬁeld experiment that randomly assigned a home-grown community policing intervention to police stations throughout rural Uganda. Drawing on administrative crime data and close to 4,000 interviews with citizens, police ofﬁcers, and local authorities, we show that community policing had limited effects on core outcomes such as crime, insecurity, and perceptions of the police. We attribute these ﬁndings to a combination of turnover, treatment non-compliance, and resource constraints. Our study draws attention to the limits of community policing’s potential to reduce crime and build trust in the developing world.
Medellín's government wanted to raise its efficacy, legitimacy, and control. The city identified 80 neighborhoods with weak state presence and competing armed actors. In half, they increased nonpolice street presence tenfold for two years, offering social services and dispute resolution. In places where the state was initially weakest, the intervention did not work, mainly because the government struggled to deliver on its promises. Where the state began stronger, the government raised opinions of its services and legitimacy. If there are indeed low marginal returns to investing in capacity in the least-governed areas, this could produce increasing returns to statebuilding.
We examine how trust shapes compliance with public health restrictions during the COVID-19 pandemic in Uganda. We use an endorsement experiment embedded in a mobile phone survey to show that messages from government ofﬁcials generate more support for public health restrictions than messages from religious authorities, traditional leaders, or international NGOs. We further show that compliance with these restrictions is strongly positively correlated with trust in government, but only weakly correlated with trust in local authorities or other citizens. The relationship between trust and compliance is especially strong for the Ministry of Health and—more surprisingly—the police. We conclude that trust is crucial for encouraging compliance but note that it may be difﬁcult to change, particularly in settings where governments and police forces have reputations for repression.
La pandemia causada por el virus del Covid-19 ha cambiado la vida humana como muy pocos acontecimientos lo habían hecho en el pasado: no solo por afectar a miles de personas en todos los continentes, sino por paralizar la vida y la economía como las conocíamos. Poco se sabía del virus cuando varias naciones europeas reportaban miles de contagios y de fallecidos. En países como Colombia, se decretaban cuarentenas nacionales y medidas sanitarias ante una situación de la que poca información se tenía. Según el World Uncertainty Index (wui) realizado por el Fondo Monetario Internacional, la pandemia del Covid-19 ha causado más incertidumbre que cualquier otra crisis sanitaria en la historia o acontecimientos recientes, como los eventos terroristas del 11 de septiembre de 2001, la crisis financiera global de 2008, la crisis del euro a finales de 2009 o el Bréxit en 2016 (Ahir, Bloom y Furceri, 2018).
Working in partnership with local police agencies, we conducted six coordinated field experiments in Brazil, Colombia, Liberia, Pakistan, the Philippines, and Uganda. We collaborated with the police to implement locally appropriate increases in community policing practices. We planned for risks involved in partnering with the police by soliciting reports of police abuse and carefully selecting the areas we worked in and the police units we partnered with. We randomly assigned areas to either the community policing practices or a control group. Our interventions reached approximately 9 million people in 516 treated areas. At the end line, we surveyed 18,382 citizens and 874 police officers and obtained crime data from the police. We conducted experiments in multiple settings with common measures to strengthen the generalizability of our findings and preregistered a joint analysis of the six studies to reduce the risk of publication bias.
Can information and communication technologies help citizens monitor their elections? We analyze a large-scale field experiment designed to answer this question in Colombia. We leveraged Facebook advertisements sent to over 4 million potential voters to encourage citizen reporting of irregularities and varied whether candidates were informed about the campaign in a subset of municipalities. Total reports, as well as evidence-backed ones, experienced a large increase. Across a wide array of measures, electoral irregularities decreased. Finally, the reporting campaign reduced the vote share of candidates dependent on irregularities. This light-touch intervention is more cost-effective than monitoring efforts traditionally used by policymakers.
Lessons from randomized evaluations on managing and preventing crime, violence, and conflict
What are the most promising strategies for reducing crime, violence, and conflict? The past decade has seen a dramatic expansion in the experimental literature designed to help answer this question. Moving beyond evaluations of individual programs, increasingly, these studies are striving to test broader hypotheses about how programs work (i.e. what are the key program components driving change) and to generate insights into human behavior (i.e. why individuals may be motivated to act in certain ways).
This evidence review, prepared by staff at the Abdul Latif Jameel Poverty Action Lab (J-PAL) and Innovations for Poverty Action (IPA) ) for the Foreign, Commonwealth and Development Office (FCDO), offers a broad review of the expansion of this literature and seeks to capture some of the emerging insights from across these studies. The review has been prepared as part of J-PAL and IPA’s Governance, Crime and Conflict Initiative, a £15- million investment by FCDO launched in 2017 to produce new research on effective policies to promote peace and good governance, reduce crime, and support individuals and communities recovering from conflict.
Since the COVID-19 outbreak, scholars and journalists have spread anecdotes of gangs and criminal organizations coming to the aid of citizens, governing in place of the state. Researchers studied if gangs respond to COVID-19 in Medellín. Despite the headlines, gang involvement in pandemic response is exceptional and mostly idiosyncratic. Surveying every low- and middle-income neighborhood in Medellin, they find that most of the support for civilians comes from state authorities and not from gangs
Medellín tiene unas estructuras de crimen organizado altamente jerárquicas y estructuradas. Con el objetivo de comprender cómo está organizada y cómo funciona esta estructura, durante los últimas tres años Innovations for Poverty Action, la Universidad de Chicago y la Universidad EAFIT han recolectado información sobre ella. Esto se ha hecho a través de entrevistas con integrantes de diferentes comunidades de la ciudad. En este documento se presentan los hallazgos de este trabajo.