Periodic Reporting for period 1 - GANGNET (Modelling the dynamics of Violent Gang-Crime: a Network approach)
Période du rapport: 2021-10-01 au 2023-09-30
Gang delinquency is one of the rising challenges to the security of urban environments and the wellbeing of European communities. Among the negative externalities generated by gangs’ activity, the propagation of inter-group violent behaviours inflicts a heavy toll on citizens in terms of violence and increased insecurity. The goal of GANGNeT is to study the propagation of criminal behaviours along groups of co-offenders. Society is made of relations, and so are criminal phenomena. However, while connections might lead to positive outcomes, they can also foster the spread and continuation of violent and illegal activities. This key aspect in the study of crime is largely unexplored. This is due to (a) lack of data and, crucially, (b) statistical complexity of modelling highly dynamical processes involving hundreds of goal-driven individuals.
The key innovation of GANGNeT is in implementation of new mixed methods from network science, economics and advanced computational techniques to link and explain two seemingly distant sub-issues: (a) the evolution of complex and highly-dynamic relational networks of incentive-driven co-offenders (b) the propagation of persistent intergroup victimization.
GANGNeT has two specific goals: (I)The development of a novel mathematical model to explain how violence propagates on co-offending networks. The model is used to construct an early-signal indicator for violence out-breaks across organized crime groups and aims to be a platform for further empirical and theoretical analysis. (II) Empirical analysis of a new, surprisingly rich crime dataset to: (II.1) explain the interplay between violence and drug dealing in the behaviour of organized crime groups. (II.2) calibrate the model in (I) for predictive policing forecasting of OCG violence.
The key innovation of GANGNeT is in implementation of new mixed methods from network science, economics and advanced computational techniques to link and explain two seemingly distant sub-issues: (a) the evolution of complex and highly-dynamic relational networks of incentive-driven co-offenders (b) the propagation of persistent intergroup victimization.
GANGNeT has two specific goals: (I)The development of a novel mathematical model to explain how violence propagates on co-offending networks. The model is used to construct an early-signal indicator for violence out-breaks across organized crime groups and aims to be a platform for further empirical and theoretical analysis. (II) Empirical analysis of a new, surprisingly rich crime dataset to: (II.1) explain the interplay between violence and drug dealing in the behaviour of organized crime groups. (II.2) calibrate the model in (I) for predictive policing forecasting of OCG violence.
During the course of the project, we performed research work in two parallel directions.
The first direction deals with understanding what is the influence of crime flows across various areas of urban landscapes. To address this problem, we first developed a mathematical theory of diffusion of behavior of complex social networks. Urban landscapes are characterized by a high degree of heterogeneity in multiple dimensions. Hence, economic opportunities, human flows and, relatedly, crime behavior are embedded within a complex multi-layer morphology. We disciplined the interaction between local (i.e. within neighborhood) and global (between neighborhoods) features of a city that may lead to an idiosyncratic spike in the crime activity within some neighborhood to propagate across a city. We derived a theorem to show that the role of network is ambiguous: contagion is pinned down to the interconnectivity between neighborhoods, and their socio-demographic properties. Depending on the connectivity, urban networks can either amplify or dilute localized shocks. The main result we obtain in this dimension is to show that strategies exclusively targeting vulnerable neighborhoods are not necessarily the most effective ones.
The second dimension we investigated is given by the mechanisms of interaction of organized crime groups (OCGs). We studied the determinants of cooperative interactions among OCGs operating in Merseyside (UK) using the complete crime dataset integrated with neighborhood-level socio-economic data and sentencing outcomes for years 2018-2022. We first addressed the puzzle of the coexistence of stable illegal markets and OCG violence. We found that, net of urban and socio-demographic factors, violence is consequential to cooperation failure. Second, as in illegal markets contracts are not enforceable, incentives to collaborate and profit-sharing mechanisms are distorted. We posited that OCGs select partners and collaborations to balance risks and opportunities. Relative to the former aspect, we showed that cooperation is differential, as it is more likely to realize between groups characterized by asymmetric control of territory. Relative to the latter, OCGs are selective in the nature of interactions, with a positive relationship between expected returns (and associated risks) and cooperation intensity. Importantly, this mechanism complements network-based strategies used by OCGs for mitigation of risks involved with partner selection.
Results from both research streams have been disseminated at general level and to selected practitioners and specialists. General audience dissemination has taken place through participation to the Intersection event organized by the E.U. Dissemination to stakeholders/practitioners has taken place through participation of 2 workshops and two ad-hoc meetings. Communication to specialists has taken place with participation to 6 world-level conference, an internal international workshop and PhD training event.
The first direction deals with understanding what is the influence of crime flows across various areas of urban landscapes. To address this problem, we first developed a mathematical theory of diffusion of behavior of complex social networks. Urban landscapes are characterized by a high degree of heterogeneity in multiple dimensions. Hence, economic opportunities, human flows and, relatedly, crime behavior are embedded within a complex multi-layer morphology. We disciplined the interaction between local (i.e. within neighborhood) and global (between neighborhoods) features of a city that may lead to an idiosyncratic spike in the crime activity within some neighborhood to propagate across a city. We derived a theorem to show that the role of network is ambiguous: contagion is pinned down to the interconnectivity between neighborhoods, and their socio-demographic properties. Depending on the connectivity, urban networks can either amplify or dilute localized shocks. The main result we obtain in this dimension is to show that strategies exclusively targeting vulnerable neighborhoods are not necessarily the most effective ones.
The second dimension we investigated is given by the mechanisms of interaction of organized crime groups (OCGs). We studied the determinants of cooperative interactions among OCGs operating in Merseyside (UK) using the complete crime dataset integrated with neighborhood-level socio-economic data and sentencing outcomes for years 2018-2022. We first addressed the puzzle of the coexistence of stable illegal markets and OCG violence. We found that, net of urban and socio-demographic factors, violence is consequential to cooperation failure. Second, as in illegal markets contracts are not enforceable, incentives to collaborate and profit-sharing mechanisms are distorted. We posited that OCGs select partners and collaborations to balance risks and opportunities. Relative to the former aspect, we showed that cooperation is differential, as it is more likely to realize between groups characterized by asymmetric control of territory. Relative to the latter, OCGs are selective in the nature of interactions, with a positive relationship between expected returns (and associated risks) and cooperation intensity. Importantly, this mechanism complements network-based strategies used by OCGs for mitigation of risks involved with partner selection.
Results from both research streams have been disseminated at general level and to selected practitioners and specialists. General audience dissemination has taken place through participation to the Intersection event organized by the E.U. Dissemination to stakeholders/practitioners has taken place through participation of 2 workshops and two ad-hoc meetings. Communication to specialists has taken place with participation to 6 world-level conference, an internal international workshop and PhD training event.
The main scientific achievement of the first research stream is a comprehensive paper (CP1). CP1 is in readable state with fully developed analytical and mathematical results.
The main innovation outputs are a novel model of diffusion of crime on networks, calibrated on a novel dataset that has not been explored in the literature yet. It advances the state of the art both in the dimensions of mathematical theory of percolation through a novel theorem allowing to discriminate between the role of connectivity and individual influence, as well as the applied field of quantitative criminology via the model and analysis. It also contributes to policing field by exploration of the novel dataset.
The results of CP1 supply public stakeholders (particularly, police forces) with a novel approach and an analytical platform to measure and rank neighborhoods in terms of their vulnerability as well as their relevance in the propagation of crime across a city.
The main scientific innovations of the second research stream is a comprehensive readable working paper (CP2). CP2 is constituted by a 74-pages theoretically-grounded study of a large crime dataset. CP2 advanced the state of the art in the literature of criminology from three main aspects:
1) Exploration of the quantitative features of the organize crime landscape of Merseyside. The exploration features the analysis of a big-data dataset which is new to the literature reconstructing a detailed set of stylized facts related to the behavior of crime organization on the territory with respect to drug market, contedibility of territories and turnaround of groups on the territories.
2) Development of a new statistical index for measuring the organized crime activity in territories. This is a synthetic index for organized crime cooperation which functions as early-signal indicator for crime of violence and crime of drug dealing which can be generalized to any urban context and therefore used for monitoring purpose by any public actor and police force with information on crime activity of organized groups.
3) Estimation of organized crime group behavior in relationship to serious offenses. A theoretical body of hypotheses is developed about the behaviour of organized crime groups. This theory advances the understanding and operative prediction capabilities on the relationship between violence, drug dealing and organized group cooperation.
The results of CP2 are operative and as such have been communicated to the stakeholder Merseyside police force as well other key police forces of Europe.
The main innovation outputs are a novel model of diffusion of crime on networks, calibrated on a novel dataset that has not been explored in the literature yet. It advances the state of the art both in the dimensions of mathematical theory of percolation through a novel theorem allowing to discriminate between the role of connectivity and individual influence, as well as the applied field of quantitative criminology via the model and analysis. It also contributes to policing field by exploration of the novel dataset.
The results of CP1 supply public stakeholders (particularly, police forces) with a novel approach and an analytical platform to measure and rank neighborhoods in terms of their vulnerability as well as their relevance in the propagation of crime across a city.
The main scientific innovations of the second research stream is a comprehensive readable working paper (CP2). CP2 is constituted by a 74-pages theoretically-grounded study of a large crime dataset. CP2 advanced the state of the art in the literature of criminology from three main aspects:
1) Exploration of the quantitative features of the organize crime landscape of Merseyside. The exploration features the analysis of a big-data dataset which is new to the literature reconstructing a detailed set of stylized facts related to the behavior of crime organization on the territory with respect to drug market, contedibility of territories and turnaround of groups on the territories.
2) Development of a new statistical index for measuring the organized crime activity in territories. This is a synthetic index for organized crime cooperation which functions as early-signal indicator for crime of violence and crime of drug dealing which can be generalized to any urban context and therefore used for monitoring purpose by any public actor and police force with information on crime activity of organized groups.
3) Estimation of organized crime group behavior in relationship to serious offenses. A theoretical body of hypotheses is developed about the behaviour of organized crime groups. This theory advances the understanding and operative prediction capabilities on the relationship between violence, drug dealing and organized group cooperation.
The results of CP2 are operative and as such have been communicated to the stakeholder Merseyside police force as well other key police forces of Europe.